O C C A S I O N A L PA P E R 258

Northern Star Canada’s Path to Economic Prosperity

Edited by Tamim Bayoumi, Vladimir Klyuev, and Martin Mühleisen

INTERNATIONAL MONETARY FUND

Washington DC

2007

©International Monetary Fund. Not for Redistribution

©2007 International Monetary Fund

Production: IMF Multimedia Services DivisionTypesetting: Julio R. Prego

Figures: Bob Lunsford

Cataloging-in-Publication Data

Northern star : Canada’s path to economic prosperity / edited by Tamim Bayoumi, Vladimir Klyuev, and Martin Mühleisen. — Washington, D.C. : International Monetary Fund, 2007.

p. cm.—(Occasional paper ; 258)

Includes bibliographical references.ISBN 978-1-58906-614-4

1. Canada — Economic conditions — 1991– 2. Canada — Economic policy. 3. Fiscal policy — Canada. 4. Monetary policy — Canada. I. Bayoumi, Tamim A. II. Klyuev, Vladimir. III. Mühleisen, Martin. IV. International Monetary Fund. V. Series: Occasional paper (International Monetary Fund) ; no. 258

HC115 .N678 2007

Price: US$30.00(US$28.00 to full-time faculty members and

students at universities and colleges)

Please send orders to:International Monetary Fund, Publication Services

700 19th Street, N.W., Washington, D.C. 20431, U.S.A.Tel: (202) 623-7430 Telefax: (202) 623-7201

E-mail: publications@imf.orgInternet: www.imf.org

The commercial printer will place here the FSC logo.It will show on bluelines

©International Monetary Fund. Not for Redistribution

iii

Contents

Preface ix

Abbreviations xi

Introduction 1Vladimir Klyuev

Part I. Macroeconomic Environment

I. Factoring in Canadian Cycles 5Alejandro Justiniano

Dynamic Factor Analysis 5Canadian-U.S. Economic Interactions 6Robustness Checks 10Conclusions 13

II. Energy, the Exchange Rate, and the Economy 14Tamim Bayoumi and Martin Mühleisen

Outlook for Energy Production 14Export Prospects and Dutch Disease 16Exchange Rate Equation 19Putting It All Together 22Conclusions 24

III. Canada and the United States 25

A. Regional Dimensions of the Canadian Economy 25Vladimir Klyuev and Rodolfo Luzio

Industrial Structure 25Response to Shocks 28Conclusions 31

B. The Canada–United States Productivity Gap 32Roberto Cardarelli

Review of the Literature 32Results from Sectoral Growth Accounting 34Productivity Growth and Trade 40

C. The Effects of U.S. Shocks 41Iryna V. Ivaschenko and Andrew Swiston

Model Description 42Data and Estimation 43Results 44Conclusions 47

©International Monetary Fund. Not for Redistribution

iv

CONTENTS

Part II. Fiscal Policies

IV. Budget Forecasting 51Martin Mühleisen, Stephan Danninger, David Hauner, Kornélia Krajnyák, and Bennett Sutton

The Institutional Environment 51Fiscal Forecasting Practices in International Comparison 59Assessing Forecast Accuracy 62Statistical Analysis of Forecast Outcomes 67Conclusions 72

V. The Pension System 74Martin Mühleisen

The Public Pension System 74Private Pension Schemes 76Pension Incomes and Retirement Trends 78Policy Issues 81

VI. Fiscal Prudence and Tax Reduction Opportunities 84

A. Jam Today or More Jam Tomorrow? The Timing of Tax Cuts 84Tamim Bayoumi and Dennis Botman

The Model and Calibration 84Implications of Cutting Transfers and Lowering Taxes 86Fiscal Spillovers 87Conclusions 90

B. Comparing Efficiency Gains from Reducing the GST andthe Personal Income Tax 90Dennis Botman

Comparing Wage, Consumption, and Income Taxation 91Ranking Tax Distortions in Canada 92Conclusions 95

Part III. Monetary and Financial Policies

VII. Disintermediation and Monetary Transmission 99Jorge E. Roldos

Financial Deregulation and Disintermediation 99Evidence from VAR Models 101Evidence from Structural Models 107Conclusions and Policy Implications 109

VIII. Inflation Targeting and Macroeconomic Volatility 111Tamim Bayoumi and Vladimir Klyuev

The Framework 111Empirical Results 112Conclusions and Policy Implications 114

IX. Competition in the Banking System 116Iryna V. Ivaschenko

Optimal Level of Bank Competition: A Review of the Literature 116Methodology and Data 117Results 117Conclusions 118

©International Monetary Fund. Not for Redistribution

v

Contents

X. Large Banking Groups and Financial System Soundness 120Gianni De Nicolò, Alexander Tieman, and Robert Corker

Banking Sector Trends 120Market-Based Financial Soundness Indicators 122Soundness Indicators in Canada and the United States 124Conclusions 125

References 126

Boxes

1.1. United States: Data Sources, Descriptions, and Transformations 11 1.2. Canada: Data Sources, Descriptions, and Transformations 12 2.1. Data Sources 20 3.1. Data Sources and Definitions for Canadian and

U.S. Regional Variations 29 3.2. Data Sources on Canada–United States Productivity Gap 33 4.1. Equalization Transfers in Canada 57 4.2. Fiscal Forecasting Arrangements 59 4.3. Data Overview 64 5.1. Canada’s Three-Tiered Pension System 76 5.2. Reform Proposals for the Quebec Pension Plan 8210.1. Changes in Canada’s Financial Sector Legislation 121

Tables

1.1. Variance Decompositions for the United States from a Factor Model 6 1.2. Variance Decompositions for Canada from a Factor Model 7 2.1. Canadian Energy Exports: Static Projection 17 2.2. Exchange Rate and Manufacturing in Energy-Exporting

Industrial Countries 20 2.3. Real Exchange Rate Equations 21 2.4. Conventional Real Exchange Rate Equations 21 2.5. Impact of Oil Production and Price Increases 24 3.1. Canada: Growth Regressions 30 3.2. Canada: Real Private Consumption Regressions 30 3.3. Canada: Real Private Investment Regressions 31 3.4. United States: Growth Regressions 31 3.5. Canada and the United States: ICT Captial Accumulation 34 3.6. Canada and the United States: Productivity Growth, 1982–2000 35 3.7. Canada and the United States: Productivity Growth, 1982–1995 36 3.8. Canada and the United States: Productivity Growth, 1995–2000 37 3.9. Canada–United States Labor Productivity Growth Gap 383.10. Canada: ICT Regression with Current and Lagged ICT

Capital Services Growth 403.11. Canada and the United States: Value Added, Shares of Total 40 4.1. Indications of the Relationship between Legislature and Executive 53 4.2. Share of Spending by Subnational Governments 55 4.3. Consolidated Central Government Expenditure Shares, 2003 56 4.4. Fiscal Policy Rules and Transparency Laws 58 4.5. Key Institutional Characteristics of the Fiscal Forecasting Process 60 4.6. Fiscal Forecasting: Quality Assurance 61 4.7. Descriptive Statistics of One-Year Budget Forecast Errors, 1995–2003 66 4.8. Results of Forecast Error Median and Mean Tests 68 4.9. Results of Efficiency Tests 694.10. Potential Factors Affecting Forecast Outcomes 71 5.1. CPPIB Operations 77

©International Monetary Fund. Not for Redistribution

vi

CONTENTS

5.2. Membership in Registered Pension Plans 78 5.3. Old-Age Income and Labor Market Participation 79 5.4. Distribution of the Elderly by Population Income Quintile, 1980 –95 79 5.5. Sources of Pretax Income for the Elderly 80 5.6. Withdrawal from Labor Force and Retirement Duration, 1999 80 5.7. Projected Increases in Old-Age Pension and Social Spending 83 7.1. Estimates of Open Economy IS Equation (1971:Q1–2005:Q2) 108 7.2. Estimates of Open Economy IS Equation (Sample Break in 1988:Q1) 109 8.1. Estimates of Monetary Model 113 8.2. Standard Deviation of Inflation, Output Gap, and Interest Rate

before and after Introduction of Inflation Targeting (IT) 115 8.3. Standard Deviation of Inflation, Output Gap, and Interest Rate

under Hypothetical Regimes 115 9.1. Size and Profitability Indicators of 25 Largest Banks 117 9.2. H-Statistics, by Country and Region 118 9.3. H-Statistics for Largest Banks, by Country and Region 119 9.4. Size and Concentration of Banking Systems across Regions, 2003 11910.1. Balance Sheet Items of Six Large Banking Groups 12110.2. Vulnerability Indicators of the Banking System 122

Figures

1.1. “Oil Factor” and Comovements in Selected Series 8 1.2. “Real Foreign Factor” and Comovements in Selected Series 9 1.3. “Exchange Rate Factor” and Comovements in Selected Series 10 1.4. “Real Domestic Factor” and Comovements in Canadian GDP 10 2.1. Oil and Natural Gas Prices 15 2.2. Net Hydrocarbon Energy Exports 15 2.3. Projected Energy Production 16 2.4. Exchange Rate and Manufacturing Activity in Resource-

Exporting Countries 18 2.5. Composition of Net Exports 19 2.6. Dynamic Forecasts of Canadian Real Exchange Rate 22 2.7. Long-Term Impact of Commodity Prices on the Exchange Rate 23 3.1. Canada: Difference between Regional and National Industry Shares 26 3.2. United States: Difference between Regional and

National Industry Shares 27 3.3. Regional GDP by Sector: Divergence from Country Average 28 3.4. Average Regional Divergence from National Industrial Structure 29 3.5. United States and Canada: Income and Productivity Indicators 32 3.6. Sectoral Contributions to the Canada–United States Aggregate Labor

Productivity Growth Gap 39 3.7. Canada: Sectoral TFP Growth and Openness to Trade 41 3.8. Real Effective Exchange Rate and Current Account Balance 42 3.9. Response of Canadian GDP to U.S. Shocks, with Immediate

and Delayed Monetary Policy Response 433.10. Response of Canadian CPI to U.S. Shocks, with Immediate

and Delayed Monetary Policy Response 443.11. Response of Canadian GDP and Inflation to U.S. Shocks,

with Immediate and Delayed Monetary Policy Response 453.12. Response of Canadian GDP and Inflation to Exchange Rate Shock,

with Immediate and Delayed Monetary Policy Response 463.13. Responses of Canadian GDP and Inflation to Changes in the

Pass-Through Coefficient, with Immediate and Delayed Monetary Policy Response 47

4.1. Influence of Subnational Governments 55 4.2. Canada: Fiscal Balance Forecast Errors 70

©International Monetary Fund. Not for Redistribution

vii

Contents

The following conventions are used in this publication:

• In tables, a blank cell indicates “not applicable,” ellipsis points ( . . .) indicate “not avail-able,” and 0 or 0.0 indicates “zero” or “negligible.” Minor discrepancies between sums of constituent figures and totals are due to rounding.

• An en dash (–) between years or months (for example, 2005–06 or January–June) indi-cates the years or months covered, including the beginning and ending years or months; a slash or virgule (/) between years or months (for example, 2005/06) indicates a fiscal or financial year, as does the abbreviation FY (for example, FY2006).

• “Billion” means a thousand million; “trillion” means a thousand billion.

• “Basis points” refer to hundredths of 1 percentage point (for example, 25 basis points are equivalent to ¼ of 1 percentage point).

As used in this publication, the term “country” does not in all cases refer to a territorial entity that is a state as understood by international law and practice. As used here, the term also covers some territorial entities that are not states but for which statistical data are main-tained on a separate and independent basis.

4.3. Impact of GDP Volatility on Forecast Quality 73 5.1. Old-Age Dependency Ratios in Selected G-7 Countries 74 5.2. Government Spending on Old-Age Pensions 75 5.3. Projected Canada Pension Plan Contribution Rates

and Pension Reserves 75 6.1. Effects on Real GDP of Immediate versus Delayed Wage Tax Cuts 86 6.2. Effects on Real GDP of Immediate versus Delayed Corporate

Income Tax Cuts 87 6.3. Effects of Alternative Parameterizations on Impact of a Cut

in Transfers and in the Wage Tax Rate on Real GDP 88 6.4. Effects of Alternative Parameterizations on Impact of a Cut

in Transfers and in the Corporate Income Tax Rate on Real GDP 89 6.5. Impact of Tax Spillovers from the Rest of the World 90 6.6. Macroeconomic Effects of a Reduction in the Goods and Services Tax 92 6.7. Macroeconomic Effects of a Reduction in Personal Income Taxation 93 6.8. Efficiency Gains from Reducing Alternative Types of Taxation 94 6.9. Efficiency Gains from Reducing the GST versus

Personal Income Taxation: Sensitivity Analysis 94 7.1. Nonfinancial Corporate Financing Sources, 1971–2005 100 7.2. Credit Flows, 1971–2005 100 7.3. Ratio of Direct to Indirect Private Lending, 1971–2005 101 7.4. Outstanding Credit, 1971–2005 101 7.5. Impulse-Response Functions from Benchmark VAR 103 7.6. Impulse-Response Functions from VAR, Including Business Loans 104 7.7. Impulse-Response Functions from VAR, Including Asset Prices 105 7.8. Impulse-Response Functions from VAR, Including Asset Prices and

Business Loans 106 8.1. Impulse-Response Functions 11410.1. Average Distance to Default (DD)of Large Banking Groups 12310.2. System DD Minus Average DD of Large Banking Groups 12410.3. Correlation between Banking and Insurance Sector DDs 12410.4. Average DD in Canada and the United States 12510.5. Risk Concentration in Canada and the United States 125

©International Monetary Fund. Not for Redistribution

This page intentionally left blank

©International Monetary Fund. Not for Redistribution

ix

Preface

Two decades ago, Canada was running current account deficits and its fiscal situation looked dismal. While inflation had come down from double digits, it remained in the 4–5 percent range, the unemployment rate was above 8 percent, and GDP growth was erratic. Gloom permeated assessments of Canada’s economic prospects.

Facing up to the challenges, Canadian authorities undertook an impressive set of reforms to set the economy on the right course. Tax reform, tight expenditure control, as well as an explicit fiscal framework that focused on reduction and, subsequently, elimi-nation of the deficit have turned the fiscal situation around. Pension reform has put the Canada and Quebec Pension Plans on a sustainable track. An inflation targeting (IT) framework, which Canada was the second in the world to introduce, has anchored infla-tion expectations and made monetary policy more predictable. The Canada-U.S. Free Trade Agreement, followed by the North American Free Trade Agreement (NAFTA), opened the economy to more foreign competition. This was accompanied by an agreement on internal trade and other measures to make the economy more flexible.

As a result, the economic situation now looks dramatically different. The stop-and-go cycle is a thing of the past. Inflation has averaged 2 percent since December 1995, when that number was adopted as the target in the IT agreement. The unemployment rate is at a 33-year low, and real GDP growth has been among the fastest in industrial countries during the last 10 years. The federal budget has recorded nine surpluses in a row, and general government debt is trending down. The current account has been in surplus since 1999. The new framework has dramatically improved macroeconomic stability, with variation in GDP growth, inflation, and interest rates much lower than previously. While the performance of many industrial countries has improved, Canada’s improvement has been particularly dramatic.

This improvement has occurred despite a series of major shocks, including the emerg-ing market financial crises of the 1990s; the collapse of the dot-com bubble later in the decade; the September 11, 2001, terrorist attacks on the United States; outbreaks of SARS (severe acute respiratory syndrome) and mad cow disease (bovine spongiform encepha-lopathy, or BSE); growing competition from the rising Asian economies, particularly China; and wide fluctuations in commodity prices. The Canadian economy has exhibited remarkable resilience in the face of these challenges. In particular, the economy has adjusted extremely well to the recent run-up in energy prices and the exchange rate, which has put differential pressures on regions and sectors, boosting the resource-rich western provinces and squeezing manufacturers in central Canada.

Despite the successful record, challenges inevitably remain. Population aging and the rising cost of medical services are putting pressure on government finances, and reining in health care costs remains an important challenge, which is by no means unique to Canada. While safe and sound, Canada’s financial sector may lack dynamism—possibly reflecting a regime that effectively precludes takeovers from home or abroad. Marginal tax rates on investment are among the highest in the world, which may help explain why the rise in productivity has been less striking than for real GDP.

The IMF enjoys a close working relationship with the Canadian authorities and has been highly supportive of Canada’s strong institutional framework and continuing struc-tural reforms, just as the Canadians have provided the IMF with much useful insight on its own future direction. This volume collects analyses of various aspects of the Canadian

©International Monetary Fund. Not for Redistribution

PREFACE

x

economy undertaken by IMF staff members in recent years. Strikingly, despite its eco-nomic size, Canada features many characteristics of a “small open economy”—it is rela-tively open; its trade is highly concentrated with the United States, making it susceptible to economic fluctuations in its southern neighbor; and it is subject to the vicissitudes of global economic conditions by virture of its large commodity exports. Canada’s success in facing challenges common to many other IMF member countries can thus provide valu-able lessons to a wide range of academics and policymakers.

This paper has benefited from the comments of colleagues in the Western Hemisphere Department (WHD) and in other departments of the IMF. In particular, the editors appre-ciate the contributions of Charles Collins and Christopher Towe. The editors are also grateful for the assistance of Alfred Go, RoseMarie Elizabeth Fonseca, Eneshi Irene Kapijimpanga, and Volodymyr Tulin in the WHD department. Linda Griffin Kean of the External Relations Department edited the manuscript and coordinated the production of the publication. The views expressed in this paper are those of the IMF staff and do not necessarily reflect the views of national authorities or IMF Executive Directors.

©International Monetary Fund. Not for Redistribution

xi

Abbreviations

bbl barrelBLS Bureau of Labor Statistics (U.S.)BSE bovine spongiform encephalopathy (mad cow disease)CAPP Canadian Association of Petroleum Producerscf/bbl cubic feet of natural gas per barrel of oilCPI consumer price indexCPP Canada Pension PlanCPPIB CPP Investment BoardDD distance to defaultEI Employment InsuranceEUB Energy and Utilities Board (Alberta)FIRE finance, insurance, and real estateGDP gross domestic productGEM Global Economic Model (IMF)GFM Global Fiscal Model (IMF)GIS Guaranteed Income Supplement GMM generalized method of momentsGPO gross product originating (by industry)GST Goods and Services TaxHP Hodrick-PrescottICT information and communication technologyIO industrial organizationLBG large banking groupmb/d million barrels a dayMCI Monetary Conditions IndexMST Manufacturers’ Sales TaxNAFTA North American Free Trade AgreementNAICS North American Industry Classification SystemNEB National Energy BoardOAS Old-Age SecurityOECD Organization for Economic Cooperation and DevelopmentOLS ordinary least squaresOPEC Organization of the Petroleum Exporting CountriesOSFI Office of the Superintendent of Financial InstitutionsPEM public expenditure managementREER real effective exchange rateROSC Report on the Observance of Standards and Codes (IMF and World Bank)RPP Registered Pension PlanRRSP Registered Retirement Savings PlanSARS severe acute respiratory syndromeSGP Stability and Growth Pact (EU)SIC Standard Industrial ClassificationSME small and medium-size enterpriseSPA Spousal Allowance TFP total factor productivity

©International Monetary Fund. Not for Redistribution

xii

TPSP tax-prepaid savings planVAR vector autoregressionWEO World Economic OutlookWTI West Texas IntermediateYBE yearly basic exemption YMPE yearly maximum pensionable earnings

ABBREVIATIONS

©International Monetary Fund. Not for Redistribution

1

Introduction

Vladimir Klyuev

Over the last decade, Canada has enjoyed robust GDP growth, declining unemployment, low

and stable inflation, and a string of fiscal and cur-rent account surpluses. These outcomes are largely the result of sound macroeconomic policies and a favorable external environment. This book analyzes Canada’s recent economic success, drawing together analyses conducted by IMF staff members. The book outlines the effects of recent fiscal, monetary, and financial policies and examines the economy’s salient features, including its close trade integration with the United States, its large commodity sector, and its decentraliza-tion and regional diversity.

Part I of this paper assesses Canada’s macroeconomic environment, including factors underlying the business cycle, the macroeconomic implications of Canada’s energy sector, and the interrelationships between Can-ada and the United States. Part II takes a closer look at Canada’s fiscal policies, including a comparison of Canadian budget forecasts with those of other indus-trial countries, an examination of the results of pension reform, and the opportunities for tax reduction afforded by Canada’s fiscal surpluses. Part III explores Canada’s monetary and financial policy environment by outlin-ing the positive results of Canada’s long experience with inflation targeting (IT), reviewing the evolution of the monetary policy transmission mechanism since the 1990s, and examining Canada’s banking system and the implications for competition and financial stability. It is based largely on background work conducted by IMF staff in the context of annual bilateral consulta-tions with Canada. While some of the frameworks (for example, fiscal) have evolved, and some analysis may not be based on the most recent data, the underlying messages remain relevant.

Macroeconomic Environment

Canada is an open economy whose trade is domi-nated by the United States. Kose (2004) finds that Canadian-U.S. free trade agreements have substan-tially increased trade and financial flows while increasing business-cycle synchronicity. In Section I of this paper, Alejandro Justiniano uses factor

analysis to demonstrate that the U.S. business cycle explains about half the variation in Canada’s real GDP and industrial production.

Canada’s impressive GDP growth over the last decade despite somewhat lackluster productivity per-formance is explained by a substantial rise in labor force participation, particularly among women. Tsounta (2006) finds that reforms in the Canadian tax and ben-efit system in the mid-1990s account for at least a third of the observed increase in female participation during 1995–2001.

Since the early 1980s, Canada, like the United States, has experienced a secular decline in the household sav-ing rate. Faulkner-MacDonagh (2004) estimates a long-run relationship between the saving rate and household net worth, inflation, interest rates, and government spending and finds this decline reflects improvement in the fiscal balance, success in fighting inflation, and increases in households’ net worth. Klyuev and Mills (2006), using an error-correction framework on saving behavior in four Anglo-Saxon economies, find that, in contrast to the experience in the United States, the decline in the Canadian household saving rate in recent years has not coincided with a rise in home-equity withdrawal.

Canada’s large commodity sector makes the coun-try susceptible to terms of trade shocks that spill over into the Canadian dollar—its “commodity currency”—which has appreciated strongly in recent years as oil and other commodity prices have soared. Lee and Mühlei-sen (2004) argue that this appreciation has largely been in line with the fundamentals. In Section II of this paper, Tamim Bayoumi and Martin Mühleisen rein-force this assessment and, in particular, argue that the relationship between the Canadian dollar and energy prices has tightened over time, reflecting the growing importance of net energy exports.

Section III explores the deep interrelationships between the Canadian and U.S. economies. One of Can-ada’s challenges is its regionally specialized economy, with manufacturing concentrated in the central prov-inces and raw materials production elsewhere. In the first segment of this section, Vladimir Klyuev and Rodolfo Luzio find that, despite some convergence over the last decade, Canadian regions are still significantly more

©International Monetary Fund. Not for Redistribution

INTRODUCTION

2

diverse than their U.S. counterparts in terms of industrial structure and their responses to macroeconomic shocks.

By making asymmetric shocks more likely, regional diversity underscores the need for flexible markets. Bay-oumi, Sutton, and Swiston (2006) find that Canada’s labor markets are flexible by international standards, with migration playing a significant role in adjustment. Flexibility is particularly notable from the province of Ontario westward, while adjustment is more sluggish in Quebec and the Atlantic provinces. This may reflect the finding by Kaufman, Swagel, and Dunaway (2003) that the Employment Insurance (EI) system discour-ages migration and impedes convergence in per capita output. Moving beyond the labor market, Bayoumi and Cardarelli (2005) find that the Canadian economy is characterized by a relatively high degree of flexibility, as demonstrated by substantial changes in the industrial structure over time, high rates both of firm entry and exit and of job creation and destruction, and speedy adjustment to macroeconomic disturbances.

Despite the close integration between the Canadian and U.S. economies, however, the labor productivity gap between the two countries has widened over the last two decades. In the second segment of Section III, Roberto Cardarelli explores the factors that have led to the Canadian–U.S. productivity gap using a sectoral growth-accounting approach. This lack of productivity convergence is also analyzed in Cardarelli and Kose (2004), who find that it is explained by Canada’s indus-trial structure rather than increased trade integration with the United States. In this section, Cardarelli con-cludes that the United States has been much more suc-cessful in shifting resources toward high-productivity industries and that differences in industrial structure explain the majority of the productivity growth gap over the second half of the 1990s.

In the final part of Section III, Iryna Ivaschenko and Andrew Swiston explore the impact of U.S. shocks on the Canadian economy, using a two-country model. They find that spillovers from U.S. activity are signifi-cant but can be mitigated by a speedy monetary policy response, while U.S. monetary policy has relatively modest effects on Canada.

Fiscal Policies

Canada has had an enviable fiscal record in recent years, moving from chronic deficits and rising public debt to nine consecutive federal budget surpluses. In Section IV, Martin Mühleisen and others emphasize that macroeconomic uncertainty makes necessary a considerable cushion to ensure that Canada’s asym-metric “balanced budget or better” objective (in place

until 2006) can be met. They compare Canada’s budget forecasts against those of other industrial countries and find that the authorities have a tendency to build that cushion partly by using conservative macroeconomic and fiscal assumptions. They suggest that the transpar-ency of the budget process could be improved.

Cardarelli (2003) finds the country’s long-term fis-cal prospects to be relatively favorable even in the face of demographic pressures. In Section V, Mühleisen reviews the current status of the pension system and points out that, despite the success of the 1998 reform, there is scope for simplifying and better targeting pub-lic pension benefits. At the same time, cost pressures within the health care system remain a challenge (De Masi and Towe, 2003).

Section VI reviews the opportunities for tax reduc-tion that have been opened by Canada’s fiscal pru-dence. Using the IMF’s Global Fiscal Model (GFM), Bayoumi and Dennis Botman demonstrate significant macroeconomic benefits of debt reduction, and Botman then uses the model to compare the efficiency gains from different tax cuts and concludes that the efficiency gains from cutting the Goods and Services Tax (GST, Canada’s value-added tax) would be relatively low.

Monetary and Financial Policies

Canada was among the pioneers of inflation target-ing. In Section VII, Jorge E. Roldos tracks changes in the monetary transmission mechanism in Canada since 1990 and shows that these changes are partly the result of a shift from indirect (bank-based) to direct (market-based) financing, which increased the respon-siveness of aggregate demand to the real interest rate. In Section VIII, Bayoumi and Klyuev examine the con-tribution of the IT regime to reducing Canada’s mac-roeconomic volatility since 1991. They attribute the enhanced stability during this period primarily to the high credibility of the new regime, which made infla-tion expectations more forward looking, rather than to changes in the monetary reaction function or tamer shocks. The final two sections in this part address the relationship between financial stability and efficiency. A key policy issue in Canada is whether to allow fur-ther mergers in a banking system dominated by six large banking groups. In Section IX, Ivaschenko finds that the Canadian banking system is relatively competi-tive and efficient despite this concentration. In Section X, Gianni De Nicolò, Alexander Tieman, and Robert Corker calculate market-based soundness indicators for the six large banks and find that the low correlation of risk profiles across these banks enhances the resilience of Canada’s financial system.

©International Monetary Fund. Not for Redistribution

Part I

Macroeconomic Environment

©International Monetary Fund. Not for Redistribution

This page intentionally left blank

©International Monetary Fund. Not for Redistribution

5

I Factoring in Canadian Cycles

Alejandro Justiniano

Canada’s recent economic history illustrates the important role of external as well as domestic mac-

roeconomic disturbances. Canada’s economy slowed in 2001, in the wake of the global slowdown, although by less than in many other countries. In 2003, the recovery was interrupted by a series of shocks that moderated growth: externally, there was an appreciation of the Canadian dollar and a case of mad cow disease that constrained agricultural exports; domestically, there were an outbreak of SARS (severe acute respiratory syndrome) and some forest fires. Growth rebounded in 2004, partly as a result of strong global commodity demand, but appreciation of the Canadian dollar led to concerns about the prospects for 2005.

Previous studies have documented the importance of U.S. real shocks on Canadian business cycles. For instance, IMF (2004a) concluded that the synchroniza-tion of real output, consumption, and investment fluc-tuations between Canada and the United States has increased over the last two decades. Other work using vector autoregression (VAR) techniques on a small set of Canadian and foreign variables also has concluded that developments in the United States strongly influence real activity and nominal variables in Canada (Schmitt-Grohé, 1998; Cushman and Zha, 1997; Burbidge and Harrison, 1985). These findings naturally lead to ques-tions about the transmission channels through which U.S. and other external shocks impact the Canadian economy. This section uses recent developments in dynamic factor models to undertake a comprehensive and broad-based analysis of the role of domestic and external shocks in the Canadian economy.

Dynamic Factor Analysis Dynamic factor analysis has a number of advantages

over the use of VARs:• A wider set of series can be analyzed. The number of

variables that can be included in a VAR is limited by the need to include lagged values of all series in the estimation. Factor analysis, in contrast, allows a wider range of series to be analyzed, allowing for a more comprehensive analysis of economic fluctuations.

• The number of shocks is determined by the data. In VAR models, the number of disturbances is by

definition equal to the number of series in the esti-mation. In factor analysis, the number of shocks is determined statistically.

• Factor analysis provides more information on the dis-turbances. Factor analysis and VARs use similar tech-niques to identify shocks. However, in contrast to the estimation of exactly identified VARs, convergence diagnostics in the estimation of factor models can be used to check if the identifying restrictions are valid.1

• Factor models provide relatively efficient forecasts. The internal dynamics of the factors and their effects across series can be used to project likely future developments in the economy. By summarizing the information contained in a large number of series, forecasts based on dynamic factor models can out-perform those obtained from VARs.2In contrast to recent work on the international trans-

mission of shocks, this study analyzes the effects of multiple shocks with a flexible specification of dynam-ics.3 Extending the earlier work of Gregory, Head, and Raynauld (1997) and Kose, Otrok, and Whiteman (2003), this paper uses dynamic factors to examine multiple domestic and external shocks affecting the Canadian economy. Moreover, a flexible specification of dynamics allows the factors to affect series contem-poraneously and with one lag. Therefore, the analysis can account for spillover effects.4

The factor model used here assumes that each series can be described using a small number of factors with series-specific dynamics plus an error term. For exam-ple, consider the case of two U.S. and two Canadian

1Convergence diagnostics in the estimation can indicate problems with the identifying assumptions. Note that it is also possible to test restrictions in overidentified VARs.

2Indeed, recent academic research suggests that factor models provide gains in the accuracy of forecasts of the data they describe, relative to small-scale VARs and other methods. See, for instance, Stock and Watson (2002).

3Much recent work in this field uses principal components to analyze the transmission of shocks across real GDP series. See, for instance, Bowden and Martin (1996); Lumsdaine and Prasad (2003); Melek Mansour (1999); and Helbling and Bayoumi (2003). This partly reflects recent advances in estimation techniques (Stock and Watson, 1998; Forni and others, 2001; and Kim and Nelson, 1999).

4The specification of dynamics is, consequently, similar to the one preferred by Kaufmann (2000) for the analysis of European business cycles.

©International Monetary Fund. Not for Redistribution

I. FACTORING IN CANADIAN CYCLES

6

series labeled yUS1, yUS2, yCN1 and yCN2. Assume these series are driven by two external and two domestic fac-tors, labeled fE1, fE2, fD1 and fD2, that affect the series both contemporaneously and with one lag.5 Then the model is

yUS1(t) fE1(t) fE1(t–1) ηUS1(t) yUS2(t) fE2(t) fE2(t–1) ηUS2(t)[ ] = A0 [ ]+ A1 [ ] + [ ] , yCN1(t) fD1(t) fD1(t–1) ηCN1(t) yCN2(t) fD2(t) fD2(t–1) ηCN2(t)

where η(t) is a series-specific error and the matrices of coefficients are given by

αsUS1,E1 αs

US1,E2 βsUS1,D1 βs

US1,D2

αsUS2,E1 αs

US2,E2 βsUS2,D1 βs

US2,D2A

s = [ ] for s = 0,1

αsCN1,E1 αs

CN1,E2 βsCN1,D1 βs

CN1,D2

αsCN2,E1 αs

CN2,E2 βsCN2,D1 βs

CN2,D2

Factors and coefficients are provided as outputs of the estimation process, based on a few identifying restrictions. Factors are identified by assuming that they

5Of course, the simplicity of this example does not highlight one of the greatest advantages of factor models: working with several (possibly hundreds of) series driven by a few common shocks.

are orthogonal both to each other and to the series-spe-cific error terms and by exclusion restrictions similar to those used in VARs. For example, this analysis assumes that Canada is a small open economy so that external factors affect economic variables in both the United States and Canada and domestic factors affect only the Canadian series. In the example above, this implies that βS

US,D1 and βSUS,D2 = 0 for the U.S. series both in the

contemporaneous and lagged coefficients (s = 0,1). In addition, the first external factor is assumed to affect the first U.S. series contemporaneously, whereas the second external factor only impacts with a lag (β0

US1,E2 = 0). A similar assumption applies to the way domestic fac-tors affect the two Canadian variables.

Canadian–U.S. Economic Interactions

To provide a comprehensive description of economic interactions within and across the United States and Canada, a large number of variables are employed in the analysis. Specifically, the estimation uses a panel of 44 quarterly series from early 1984 to early 2004, comprising world prices for oil and other commodities, 18 U.S. real and nominal series, and 24 real and nominal

Table 1.1. Variance Decompositions for the United States from a Factor Model

External External Factor 1, Factor 2,Series Transformation Median Median

U.S. producer price index intermediate goods LD 0.82 0.01U.S. industrial production index LDHP 0.03 0.65U.S. unemployment rate DHP 0.02 0.26U.S. shares price index (nominal) LD 0.07 0.04U.S. GDP LDHP 0.01 0.58U.S. consumption LDHP 0.03 0.25U.S. investment LDHP 0.01 0.23U.S. government LDHP 0.04 0.05U.S. exports (goods) LDHP 0.01 0.14U.S. imports (goods) LDHP 0.01 0.51U.S. hours LDHP 0.07 0.33U.S. labor productivity LDHP 0.02 0.32U.S. capacity utilization rate DHP 0.03 0.71U.S. CPI (all goods) LD 0.48 0.03U.S. producer price index, finished goods LD 0.68 0.08U.S. unit labor costs LD 0.05 0.10U.S. federal funds rate D 0.15 0.45U.S. M2 LD 0.08 0.06World price of oil (non-OPEC countries) LD 0.65 0.03World commodity price index LD 0.17 0.11

Source: IMF staff calculations.Notes: CPI = consumer price index; D = first difference; LD = log of first difference; DHP = deviation from

Hodrick-Prescott (HP) trend; LDHP = log of deviation from HP trend; OPEC = Organization of Petroleum Exporting Countries.

©International Monetary Fund. Not for Redistribution

7

Canadian variables.6 All series except interest rates are included in terms of their logarithms. Real variables are detrended by calculating deviations from a Hodrick-Prescott trend (with the standard smoothing factor of 1,600), while prices and monetary aggregates are mea-sured as rates of change (that is, the change in the loga-rithm). Further details and sources for the dataset and detrending methods are provided in Boxes 1.1 and 1.2.7

Bayesian analysis resulted in a preferred model including factors that broadly reflect international oil prices, the U.S. cycle, the exchange rate, the non-oil producer and commodity prices, and a Canadian cycle. This model—involving two external and two “domestic”

6This implies that the models and matrices described on the pre-vious page would each consist of 44 rows. The U.S. and Cana-dian series include main national accounts aggregates (real GDP, consumption, investment, government consumption, exports, and imports), other measures of real activity (industrial production, unemployment, hours worked, labor productivity), prices at differ-ent stages of production, interest rates, other financial aggregates, and, in the case of Canada, real exchange rates, prices of exports and imports, and price indices for oil and non-oil commodities.

7For the estimation, the data were also standardized, as is custom-ary in factor analysis, to prevent giving undue weight to the most volatile components in the data.

factors (one of which is associated with the exchange rate, non-oil commodity prices, and producer prices)—resulted in the largest Bayes Factor out of a wide range of estimated models.8 The results also indicate that the factors follow an autoregressive process with three lags, implying potentially quite complex dynamics, and that the factors affect the series contemporaneously and with a one-quarter lag.9

The factors are estimated with fairly narrow error bands, although a widening of the bands over time suggests that the Canadian cycle may be playing a diminishing role (see Justiniano, 2005, Figure I.1). Decompositions that analyze the relative contribution of each factor to fluctuations in individual series, as well as examination of plots of the factors, suggest that external and exchange rate disturbances play a significant role in explaining Canadian fluctuations (Tables 1.1 and 1.2). That said, the explanatory power

8In the Bayesian setting adopted here, the Bayes Factor (the ratio of the posterior model probabilities) corresponds to the ratio of mar-ginal likelihoods. See Kass and Raftery (1995) for an overview of Bayes factors; Geweke (1999) for the method used here to compute the marginal density; and Lopes and West (2004) and Justiniano (2004) for a discussion of these techniques in factor analysis.

9Formal statistical methods did not validate additional lags.

Table 1.2. Variance Decompositions for Canada from a Factor Model

External External Domestic Domestic Factor 1, Factor 2, Factor 1, Factor 2,Series Transformation Median Median Median Median

Canadian real exchange rate LD 0.02 0.03 0.53 0.01Canadian GDP LDHP 0.03 0.42 0.01 0.47Canadian industrial production index LDHP 0.03 0.48 0.01 0.38Canadian nominal exchange rate LD 0.04 0.04 0.76 0.01Canadian labor productivity LDHP 0.02 0.12 0.01 0.43Canadian consumption LDHP 0.07 0.19 0.05 0.16Canadian government LDHP 0.01 0.07 0.01 0.05Canadian investment LDHP 0.01 0.10 0.02 0.02Canadian exports (goods) LDHP 0.02 0.44 0.01 0.02Canadian imports (goods) LDHP 0.01 0.49 0.02 0.07Canadian hours LDHP 0.05 0.25 0.02 0.25Canadian capacity utilization rate DHP 0.02 0.48 0.01 0.02Canadian unemployment rate DHP 0.01 0.26 0.01 0.01Canadian unit labor costs LD 0.06 0.53 0.01 0.02Canadian M2 LD 0.02 0.05 0.02 0.03Canadian bank rate D 0.03 0.36 0.06 0.05Canadian CPI (all goods) LD 0.10 0.05 0.02 0.05Canadian CPI (minus volatile components) LD 0.02 0.04 0.02 0.02Canadian producer price index, excluding oil LD 0.03 0.15 0.26 0.02Canadian commodity price index: energy LD 0.68 0.03 0.01 0.02Canadian commodity price index: nonenergy LD 0.05 0.08 0.07 0.04Canadian export prices LD 0.44 0.03 0.25 0.01Canadian import prices LD 0.06 0.09 0.66 0.03Canadian shares price index (nominal) LD 0.02 0.03 0.15 0.04

Source: IMF staff calculations.Notes: CPI = consumer price index; D = first difference; LD = log of first difference; DHP = deviation from Hodrick-Prescott (HP) trend;

LDHP = log of deviation from HP trend.

Canadian–U.S. Economic Interactions

©International Monetary Fund. Not for Redistribution

I. FACTORING IN CANADIAN CYCLES

8

of each factor varies substantially across variables and, to some extent, also over time. The discussion below provides an overview.

The first external factor can be interpreted as fluctuations in the world price of oil in U.S. dollars and in U.S. producer prices of intermediate inputs (Figure 1.1). It accounts for 65 percent and 82 percent of the respective variances and tracks these series closely. In the United States, this “oil factor” explains much of the variation in prices and in the federal funds rate. In Canada, it accounts for a large amount of the variation in export and energy prices and some 10 percent of fluctuations in the headline consumer price index (CPI), but a smaller proportion of fluctua-tions in core CPI inflation, Canadian interest rates, and nonenergy commodity prices.10 Consistent with

10The more limited impact on Canadian inflation compared to its U.S. counterpart presumably reflects the fact that oil prices are measured in the U.S. currency and hence affect U.S. relative prices more directly.

post-1985 results reported in Kose, Otrok, and White-man (2003), there is little interaction between oil prices and real variables.

The second external factor, the “real foreign factor,” which tracks the U.S. cycle, accounts for almost 60 percent of the deviations of U.S. real GDP from trend (Figure 1.2). It captures the recessions (and subsequent recoveries) of 1990 and 2001, as well as the slowdown in 1995, and it can explain about half of the changes in the federal funds rate (which tends to lead the cycle), particularly since 1987. This real foreign factor also explains about half the movements in U.S. imports and one-quarter of consumption fluctuations.

This real foreign factor has a large influence on Canadian real GDP and industrial production, explain-ing about half their variance. The link with downturns in Canadian real GDP is particularly striking, whereas the synchronization of recoveries is less close—indeed, this factor often leads Canada’s upturns. Interestingly, the factor suggests that the 2001 downturn in Cana-dian real GDP was less than expected given the U.S. slowdown. More recently, however, the recovery of Canadian real GDP has lagged behind the real foreign factor.

The results emphasize the role of trade linkages for the transmission of U.S. cyclical shocks. The impor-tance of trade linkages in explaining the synchroni-zation of fluctuations between the United States and Canada is clear from the fact that the real foreign fac-tor explains about half of the variation in Canadian exports and imports. This relationship appears to have increased during the 1990s, plausibly reflecting greater economic integration over this period.

The real foreign factor also explains about one-third of the fluctuations in Canada’s bank rate, particularly since the mid-1990s. However, these two series behaved quite differently in 1991—the inception of the Bank of Canada’s IT regime—and, to a lesser extent, more recently in 2002−03. Despite the factor’s important role for the Canadian real economy, its impact on inflation is quite limited, however, echoing the conclusions from other studies that have encountered difficulties in estab-lishing a stable relationship between capacity measures and inflation.11

The third factor, the “exchange rate factor,” closely tracks movements in Canada’s exchange rate and non-oil producer and commodity prices (Figure 1.3). As might be expected, this factor is closely associated with movements in import prices and, to a lesser degree, export prices. However, the influence on fluctuations in headline and core CPI is limited, suggesting that

11Demers (2003) documents the instability of the Phillips Curve (inflation versus unemployment) in Canada and finds that measures of cyclical activity are not linked to the evolution of inflation during most of our sample period. Similar observations are discussed in Box 2 in IMF (2004a).

Source: IMF staff calculations.Note: Each series is divided by its standard deviation and its

mean is removed. Vertical axes are therefore measured in standard deviations.

–3

–2

–1

0

1

2

3Oil price

Oil factor

Export prices

–3

–2

–1

0

1

2

3

1984 90 96 2002

1984 90 96 2002

World Price of Oil

Canadian Export Prices

Oil factor

Figure 1.1. “Oil Factor” and Comovements in Selected Series(Standard deviations)

©International Monetary Fund. Not for Redistribution

9

pass-through from import prices subsides as goods move down the production chain. This exchange rate factor also displays some comovements with Canada’s bank rate in the 1980s and mid-1990s, excluding the 1990−91 period when IT was first adopted. Nonethe-less, this relationship seems to weaken considerably after 1998, about the time that the Bank of Canada abandoned the Monetary Conditions Index as an indi-cator of monetary policy.

The last factor, the “real domestic factor,” cor-responds to domestic disturbances responsible for

Canada’s cycle (Figure 1.4). The factor explains about half the fluctuations in Canadian real GDP and about one-third of movements in industrial production.12 There was a close link between this factor and fluctuations in Canada’s real GDP through the mid-1990s. Subse-quently, however, the two series became less correlated, as did other variables that were previously well explained

12The four factors explain close to 95 percent of the variation in Canadian real GDP and industrial production, with the U.S. and Canadian cycles explaining almost 90 percent of the variance.

Source: IMF staff calculations.Note: Each series is divided by its standard deviation and its mean is removed. Vertical axes are therefore

measured in standard deviations.

–3

–2

–1

0

1

2

3

1984 90 96 2002

U.S. GDP

Real foreign factor

U.S. GDP

–3

–2

–1

0

1

2

3

1984 90 96 2002

U.S. imports

Real foreign factor

U.S. Goods Imports

–3

–2

–1

0

1

2

3

1984 90 96 2002

Federal funds rate

Real foreign factor

U.S. Federal Funds Rate

–3

–2

–1

0

1

2

3

1984 90 96 2002

Canadian GDPReal foreign factor

Canadian GDP

–3

–2

–1

0

1

2

3

1984 90 96 2002

Canadian exports

Real foreign factor

Canadian Exports of Goods

–3

–2

–1

0

1

2

3

1984 90 96 2002

Bank of Canada rate Real foreign factor

Canadian Bank Rate

Figure 1.2. “Real Foreign Factor” and Comovements in Selected Series(Standard deviations)

Canadian–U.S. Economic Interactions

©International Monetary Fund. Not for Redistribution

I. FACTORING IN CANADIAN CYCLES

10

by this factor such as industrial production, labor produc-tivity, and hours worked.13 Finally, the real foreign factor appears to lead this domestic real factor somewhat. The correlation coefficients of the first and second lag of the real foreign factor with the real domestic factor are 0.17 and 0.34, respectively. This could indicate that the impact of U.S. fluctuations may be underestimated even under this flexible dynamic specification.

Robustness Checks

The results appear generally robust to changes in the way the data are measured. This was exam-ined by reestimating the model with real variables

13As with the real foreign factor, this real domestic factor has very limited effects on CPI inflation. Indeed, it explains less than 5 per-cent of the variance in inflation. This shift is also reflected in the precision with which this factor is estimated, evident from widening error bands reported in Justiniano (2005, Figure I.1).

measured as rates of change, rather than as deviations from trend. Statistical methods indicate that the same model structure—four factors with extremely similar features—remain valid. More generally, the results were extremely similar to the benchmark case with the following exceptions: • The oil factor now explains a greater share of consump-

tion. This is particularly true for the United States.• The spillovers from the U.S. cycle to Canadian real

GDP and industrial production are lower.14 One pos-sible explanation for this is that first differencing makes it more difficult to identify spillovers, as there is a greater degree of noise in the data. Estimating the model with and without lags reveals

the importance of spillover effects from external shocks. The model was reestimated excluding the lags in the impact of factors on individual series to explore the importance of this assumption to the results. Com-paring the variance decompositions obtained with and without a lag indicates the following qualitative differences: • The share of the variance explained by the U.S.

real foreign factor in the United States falls when the lag is excluded. This is particularly true for

14Variance shares for Canadian real GDP and industrial pro-duction explained by the real foreign factor are 15 and 22 percent, respectively. Curiously, the lower variance shares for Canada’s real GDP cannot be attributed to difficulties in explaining the volatility of Canadian trade volumes. Indeed, for real exports the proportion of the variance accounted for by the factor is higher in growth rates (56 percent, compared to 46 percent).

Source: IMF staff calculations.Note: Each series is divided by its standard deviation and its

mean is removed. Vertical axes are therefore measured in standard deviations.

–3

–2

–1

0

1

2

3

1984 90 92 9486 88 96 98 022000

Canadian GDP

Real domestic factor

Canadian GDP

Figure 1.4. “Real Domestic Factor” and Comovements in Canadian GDP(Standard deviations)

Source: IMF staff calculations.Note: Each series is divided by its standard deviation and its

mean is removed. Vertical axes are therefore measured in standard deviations.

–3

–2

–1

0

1

2

3

–3

–2

–1

0

1

2

3

1984 90 96 2002

1984 90 96 2002

Nominal Exchange Rate (Can$ per US$)

Canadian Import Prices

Import prices

Exchange rate factor

Exchange rate

Exchange rate factor

Figure 1.3. “Exchange Rate Factor” and Comovements in Selected Series(Standard deviations)

©International Monetary Fund. Not for Redistribution

11

Box 1.1. United States: Data Sources, Descriptions, and Transformations

Series Transformation Description Source

US capacity utilization rate

DHP and D Manufacturing survey: capacity utilization rate (SA, %)

OECD

US consumption LDHP and LD Private final consumption expenditure (SAAR, millions chained 2000 US$)

OECD

US CPI (all goods) LD All urban consumers, all items Federal Reserve Bank of St. Louis

US exports (goods) LDHP and LD Exports of goods (SAAR, billions chained 2000 US$)

OECD

US federal funds rate D Effective federal funds rates (averages of daily values)

Federal Reserve Bank of St. Louis

US GDP LDHP and LD Gross domestic product (SAAR, billions chained 2000 US$)

OECD

US government LDHP and LD Government final consumption expenditure (SAAR, millions chained 2000 US$)

OECD

US hours LDHP and LD Total private, quarterly averages (SA, hours) Bureau of Labor Statistics

US imports (goods) LDHP and LD Imports of goods (SAAR, billions chained 2000 US$)

OECD

US industrial productionindex

LDHP and LD Industrial production index (SA, 1997 = 100) Federal Reserve Board

US investment LDHP and LD Gross fixed capital formation (SAAR, millions chained 2000 US$)

OECD

US labor productivity LDHP and LD Labor productivity index of the total economy

OECD

US M2 LD Money stock, M2 Federal Reserve Board

US producer price indexfinished goods

LD PPI finished goods (SA, 1982 = 100) Bureau of Labor Statistics

US producer price indexintermediate goods

LD PPI intermediate materials, supplies and components (SA, 1982 = 100)

Bureau of Labor Statistics

US shares price index(nominal)

LD Standard & Poor’s 500 Composite (1941–43 = 10)

Wall Street Journal

US unemployment rate DHP and D Standardized unemployment rate (SA, %) OECD

US unit labor costs LD Unit labor cost, business sector (SA, 1992 = 100)

Bureau of Labor Statistics

World commodity price index

LD Spot commodity price index: all commodities (1967 = 100)

Commodity Research Bureau

World price of oil (non-OPEC countries)

LD Average crude oil spot price: total non-OPEC (US$/barrel)

Department of Energy

Notes: CPI = consumer price index; PPI = producer price index; D = difference; LD = log differences; DHP = deviations from Hodrick-Prescott (HP) trend; LDHP = log-deviations from HP trend; SA = seasonally adjusted; SAAR = seasonally adjusted annualized rates or levels; OPEC = Organization of the Petroleum Exporting Countries; OECD = Organization for Economic Cooperation and Development.

Robustness Checks

©International Monetary Fund. Not for Redistribution

I. FACTORING IN CANADIAN CYCLES

12

Box 1.2. Canada: Data Sources, Descriptions, and Transformations

Series Transformation Description Source

CDN bank rate D Official discount rate of the Bank of Canada (monthly average, %)

Bank of Canada

CDN capacity utilization rate DHP and D Manufacturing survey: capacity utilization rate (NSA, %) OECD

CDN commodity price index: energy

LD Commodity price index, energy (1982–90 = 100) Bank of Canada

CDN commodity price index: non-energy

LD Commodity price index, total excluding energy (1982–90 = 100)

Bank of Canada

CDN consumption LDHP and LD Private final consumption expenditure (millions chained 1997 Can$, SAAR)

OECD

CDN CPI (all goods) LD CPI, all items (NSA, 1992 = 100) Statistics Canada

CDN CPI (minus volatile components)

LD CPI, all items excluding 8 volatile components and indirect taxes (NSA, 1992 = 100)

Bank of Canada

CDN export prices LD Export price index (SA, 1992 = 100) Statistics Canada

CDN exports (goods) LDHP and LD Exports of goods (SAAR, millions chained 1997 Can$, SAAR)

OECD

CDN GDP LDHP and LD Gross domestic product (millions chained 1997 Can$, SAAR)

OECD

CDN government LDHP and LD Government final consumption expenditure (millions chained 1997 C$, SAAR)

OECD

CDN hours LDHP and LD Actual hours worked during reference week, all sectors (SA, 1,000s of hours)

Statistics Canada

CDN import prices LD Import price index (SA, 1992 = 100) Statistics Canada

CDN imports (goods) LDHP and LD Imports of goods (SAAR, millions chained 1997 Can$, SAAR)

OECD

CDN industrial production index LDHP and LD Industrial production index (SA, 2000 = 100) OECD

CDN investment LDHP and LD Gross fixed capital formation (millions chained 1997 Can$, SAAR)

OECD

CDN labor productivity LDHP and LD Labor productivity index of the total economy OECD

CDN M2 LD M1 plus all checkable notices and personable term deposits Bank of Canada

CDN nominal exchange rate LD U.S. dollar noon spot rate (Can$/US$) Bank of Canada

CDN producer price index excluding oil

LD PPI, total excluding petroleum/coal products (NSA, 1997 = 100)

Statistics Canada

CDN real exchange rate LD Broad real effective exchange rate index (2000 = 100) JPMorgan

CDN share price index (nominal) LD S&P/TSX: 60 index (1/29/1982 = 100) Bank of Canada

CDN unemployment rate DHP and D Standardized unemployment rate (SA, %) OECD

CDN unit labor costs LD Unit labor cost, manufacturing (SA, 2000 = 100) OECD

Notes: CPI = consumer price index; PPI = producer price index; D = difference; LD = log differences; DHP = deviations from Hodrick-Prescott (HP) trend; LDHP = log-deviations from HP trend; S&P = Standard and Poor’s; TSX = Toronto Stock Exchange; SA = seasonally adjusted; SAAR = seasonally adjusted annualized rates or levels; NSA = not seasonally adjusted; OECD = Organization for Economic Cooperation and Development.

©International Monetary Fund. Not for Redistribution

13

“sluggish” variables such as the unemployment rate and investment.

• Without lags, the proportion of the variation in Cana-dian real GDP (as well as industrial production and unit labor costs) attributed to the U.S. cycle falls.15

This indicates that lags matter in the effects of U.S. activity on the Canadian economy.

Conclusions

The results of this analysis suggest the following: • Four factors explain a large amount of the fluctua-

tions across a wide range of macroeconomic series in Canada and the United States. For instance, they

15In contrast, variance shares remain largely unchanged for Cana-dian exports, labor productivity, and the capacity utilization rate.

account for roughly 95 percent of the variance in Canadian real GDP and industrial production. These factors seem to correspond to world oil price shocks, the U.S. cycle, an exchange rate and non-oil price shock, and a domestic Canadian cyclical factor.

• The fraction of the variance accounted for by fac-tor varies substantially across series. Fluctuations in Canadian real GDP are about equally explained by external and domestic cycles, while the role of exter-nal factors is even larger for other real series and for policy interest rates. Furthermore, the analysis indi-cates that the importance of the real domestic factor declined during the 1990s.

• These results appear relatively robust to alternative methods of detrending the data. In addition, allow-ing for differences in the speed at which factors affect specific series is important for distinguishing spillover effects.

Conclusions

©International Monetary Fund. Not for Redistribution

14

II Energy, the Exchange Rate, and the Economy

Tamim Bayoumi and Martin Mühleisen

The boom in global energy markets has again focused attention on Canada’s role as a major

energy producer. The rapid rise in oil and gas prices in the past few years has lifted energy exports back on par with other commodity exports. It has also set the stage for increased energy production and the development of nontraditional energy reserves, especially Alberta’s oil sands.

This section explores the macroeconomic conse-quences for Canada of expanding oil sands produc-tion, with a particular focus on the exchange rate. It describes the potential benefits of higher energy exports and, given that North American energy markets are expected to remain tight for the foreseeable future, the possible impact of further improvements in the terms of trade. This section also discusses the risk that these gains could be offset by a decline in exports of other tradable goods, owing to a possible appreciation of the exchange rate (“Dutch disease”).

Included here is an exchange rate equation that relates the Canadian dollar’s value to energy and nonenergy commodity trade, with a particular focus on changes in the composition of trade over time. Although the close link between nonenergy commodity prices and the value of the Canadian dollar has been well established in earlier studies, the role of energy prices has remained unclear. This analysis gauges the degree to which the recent appreciation of the exchange rate reflects stron-ger fundamentals, and estimates the likely impact of the development of oil sands production. The results suggest that a permanent increase of net energy exports lifts the equilibrium value of the Canadian dollar, but that the impact on nonenergy exports remains modest.

Outlook for Energy Production

Canada is the world’s third largest producer of natural gas and the eleventh largest producer of oil, and there-fore has benefited enormously from the recent run-up in energy prices (Figure 2.1). Canadian gas exports reached Can$27 billion in 2004 (2 percent of GDP), a 150 percent increase from their value in the late 1990s, with all but 5 percentage points of revenue gains result-ing from higher prices. Oil exports also improved on

the backs of rising prices and higher production, but a concomitant rise in the value of petroleum imports, especially into eastern Canada, kept net oil exports at around 0.7 percent of GDP (Figure 2.2).

Canada’s future as an energy-exporting nation increasingly rests on the development of unconven-tional oil and gas production. Despite record drilling in the Western Canada Sedimentary Basin—Canada’s most important source of crude oil and natural gas— production in that area appears to have peaked. Activity in offshore locations and the Northern Territories is being stepped up but will likely prove insufficient to stave off the decline in traditional oil and gas reserves over the medium to long term (National Energy Board (NEB), 2003). Canada still has access to large reserves of coal, and especially to coal bed methane gas which, if exploitation proves commercially feasible, could boost gas production for years to come. In the near term, however, oil production from Canada’s tar sands offers the greatest potential for expanding energy exports.

Producing Oil from Tar Sands

Canada’s tar sands—located mostly in the province of Alberta—contain one of the largest known hydro-carbon deposits in the world.1 Extracting oil from tar sands requires large amounts of capital and energy, and until recently, this made it prohibitively expensive to tap most of these resources. Recently, however, extraction from some locations has proven commercially viable, using one of two approaches:• Mining and upgrading: Deposits close to the surface

are mined and mixed with water to produce a slurry from which bitumen can be extracted. Bitumen must be blended with a diluent to be transportable and must be upgraded with hydrogen to create an acceptable feedstock for conventional oil refineries. The upgrad-ing process uses electrical power, natural gas as a source of heat, and hydrogen for hydroprocessing.

• “In situ” production: Deposits underground are typi-cally exploited by directing steam through drilled

1Tar sands consist of quartz sand, silt, clay, water, and 10–12 per-cent bitumen.

©International Monetary Fund. Not for Redistribution

15

wells into a bitumen reservoir. The steam reduces the bitumen’s viscosity, allowing a mixture of heated oil and water to be pumped to the surface through another set of pipes. In situ output has lower viscosity than mined bitumen and is therefore often blended with lighter products, rather than upgraded, before being shipped to downstream refineries.2The tightening of global energy markets has made

oil sands production much more profitable. Techno-logical improvements brought down supply costs in the 1980s and early 1990s, although a comparison by Birol and Davie (2001) suggests that production costs from most other sources remain significantly lower.3Although the increase in natural gas prices reportedly raised the long-term break-even point for oil sands proj-ects to US$30–35/bbl during 2005, from about US$25/bbl in 2004, it remains well below long-term oil price expectations.

These developments led to a significant upward revi-sion in estimates of Canada’s commercially accessible oil reserves. According to the Alberta Energy and Utili-ties Board (EUB), tar sands containing about 174 bil-lion bbl of oil are classified as recoverable with current technologies. Together with conventional oil fields, this leaves Canada with about one-sixth of the estimated 1,277 billion bbl of global oil reserves, second only to Saudi Arabia’s 262 billion bbl (Radler, 2002).4

2Integrated mining/upgrading operations require around 900 cubic feet of natural gas per barrel of oil (cf/bbl) produced. The in situ process requires about 1,000–1,200 cf/bbl, with an additional 650 cf/bbl for upgrading (Schroeder, Mui, and Maxwell, 2005).

3Supply costs are equal to the price needed to cover capital expen-ditures, operating costs, royalties, and taxes, and they typically allow for a 10 percent return on investment over the lifespan of the project (30–40 years). For a comparison of production versus supply costs, see Bayoumi and Mühleisen (2006, Figure 3).

4Estimates of the size of energy reserves are highly uncertain, given geological uncertainties and intrinsic measurement problems.

As a result, investment in the Canadian oil sands is expected to take off sharply. A recent industry survey suggests that capital spending in the oil sands sector could amount to Can$8.5 billion (¾ percent of GDP) a year by 2008, plus an additional Can$15 billion in replacement investment over 2005–15 (Alberta Employment, Immigration and Industry, 2006). This is a relatively conservative estimate, given that the size of existing investment proposals has been discounted based on the project’s position in the approval and implementation process, leaving the projected peak of investment activity well below its nominal value of around Can$14 billion.5

Obstacles and Possible Solutions

Development of Canada’s oil sands could be con-strained by four fundamental factors. While possible solutions exist for all of them, their realization will require time, technological progress, and considerable investment.• Skilled labor: Past projects have suffered from severe

cost overruns, owing in part to difficulties in attract-ing workers in a tight Albertan labor market. The industry is seeking to attract skilled workers, and internal trade and immigration restrictions appar-ently are causing some concern, at least in individual cases.

There has also been a discussion on the extent to which oil sands can be compared to traditional reserves, given the resource- and time-intensive nature of their exploitation (Reynolds, 2005).

5It should be assumed that additional projects will be forthcom-ing, increasing projected investment levels after 2008.

Source: IMF staff estimates.

1991 95 99 2003

Per

thou

sand

cub

ic m

eter

s

Per

barr

el

0100200300400500600700

010203040506070

Crude oil (right scale)

Natural gas (left scale)

Figure 2.1. Oil and Natural Gas Prices(U.S. dollars)

Source: Haver Analytics.

–2

–1

0

1

2

3

4

Coal and other products

Natural gas

Crude petroleum

0420019895928986831980

Figure 2.2. Net Hydrocarbon Energy Exports(Percent of GDP)

Outlook for Energy Production

©International Monetary Fund. Not for Redistribution

II ENERGY, THE EXCHANGE RATE, AND THE ECONOMY

16

• Other production inputs and infrastructure: Given the geographic concentration of the oil sands industry, providing additional amounts of electricity, water, diluents, and other critical production inputs to feed an expansion of output is a key challenge. Augment-ing the existing transportation and pipeline capacity will require substantial resources.

• Natural gas: The demand for natural gas by oil sands operations is expected to roughly double by 2015, contributing to a tightening in the North American gas market and adding to the cost of oil sands pro-duction. Coal bed methane gas and gasification of residual hydrocarbons begin to serve as alternative energy sources, but greenhouse gas emissions could increase as a result.

• Greenhouse gas emissions: Notwithstanding consid-erable improvements in the energy efficiency of oil sands projects, the anticipated production increase complicates Canada’s efforts to meet its emissions reduction objectives. The government has agreed to limit compliance costs for large single emitters to the cost equivalent of about Can$0.25/bbl through 2012. However, uncertainty about the success of emissions-reducing technologies and the costs of future environ-mental regulation could become a limiting factor.6

Export Prospects and Dutch Disease

Production and Exports

The capital already committed to oil sands projects is expected to provide a significant boost to Canadian oil production. Canada produced 2.4 million barrels of crude oil per day (mb/d) in 2004, the bulk of it from conventional oil fields. However, oil sands production has rapidly increased—doubling since the late 1990s—and now stands at 1 mb/d, with most forecasters pro-jecting a further tripling to 2½–3 mb/d by 2015. Further increases are projected thereafter, depending on supply and demand conditions in the oil market and the pace of technological progress in the interim.

Within the next 15 years, additional oil exports could add the equivalent of 1 percent of GDP to Canada’s energy trade surplus at current prices. This estimate is based on a simple static calculation, using assumptions from the NEB’s comprehensive forecast of Canada’s energy supply (NEB, 2003), with the exception that oil sands projections are based on the Canadian Association of Petroleum Producers’ moderate case (CAPP, 2005), which is around the midpoint of the range of recent

6Canada is exploring the feasibility of sequestering CO2 emis-sions underground, building on its experience with advanced oil- recovery technologies. A large pilot with international participation (the Weyburn Project) is currently under way in the province of Saskatchewan.

forecasts. Non-oil GDP would grow at 2#/4 percent, non-energy exports and imports would remain unchanged relative to non-oil GDP, and relative prices, including the exchange rate, would also remain constant.7

Under these assumptions, exports of crude oil could increase by about 2 mb/d by 2020 (Figure 2.3). For the purposes of this simulation, the resulting excess demand for gas relative to the NEB’s projection is assumed to be met from alternative sources, but the downward trend in gas production after 2015 would not be abated.

Total oil export revenues would rise to 4 percent of GDP by 2020, leading to an improvement in the net energy trade balance of about 1 percent of GDP (Table 2.1). The price sensitivity is such that an average increase in real prices of 1 percent a year would yield an additional ¾ percent of GDP in net export revenues by 2020.

Are There Risks of Dutch Disease?

The beneficial economic impact of oil production could be lessened if a rising exchange rate were to negatively affect other tradable goods sectors. Over-all, higher production and export of crude oil should boost the Canadian economy, both by offering high-value-added employment opportunities and by raising national income through additional foreign exchange earnings. However, the nonenergy tradable sector could come under increased pressure if current account sur-

7Canada will likely remain a price taker in the international mar-ket for oil. Although reserves are likely to last for many decades, the long lead times and high costs of expanding oil sands production suggest that additional Canadian oil to market will not meaning-fully affect prices on the global marketplace (Söderbergh, 2005; and Reynolds, 2005).

Sources: National Energy Board, Canadian Association of Petroleum Producers; and IMF staff projections.

0

1

2

3

4

5

6

19 17 15 13 11 09 07 05 03 200115

16

17

18

19

20

21

22

Mill

ion

barr

els

a da

y

Billi

on c

ubic

feet

a d

ayGas (right scale)

Crude oil (left scale)

of which oil sands(left scale)

Figure 2.3. Projected Energy Production

©International Monetary Fund. Not for Redistribution

17

pluses and rising capital inflows put upward pressure on the exchange rate (Dutch disease). Moreover, the economy’s sensitivity to global oil market conditions will increase as the share of oil production in total GDP rises.8

The literature suggests that Dutch disease effects in resource-exporting industrial countries tend to be small. The effect is most often associated with the Netherlands, the United Kingdom, and Norway, all of which have developed energy reserves in the North Sea, and Australia and Canada, which are major commodity exporters. The first three countries experienced both exchange rate appreciation and a contraction in manu-facturing output and employment during the period when their energy exports soared (Figure 2.4). How-ever, although some analyses have reported a statisti-cally significant correlation of exchange rate effects with manufacturing activity, causality has been more difficult to establish.9 In any case, the effects were of a size that led most authors to conclude that the manufac-

8Following Corden’s (1981) seminal paper on Dutch disease, two out of three possible factors apply in Canada’s case: technological progress in exploiting oil sands and an exogenous price increase. The third factor, a windfall discovery, does not apply, given that the rough extent of the oil sands region was already established in 1882 by the Geological Survey of Canada.

9See Hutchison (1994); Bjørnland (1998); Spatafora and Warner (1999); and an overview by Stevens (2003). More recently, Stijns

turing decline in these countries was part of a longer-term deindustrialization trend and was not specifically linked to higher energy exports. Indeed, some stud-ies suggest manufacturing output has benefited from higher demand induced by energy revenues.

There are also few signs that natural resource exports have persistently hurt Canada’s manufacturing sector. Although Canada’s energy exports have risen steadily relative to GDP since 1980 (Figure 2.5), the real effective exchange rate (REER) has declined over that period, both output and exports of manufacturing goods have been stable, and the decline in the share of manufacturing employment has been smaller than in other resource-exporting industrial countries. Within this long-term trend, there was a steep increase in man-ufacturing activity between 1994 and 2000, succeeded by a similarly dramatic decline, which may still be ongoing. However, these shifts have not been closely correlated with movements in the REER, which only began to appreciate in 2002. Instead, the timing sug-gests that Canadian manufacturing remains dependent on the U.S. economic cycle, especially in view of the relatively close correlation between U.S. investment and Canadian goods exports.

(2003) presents some evidence for the presence of Dutch disease effects using a gravitational trade model.

Table 2.1. Canadian Energy Exports: Static Projection

Forecast______________________________________CAPP, Moderate Case 2001 2002 2003 2004 2005 2010 2015 2020

Oil and gas productionCrude oil (mb/d) 2.2 2.4 2.5 2.6 2.5 3.2 3.9 4.8

of which: Oil sands 0.7 0.7 0.9 1.0 1.0 1.9 2.7 3.8Gas (bcf/d) 17.4 17.4 16.9 17.4 17.5 18.8 20.2 19.2

Oil and gas exportsCrude oil (mb/d) 1.3 1.4 1.5 1.6 1.5 2.1 2.6 3.5Gas (bcf/d) 10.6 10.5 9.3 9.7 9.6 10.1 10.6 8.8

(in percent of GDP)Energy exports 5.0 4.3 5.0 5.3 6.4 6.5 6.4 6.4Oil and gas exports 4.7 4.1 4.8 5.1 6.2 6.4 6.3 6.3

of which: Crude oil 1.4 1.6 1.7 2.0 2.6 3.1 3.4 4.0of which: Natural gas 2.3 1.6 2.1 2.1 2.7 2.4 2.2 1.6

Electricity 0.4 0.2 0.2 0.2 0.1 0.1 0.1 0.1

Energy imports 1.6 1.4 1.6 1.9 2.3 2.1 1.7 1.5of which: Crude oil 1.2 1.0 1.1 1.3 1.7 1.6 1.2 1.1

Trade balance 6.4 5.0 4.7 5.1 4.3 4.6 4.9 5.1Net energy exports 3.4 2.8 3.3 3.3 4.0 4.4 4.7 4.9

Net oil and gas exports 3.1 2.7 3.2 3.2 3.9 4.3 4.6 4.8

Sources: IMF staff calculations, based on CAPP (2005); NEB (2003); and Timilsina, LeBlanc, and Walden (2005).Notes: CAPP = Canadian Association of Petroleum Producers; mb/d = millions barrels a day; bcf/d = billion cubic feet a day.

Export Prospects and Dutch Disease

©International Monetary Fund. Not for Redistribution

II ENERGY, THE EXCHANGE RATE, AND THE ECONOMY

18

Source: Organization for Economic Cooperation and Development (OECD).Notes: Shaded areas indicate periods of a rising share of energy exports in GDP. The charts depict the real effective exchange rate (REER, left scale)

for major commodity-exporting industrial countries, as well as the output and employment share of each country’s manufacturing sector (right scale).

20

40

60

80

100

120

140

160

10

12

14

16

18

20

22

24

1970 75 80 85 90 95 2000 04

Canada

Inde

x, 2

000

= 1

0C

In p

erce

nt80

90

100

110

120

130

10

15

20

25

30

35

1970 75 80 85 90 95 2000 04

Netherlands

Inde

x, 2

000

= 1

0C

In p

erce

nt

REER

REER (left scale)

REER

Output share(right scale)

Output share

Employment share(right scale)

Employment share

85

90

95

100

105

110

115

120

6

9

12

15

18

21

24

27

1970 75 80 85 90 95 2000 04

Norway

Inde

x, 2

000

= 1

0C

In p

erce

nt

70

80

90

100

110

120

5

10

15

20

25

30

35

1970 75 80 85 90 95 2000 04

United Kingdom

Inde

x, 2

000

= 1

0C

In p

erce

nt

REEROutput share

Output share

Employment share

Employment share

REER

80

100

120

140

160

180

6

11

16

21

26

31

1970 75 80 85 90 95 2000 04

Australia

Inde

x, 2

000

= 1

0C

In p

erce

nt

Output share

Employment share

Figure 2.4. Exchange Rate and Manufacturing Activity in Resource-Exporting Countries

©International Monetary Fund. Not for Redistribution

19

Exchange Rate Equation

For a more detailed analysis of the nexus between energy exports and the exchange rate, a new exchange rate equation is developed here. The model revisits earlier studies that identified nonenergy commod-ity trade as one of the key drivers of the Canadian exchange rate.10 Having found a specification that assesses the potential impact of higher energy exports on the exchange rate, the results of the equation are fed into Oxford Economic Forecasting’s global economic model for a simulation of the overall impact of higher oil production in Canada.

Theoretical Considerations

Commodity exports affect the exchange rate through changes both in their volume and in the terms of trade. Commodities are generally exchanged at a single world price, with quantities determined by production capac-ities. This implies that commodity trade (in foreign currency) is largely independent of exchange rate con-siderations. The Canadian trade balance in U.S. dollars can thus be written as:

TBUS$ = pcCOM + px(+)XNC(–) – pMNC(+), (1)

where TBUS$ is the trade balance; pc and COM are the (U.S. dollar) price and real net exports of com-modities, respectively; pX and XNC are the U.S. dollar price and volume of noncommodity exports; and pMand MNC are the U.S. dollar price and volume of non-commodity imports. Superscripts (+) and (−) indicate

10See, for example, the exchange rate equation reported in Amano and van Norden (1993). Similar approaches are discussed in Helli-well and others (2005); Issa, Lafrance, and Murray (2006); and Bailliu and King (2005).

whether a particular variable would have a positive or negative impact on the trade balance under a Canadian dollar appreciation. Because the Canadian commodity trade balance is positive, higher commodity prices (or volumes) improve the trade balance, allowing a larger deficit in the trade of other goods. This occurs through a real exchange rate appreciation, which boosts real imports and reduces real exports by raising the price of Canadian noncommodity goods.

The impact of a shock in net commodities trade on the exchange rate depends on the size of the net commodi-ties trade relative to noncommodity imports. Standard elasticities imply that a rise in U.S. dollar noncommod-ity export prices is approximately offset by the fall in export volumes. As a result, the exchange rate apprecia-tion primarily affects the U.S. dollar value of noncom-modity imports. Accordingly, in the model reported here, the ratio of net commodity exports divided by the size of noncommodity imports (the variable tbcom) is used as a measure of the relative importance of shocks in commodity trade for exchange rate adjustment.11

Empirically, it is likely that commodity export vol-umes affect exchange rates more over the longer term. In the short term, exchange rates primarily respond to unanticipated news, as implied by the “random walk” model. Over the same time scale, commodity prices are also close to random walks that incorporate informa-tion on underlying conditions—for example, news of bad weather will affect the spot price of a commodity well before it impacts physical exports.

Accordingly, the estimated exchange rate equation incorporates a short-term “dynamic” model that relates the change in the exchange rate to changes in commod-ity prices and a long-term “error-correction” mecha-nism that relates the exchange rate level to commodity prices and volume. More specifically:

Δet = φΔptctbcomt–1 + α'Xt

– λ(et–1 – θtbcomt–1 – β'Zt–1), (2)

where e is the logarithm of the real exchange rate, pct is

the logarithm of the price of commodities, and X and Zrepresent other short- and long-term explanatory vari-ables, respectively. The long-run model, involving the expression in parentheses, relates the logarithm of the exchange rate to tbcom and other long-term exchange rate determinants. By contrast, in earlier work at the Bank of Canada, the exchange rate was related to the logarithm of the price of commodities only through the long-run equation:12

Δet = α'Xt – λ(et–1 – θpct–1 – β'Zt–1). (3)

11See Bayoumi, Faruqee, and Lee (2005) for a formal model of this approach.

12See Amano and van Norden (1993) and Helliwell and others (2005) for estimated exchange rate equations, and Bailliu and King (2005) for a survey of this work.

Source: OECD.

Crude materials

Other

Energy

Manufacturing/machinery

–4–3–2–1012345

2000959085807570651960

Figure 2.5. Composition of Net Exports(Percent of GDP)

Exchange Rate Equation

©International Monetary Fund. Not for Redistribution

II ENERGY, THE EXCHANGE RATE, AND THE ECONOMY

20

Comparing equations (2) and (3), the new specification differs from the earlier one in two respects:• It relates the short-term rate of change in the exchange

rate to the contemporaneous rate of change of com-modity prices, rather than only through the error- correction mechanism.

• As discussed, the impact of changes in commod-ity markets in both the short and long term depend on the prevailing level of tbcom, that is, the ratio of the commodity trade balance to noncommodity imports.

Empirical Results

Differentiating between trade in energy and non-energy commodities, the estimated model specification becomes

Δet = φnecΔptnectbnect–1 + φencΔpt

enctbenct–1+ αiidifft–1 +αexchΔet–1– λ(et–1 – θnectbnect–1 – θenctbenct–1) +εt. (4)

The dependent variable is the CPI-based real exchange rate against the U.S. dollar.13 The variable pnec refers to the log of nonenergy commodity prices and tbnec to net exports as a ratio of noncommodity imports; penc and tbenc are energy prices and net exports, normalized in the same manner.14 The dynamic equation includes the short-term interest rate differential between the United

13This is the definition of the real exchange rate used in earlier work. The U.S. real exchange rate was preferred over the multilateral REER because, although both series move closely together, data for the U.S.-specific series can be more easily constructed back to the 1970s, allowing the analysis to encompass the oil shocks during that decade.

14In the case of nonenergy commodities, it proved impossible to match the data on the trade balance with the corresponding price series, as the implied negative correlation between prices and vol-umes was implausibly high. Based on long-term trends, the ratio of real net nonenergy commodity exports to real noncommodity imports was assumed constant at 22.5 percent.

States and Canada (idiff). As is standard in this work, this was lagged to avoid simultaneity bias. A lagged dependent variable was also added to account for the serial correlation created by time-averaging the real exchange rate.15 The equation was estimated using quarterly data from 1972 on, and the data construction is described in Box 2.1.

The results from the basic specification suggest that energy and nonenergy commodities both play a significant role in explaining exchange rate trends (Tables 2.2–2.4). In the dynamic equation, oil and

15Time-averaging series that are random walks induce a moving average coefficient in the error term of !/4.

The data were constructed as follows:• Canadian–U.S. dollar real exchange rate: quarterly

averages of the Can$/US$ exchange rate divided by the relative GDP deflators.

• Canada–United States interest rate differential: difference in interest rates on 90-day commercial paper.

• Canadian net energy (nonenergy) commodity exports as a ratio of noncommodity imports: dif-ference between exports and imports of energy products (of agricultural and fishing products and forestry products) divided by total goods imports less imports of agricultural and fishing products, energy products, and forestry products. (Source: Statistics Canada, Balance of Payments Statistics,various years.)

• Relative price of energy (nonenergy) commodi-ties: price of Canadian energy (nonenergy) com-modities in U.S. dollars divided by the U.S. GDP deflator. (Source for commodity price series: Bank of Canada.)

Box 2.1. Data Sources

Table 2.2. Exchange Rate and Manufacturing in Energy-Exporting Industrial Countries

Canada Netherlands Norway United Kingdom Australia

Period of export run-up 1980–present 1963–86 1971–2000 1974–83 1980–85

Change in net energy exports (percent of GDP) 2.9 4.8 24.6 6.6 2.8

Average annual change during export run-upReal effective exchange rate –0.1 1.2 0.0 1.4 –1.9Manufacturing output share 0.0 –0.3 –0.4 –0.9 –0.4Manufacturing employment share –0.2 –0.5 –0.3 –0.8 –0.7

Sources: Haver Analytics; and IMF staff calculations.

©International Monetary Fund. Not for Redistribution

21

Table 2.3. Real Exchange Rate Equations

General Preferred Specification Model

Dynamic modelChange in log of nonenergy commodity prices (φnec)1 0.20 (.11)+ }Change in log of energy prices (φenc)1 0.50 (.19)**

0.37 (.09)**

Short-term interest rate differentials (αi) 0.57 (.13)** 0.59 (.13)**Lagged dependent variable (αexch) 0.30 (.08)** 0.30 (.07)**

Error correction mechanismCoefficient on error correction mechanism (λ) –0.12 (.03)** –0.11 (.03)**Nonenergy commodity trade balance (θnec)2 1.07 (.21)** }Energy trade balance (θenc)2 0.91 (.55)+

1.10 (.21)**

R-squared 0.35 0.34Durbin-Watson statistic 2.04 2.04Schwartz criterion –4.97 –5.03

Notes: Constant terms not reported. Standard errors are provided in parentheses; +, *, and ** represent coefficients that are significant at the 10, 5, and 1 percent significance levels, respectively.

1Logarithm of change in prices multiplied by lagged trade balance as a ratio of noncommodity imports.2As a ratio of noncommodity imports. Real nonenergy commodity imports are assumed constant at

22.5 percent of real noncommodity imports.

Table 2.4. Conventional Real Exchange Rate Equations

Basic Specifications Split Energy Coefficient

Dynamic modelShort-term interest rate differential (αi) 0.52 (.13)** 0.57 (.14)**Lagged dependent variable (αexch) 0.41 (.08)** 0.30 (.08)**

Error correction mechanismCoefficient on error correction mechanism (λ) –0.13 (.03)** –0.11 (.04)**Log of nonenergy commodity prices (τnec) 0.31 (.06)** 0.31 (.07)**Log of energy commodity prices (τenc) –0.05 (.03) –0.11 (.04)**Log of energy commodity prices after 1990 (τenc,90) — 0.26 (.08)**R-squared 0.26 0.32Durbin-Watson statistic 2.02 1.99Schwartz criterion –4.91 –4.92

Notes: Constant terms are not reported. Standard errors are provided in parentheses; +, *, and ** represent coefficients that are significant at the 10, 5, and 1 percent significance levels, respectively.

non-oil commodity prices as well as interest rate dif-ferentials are correctly signed and significant at the 10 percent level or better, and the lagged dependent variable is around its expected value of 0.25. In the error-correction mechanism, the coefficients for the energy and nonenergy commodity trade balance were also significant.

As expected, the long-term coefficients on the trade balances are similar and also several times larger than their short-term equivalents. A plausible explanation for this difference is that supply responses often reverse

movements in real prices over time, and therefore short-term price movements are discounted in markets. At −0.12, the coefficient on the error-correction mecha-nism implies that it takes about two years to reduce a deviation from the long-term trend by 50 percent. This half-life estimate is similar to other Canadian exchange rate equations but larger than estimated in many other models (Bayoumi, Faruqee, and Lee, 2005). This is encouraging, as the slow return of real exchange rates to equilibrium (generally defined as a fixed purchas-ing power parity value) has been an important focus of

Exchange Rate Equation

©International Monetary Fund. Not for Redistribution

II ENERGY, THE EXCHANGE RATE, AND THE ECONOMY

22

analysis for exchange rate models after being identified as one of the six puzzles of international economics (Obstfeld and Rogoff, 2000).

In the preferred specification, the model is made more parsimonious by imposing the same coeffi-cients on shocks to both energy and nonenergy com-modities. Such a restriction, which is easily accepted by the data, is appropriate because both coefficients are already scaled by their respective trade bal-ances (tbnec and tbenc) and hence are in the same “units.” In the dynamic model, the resulting coeffi-cient of 0.37 on energy and nonenergy prices implies a slightly less than ½ percent increase in the real exchange rate for every 1 percentage point increase in either price. At 1.1, the error-correction coefficient for the commodity trade balance implies that the long-run impact of volume changes is almost three times the short-term effect of price changes. All of the coefficients in this preferred model are signifi-cant at the 1 percent level.

Interpretation

The results of the exchange rate equation suggest that most of the recent appreciation of the Canadian dollar against its U.S. counterpart reflects underlying factors. A dynamic projection explains over three-quarters of the real appreciation of more than 30 percentage points between 2002:Q1 and 2005:Q2, leaving an unexplained gap of about 5 percentage points that could well reflect generalized U.S. dollar weakness over this period or the anticipation of higher production from oil sands (Figure 2.6). The estimate of the gap’s size is relatively insensitive to variations of one to two years in the start-ing date of the projection.

The model implies a limited exchange rate appre-ciation due to expected increases in the volume of oil exports from Alberta’s oil sands. As discussed above, such production is expected to add some 1 percent of GDP to net oil exports over 15 years, increasing net oil exports as a ratio of noncommodity imports by some 4 percentage points. Given the long-run semi-elasticity on this ratio of around unity, this implies a real appre-ciation of only around 4 percent—much less than the implied appreciation since early 2002, although similar to the unexplained gap over that period.

The new specification compares relatively favorably to a conventional one in which commodity prices are included in the long-run model with no adjustment for the commodity trade balance. Table 2.4 reports results from a conventional equation that drops the contem-poraneous changes in commodity prices and includes only the logarithms of energy and nonenergy prices in the error-correction mechanism:

Δet = αiidifft–1 + αexchΔet–1–λ(et–1 – τnecPt–1

nec – τencPt–1enc) + εt. (5)

In one regression, all of the coefficients are fixed over time, while in the other the coefficient on energy prices is allowed to change in 1990:Q1, to proxy the growing importance of energy exports in Canadian trade.16 The results confirm that the equation that allows the coefficient on energy prices to change in 1990:Q1 fits the data considerably better—all of the coefficients are significant at the 1 percent level—although the R2 remains below the preferred specification in Table 2.3.17

As can be seen in Figure 2.6, the split-coefficient equation implies an appreciation since early 2002 that is almost identical to the specification discussed above. This largely reflects the similarity of the long-run coef-ficient on energy prices in this specification and in the preferred specification reported in Table 2.3. This like-ness in coefficients is not, however, typical. This can be seen in Figure 2.7, which graphs the long-term coef-ficients on energy and nonenergy commodity prices in the preferred specification (which vary with the value of the commodity trade balance) and the split- coefficient specification (where the coefficient on energy prices jumps in 1990:Q1).

Putting It All Together

In addition to any balance of payments benefits, the oil sands impact will be amplified by the large amount of factor inputs needed to support rising pro-duction levels. Significant capital equipment as well

16Bailliu and King (2005) suggest that this modification produces considerably better empirical results.

17Root mean-squared errors on dynamic projections since 1990 provide a similar ranking of the three models.

Source: IMF staff estimates.

70

80

90

100

110

050403022001

Actual

Preferred model

Split-coefficient model

Figure 2.6. Dynamic Forecasts of Canadian Real Exchange Rate(Index, 2001:Q4 = 100)

©International Monetary Fund. Not for Redistribution

23

as energy, transport, and urban infrastructure must be in place before oil can be produced. Unlike in countries with more conventional energy reserves, the production process itself is also very resource-intensive, reflecting both energy and labor inputs and the need for large replacement investment. Timilsina and others (2005) use input-output tables to esti-mate that each dollar spent on oil sands investment or production yields Can$3.35 or Can$1.74 of GDP, respectively.18

Oxford Economic Forecasting’s global macro model provides a tool for assessing the overall consequences of rising Canadian oil production. The model uses a

18Assuming that annual expenditure for investment and pro-duction rises to around Can$50 billion per year by 2015 (in 2004 prices), their calculation suggests that the level of real GDP would be shifted upward by 7¼ percent.

traditional adaptive-expectations approach with a Cobb-Douglas production function and imperfectly competitive labor markets. Long-term growth is driven by demographics and productivity growth. The model also includes a block of equations for energy produc-tion and exports in Canada, which makes modeling the stepped-up production activity relatively straightfor-ward. Two scenarios were studied:• Higher oil sands output: In addition to a stronger oil

production forecast, investment and factor input data from Timilsina, LeBlanc, and Walden (2005) were added to the forecast for investment and other GDP. In addition, an exogenous path was provided for global oil prices (from the IMF’s World Economic Outlook). The model was calibrated to ensure that the Oxford model’s projection of the current balance and exchange rate was consistent with the results of the previous section. The simulations confirm a positive but relatively small impact of oil sands production on Canadian GDP, mainly resulting from the effect of higher export incomes on domestic consumption (Table 2.5).

• Higher output and energy prices: In addition to higher oil sands output, the second scenario assumes an increase in the world oil price to US$110 per bar-rel by 2020, with a concomitant increase in gas and other energy prices. Although this would yield an additional increase of 1 percent of GDP in the current account balance, the change in both the exchange rate and nonenergy exports is more than twice as strong compared to baseline as in the first scenario—in part because of falling demand abroad. With consump-tion also growing less strongly, the overall impact of higher oil prices on GDP is slightly negative in this simulation.Both scenarios may be optimistic in that they assume

only a gradual buildup of real wages in the face of tight labor markets. The economy’s productive capacity ini-tially increases (over baseline) as a result of the shift toward oil sands production, implicitly assuming that sufficient resources can be moved to high- productivity energy activities. Although the effect diminishes after oil sands investment is assumed to plateau on a higher level, potential output remains above baseline in both scenarios. Labor pressures are more intense in the first scenario, resulting in a real wage increase of around 1 percent over 15 years. Real wage increases are smaller in the second scenario, as higher energy prices impact negatively on demand.

To test for consistency, the IMF’s Global Economic Model (GEM) was used to run a similar exercise.19

Using the same production figures, GEM projects that rising oil sands exports could lift GDP by about 4 per-cent by 2020—a marginally higher but still moderate

19See Laxton and Pesenti (2003) and Bayoumi and others (2004) for a description of GEM.

Source: IMF staff estimates. Note: The charts compare θ multiplied by the ratio of net energy

and nonenergy commodity exports to noncommodity imports, respectively, from equation (4), with tenc from equation (5).

Energy Prices

Nonenergy Commodity Prices

–0.20

–0.15

–0.10

–0.05

0

0.05

0.10

0.15

0.20

0.1

0.2

0.3

0.4

0.5

0.6

1973 77 81 85 89 93 97 2001 05

1973 77 81 85 89 93 97 2001 05

Preferred model

Split-coefficient model

Preferred model

Split-coefficient model

Figure 2.7. Long-Term Impact of Commodity Prices on the Exchange Rate

Putting It All Together

©International Monetary Fund. Not for Redistribution

II ENERGY, THE EXCHANGE RATE, AND THE ECONOMY

24

Table 2.5. Impact of Oil Production and Price Increases(Difference to baseline)

Higher Oil Sands Output Higher Energy Prices ______________________________ ______________________________ 2010 2015 2020 2010 2015 2020

GDP (percent) 0.45 0.53 1.07 –0.04 –0.39 –0.53GDP growth (percentage points) 0.11 0.01 0.08 0.03 –0.08 –0.08Consumption (percent) 0.27 0.75 1.26 –0.04 0.25 0.70Real wages (percent) 0.34 0.65 1.01 0.04 0.15 0.21Real effective exchange rate (percent) 1.32 2.64 3.97 3.39 6.85 10.40Export of nonenergy goods (percent) –0.79 –1.74 –1.97 –2.07 –4.12 –6.32Export of services (percent) –0.65 –1.21 –1.34 –1.71 –3.34 –5.86Current account balance (percent of GDP) 0.48 0.38 1.08 1.00 1.17 1.90

Sources: Oxford Economic Analysis; and IMF staff calculations.

beneficial effect. Output of the non-oil tradable sector shrinks as expected (by 5 percent relative to baseline) as the REER appreciates by about 3 percent. The dif-ference may in part reflect the absence of a specific energy sector in GEM—the increase in oil sands output was instead modeled by raising the available supply of land used as a production factor.20

20A multilateral exchange rate model applied to Canada by Lee and Mühleisen (2004) also finds that the stronger path for oil exports could increase the exchange rate’s sensitivity to energy prices: a 10 percent oil price increase would push up the equilibrium exchange rate by 1.20 percent instead of 1.07 percent.

Conclusions

The expansion of oil sands production is likely to have a small beneficial impact on the Canadian economy. The potential gains in export revenues could result in some upward pressure on the exchange rate, but this is not expected to be large enough to significantly affect nonen-ergy exports. Rising factor demand from oil sands explo-ration and production would likely also boost domestic output. However, the anticipated production increases are relatively small on a global scale, and even a growing energy sector will remain only one of several pillars in a well-diversified Canadian economy.

©International Monetary Fund. Not for Redistribution

25

III Canada and the United States

Canada is an open economy whose trade is domi-nated by the United States. Indeed, the signing of

the Canada-U.S. Free Trade Agreement in 1988, and its successor, the North American Free Trade Agreement in 1993 led to substantially increased trade, financial flows, and increasing business-cycle synchronicity. This section examines three important characteristics of the Canadian economy that significantly affect its relationship with the United States.

Canada has a regionally specialized economy, with manufacturing concentrated in the central provinces and raw materials production elsewhere. The first seg-ment of this section examines how Canada’s regional economies have evolved over time and how they have responded to macroeconomic shocks.

Canada continues to have a large labor productiv-ity gap with the United States. The second segment of this section examines the reasons for this continuing difference.

The third segment uses a small two-country mon-etary model to examine spillovers from U.S. activity and to assess how they can be mitigated by monetary responses in both the United States and Canada.

A. Regional Dimensions of the Canadian EconomyVladimir Klyuev and Rodolfo Luzio

The Canadian economy is highly diverse across regions. This analysis focuses on two aspects of this diversity—differences in the industrial structure and differences in responses to common shocks—and com-pares these with differences across U.S. regions.

Industrial Structure

We first examine differences in industrial struc-ture across regions. Figure 3.1 reports the weights of five industries—agriculture, construction and utilities, manufacturing, mining, and services—across Cana-dian regions as deviations from the average for Canada as a whole. The regions are British Columbia, Prai-ries, Ontario, Quebec, and Atlantic provinces. (Box 3.1 lists the Canadian provinces and U.S. states included in the regional classifications and also outlines the data sources used here.)

Differences in industrial structure are large and persistent. Compared with the rest of Canada, the provinces of Ontario and Quebec are more heav-ily based on manufacturing, while mining (which includes the energy sector) plays a heavier role in the Prairies. The Atlantic provinces and British Colum-bia are particularly strong in services, except that offshore gas exploration has recently strengthened the role of mining in the east.

The degree of regional economic diversity appears larger in Canada than in the United States. The Prairies’ specialization in mining is larger than in the U.S. South-west, and British Columbia is more focused on services relative to the rest of Canada than New England and the Mideast in the United States (Figure 3.2). On the other hand, the U.S. Great Lakes region seems more concen-trated on manufacturing than either Ontario or Quebec.

To quantify the degree of dispersion, we calculate the average of absolute deviations between regional and national shares in the five industries.1 This measure is plotted for each region in Canada and the United States (Figure 3.3). In addition, Figure 3.4 shows unweighted and weighted (by regional GDP) averages of regional dispersions for both countries.

The charts confirm that the Canadian economy is regionally more diversified. In addition, we observe the following:• The evolution over time of the national dispersion

measures in the two countries is quite similar. A flat trend in the 1980s gave way to a short period of diver-gence in the early 1990s, followed by fairly rapid convergence coinciding with the Internet and com-munications revolution in the middle to late 1990s.2

• Unlike the United States, Canada has one main out-lier in economic structure: the resource-rich Prairies. The U.S. Southwest was almost as much an outlier in the United States in the 1970s, but its industrial structure has since shifted to be much more similar to that of the rest of the country.

1We also calculate the square root of the average squared devia-tions—a measure that gives a bigger weight to outliers than the aver-age absolute deviation. The two measures tell essentially the same story, and so we report only the latter.

2The United States also experienced convergence in the late 1970s—a period for which Canadian data are not available. This may contribute to the impression (conveyed by Figures 3.1 and 3.2) that some U.S. regions seem to have experienced faster convergence than their Canadian counterparts.

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

26

Sources: Statistics Canada; and IMF staff calculations.Note: Atlantic provinces include New Brunswick, Newfoundland and Labrador, Nova Scotia, and Prince Edward Island. The Prairies comprise Alberta,

Manitoba, and Saskatchewan.

Atlantic Provinces

–10–8–6–4–202468

101214

ServicesMiningManufacturingConstruction and utilitiesAgriculture

022000989694929088861984

Quebec

British Columbia

–10–8–6–4–202468

101214

022000989694929088861984

–10–8–6–4–202468

101214

022000989694929088861984

Ontario

–10–8–6–4–202468

101214

022000989694929088861984

Prairies

–10–8–6–4–202468

101214

022000989694929088861984

Figure 3.1. Canada: Difference between Regional and National Industry Shares(In percent)

©International Monetary Fund. Not for Redistribution

27

Sources: Bureau of Economic Analysis; and IMF staff calculations.

New England

Southeast

Great Lakes

Rocky Mountains

Southwest

Far West

Plains

Mideast

ServicesMiningManufacturingConstruction and utilitiesAgriculture

–10

–5

0

5

10

15

2001989592898683801977–10

–5

0

5

10

15

2001989592898683801977

–10

–5

0

5

10

15

2001989592898683801977–10

–5

0

5

10

15

2001989592898683801977

–10

–5

0

5

10

15

2001989592898683801977-10

-5

0

5

10

15

2001989592898683801977

–10

–5

0

5

10

15

2001989592898683801977-10

-5

0

5

10

15

2001989592898683801977

Figure 3.2. United States: Difference between Regional and National Industry Shares(In percent)

A. Regional Dimensions of the Canadian Economy

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

28

• Even excluding the Prairies, regional dispersion in Canada is higher than in the United States. The aver-age absolute deviation for Canada excluding the Prai-ries region is 2.1, which exceeds the value of 1.5 measured for the United States.3

Response to Shocks

We next examine how these differences in struc-ture affect regional economies’ responses to aggre-gate shocks. For all of Canada’s regions, we study the

3The weighted average, however, is 1.5 for both countries. This is not surprising, given the economic size of Ontario and Quebec.

response of real GDP growth, real private consumption, and real investment (annual data) to a set of explana-tory variables that capture both domestic and external factors (Tables 3.1–3.3). These include changes in real GDP in both the United States and the rest of Canada, the real effective exchange rate, the real price of oil, and the real short-term interest rate.

The regression results demonstrate that Canadian regions are influenced heavily by both domestic and external factors, albeit in very divergent ways:• The links to the United States are particularly power-

ful in Ontario and the Prairies. In these regions, U.S. growth is more important than growth in the rest of Canada, likely reflecting their concentration on

Canada, 1984–2003

0

1

2

3

4

5

6

7British Columbia QuebecOntarioPrairies Atlantic Provinces

032001999795939189871984

United States, 1977–2003

Sources: Statistics Canada; Bureau of Economic Analysis; and IMF staff calculations.

0

1

2

3

4

5

6

7Far WestRocky MountainsSouthwestSoutheast

PlainsGreat LakesMideastNew England

03200199979593918987858381791977

Figure 3.3. Regional GDP by Sector: Divergence from Country Average(Average absolute difference, in percent, across sectors)

©International Monetary Fund. Not for Redistribution

29

exports to the United States of either manufactures or raw materials.

• Links with the rest of Canada are strongest in Que-bec, followed by Ontario and British Columbia. The weakest links are in the resource-rich Prairies and Atlantic provinces.

• An increase in the price of oil is beneficial for the Prai-ries, but has a negative impact on all the other regions and no major impact on the country as a whole.

• Real exchange rate appreciation and real inter-est rate increases are particularly negative for the eastern provinces, for the manufacturing-based central provinces, and for the country as a whole. The impact of exchange rate appreciation and the real interest rate is limited in British Columbia and the Prairies, but this may reflect a link between commodity prices, the exchange rate, and inflation that could not be resolved in the context of this exercise.Regressions using real private consumption and

investment show a similar pattern. Consumption and investment in all regions are negatively affected by a contemporaneous increase in the real interest rate. Higher oil prices boost consumption and investment in the Prairies but slow them in other regions. Exchange rate appreciation increases investment in the Atlantic provinces and consumption in British Columbia, but does not appear to affect private demand in the other provinces or across the country as a whole.

Similar regressions indicate that the United States is more domestically integrated and less susceptible to external shocks. In particular, real GDP growth in each region is positively and significantly correlated with growth elsewhere in the United States (Table 3.4). Changes in the real effective exchange rate do not have a significant impact in any region. Higher oil prices depress output in a number of regions, and the positive impact on the Southwest is not stastically significant.

Regional diversity underlines the importance for Canada of maintaining flexibility in responding to shocks, particularly for Ontario, British Columbia, and the Prairies. External shocks have their biggest impact

Sources: Bureau of Economic Analysis; Statistics Canada; and IMF staff calculations.

1.4

1.8

2.2

2.6

3.0U.S. weightedU.S. unweightedCanada weightedCanada unweighted

2001989592898683801977

Figure 3.4. Average Regional Divergence from National Industrial Structure(In percent)

A. Regional Dimensions of the Canadian Economy

Box 3.1. Data Sources and Definitions for Canadian and U.S. Regional Variations

Real interest rateAverage overnight rate adjusted for average CPI

inflationReal oil price (U.S.)West Texas Intermediate (WTI) price adjusted for U.S.

GDP deflatorReal oil price (Canada)World Economic Outlook (WEO) oil price series

adjusted for the US$/Can$ exchange rate and the Canadian GDP deflator

Real effective exchange rate (REER)Source: JP MorganTime period1983–2004 (dictated by availability of Canadian GDP

data by region)Canadian regionsAtlantic provinces: New Brunswick, Newfoundland and

Labrador, Nova Scotia, and Prince Edward IslandBritish ColumbiaOntarioPrairies: Alberta, Manitoba, and SaskatchewanQuebec

U.S. regions The eight U.S. regions referred to here are those

defined by the Bureau of Economic Analysis of the U.S. Department of Commerce.

Far West: Alaska, California, Hawaii, Nevada, Oregon, Washington

Great Lakes: Illinois, Indiana, Michigan, Ohio, Wisconsin

Mideast: Delaware, District of Columbia, Maryland, New Jersey, New York, Pennsylvania

New England: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont

Plains: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, South Dakota

Rocky Mountains: Colorado, Idaho, Montana, Utah, Wyoming

Southeast: Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Virginia, West Virginia

Southwest: Arizona, New Mexico, Oklahoma, Texas

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

30

on these provinces, reflecting their external orienta-tion, although the adjustment needs are often diametri-cally opposed, given their highly different economic structures. The recent rise in oil and other commodity

prices demonstrates this divergence, boosting activity in Alberta and other western provinces while depressing conditions—through exchange rate appreciation—in the manufacturing heartland of Ontario.

Table 3.1. Canada: Growth Regressions

Dependent variable: British AtlanticReal GDP growth in the region Columbia Prairies Ontario Quebec Provinces Canada

Lagged dependent variable 0.074 0.114 0.218 0.035 0.145 0.336** (0.164) (0.215) (0.111) (0.143) (0.172) (0.069)Growth in Canada outside region 0.435** 0.257 0.429 0.708* 0.274* — (0.188) (0.319) (0.294) (0.344) (0.145) Growth in the United States — 0.651* 1.160** 0.095 — 0.945** (0.353) (0.244) (0.367) (0.110)Real interest rate (lagged) 0.154 0.084 −0.273 −0.170 −0.309* −0.235** (0.219) (0.222) (0.195) (0.121) (0.148) (0.094)Change in REER (lagged) 0.017 0.018 −0.131** −0.123* −0.101 −0.126** (0.069) (0.110) (0.063) (0.059) (0.062) (0.029)Change in oil price (lagged) −0.005 0.042** −0.009 −0.011 −0.032** −0.003 (0.013) (0.016) (0.009) (0.007) (0.010) (0.005)R-squared 0.23 0.37 0.84 0.83 0.51 0.78Durbin-Watson statistic 2.38 2.20 1.86 2.52 1.48 2.70

Source: IMF staff calculations.Notes: Newey-West heteroscedasticity-consistent standard errors in parentheses; REER = real effective exchange rate.* = significant at 10 percent.** = significant at 5 percent.

Table 3.2. Canada: Real Private Consumption Regressions

Dependent variable: Real private consumption British Atlanticgrowth in the region Columbia Prairies Ontario Quebec Provinces Canada

Lagged dependent variable 0.345** 0.084 0.328* 0.161 −0.013 0.370** (0.081) (0.186) (0.181) (0.117) (0.208) (0.091)Growth in Canada outside region 0.152 0.533** 0.345* 0.260* 0.534** — (0.097) (0.171) (0.188) (0.127) (0.167)Growth in the United States — −0.035 0.627** 0.525** — 0.704** (0.194) (0.201) (0.156) (0.139)Real interest rate −0.029 −0.238 −0.278** −0.255** −0.082 −0.283** (0.070) (0.169) (0.103) (0.094) (0.128) (0.069)Change in REER (lagged) 0.121 0.135 0.027 0.020 0.065 0.034** (0.051)** (0.077) (0.092) (0.058) (0.087) (0.037)Change in oil price (lagged) −0.025 0.017* −0.011* −0.004 −0.014 −0.007 (0.016) (0.009) (0.006) (0.004) (0.008) (0.006)

R-squared 0.54 0.64 0.81 0.82 0.69 0.76 Durbin-Watson statistic 2.06 2.17 1.85 1.50 1.47 2.26

Source: IMF staff calculations.Notes: Newey-West heteroscedasticity-consistent standard errors in parentheses; REER = real effective exchange rate.* = significant at 10 percent.** = significant at 5 percent.

©International Monetary Fund. Not for Redistribution

31

Conclusions

Although they are gradually converging, Canadian provinces still exhibit considerably diverse economic structures. This diversity contributes to differential

responses to domestic and external shocks. In particu-lar, growth in Ontario—which comprises over one-third of the Canadian economy—is closely linked to U.S. growth, and is also negatively affected by interest

A. Regional Dimensions of the Canadian Economy

Table 3.3. Canada: Real Private Investment Regressions

Dependent variable:Real private investment British Atlanticgrowth in the region Columbia Prairies Ontario Quebec Provinces Canada

Lagged dependent variable 0.113 0.125 0.340 0.445* −0.173 0.331 (0.293) (0.223) (0.207) (0.218) (0.144) (0.196)Growth in Canada outside region 0.401 1.393 −0.555 −0.886* 1.782* — (0.606) (1.548) (1.255) (1.084) (0.877)Growth in the United States — 0.711 3.615** 3.028** — 2.309** (3.310) (1.268) (1.315) (0.753)Real interest rate −1.035 −0.680 −1.452 −1.521** −1.553* −1.325** (0.802) (1.549) (0.863) (0.565) (0.841) (0.596)Change in REER (lagged) 0.805 0.011 −0.075 −0.035 0.748* 0.061 (0.438) (0.728) (0.563) (0.383) (0.403) (0.306)Change in oil price (lagged) −0.033 0.195** −0.030 −0.100** −0.058 0.002 (0.096) (0.088) (0.045) (0.042) (0.122) (0.053)R-squared 0.15 0.32 0.58 0.65 0.31 0.52Durbin-Watson statistic 1.96 1.83 2.30 1.98 2.28 2.37

Source: IMF staff calculations.Notes: Newey-West heteroscedasticity-consistent standard errors in parentheses, REER = real effective exchange rate.* = significant at 10 percent.** = significant at 5 percent.

Table 3.4. United States: Growth Regressions

Dependent variable: New Great Real GDP growth in the region England Mideast Lakes Plains Southeast Southwest Rockies Far West

Lagged dependent variable 0.225 −0.001 −0.073 −0.123 −0.053 0.247** 0.497** 0.528** (0.145) (0.134) (0.070) (0.121) (0.072) (0.106) (0.139) (0.163)

Growth in U.S. outside region 1.265** 0.826** 0.999** 0.987** 0.766** 1.058** 0.661** 0.88** (0.254) (0.196) (0.152) (0.231) (0.106) (0.290) (0.255) (0.114)

Real interest rate (lagged) 0.214 0.282 −0.159 −0.251* −0.018 −0.536** −0.380 0.009 (0.275) (0.180) (0.204) (0.122) (0.117) (0.234) (0.224) (0.224)

Change in REER (lagged) −0.023 −0.060 0.044 −0.026 0.007 0.054 0.029 0.022 (0.060) (0.044) (0.053) (0.047) (0.026) (0.034) (0.036) (0.038)

Change in oil price (lagged) −0.006 0.009 −0.016 0.006 −0.016** 0.028 −0.006 −0.036** (0.020) (0.010) (0.012) (0.009) (0.006) (0.018) (0.011) (0.014)

R-squared 0.77 0.73 0.73 0.68 0.82 0.62 0.65 0.76Durbin-Watson statistic 1.43 1.06 1.07 1.99 1.52 1.49 1.26 1.83

Source: IMF staff calculations.Notes: Newey-West heteroscedasticity-consistent standard errors in parentheses; REER = real effective exchange rate.* = significant at 10 percent.** = significant at 5 percent.

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

32

rate increases, real currency appreciation, and higher oil prices. On the other hand, growth in the western provinces is boosted by increased oil prices and appears to be less sensitive to changes in the exchange rate or the interest rate.

B. The Canada–United States Productivity Gap

Roberto Cardarelli

Despite the close integration between the Canadian and U.S. economies, the labor productivity gap between the two countries has widened over the last two decades (Figure 3.5). While a greater utilization of labor resources has allowed Canada to narrow the gap with the United States in terms of per capita income since the mid-1990s, convergence has been held back by the more modest pace of Canadian labor productivity growth.

This analysis explores the factors that have led to the Canada–United States productivity gap using a sectoral growth-accounting approach. It builds on the approach of Faruqui and others (2003) to construct a sectoral database with comparable data on value added, labor, and capital inputs for 23 industries during 1981–2000, in order to assess the extent to which this gap reflects differences in the industrial structure of the two countries.4

The chapter’s main results are these: • The labor productivity growth gap during 1995–2000

largely reflects the performance of two key service sectors, trade and “finance, insurance, and real estate” (FIRE). The manufacturing sector, and in particular industries related to information and communication technology (ICT), also continued to contribute to the gap, but no more than in the previous period.

• Differences in industrial structure explain the major-ity of the productivity growth gap over the second half of the 1990s. The United States appears to have been more successful than Canada in shifting resources toward high-productivity sectors.

• The lower contribution from ICT capital accumu-lation to productivity growth in Canada may also reflect differences in realizing the productivity ben-efits of ICT investments. In particular, Canadian pro-ductivity growth may have been held up by the delays in introducing organizational changes necessary to complement ICT capital.

• The increased economic integration with the United States has allowed Canadian firms to benefit from

4Data for Canada are from Statistics Canada, while data for United States are from several industry data sources, including the database used by Jorgenson, Ho, and Stiroh (2002) in their latest study on the U.S. productivity performance. See Box 3.2 for a description of the database.

economies of scale and technology transfers, some-thing that appears to have positively contributed to their productivity performance over the last two decades.

Review of the LiteratureA copious number of studies have sought to analyze

the factors behind the productivity gap between Canada and the United States (for a survey, see Crawford, 2002; and Macklem, 2003). Among the explanations offered are the following:• Different-sized ICT-producing sectors: Some stud-

ies attribute most of the post-1995 acceleration of labor productivity in the United States to the excep-tional total factor productivity (TFP) performance of the ICT-producing sector (for example, Gordon, 2003; and Harchaoui and Tarkhani, 2002). Given its smaller ICT-producing sector, these studies suggest that Canada is at a relative disadvantage in reaping the benefits of the ICT productivity wave.5

• Different contributions from ICT capital accu-mulation: The widespread adoption of ICT capi-tal assets has been regarded as a key factor behind the strong labor productivity growth in the United States.6 Harchaoui and Tarkhani (2002) show that Canada’s business sector also experienced solid

5Harchaoui and Tarkhani (2002) show that the size of Canada’s ICT-producing sector increased from around 2½ percent of GDP in 1981 to around 4 percent of GDP on average over the second half of the 1990s, but remained below the U.S. share, which was around 6 percent of GDP over this period.

6See, for example, Oliner and Sichel (2002) and Jorgenson, Ho, and Stiroh (2002).

Source: OECD. Notes: Labor productivity: GDP in millions of 1999 US$

(converted at purchasing power parity) per hour worked. Labor utilization: hours worked per person.

75

80

85

90

95

100

105

110Labor productivityLabor utilization Per capita income

2000979491888582791976

Figure 3.5. United States and Canada: Income and Productivity Indicators(United States = 100)

©International Monetary Fund. Not for Redistribution

33

growth in ICT capital services during 1981–2000, at levels comparable to if not higher than the United States (Table 3.5).7 Nonetheless, the contribu-tion from ICT capital deepening to labor produc-tivity growth is generally estimated to be lower in Canada than in the United States, mainly reflecting the lower estimated marginal productivity of ICT capital and the lower ICT capital intensity in Canada.8

7The faster growth in ICT capital services in Canada might be partly explained by differences in the capital asset depreciation rates used by Statistics Canada and the U.S. Bureau of Labor Statistics. In particular, Statistics Canada uses higher depreciation rates for ICT assets, something that might lead to a faster growth of their capital services (see Ho, Rao, and Tang, 2003).

8See Khan and Santos (2002); Harchaoui and Tarkhani (2002); Armstrong and others (2002); and Ho, Rao, and Tang (2003). Both Armstrong and others and Harchaoui and Tarkhani find that ICT capital deepening has contributed around ¼ percent to the average annual labor productivity growth in Canada during 1995–2000, up

• Differences in the share and productivity performance of small and medium-size enterprises (SMEs): In the Canadian manufacturing sector, SMEs (firms with fewer than 500 employees) accounted for 75 percent of total manufacturing employment, compared to around 60 percent in the United States in 1997. Not only has the relative role of SMEs in the Canadian economy increased over the last two decades, but some studies have found these firms to be less pro-ductive relative to their U.S. counterparts.9

only slightly compared to 1981–2000. The equivalent figure for the United States is estimated between ½ percent (Oliner and Sichel, 2002) and �/3 percent (Jorgenson, Ho, and Stiroh, 2002).

9Baldwin and Tang (2003) show that around ¼ percentage point of the labor productivity gap in the manufacturing sector in 1997 was due to the larger share of SMEs in Canada compared to the United States, and ½ percentage point was due to lower productivity of SMEs in Canada.

B. The Canada–United States Productivity Gap

Data for Canada

Real (chain Fisher weighted) value added, hours worked, labor input, and capital services data were obtained from Statistics Canada and were based on a Standard Indus-trial Classification (SIC) 80 industry classification. For the manufacturing sectors, however, the data ended in 1997. They were extrapolated to 2000 using the growth rates from the KLEMS input and output database, which follows a North American Industry Classification System (NAICS) classification (starting from 1997).

Comparing the 1997 industries’ value added based on the two industry classifications shows that the difference is generally around 15 percent, except for “other manu-facturing sector” and “furniture and fixture,” for which the difference is around 30 percent. The results for these sectors should then be interpreted with greater caution than others.

Data for the United States

Industry data for the United States follow the U.S. SIC 87 industry classification.

Real (chain Fisher weighted) value added industry data for the United States were obtained from the Bureau of Economic Analysis’s “gross product originating” by industry (GPO). Because these figures are on a market-price basis, value-added data at basic prices were obtained by subtracting the indirect business tax and nontax liabil-ity from GPO.

GDP by industry is obtained from industry components of domestic income, which, as is well known, tend to fall short of GDP measured on an expenditure basis. The dif-ference is named “statistical discrepancy” and is attrib-uted to the industries based on their share of total GDP.

Hours worked, labor input, and capital services were obtained from Jorgenson, Ho, and Stiroh (2002) and are

based on methodologies that are largely comparable with those adopted by Statistics Canada.

Industry Classification

To obtain the same level of industry classification for the two countries, a number of subsectors were aggregated into larger sectors. As an example, a “mining” sector was obtained for the United States by aggregating four subsec-tors (namely, metal mining, coal mining, petroleum and gas, and nonmetallic mining).

The aggregation was needed also to obtain comparable sectors for the two countries. For example, U.S. SIC 87 classification places computers and office equipment in industrial machinery, while Canada SIC 80 classifica-tion places it in electrical and electronic equipment. For the sake of comparison, the two sectors were aggregated into one large sector, which is taken as a proxy for the information communications and technology sector in this analysis.

The following aggregation criteria were utilized: sub-industries’ value added were aggregated using value-added shares as weights; labor and capital inputs were aggregated using relative shares in aggregate labor com-pensation and capital income, respectively, as weights; and hours worked were aggregated through the unweighted sum.

Even though these aggregation procedures produce rea-sonably comparable sectors for the two countries, some minor differences still persist, particularly in the service sectors. In addition to the different treatment of eating and drinking places (as noted in footnote 16), another dif-ference worth mentioning regards postal services, which are placed in the communication sector for Canada but in the transportation sector for the United States.

Box 3.2. Data Sources on Canada–United States Productivity Gap

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

34

• Differences in the share and income of self-employed: The difference in labor productivity growth between Canada and the United States in the 1990s has also been attributed to faster growth in self-employment in Canada and the poorer income performance of this group compared with the self-employed in the United States (Baldwin and Chowhan, 2003).Less relevant factors include:

• Differences in national accounts statistics: While differences still remain, the methodology used by national statistical agencies to measure labor and TFP has been converging. In particular, both the U.S. Bureau of Labor Statistics and Statistics Canada now use hedonic prices and include purchases of com-puter software in the national account measures of investment.

• Differences in the regulatory burden in labor and product markets: Gust and Marquez (2004) find that countries with a more burdensome regulatory frame-work tend to have lower TFP growth. However, not-withstanding the difficulties in building comparable indexes of regulatory burden across countries, empiri-cal evidence does not reveal a large difference between Canada and the United States in terms of labor and product market legislation and institutions.10

10Based on the regulatory variables they use, Canada is lagging the United States according to the OECD’s employment protection legislation index but is leading the United States according to the World Economic Forum’s regulatory burden index.

Few studies have examined the contribution of different industries to the business sector labor productivity gap between Canada and the United States. The majority of the literature has either focused on the labor productivity gap in the manufacturing sector or used the growth accounting framework at an aggregate level. This analysis probes the extent to which pro-ductivity differences between the two countries reflect differences in their industrial structure and the perfor-mance of specific industries.

Results from Sectoral Growth Accounting

The analysis uses a traditional growth accounting framework and attributes labor productivity growth (value added per hours worked, yt) to the contribution of three factors: the improvement in labor quality (Ht),weighted by the labor income share of value added (αt); capital deepening (proxied by the flow of capital services per hours worked, kt), weighted by the capital income share of value added (βt); and TFP (denoted by At):11

yt = αtkt + βtHt + At. (1)

Labor and capital inputs for both Canada and the United States are adjusted for quality changes using the same methodology. In particular, labor quality (Ht) is the

11The dot over the variables denotes percentage growth rates. For a more detailed discussion of the methodology, see Jorgenson, Ho, and Stiroh (2002).

Table 3.5. Canada and the United States: ICT Capital Accumulation (In percent)

1981–1995 1995–2000 ____________________ ____________________ Canada United States Canada United States

Investment (average rate of growth)1

Computers 25.8 28.0 39.8 39.3Software 19.2 16.6 10.2 19.8Communications 4.5 4.8 17.9 12.2

Capital services (average rate of growth)2

All assets 3.0 3.4 4.3 5.4ICT 16.9 14.9 18.4 21.3

Computers 27.1 23.9 32.9 41.8Software 14.5 15.0 7.2 16.4Communications 7.4 6.3 12.2 8.5

ICT share of capital income2 6.3 10.9 8.3 15.3

ICT share of capital stock3 3.9 7.0 6.4 11.7

1Source: Haver Analytics.2Source: Harchaoui and Tarkhani (2002).3Sources: Armstrong and others (2002) for Canada, and U.S. Bureau of Economic Analysis for the

United States.Notes: Values are for 1981 and 2000; ICT = information and communication technology.

©International Monetary Fund. Not for Redistribution

35

difference between the growth of hours worked and the growth of labor input, obtained by weighting the hours of different types of labor (in terms of educa-tional attainment, age, and gender) by their marginal productivity (proxied by their relative compensation). Similarly, capital services are obtained by weighting the growth rates of different capital assets, using their esti-mated marginal productivity (proxied by rental prices). Within this framework, the estimates of labor and capi-tal inputs capture the effect of substituting toward inputs with a higher marginal productivity (such as ICT capital and more highly educated labor). In turn, this allows the estimates of TFP to better proxy the impact of technical and organizational changes on productivity.12

12Industry value-added measures of TFP reflect technological changes only, assuming that the production function is separable in primary (capital and labor) and intermediate inputs. Loosely speaking, this is equivalent to assuming that firms’ decisions on the capital-labor mix are not affected by decisions regarding inter-mediate inputs. If this is not the case, industry value-added TFP would capture not only technical changes, but also the productivity

Tables 3.6–3.8 show Canadian and U.S. industries’ average labor productivity growth for the entire busi-ness sector, aggregated over the 23 sectors considered,13

and the contribution of the three proximate causes. The main results may be summarized as follows:

improvements that derive from more efficiently produced interme-diate inputs. Hence, while TFP would still reflect technological changes at an aggregate level, its breakdown across industries would be affected. OECD (2001b) suggests interpreting more widely the industry value-added measure of TFP as an indication of an industry’s ability to translate technical changes into overall income.

13Consistent with OECD (2001b), total labor productivity growth is obtained as the weighted average of labor productivities across industries using value-added shares as weights, plus a reallocation term that reflects the economy’s ability to move labor resources to those sectors with a higher-than-average level of labor productivity (see equation 2). Given that aggregate TFP and contributions from labor quality and capital deepening are obtained as weighted aver-ages of industries figures using value-added shares as weights, their sum is different than total labor productivity, the difference being the reallocation factor.

B. The Canada–United States Productivity Gap

Table 3.6. Canada and the United States: Productivity Growth, 1982–2000 (In percent)

Canada United States ______________________________________ _____________________________________ Contribution from: Contribution from: ________________________ ________________________ Labor Labor Labor Labor productivity Capital quality TFP productivity Capital quality TFP

Agriculture 3.7 0.1 0.5 3.1 4.3 0.3 0.3 3.7Mining 2.2 2.1 0.2 –0.1 1.8 2.9 0.2 –1.3Construction –0.6 0.1 0.3 –1.0 –0.1 –0.1 0.4 –0.4Food, beverage, and tobacco 1.3 0.5 0.1 0.7 0.6 0.9 0.2 –0.5Rubber and plastic 2.2 0.0 0.1 2.1 4.3 0.6 0.3 3.3Textiles, apparel, and leather 2.5 0.5 0.5 1.5 3.0 1.0 0.5 1.5Lumber and wood 2.0 0.3 0.4 1.3 0.1 –0.6 0.4 0.2Furniture and fixtures 1.3 –0.5 0.2 1.7 1.0 0.3 0.6 0.2Paper and allied products 3.2 1.9 0.2 1.1 1.5 1.0 0.3 0.2Printing and publishing –0.7 0.3 0.3 –1.3 –1.0 1.1 0.3 –2.3Primary metal 5.2 0.7 0.3 4.3 2.1 0.7 0.4 1.0Fabricated metal 1.5 –0.4 0.2 1.6 2.2 0.4 0.4 1.3Industrial machinery and

electrical equipment 5.8 1.3 0.3 4.2 11.0 1.3 0.5 9.2Transportation equipment 4.2 1.1 0.1 3.0 2.2 0.5 0.2 1.5Nonmetallic mineral products 2.2 –0.4 0.2 2.4 2.2 0.3 0.4 1.5Chemicals and chemical products 4.1 0.1 0.2 3.8 3.7 1.4 0.3 2.0Other manufacturing industries 1.4 0.4 0.3 0.7 5.7 1.1 0.4 4.2Transportation 2.3 0.5 0.4 1.4 1.7 –0.1 0.3 1.5Communications 3.1 1.8 0.8 0.5 3.9 2.8 0.2 0.9Utilities 1.3 0.1 0.1 1.1 3.1 2.0 0.1 0.9Trade 2.4 0.5 0.5 1.3 3.5 1.2 0.2 2.0Finance, insurance, and real estate 1.7 1.5 0.5 –0.3 1.2 1.5 0.2 –0.5Other services –0.2 0.8 0.7 –1.7 –0.2 0.6 0.2 –1.1

Total 1.6 0.8 0.4 0.6 1.9 1.1 0.3 0.8

Source: IMF staff estimates.Note: TFP = total factor productivity.

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

36

• Annual growth in Canadian aggregate labor produc-tivity averaged 0.3 percentage point less than that in the United States over the whole period, but the gap in growth rates widened to an average 0.8 percentage point during 1995–2000. These estimates are broadly consistent with the estimates obtained using growth accounting at an aggregate level (see Macklem, 2003).

• During 1995–2000, the labor productivity gap between the two countries widened not only in the ICT-producing sector, but also in sectors that intensively used ICT capital.14 Canada’s non-ICT-producing manufacturing industries appear to have performed as well, if not better, than their U.S. coun-terparts. However, a gap emerged in sectors that have been most intensively using new technologies, such as trade and finance, insurance, and real estate (the FIRE sector). In particular, labor productivity growth in Canada’s trade sector was well below that in the United States, reflecting shortfalls in both TFP and

14The ICT-producing sector is proxied by the industrial machin-ery and electrical and electronic product sectors.

capital deepening. A gap also opened in the FIRE sector, reflecting a smaller contribution of capital deepening than in the United States.

Industrial Structure and the Aggregate Labor Productivity Growth Gap

Using growth accounting at a sectoral level allows the aggregate labor productivity growth gap between Canada and the United States to be disaggregated into three components, which correspond to the three terms on the right-hand side of equation (2): • a “direct” effect, which reflects the contribution from

industry i’s different labor productivity growth per-formance, weighted by its average value-added share (vai);

• a “structural” effect, which reflects the contribution from industry i’s different relative size across the two countries, weighted by its average labor productivity growth; and

• a “reallocation factor,” which reflects the different ability of the two economies to direct labor resources (hours worked, hi) toward sectors with a value-added

Table 3.7. Canada and the United States: Productivity Growth, 1982–1995(In percent)

Canada United States ______________________________________ _____________________________________ Contribution from: Contribution from: _________________________ _________________________ Labor Labor Labor Labor productivity Capital quality TFP productivity Capital quality TFP

Agriculture 3.0 –0.7 0.6 3.1 3.7 0.1 0.4 3.2Mining 3.3 1.8 0.2 1.2 3.0 3.1 0.3 –0.4Construction –0.6 0.2 0.3 –1.1 –0.1 –0.3 0.5 –0.3Food, beverage, and tobacco 1.6 0.3 0.3 1.0 2.5 0.7 0.2 1.5Rubber and plastic 3.2 0.5 0.3 2.4 4.1 0.3 0.4 3.4Textiles, apparel, and leather 2.5 0.6 0.6 1.3 3.1 0.7 0.6 1.8Lumber and wood 1.9 0.2 0.4 1.3 0.5 –0.9 0.5 0.9Furniture and fixtures 0.8 –0.1 0.3 0.6 0.9 0.2 0.7 0.0Paper and allied products 3.1 1.8 0.4 0.9 1.5 0.9 0.4 0.2Printing and publishing –1.2 0.4 0.4 –2.0 –1.4 0.8 0.3 –2.6Primary metal 5.8 0.7 0.4 4.6 1.9 0.7 0.4 0.8Fabricated metal 1.3 –0.1 0.3 1.0 2.5 0.4 0.4 1.7Industrial machinery and

electrical equipment 5.4 1.1 0.7 3.6 9.0 0.9 0.5 7.6Transportation equipment 4.1 0.8 0.3 3.0 2.2 0.4 0.2 1.5Nonmetallic mineral products 1.3 –0.2 0.3 1.2 2.6 –0.1 0.4 2.2Chemicals and chemical products 4.2 0.1 0.3 3.8 4.2 1.2 0.3 2.7Other manufacturing industries 1.6 0.7 0.4 0.4 5.3 1.0 0.4 4.0Transportation 2.5 0.4 0.5 1.7 1.5 –0.6 0.3 1.7Communications 2.9 1.3 0.3 1.2 4.5 3.0 0.2 1.3Utilities 0.5 0.4 0.2 –0.1 3.0 1.8 0.2 1.1Trade 1.9 0.4 0.5 1.0 2.3 1.0 0.3 1.0Finance, insurance, and real estate 1.4 1.6 0.6 –0.8 0.7 1.3 0.2 –0.9Other services –0.4 0.9 0.7 –1.9 –0.4 0.5 0.2 –1.0

Total 1.6 0.8 0.5 0.5 1.6 0.9 0.3 0.6

Source: IMF staff estimates.Note: TFP = total factor productivity.

©International Monetary Fund. Not for Redistribution

37

share that exceeds the labor compensation share (lsi)(that is, toward sectors with higher-than-average labor productivity level).15

vai,Can + vai,USyCan – yUS = ∑(––––––––––––––)(yi,Can – yi,US)i 2

yi,Can + yi,US +∑(–––––––––––)(vai,Can – vai,US)i 2

+[∑(vai,Can – lsi,Can,US)i

15See Faruqui and others (2003) for a similar decomposition for-mula. Their conclusion is that most of the business sector labor productivity growth gap between Canada and the United States during 1987–2000 is explained by the direct effect. Moreover, the manufacturing sector is the main contributor to the aggregate gap during 1996–2000, while the service sector is more relevant during 1987–96. However, these results are obtained at a rather coarse level of disaggregation (four large sectors are identified: primary, manu-facturing, construction, and services), something that (as admitted by the same authors) might conceal the contribution from the struc-tural and reallocation effects.

+∑(vai,Can – Isi,US).hi,Can

i

–∑(vai,US – lsi,US).hi,US] (2)

This decomposition shows that a significant part of the widening labor productivity gap between Canada and the United States during 1995–2000 is explained by structural differences between the two economies. Table 3.9 shows that the negative contribution from the direct effect has remained relatively constant over the two periods, whereas the other two effects became a negative contributor after 1995. This seems to suggest that the widening of the Canada–United States labor productivity gap over the second half of the 1990s was mostly due to a shift in the relative pattern of industry specialization. In other words, rather than having become less productive than the United States, Canada has tended to be less successful in directing resources toward high-productivity sectors.

B. The Canada–United States Productivity Gap

Table 3.8. Canada and the United States: Productivity Growth, 1995–2000 (In percent)

Canada United States ______________________________________ _____________________________________ Contribution from: Contribution from: __________________________ __________________________ Labor Labor Labor Labor productivity Capital quality TFP productivity Capital quality TFP

Agriculture 5.3 2.2 0.3 2.8 3.2 0.6 0.2 2.4Mining –0.8 2.5 0.1 –3.4 –1.0 1.9 0.0 –2.9Construction –0.4 0.0 0.2 –0.6 –0.5 0.5 0.2 –1.2Food, beverage, and tobacco 0.7 1.0 –0.4 0.1 –0.7 1.3 0.1 –2.2Rubber and plastic –0.6 –1.1 –0.3 0.8 4.3 1.4 0.2 2.8Textiles, apparel, and leather 2.9 0.0 0.1 2.8 2.5 1.7 0.2 0.6Lumber and wood 1.5 1.3 0.3 0.0 0.2 0.5 0.2 –0.5Furniture and fixtures 2.6 –1.4 –0.2 4.2 1.6 0.5 0.3 0.9Paper and allied products 2.2 1.9 –0.2 0.5 –1.2 1.1 0.2 –2.4Printing and publishing 0.4 0.1 0.0 0.3 –0.3 1.5 0.2 –2.0Primary metal 2.8 0.2 –0.1 2.7 1.8 0.7 0.2 0.9Fabricated metal 1.9 –1.0 0.0 2.9 1.4 0.6 0.3 0.5Industrial machinery and electrical equipment 7.4 1.5 –0.4 6.3 16.9 2.3 0.4 14.2Transportation equipment 3.7 1.8 –0.3 2.2 2.0 0.8 0.1 1.1Nonmetallic mineral products 3.0 –1.2 –0.1 4.3 1.3 1.3 0.3 –0.2Chemicals and chemical products 3.9 –0.2 –0.1 4.1 2.3 1.9 0.2 0.2Other manufacturing industries 1.0 –0.8 0.0 1.7 6.2 1.3 0.4 4.5Transportation 1.7 0.8 0.3 0.6 1.7 0.9 0.2 0.5Communications 3.3 2.7 2.0 –1.3 2.3 2.2 0.1 0.1Utilities 3.3 –1.0 –0.2 4.5 4.0 2.6 0.1 1.4Trade 3.0 0.8 0.6 1.6 5.4 1.7 0.1 3.6Finance, insurance, and real estate 2.7 1.3 0.2 1.2 2.8 2.0 0.2 0.6Other services 0.6 0.5 0.7 –0.7 0.0 1.0 0.3 –1.3

Total 1.8 0.8 0.3 0.9 2.6 1.5 0.2 1.1

Source: IMF staff estimates.Note: TFP = total factor productivity.

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

38

The widening of the aggregate labor productivity gap between Canada and the United States after 1995 was driven by two major service sectors, trade and FIRE. Figure 3.6 shows the contribution of industries in these sectors to the aggregate labor productivity growth gap in the two subperiods, 1981–1995 and 1995–2000. Each industry’s contribution is given by the sum of its contribution to the direct, structural, and reallocation effects. While the negative contribution from the ICT-producing sector increased only slightly in the second sub period, the negative contribution from the trade and FIRE sectors rose significantly. The negative contri-bution from the ICT-producing manufacturing and trade sectors mainly reflected lower labor productiv-ity growth, whereas the negative contribution from the FIRE sector was largely the result of the lower relative size of the sector in Canada.16

The Contribution from ICT Capital Accumulation

The more muted contribution of ICT capital accu-mulation to productivity growth in Canada may reflect the different timing of Canadian and U.S. ICT invest-ments. ICT investments affect both labor productivity, through capital deepening, and TFP growth, by induc-ing additional investments in intangible assets such as

16The comparison between the trade sectors in the two coun-tries may be blurred by the fact that eating and drinking places are included in the U.S. trade sector, while they are part of the “other service” sector for Canada (as part of the “accommodation, food, and beverage” subsector). However, the results do not change sub-stantially when the Canadian trade sector is adjusted to include that fraction of the “accommodation, food, and beverage” sector attribut-able to eating and drinking places.

organizational changes and knowledge accumulation. However, the payoff from these investments in terms of measured output can be delayed considerably, given the time and resources required to reorganize production after investing in ICT capital.

Empirical evidence is supportive of the existence of relatively long lags between ICT capital accumula-tion and TFP growth in Canada. Following Basu and others (2003), a simple ordinary least squares (OLS) regression is run to relate TFP average growth during 1995–2000 to ICT capital service growth in the 1980s, the mid-1990s, and the late 1990s, taking each industry as a cross-sectional observation. Table 3.10 shows that Canadian TFP growth in the late 1990s is negatively correlated to ICT capital investments made during both halves of the 1990s but is positively correlated to ICT investments made in the 1980s.17 For the United States, the results are qualitatively similar to Basu and others (2003), with post-1995 TFP growth negatively corre-lated to ICT capital accumulation over the same period but positively correlated to ICT capital accumulation during the 1980s and early 1990s. This result suggests that Canada’s slower TFP acceleration during 1995–2000 may reflect the fact that Canadian firms invested in complementary capital later than U.S. firms and/or that this process took longer for Canadian firms. It also

17As stressed by Basu and others (2003), the OLS regression and the simple specification do not show causation but only indicate whether a correlation exists between lagged ICT capital and TFP growth. The results in Table 3.10 are robust to the exclusion of key sectors, such as the ICT-producing, trade, and FIRE sectors.

Table 3.9. Canada–United States Labor Productivity Growth Gap(In percent)

Contribution from: 1982–2000 1982–1995 1995–2000

Direct effect –0.2 –0.1 –0.2of which:

Industrial machinery and –0.1 –0.1 –0.2 electrical and electronic equipment Transportation equipment 0.1 0.0 0.1Other manufacturing industries 0.0 0.0 –0.1Communications 0.0 0.0 0.0Utilities –0.1 0.0 0.0Trade –0.2 0.0 –0.4Finance, insurance, and real estate 0.1 –0.1 0.0Other services 0.0 0.0 0.1

Structure effect 0.0 0.1 –0.2Reallocation effect 0.0 0.1 –0.2

Source: IMF staff estimates.

©International Monetary Fund. Not for Redistribution

39

Source: IMF staff calculations.

–0.6% –0.3% 0% 0.3%

–0.6% –0.3% 0% 0.3%

Mining

Other services

Primary metal

Transportation

Transportation equipment

Paper and allied products

Lumber and wood

Construction

Printing and publishing

Furniture and fixture

Agriculture

Rubber and plastic

Food, beverage, and tobacco

Nonmetal mineral products

Textiles, apparel, and leather

Chemical and chemical products

Utilities

Communications

Fabricated metal

Finance, insurance, and real estate (FIRE)

Other manufacturing industries

Trade

Industrial machinery and electrical and electronic equipment

Other services

Transportation equipment

Utilities

Agriculture

Paper and allied products

Lumber and wood

Primary metal

Food, beverage, and tobacco

Transportation

Mining

Nonmetal mineral products

Chemical and chemical products

Fabricated metal

Printing and publishing

Furniture and fixture

Textiles, apparel, and leather

Construction

Communications

Rubber and plastic

Other manufacturing industries

Finance, insurance, and real estate (FIRE)

TradeIndustrial machinery and electrical and electronic equipment

1981–1995

1995–2000

Figure 3.6. Sectoral Contributions to the Canada–United States Aggregate Labor Productivity Growth Gap(In percent)

B. The Canada–United States Productivity Gap

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

40

suggests, however, that Canadian firms might benefit from faster TFP growth in the near future.18

18Leung (2004) finds that investments in new technologies in Canada have their strongest impact on TFP growth only after three years. Robidoux (2003) also suggests that the contribution of ICT capital to the acceleration of TFP growth in the service sector in the

Productivity Growth and Trade

The labor productivity gap between Canada and the United States widened despite the marked deepening of trade linkages between the two countries. Some have argued that this reflects Canada’s increased specializa-tion in the natural-resource-based manufacturing sec-tors, where Canada has had a comparative advantage (see, for example, Jackson, 2003). This reallocation of resources could have dampened the aggregate growth of Canadian productivity.

However, the labor productivity gap is unlikely to be related to increased economic integration. First, the results attribute nearly the entire structural labor productivity gap to the service (nontradables) sector. Second, during the second half of the 1990s, Canada’s tradables sector seems to have evolved rapidly in the direction of high-tech production. In particular, while still lower than in the United States, Canada’s ICT-producing sector increased its share of aggregate GDP over this period (Table 3.11). Finally, several studies show that, during the 1990s, Canadian trade increased mainly in two-way trade in similar products, rather than across different industries.19

On the contrary, Canadian TFP growth is positively correlated with trade. Plotting average TFP growth against (1) the degree of vertical specialization and (2) the degree of trade exposure of Canadian sectors during 1981–2000 shows that TFP growth is posi-

late 1990s reflects the successful incorporation of ICT into produc-tion and management processes during the 1980s and early 1990s.

19See Dion (1999) and Acharya, Sharma, and Rao (2003).

Table 3.11. Canada and the United States: Value Added, Shares of Total (In percent)

Canada United States _____________________ _____________________ 1981–1995 1995–2000 1981–1995 1995–2000

Agriculture 3.2 2.5 2.3 1.8Mining 6.9 5.6 2.5 1.5Construction 8.5 6.9 5.3 5.3Manufacturing 24.0 25.1 22.4 19.5

of whichIndustrial machinery and

electrical and electronic equipment 2.7 2.9 5.0 4.4Transportation 5.5 5.2 4.0 3.8Communications 3.9 3.7 3.1 3.1Utilities 4.5 4.5 3.2 2.6Trade 14.7 14.3 16.4 16.0Finance, insurance, and real estate 12.8 14.3 19.0 21.1Other services 15.9 18.0 21.9 25.3

Source: IMF staff estimates.

Table 3.10. Canada: ICT Regression with Current and Lagged ICT Capital Services Growth

Canada United States

Constant 1.79 –0.80 (0.73) (0.93)

ICT_CAP_95–00 –0.24 –2.41 (0.54) (1.03)

ICT_CAP_90–95 –1.53 2.50 (1.02) (1.09)

ICT_CAP_80–90 1.82 0.43 (1.48) (0.46)

R-squared 0.11 0.24Observations 24 21

Source: IMF staff estimates.Notes: ICT = information and communication technolo-

gies; dependent variable: average annual TFP growth in 1995–2000. White heteroscedasticity–consistent standard errors in parentheses.

©International Monetary Fund. Not for Redistribution

41

tively correlated with openness to trade (Figure 3.7).20

In particular, the ICT-producing and transportation equipment sectors seem to have benefited most from exposure to trade and interindustry specialization dur-ing this period. Moreover, the extent of the correla-tion has increased since the inception of the free trade agreement with the United States.

20Both indexes of openness to trade are from Dion (1999) (the data, originally up to 1996, have been extrapolated to 2000). The degree of vertical specialization measures the extent to which an industry’s trade is accounted for by inputs that are imported and embodied in exports. The degree of trade exposure is the algebraic sum of three different indicators: the share of an industry’s exports in its gross output (capturing its degree of export orientation); the share of an industry’s imported intermediate inputs in its gross out-put (capturing the exposure of an industry on the cost side); and the share of an industry’s competing imports in the domestic markets for its core products (measuring the exposure to foreign penetration of the domestic market).

C. The Effects of U.S. Shocks

Iryna V. Ivaschenko and Andrew Swiston21

The close integration of the Canadian and U.S. econ-omies means that U.S. shocks are quickly transmitted across the border. Canada’s exports to the United States account for about 85 percent of total Canadian exports and about 35 percent of its GDP, and investment flows and financial markets are also closely linked. As documented by Kose (2004), the increased trade and financial linkages that resulted from the 1989 Canada-U.S. free trade agreement have significantly increased the impact of the U.S. business cycle on Canada, but Canada’s vulnerability to U.S. shocks also stems from

21The authors are grateful to Douglas Laxton for invaluable help with the model.

C. The Effects of U.S. Shocks

–0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4–2–1012345678

Trade Exposure and TFP Growth by Sector, Average 1981–2000T

FP g

row

th (

in p

erce

nt)

Trade Exposure

Fabricated metal Furniture and fixture

Mining

0 0.1 0.2 0.3 0.4 0.5 0.6–2–1012345678

Vertical Integration and TFP Growth by Sector, Average 1981–2000

TFP

gro

wth

(in

per

cent

)

Vertical Integration

Transportation equipmentLumber and wood Furniture and fixture

Mining

Source: IMF staff calculations.Note: TFP = total factor productivity.

Primary metal

Textiles, apparel, and leather

Figure 3.7. Canada: Sectoral TFP Growth and Openness to Trade

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

42

its relatively small size and the importance of its natu-ral resource sector.22

The recent strength of the Canadian dollar inten-sified interest in the impact of U.S. shocks on the Canadian economy and monetary policy. The vigor of Canada’s net exports during 2004 was surprising, given the 30 percent appreciation of the Canadian-U.S. dol-lar exchange rate during 2003 and 2004 (Figure 3.8). Competing explanations for the modest impact on trade have been offered, including that it reflects an increase in the flexibility and efficiency of Canadian industry, a decline in exchange rate pass-through, or delays in the usual adjustment process.

These issues are explored here using a small two- country macroeconomic model. In particular, the model is used to investigate how changes in the exchange rate pass-through affect the transmission of an exchange rate shock to the real economy. The results offer two key insights:• The Canadian economy responds significantly to

U.S. macroeconomic and policy shocks, as well as to exchange rate shocks. However, there is considerable scope for a monetary policy response to ameliorate the effects of these shocks.

• The strength of net exports in 2004 appears to be at least partially related to a decline in the pass-through coefficient. The weakening of pass-through—if it is sustained—would significantly reduce the impact of exchange rate shocks on both GDP and inflation.

22In 2003, Canada’s GDP was equivalent to about 8 percent of U.S. GDP, and domestic absorption accounted for 7 percent of that in the United States.

Model Description

The model employed for this study is a two- country version of a small open economy model.23 Each econ-omy is characterized by three equations—an IS curve, an expectation-augmented Phillips curve, and a mon-etary policy reaction function. Canada is assumed to face both domestic and external shocks (that is, from the United States), while the United States is assumed to be large enough to be essentially a closed economy. The model allows U.S. output shocks to feed into the Canadian IS equation, while real exchange rate shocks affect both the IS equation and the Phillips curve. For the sake of simplicity, the effects of fiscal policy are not modeled. There are four equations for Canada.• IS curve:

ygapt = βlead ygapt+1 + βlag ygapt–1

– βRR(RRt–1 – RR*t–1) + βzgap zgapt–1

+ βUS ygaptUS + t

gap, (1)

where ygap is the Canadian output gap; RR is the Cana-dian real short-term interest rate; RR* is the equilibrium real interest rate for Canada; ygapUS is the U.S. output gap; zt is the Canadian-U.S. dollar exchange rate, in real terms; z*

t is the equilibrium real exchange rate; and zgapt = zt – z*

t is the exchange rate gap.• Phillips curve:

πt = αlead π4t+4 + (1 – αlead) π4

t–1+ αygap ygapt –1 + αz (zt – zt–1) + uπ

t, (2)

where π is the annualized quarterly inflation rate, andπ4 is the four-quarter inflation rate.24

• Monetary policy reaction function:

RSt = γlagRS RSt–1 + (1– γlag

RS)[RR*t

+ π4t + γπ(π4

t+4 – π*t+4)

+ γgap ygapt] + utRS, (3)

where RS is the target for the nominal overnight rate (the Canadian monetary policy rate). The terms uygap, uπ, and ut

RS are error terms. This is equivalent to assuming that the Bank of Canada allows the real short-term interest rate to deviate from its “equilibrium” level depending on whether the inflation rate that is expected to prevail four quarters ahead deviates from the target, π*, or whether output deviates from potential. Interest rate smoothing is permitted, however—the short-term interest rate is set with reference to its value last period.• Real exchange rate (uncovered interest rate parity):

zt = zte+1 – (RRt – RRt

US – P*t)/400 + εz

t, (4)

23Lane (1999) provides a review of the new open economy litera-ture. See Berg, Karam, and Laxton (2006) for a description of the model used here.

24The results of estimating the model with either core or headline inflation are qualitatively the same.

Sources: Haver Analytics; and IMF Information Notice System.Note: REER = real effective exchange rate.

70

80

90

100

110

1990 92 94 96 98 2000 02 04543210

–1–2–3–4–5

Inde

x, 1

990

= 1

00

Perc

ent

of G

DP

Current account balance(right scale, inverted)

REER

Figure 3.8 Real Effective Exchange Rate and Current Account Balance

©International Monetary Fund. Not for Redistribution

43

where ρ*t is an interest rate premium. Exchange rate

expectations are assumed to follow an autoregressive process:

zet+1 = δzzt+1 + (1–δz)zt–1. (5)

Similar equations are assumed for the U.S. econ-omy. However, because the U.S. economy is assumed not to be affected by Canadian shocks, the U.S. IS curve includes neither the foreign output gap nor the Canadian-U.S. dollar exchange rate. Moreover, the U.S. Phillips curve does not include the exchange rate.

Data and Estimation

The model is computationally tractable and provides for a close integration with the IMF’s medium-term forecasting framework. Owing to its simplicity, the model can be easily applied to medium-term economic forecasts provided for the World Economic Outlook,

for example, to consider the effects of different policy responses or the validity of model assumptions.25 It also facilitates the application of sophisticated estimation and simulation techniques that have been developed in recent years.

The model employs Bayesian estimation techniques. This approach incorporates prior knowledge about parameter values, which is especially useful given the short data sample. It also provides information on the distribution of model parameters and shocks. All shocks are modeled as first-order autoregressive processes with normally distributed error terms, with the sole excep-tion of the exchange rate shock, which is assumed to be normally distributed. In addition, all data are assumed to include some parameterized measurement error to account for data uncertainty related to the possibility of future revisions.

25See Coats, Laxton, and Rose (2003) and references therein.

C. The Effects of U.S. Shocks

Source: IMF staff calculations.

–1.75

–1.25

–0.75

–0.25

0.25

0.75

1.25

U.S. output gapincreases by 0.5%

U.S. output gapincreases by 1%

U.S. output gapincreases by 2%

181614121086420

–0.75

–0.25

0.25

0.75

1.25

U.S. inflation rises by3 percentage points

U.S. inflation rises by2 percentage points

U.S. inflation rises by1 percentage point

181614121086420

Shock: U.S. output gap

Shock: U.S. CPI inflation Shock: U.S. CPI inflation

Immediate Policy Response

–1.75

–1.25

–0.75

–0.25

0.25

0.75

1.25

U.S. output gapincreases by 0.5%

U.S. output gapincreases by 1%

U.S. output gapincreases by 2%

181614121086420

–0.75

–0.25

0.25

0.75

1.25

U.S. inflation rises by1 percentage point

U.S. inflation rises by2 percentage points

U.S. inflation rises by3 percentage points

181614121086420

Shock: U.S. output gap

Policy Response Delayed Four Quarters

Figure 3.9. Response of Canadian GDP to U.S. Shocks, with Immediate and Delayed Monetary Policy Response(Canadian GDP, percent deviation from baseline)

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

44

The model uses quarterly data from 1996 through the third quarter of 2004. The sample period was cho-sen to exclude transition effects from Canada’s adop-tion of an inflation target in 1991. The model was first estimated using historical data and then simulated over the forecast horizon. The equilibrium values of variables (that is, the starred variables) were deter-mined using a version of the Hodrick-Prescott filter (except for the Canadian inflation target, which is set at 2 percent).

Results The simulation results indicate that real shocks to the

U.S. economy significantly affect both Canadian GDP and inflation. A 1 percentage point increase in U.S. GDP, which stems from a temporary demand shock linked to the IS curve, raises Canadian GDP by almost !/2 per-

cent and inflation by !/3 percentage point (Figures 3.9 and 3.10). Conversely, a permanent 1 percent reduction in U.S. potential output—which would be equivalent to a negative U.S. supply shock—reduces the level of GDP in Canada by about 1!/2 percent and inflation by 1 percentage point over six quarters.

The model can be used to illustrate the costs of delay-ing the monetary policy response to external and other shocks. For example, if Canadian monetary policy were assumed to respond with a four-quarter lag (rather than immediately as is implied by the monetary reaction function described above), the impact of the temporary U.S. demand shock would be about !/4 percent larger over four quarters, and CPI inflation would be cor-respondingly higher (see Figures 3.9 and 3.10). More-over, the delayed monetary policy response requires a larger interest rate movement—in this case the Bank of Canada is required to raise its policy rate by about

Source: IMF staff calculations.

–1

–0.75

–0.50

–0.25

0

0.25U.S. output gap

increases by 0.5%

U.S. output gapincreases by 1%

U.S. output gapincreases by 2%

181614121086420

Shock: U.S. output gap

Immediate Policy Response

–1.00

–0.75

–0.50

–0.25

0

0.25U.S. output gap

increases by 0.5%

U.S. output gapincreases by 1%

U.S. output gapincreases by 2%

181614121086420

Shock: U.S. output gap

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

U.S. inflation risesby 3 percentage points

U.S. inflation risesby 2 percentage points

U.S. inflation risesby 1 percentage point

181614121086420

Shock: U.S. CPI Inflation

Policy Response Delayed Four Quarters

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

181614121086420

Shock: U.S. CPI Inflation

U.S. inflation risesby 3 percentage points

U.S. inflation risesby 2 percentage points

U.S. inflation risesby 1 percentage point

Figure 3.10. Response of Canadian CPI to U.S. Shocks, with Immediate and Delayed Monetary Policy Response(Percentage point deviation of four-quarter Canadian inflation rate from baseline)

©International Monetary Fund. Not for Redistribution

45

C. The Effects of U.S. Shocks

!/2 percentage point more in the quarter of the move than otherwise.

The speed of the monetary policy response is also important in determining how U.S. inflation shocks affect the Canadian economy. For example, the impact of a 1 percentage point U.S. inflation shock—which is modeled as a shock to the U.S. Phillips curve—on Canada’s GDP and inflation would be roughly halved by an immediate response from the Bank of Canada versus a delayed response (Figure 3.11).

The model also can be used to illustrate that U.S. monetary policy has a relatively modest effect on Canada. For example, even if the monetary policy reaction were delayed by four quarters, a 100-basis-point increase in the federal funds rate would reduce Canadian GDP by only about 0.1 percent over six quarters. Again, allowing an immediate policy response would halve this (already small) effect, with the impact on Canadian inflation being negligible in both cases.

Source: IMF staff calculations.

Canadian GDP

–0.15

–0.10

–0.05

0

0.05

0.10

Immediate policy response

Delayed policy response

191817161514131211109876543210

Perc

ent

devi

atio

n fr

om b

asel

ine

Canadian Inflation

–0.15

–0.10

–0.05

0

0.05

0.10

Immediate policy response

Delayed policy response

191817161514131211109876543210

Perc

enta

ge p

oint

dev

iatio

n of

4-q

uart

erC

PI in

flatio

n ra

te fr

om b

asel

ine

Figure 3.11. Response of Canadian GDP and Inflation to U.S. Shocks, with Immediate and Delayed Monetary Policy Response(Shock: U.S. interest rates, 100-basis-point increase)

©International Monetary Fund. Not for Redistribution

III CANADA AND THE UNITED STATES

46

Source: IMF staff calculations.

Canadian GDP

–2.0

–1.5

–1.0

–0.5

0

0.5

1.0

Immediate policy response

Delayed policy response

191817161514131211109876543210

Perc

ent

devi

atio

n fr

om b

asel

ine

Canadian Inflation

–1.5

–1.0

–0.5

0

0.5

191817161514131211109876543210

Perc

enta

ge p

oint

dev

iatio

n of

4-q

uart

erC

PI in

flatio

n ra

te fr

om b

asel

ine

Immediate policy response

Delayed policy response

Figure 3.12. Response of Canadian GDP and Inflation to Exchange Rate Shock, with Immediate and Delayed Monetary Policy Response(Shock: 20 percent appreciation of exchange rate)

The impact on Canada of exchange rate shocks is found to be relatively strong, however, reflecting Canada’s export dependency. Simulation results sug-gest that a 20 percent real appreciation of the Canadian dollar against the U.S. dollar reduces Canada’s GDP by about 1 percent over four quarters, even if the Bank of Canada immediately reduces its target rate by 1½ per-cent (Figure 3.12). This effect almost doubles if rate hikes are delayed.

However, the model also illustrates that the effect of exchange rate shocks depends significantly on the

degree of pass-through. If the pass-through coefficient, βzgap, in equation (1) is lowered by half, the impact of an exchange rate appreciation on GDP is reduced by about a quarter, and the effect on the rate of inflation by nearly a half (Figure 3.13). This result is consistent with the view of a number of analysts in Canada that the seemingly modest response of Canadian net exports and growth in 2004 to the strong appreciation of the Canadian dollar over the previous two to three years is partially accounted for by a decline in exchange rate pass-through.

©International Monetary Fund. Not for Redistribution

47

C. The Effects of U.S. Shocks

Source: IMF staff calculations.

Canadian GDP

–0.10

–0.05

0

0.05

0.10

Pass-through coefficientrises by 50%

Pass-through coefficientrises by 100%

191817161514131211109876543210

Perc

ent

devi

atio

n fr

om b

asel

ine

Canadian Inflation

–0.05

–0.03

–0.01

0.01

0

0.03

0.05

191817161514131211109876543210

Perc

enta

ge p

oint

dev

iatio

n of

4-q

uart

erC

PI in

flatio

n ra

te fr

om b

asel

ine

Pass-through coefficientdeclines by 100%

Pass-through coefficientrises by 50% Pass-through coefficient

declines by 100%

Pass-through coefficientrises by 100%

Figure 3.13. Responses of Canadian GDP and Inflation to Changes in the Pass-Through Coefficient, with Immediate and Delayed Monetary Policy Response

Conclusions

Using a simple two-country version of the small open economy model, this analysis gauges the impact on Canada of shocks to the U.S. economy. The results suggest that monetary policy can play

an important role in reducing the effect of U.S. and exchange rate shocks on the Canadian economy. They also indicate that the exchange rate pass-through coefficient plays a significant role in determin-ing the impact of exchange rate shocks on Canada.

©International Monetary Fund. Not for Redistribution

This page intentionally left blank

©International Monetary Fund. Not for Redistribution

Part II

Fiscal Policies

©International Monetary Fund. Not for Redistribution

This page intentionally left blank

©International Monetary Fund. Not for Redistribution

51

IV Budget Forecasting

Martin Mühleisen, Stephan Danninger, David Hauner, Kornélia Krajnyák, and Bennett Sutton1

Canada’s strong fiscal record in recent years rests on a proven budgetary framework, including a

well-established forecasting process. Canadian public finances are highly transparent, and the prudent fis-cal policies of recent governments have enjoyed wide-spread public support. Beginning in the mid-1990s, these policies produced a string of budget surpluses that helped reduce federal debt (measured by the accumu-lated deficit) from almost 70 percent of GDP in 1996 to close to 40 percent in 2004. At that point, following a number of years with much larger than anticipated surpluses, the fiscal forecasting process went under review to ensure that “the [federal] government con-tinues to use the most up-to-date economic and fis-cal forecasting methods, and to benchmark Canadian practices against the best in the world” (Department of Finance Canada, 2004, p. 67). The IMF assisted this process by comparing Canadian central government budget forecasting with that of a benchmark group of other industrial countries.1

This section summarizes the result of this com-parison. The benchmark group consists of most other G-7 countries, as well as Australia and New Zea-land (two commodity-exporting countries) and the Netherlands, Sweden, and Switzerland (smaller indus-trial countries with advanced budget practices).2 The analysis follows a two-pronged approach, covering both structural and quantitative aspects. The section first compares the institutional environment for fiscal forecasting and forecasting processes across the bench-mark group. It then provides a description of budgetary forecast outcomes and presents the results of statistical analyses that, among other things, test for forecast bias and identify links between structural characteristics and forecast errors.

The results suggest that fiscal forecasting in Canada has been governed by one of the strongest institutional

1The authors are grateful to colleagues in the participating coun-tries as well as in the IMF’s European and Asia and Pacific Depart-ments for their cooperation in acquiring and analyzing the data for this study.

2Japanese fiscal policy in the middle to late 1990s was largely implemented through supplementary budget requests, which would complicate a comparison of its budget projections with other coun-tries. Japan was therefore not included in the benchmark group.

frameworks among the countries studied. Although Canada has no formal fiscal rule, recent governments have aimed for a modest surplus, which evolved into a de facto fiscal target. In support of this objective, Can-ada adopted a conservative approach to budgeting, with explicit prudence and contingency factors and a strong commitment to transparency and accountability. One particular strength is the explicit use of macroeconomic projections from a wide range of private forecasters for preparation of the budget. However, forecasts of fiscal variables are compiled by the Department of Finance with little participation from nongovernmental agen-cies. As is the case for many other countries, Canada could enhance public understanding of budgetary fore-casts by providing more information on the assump-tions and methods underlying the translation of the macroeconomic outlook into fiscal projections.

Quantitative analysis suggests that Canada’s budget projections of macroeconomic and fiscal aggregates were more cautious than in other countries between the mid-1990s and 2004. Measures for the distance between budget projections and actual outcomes were among the highest in the benchmark group. More-over, forecast errors for both revenue and expenditure aggregates were consistently on the conservative side, making Canada the country that has most strongly underestimated its fiscal balance since 1995. Empirical tests indicate that the forecast errors are significantly different from zero and that both public and private forecasters were repeatedly surprised by the strength of the Canadian economy and fiscal performance, par-ticularly in the late 1990s. Indeed, given the close link between tax revenues and the macro economy, stronger than expected growth appears to account for a consid-erable part of fiscal overperformance. The relatively volatile macroeconomic environment as well as institu-tional factors have also likely contributed to Canada’s conservative forecast bias.

The Institutional Environment

A country’s budget forecasting practices depend importantly on the legal and institutional structures governing fiscal policy. These structures need to be

©International Monetary Fund. Not for Redistribution

IV BUDGET FORECASTING

52

taken into account when comparing forecasting prac-tices across countries, particularly as they can influ-ence the accuracy of budget projections in a number of ways. This subsection looks at three factors that characterize the fiscal environment: (1) the distribu-tion of fiscal authority between the legislature and the executive; (2) fiscal relations between the central and subnational governments; and (3) the presence of fiscal rules and other constraints limiting fiscal policy discretion.

Distribution of Fiscal Authority

The distribution of fiscal authority between the exec-utive and legislative branch may affect the nature and quality of budget forecasts. For example, if substan-tial fiscal authority rests with the legislature, policy assumptions underlying the fiscal forecast of the execu-tive branch may differ from fiscal measures taken, and the forecast quality could suffer correspondingly. Alter-natively, the executive could face incentives to produce biased forecasts in order to influence the behavior of the legislature. For example, the executive could pro-vide conservative revenue forecasts to keep spending pressures under control. By contrast, there would likely be fewer incentives for biased forecasts in cases where the legislature tends to approve the budget as drafted.

In Canada, the legislature has largely been focused on optimizing the budget process, as opposed to taking an active role in formulating the budget. The budget process reflects international best practices in many areas. For example, an OECD-World Bank survey (OECD and World Bank, 2003) finds that 19 out of 20 key aspects of the Canadian budget process are reg-ulated by the constitution or by law (Table 4.1). Among the countries in the benchmark group, only the United States achieves a similar score.3 Moreover, Canada adheres to 10 of 13 OECD Best Practices in budget reporting, which is matched only by New Zealand and the United States.

Canadian budgets are usually passed as submitted without any changes. This appears to be a common feature in Westminster-style parliamentary systems and in other countries where the executive enjoys reliable support in the legislature, such as Australia, New Zea-land, the United Kingdom, and Sweden (OECD and World Bank, 2003). The role of Canada’s parliament is circumscribed by the following:• Parliament receives the budget relatively late, less

than two months before the start of the fiscal year. A quarter of the fiscal year has typically elapsed by the time the budget is approved. In contrast, legisla-tures of other countries receive the budget two to six months before the new fiscal year; it is even earlier in

3Switzerland was not part of the OECD-World Bank survey.

the United States. The late budget submission may be partly attributable to the use of accrual accounting, which requires the use of information that becomes available late in the fiscal year.

• Only a relatively small part of total expenditure is funded by appropriation laws. As mandatory spending in Canada does not require annual funding legislation, new appropriations cover only about 30−40 percent of spending. This is similar to arrangements in Aus-tralia and the United States, but contrasts sharply with other countries. In the United Kingdom, appro-priation laws cover 70–80 percent of total expen-diture, and coverage can reach 90−100 percent in continental Europe.

• As in many parliamentary systems, the Canadian legislature has limited powers to change the submit-ted budget. Parliament can reduce, but not increase, funding for line items, but has otherwise only the choice of approving or rejecting the government’s spending proposals. Only parliaments in Australia and New Zealand—which have to approve or reject the budget as a whole—are more constrained. Some restrictions also apply in France and Switzerland, while legislatures in Germany, Italy, the Netherlands, Sweden, and the United States are free to change every aspect of the budget proposal.

• The executive would suffer strong consequences if parliament voted against any budget proposal. The budget vote is considered a vote of confidence in many countries, but political tradition in Canada (as in Australia, New Zealand, and the United Kingdom) goes further. In these countries, the executive would customarily step down if parliament voted against any single aspect of the budget (OECD, 2002).As a result, there is little indication that executive-

legislative relations should affect the accuracy of Cana-dian budget forecasts more than in other countries. The legislature’s limited role in the annual budget process appears to provide few incentives for providing biased forecasts; it also constrains the potential loss of forecast quality resulting from modifications prior to passage. At the same time, relatively stringent process rules and reporting requirements would seem conducive to fore-cast accuracy.

Fiscal Relations with Subnational Governments

The structure of intergovernmental relations also has implications for budgetary forecasting. From a tech-nical perspective, the volatility of fiscal outcomes at the center is likely to be higher if significant trans-fers to the subnational levels are provided on a cost- sharing rather than a block-grant basis, given the scope for adjustments after the fact. However, there are also circumstances that may contribute to a deliberate bias in fiscal projections, such as when fiscal targets are set

©International Monetary Fund. Not for Redistribution

53

Table 4.1. Indicators of the Relationship between Legislature and Executive

Nether- New United United Australia Canada France Germany Italy lands Zealand Sweden Kingdom States

Aspects of budget process regulated by the constitution or by law

Public funds can be spent only in programs as authorized by legislation x x x x x x x x . . . x

The budget and financial reporting covers all central government transactions (including extrabudgetary transactions) x x x x x x . . . x

All budget transactions to be shown in gross terms x x x x x x . . .

The minister in charge of government finances has effective power over budget management x x x x x x . . . x

Individual government organizations are held accountable for the funds they collect and/or use x x x x x . . . x

Individual ministers are held accountable for the funds they collect and/or use x x x x x x . . . x

Requirements for independently audited financial accounting reports x x x x x x . . . x

Requirements for independently audited financial reports x x x x . . . x

Conditions for use of contingency or reserve provisions x x x x . . . x

Definition of public money x x x x x x x x . . . x

Rules for the creation of extrabudgetary funds to special cases, authorized by separate statute x x x x x x . . . x

Authorize the government accounts into which all public money must be paid and from which expenditures are made only by appropriation of the parliament x x x x x x x x x x

Roles for the parliament and the executive in the budget process and the relationship between the two branches with respect to budget responsibilities x x x x x x x x . . . x

The form and structure of the annual budget law (or finance bill) to be voted by parliament x x x x x x x . . . x

The definition of main headings and accounts in the annual budget law x x x x x x x . . . x

The definition of the budget deficit and surplus x x x x x x x . . . x

Legal basis for formulation and execution of budget, including the role and authorities of the ministry of finance/treasury and/or the central budget authority x x x x x x . . . x

Administrative/judicial sanctions for infractions of budget legislation x x x x x . . . x

The basis for management (internal) control and internal audit x x x x x x x . . . x

Authorities and responsibilities for issuing and reporting on government guarantees x x x x x x x . . . x

Th

e In

stitutio

nal E

nviro

nm

en

t

©International Monetary Fund. Not for Redistribution

IV

BU

DG

ET

FO

RE

CA

ST

ING

54

Table 4.1 (concluded)

Nether- New United United Australia Canada France Germany Italy lands Zealand Sweden Kingdom States

OECD Best Practices on central government budget reporting met

General overview of revenue and expenditure x x x x x x x x x x

Detailed estimates of revenue and expenditure x x x x x x x x x

Citizen’s guide x x x

Pre-budget report (general budget policy, aggregates) x x x x x x

Long-term (10 to 40 years) outlook for public finances x x x x

Midyear report(s) on fiscal outlook x x x x x x x x

Report on tax expenditures x x x x x x x x

Statement of government assets x x x x x x x x

Special reports for old-age program finances x x x x x

Special reports for civil service pension x x x x x

Special reports on government debt x x x x x

Special reports on contingent liabilities x x x

Pre-election report x x

Number of months budget is submitted to legislature before the new fiscal year <2 <2 2–4 4–6 2–4 4–6 <2 2–4 <2 >6

Percentage share of expenditure for which appropriation acts need to be passed 20–30 30–40 . . . 90–100 . . . 90–100 90–100 90–100 70–80 30–40

Sources: OECD and World Bank (2003).

©International Monetary Fund. Not for Redistribution

55

at the central government level even though the central government has limited control over the behavior of subnational governments.

Canadian provinces enjoy substantial financial inde-pendence, but transfers to provinces account for an important share of central government spending:• The center’s share in general government is smaller

in Canada than in any of the comparator countries. Combined, Canada’s subnational governments are about as large as the central government (Table 4.2). This reflects the comparatively high number of policy responsibilities falling on subnational governments, including the country’s universal health care system.

• Provinces have a high share of own-source revenues (85 percent), including from tax revenues shared with the central government (Figure 4.1). They are also free to determine their overall fiscal aggregates as well as most expenditure allocations—among the benchmark countries, only the subnational govern-ments in Sweden and the United States have as much leeway. Canadian provinces also can borrow without federal limits—as in France, the Netherlands, New Zealand, and Sweden.4

• However, transfers to other levels of government are a more important budget item in Canada than in other benchmark countries.5 Intergovernmental transfers are substantial even when compared to

4See OECD and World Bank (2003). In the United States, many states have fixed limits in their constitutions. Similar to most com-parator countries, the Canadian federal government does not guar-antee the debt of subnational governments.

5Equalization transfers (to reduce economic disparities among provinces) and transfers for health and social spending are the most important transfers, amounting to 1 and 3 percent of GDP, respectively.

GDP, and the relatively small size of the center fur-ther inflates their size relative to other federal expen-ditures (Table 4.3).Uncertainties about revisions to the level of inter-

governmental transfers and shared tax revenues may pose difficulties for fiscal forecasting. Fiscal arrange-ments in Canada provide for considerable payments from the federal government to the provinces. Under some arrangements, the final amounts are usually not determined by the end of a given fiscal year, giving rise to adjustments in subsequent years. Due to the relatively large size of transfers relative to other gov-ernment expenditures, revisions can sometimes have a notable impact on the federal fiscal forecast:• The amount of equalization transfer payments, which

are provided as unconditional block grants, was until recently subject to considerable uncertainty. Prior to 2004, these transfers were subject to significant adjustments, owing to statistical revisions of provin-cial tax bases and population size (Box 4.1). In 2004, however, the government reached an agreement with the provinces to place equalization transfers on a more predictable basis, including by eliminating retroactive adjustments to the overall amount of transfers.

• Adjustments also are made necessary because the central government collects personal and corporate income taxes on behalf of nine and seven prov-inces, respectively, as well as Canada Pension Plan payroll contributions. These collections represent about 35 percent of federal revenue. Gross income and payroll tax revenues are divided on a prelimi-nary basis throughout the year, but the actual split is only known after all relevant tax returns are assessed—usually toward the end of the following fiscal year.

Table 4.2. Share of Spending by Subnational Governments1

Federal countriesAustralia n.a.Canada 56.5Germany 36.1Switzerland n.a.United States 40.0

Unitary countriesFrance 18.6Italy 29.7Netherlands 34.2New Zealand n.a.Sweden 43.4United Kingdom 25.9

Source: Joumard and Kongsrud (2003).1Percent of general government spending, national accounts

basis, 2001.

Sources: OECD and World Bank (2003).

0 20 40 60 80 10002468

101214161820

Australia

Canada

Germany

ItalyNetherlands

New Zealand

SwedenFrance

United States

Num

ber

of 2

8 po

licy

resp

onsi

bilit

ies

Own-source revenue as percent of total revenue

Figure 4.1. Influence of Subnational Governments

The Institutional Environment

©International Monetary Fund. Not for Redistribution

IV

BU

DG

ET

FO

RE

CA

ST

ING

56

Table 4.3. Consolidated Central Government Expenditure Shares, 20031

(In percent of total expenditure)

New United United Rank, Australia Canada France2 Germany Italy3 Netherlands Zealand Sweden2 Switzerland Kingdom2 States Canada

General public services 26.6 30.3 7.7 13.7 20.6 22.1 9.0 23.5 14.3 5.2 12.2 1 of which: Public debt transactions 5.0 9.4 5.9 5.8 13.9 5.5 5.4 7.4 3.6 ... 9.2 2

Defense 6.5 5.8 5.2 3.6 2.6 3.6 2.9 5.7 4.9 7.4 19.1 4 Public order and safety 0.9 3.0 1.7 0.4 4.4 3.7 4.1 3.2 0.6 5.1 1.4 6 Economic affairs 6.2 6.0 10.0 6.7 5.7 6.0 6.8 9.4 11.5 6.9 7.0 8 Environment protection 0.2 0.6 0.2 0.1 1.7 0.4 0.0 0.5 ... ... ... 3 Housing and community amenities 0.7 1.3 1.0 0.9 1.7 0.5 1.5 0.6 0.8 1.3 2.0 4 Health 14.2 2.7 16.9 19.3 12.7 10.4 16.5 2.9 21.1 16.4 23.4 10 Recreation, culture, and religion 0.9 1.5 0.7 0.1 2.0 0.9 2.2 0.8 0.5 1.4 0.2 4 Education 9.3 2.1 10.0 0.4 10.5 10.8 21.2 6.4 2.6 12.5 2.6 9 Social protection 34.5 46.6 44.3 54.8 38.2 41.5 35.9 47.2 43.7 42.5 32.0 3

Memorandum item: Transfers to subnational governments4 27.2 31.4 8.6 14.3 12.9 26.1 1.0 13.7 27.5 19.1 19.1 1

Sources: IMF, Government Financial Statistics; OECD Revenue Statistics; and IMF staff calculations. 1Data are provided on a comparable basis, and therefore are not necessarily consistent with central government figures as reported by individual countries. 22002. United Kingdom: General government. 32000.4Includes general and earmarked transfers. Data for Italy and the United Kingdom are from 2000; data for Sweden, Switzerland, and the United States are from 2001; data for the rest of the countries

are from 2002.

©International Monetary Fund. Not for Redistribution

57

Fiscal Rules and Other Constraints

Fiscal policy rules may improve fiscal discipline, but the costs of violating budget targets may also lead to cautionary biases. Governments that face incentives to improve their budget planning and implementation pro-cesses by implication have better prospects for meeting fiscal forecasts.6 On the other hand, asymmetric conse-

6For example, the introduction of fiscal policy constraints in euro area countries led to the adoption of binding multiyear targets, supplemented with more detailed descriptions of countries’ fiscal plans.

quences for not meeting budget targets may lead to the incorporation of both explicit and implicit prudence fac-tors in the forecast (see, for example, Zellner, 1986).

Unlike in many other countries, fiscal policy in Can-ada is not constrained by budget rules imposed under the constitution or the law (Table 4.4).7 Most advanced countries have adopted some form of rule, including targets for both the overall balance or expenditure, and require embedding fiscal plans within a medium-

7Canada’s “Fiscal Spending Control Act” was in force only between 1991 and 1994.

Equalization transfers are designed to reduce dis-parities in tax-raising capacity between provinces. The transfers are provided as general-purpose block grants, channeling federal funds to provinces with below- average revenue- raising capacity. The definition of “revenue- raising capacity” is based on a comparison between per capita revenue raised and the per capita revenue each individual province could raise if it levied national aver-age tax rates on each of the sources of provincial revenue. Each province’s revenue-raising capacity is then com-pared to that of the average of the five middle-income provinces (British Columbia, Manitoba, Saskatchewan, Ontario, and Quebec) on a per capita basis. Total equal-ization entitlements are determined as:

_ _∑Eij = ∑τj(Bj / P – Bij / Pi) · Pi ,

j j

where Eij = entitlement under revenue source j in province i

Bj = the tax base for revenue source j in the representative provinces

P = the population of the representative provincesBij = the tax base for revenue source j in province iPi = the population of province iτj = the national average tax rate for revenue source j

The size of equalization transfers has been subject to considerable uncertainty. Initially, inputs to the formula determining entitlements are based on estimates for the current fiscal year. As these data are revised in subsequent years—for example, if a new census is taken or final tax revenue data become available—entitlements are modi-fied and positive or negative ex post payments are made (see Table). Between FY 2000−01 and FY 2003−04, the magnitude of adjustments ranged between −21 percent to 8 percent of annual transfers, equivalent to a margin of up to !/8 percent of GDP.

In October 2004, the government announced a new equalization framework. This included a new legislated level of overall equalization entitlements starting in 2005–06, with a built-in growth rate of 3.5 percent annually.

Box 4.1. Equalization Transfers in Canada

Source: Krelove, Stotsky, and Vehorn (1997).

Calculation of 2000–01 Equalization Transfers (in billions of Canadian dollars)

Nfld. P.E.I. N.S. N.B. Que. Man. Sask. Total

Payments through February 2001 1.0 0.2 1.2 1.1 4.3 1.1 0.1 9.0Third estimate (February 2001) 1.1 0.3 1.3 1.2 5.4 1.2 0.2 10.8Fourth estimate (October 2001) 1.1 0.3 1.3 1.2 5.4 1.2 0.3 10.8Fifth estimate (February 2002) 1.1 0.3 1.4 1.3 5.2 1.3 0.3 10.8Sixth estimate (October 2002) 1.1 0.3 1.4 1.3 5.3 1.3 0.2 10.9Seventh estimate (February 2003) 1.1 0.3 1.4 1.3 5.3 1.3 0.2 10.9Final estimate (September 2003) 1.1 0.3 1.4 1.3 5.4 1.3 0.2 10.9

Source: Department of Finance.Notes: Nfld. = Newfoundland; P.E.I. = Prince Edward Island; N.S. = Nova Scotia; N.B. = New Brunswick; Que. = Quebec; Man. = Manitoba;

Sask. = Saskatchewan.

The Institutional Environment

©International Monetary Fund. Not for Redistribution

IV BUDGET FORECASTING

58

term framework.8 The monitoring of these objec-tives is usually accompanied with rigorous reporting requirements for comparing plans with outturns. For example, the EU Commission mandates that Stability Reports include a section on the general economic pol-icy strategy, macroeconomic forecasts, and budgetary projections, as well as a series of standardized tables to enable the evaluation of the projections. In Aus-tralia, New Zealand, and the United Kingdom, fiscal planning is guided by legislation specifically aimed at enhancing transparency and accountability.

Between 1998 and 2006, Canada followed a de facto fiscal rule of “budget balance or better,” with perfor-mance observed on a relatively stringent basis and defined specific fiscal targets aimed at achieving this goal.9 The political commitment to this target, whose asymmetry was derived from long-term fiscal sustainability consid-erations, gave it a role similar to quantitative fiscal policy targets in a rules-based system. The target appeared even stronger than in many countries, both because perfor-mance was observed on an annual basis instead over the medium term and because the target was expressed in nominal terms and thus more difficult to achieve during a downturn than a GDP ratio. Forecasting performance has also been closely monitored and plays an important

8See Kopits and Symansky (1998) and Dabán and others (2003) for a detailed discussion of fiscal policy rules. The EU’s Stability and Growth Pact mandates that deficits do not exceed 3 percent and that the debt-to-GDP ratio remains less than 60 percent. Medium-term targets must be authorized by the legislature in Italy and the United States.

9The conservative government that took power in 2006 abandoned this rule in favor of targeting (without committing to) a moderate surplus of about ¼ percent of GDP.

role in assessing the government’s track record in imple-menting its policy plans.

Canada also adheres to a strict budget planning framework. Along with adoption of the new fiscal tar-gets, fiscal forecasting practices were fundamentally overhauled in the mid-1990s. A key objective of these reforms was to improve the credibility of economic and fiscal forecasts in response to a rapid buildup of public debt. Financial markets had begun to discount the government’s fiscal policy plans because economic assumptions were consistently overoptimistic. A sum-mary description of the current organization of the forecasting process is given in Box 4.2.

During the years under study, Canada placed sig-nificant emphasis on prudent forecasts, which could have affected forecast accuracy. While macroeco-nomic forecasts are obtained from a panel of private sector forecasters, fiscal forecasts contain an explicit cautionary bias—the so-called prudence factor.10 In addition, the budget included a contingency reserve to cushion against unforeseen economic developments. In 2004, the prudence factor and the contingency reserve amounted to Can$1 billion and Can$3 billion, respec-tively, for both the 2004–05 and 2005–06 budget pro-jections. Although on a smaller scale than in Canada, other countries also use cautious economic assumptions or specific reserves. For example, in the Netherlands, formal arrangements have been in place for utilizing

10Prudence was incorporated into the fiscal projections used for the budgets from 1994 until 1998 by explicitly adopting economic assumptions that were more pessimistic than the average of the pri-vate sector economic forecasts, including higher interest rates and weaker economic growth.

Table 4.4. Fiscal Policy Rules and Transparency Laws

Type of Rule _______________________________________________ Fiscal Transparency Deficit/debt Golden rule Expenditure ceiling Law

Australia — — — YesCanada —1 — — —France SGP2 — — —Germany SGP2 Yes — —Italy SGP2 — — —Netherlands SGP2 — Real —New Zealand — — — YesSweden 2% surplus — Nominal —Switzerland — — Nominal —United Kingdom Debt Yes — YesUnited States — — — —

Source: IMF staff.1Canada adopted a fiscal target of “balance or better” from 1998 until 2006. The target was supported by a

strong public consensus, providing many of the characteristics of a fiscal rule.2SGP: Stability and Growth Pact (3 percent deficit and 60 percent debt ceiling).

©International Monetary Fund. Not for Redistribution

59

Fiscal Forecasting Practices in International Comparison

In 1994 and 1995, Canada implemented significant changes to the budget formulation process. The govern-ment adopted a new public expenditure management sys-tem, a two-year rolling planning horizon was introduced, and the forecasting process was revamped. This system was refined in 1999 by publishing five-year fiscal fore-casts in the fiscal midyear reports and by being more explicit about prudent planning assumptions in fiscal forecasts.

For the macroeconomic forecast, the Department of Finance surveys approximately 20 private sector fore-casters each quarter after the National Accounts are released. Average annual private sector forecasts of real GDP growth, inflation, labor market indicators, and interest and exchange rates form the basis of the government’s macroeconomic assumptions. To ensure model consistency, the department may refine these assumptions in meetings with outside economists. The department feeds the assumptions thus gained into its internal macroeconomic model (the Canadian Economic and Fiscal Model) to construct aggregate revenue and expenditure projections consistent with the private sec-tor forecast.

The detailed revenue and expenditure forecast is pro-duced by the Department of Finance and respective spending agencies. Within the department, the forecasts are principally generated by the Fiscal Policy Division, although some smaller elements of the revenue forecast, for example, the low-income rebate of the VAT, are fore-cast by the Department’s Tax Policy Branch using micro-simulation models. Similarly, the department’s Economic Development and Corporate Finance Branch and certain Crown corporations are also consulted and provide infor-mation to help formulate the nontax revenue component of the revenue forecast. Other departments provide spend-ing forecasts based on three-year business plans, which are reviewed by the Treasury Board Secretariat.

Since 1999, five-year fiscal forecasts have been pre-pared by private sector forecasters and are published in the Economic and Fiscal Update published in the fall. These forecasts cover broad fiscal aggregates on a general government basis. Based on this forecast, central govern-ment projections are again provided by the Department of Finance, with the 2004 Update presenting details on how the central government data have been derived from the private sector’s general government forecast.

Box 4.2. Fiscal Forecasting Arrangements

funds from unexpected overperformance of the fiscal balance (Blöndal and Kristensen, 2002).

In addition to fiscal rules, expenditure discretion in Canada is constrained by relatively high debt service costs and other nondiscretionary expenditure. In particular, the share of interest payments is the second highest among the 11 countries, despite the recent decline in public debt, while the share of social protection is the third highest (see Table 4.3).11 Moreover, as noted, the share of transfers to other levels of government is far higher in Canada than in most of the benchmark countries.

Fiscal Forecasting Practices in International Comparison

The importance of fiscal forecasts for budget plan-ning purposes raises process and transparency issues. While solid technical capacities are necessary for high-quality forecast outcomes, forecasting performance also tends to be boosted by an open budget preparation process, including the involvement of nongovernmen-tal agencies, public access to information, and regular reviews of forecasting performance (IMF, 2001). This subsection contrasts technical aspects of Canada’s fis-

11The share of interest payments decreased from 20 percent in 1990 to 9 percent in 2003.

cal forecasting arrangements with other countries and assesses its transparency aspects.12

The role of fiscal forecasts in the Canadian budget process is similar to practices in other benchmark coun-tries (Table 4.5).13 In the majority of these countries, responsibility for budget preparation is assigned to one government agency (the ministry of finance or treasury), but the process usually involves collaboration with other government agencies. Forecasts are framed within a medium-term horizon in all countries, mostly in the form of a rolling three- to five-year forecasting framework (for example, euro area countries are required to prepare five-year fiscal plans), although the period for which fiscal plans are binding, or for which greater detail is presented, is typically much shorter. In Canada, budget preparation is based on a two-year framework, although since 1999 the government has also prepared five-year fiscal forecasts as part of the mid-year fiscal update.

Canada relies more than other countries on macro-economic forecasts by private forecasters (Table 4.6; see

12For a more complete analysis, see Mühleisen and others (2005).

13Sources for this information include country responses to a short staff questionnaire, an OECD and World Bank survey on budget institutions (OECD and World Bank, 2003), and available IMF Reports on the Observance of Standards and Codes. The ques-tionnaire covered the development and organization of the fore-casting process, as well as arrangements for quality control and transparency.

©International Monetary Fund. Not for Redistribution

IV BUDGET FORECASTING

60

Table 4.5. Key Institutional Characteristics of the Fiscal Forecasting Process

Characteristics of the Forecasting ProcessBudget

AuthorityForecasting

HorizonMacroeconomic

forecast Revenue and expenditure

Australia Treasury, Department of Finance and Administration

Rolling three-year budget forecast

Treasury internal: based on extensive consultation process

Government internal; revenue derived from interaction between spreadsheet-based forecast and econometric model; expenditure supplied by spending agencies

Canada Finance Department and Treasury Board Secretariat

Rolling two-year budget forecasts; aggregate fiscal forecasts for five years

Average of private forecasters

Revenue and expenditure: two-year budget forecast prepared internally by experts group and respective spending agencies; five-year forecast in midyear based on forecast of private sector

Germany Ministry of Finance (MoF)

Five-year (SGP) MoF internal: after consultations with forecasting agencies

Revenue: based on consensus among expert group with nongovernmental participation; expenditure: government internal supplied by spending agencies

Netherlands Ministry of Finance

Rolling three-year budget forecast; five years at aggregate level (SGP)

Independent public agency for coalition period; MoF otherwise

Revenue: by independent public agency for four-year coalition plan; MoF internal revenue forecast for individual budget years; expenditure forecasts by spending agencies

Sweden Ministry of Finance

Rolling three-year budget forecast; five years at aggregate level (SGP)

MoF internal: model-driven, benchmarked against other public sector forecasts

MoF internal; revenue model driven benchmarked against other public sector forecasters; expenditure prepared by spending agencies

Switzerland Ministry of Finance

Rolling three-year budget forecast

Forecast by expert group comprising MoF, central bank, and statistical office

Government internal; revenue: iterative process between different departments in the MoF; expenditure supplied by spending agencies

UnitedKingdom

Treasury Five-year budget forecast; aggregate long-term projections

Treasury: iterative process between econometric model and micro-based fiscal forecasts

Government internal; revenue: iteration between Treasury macromodel and micro-based expert models in revenue department; expenditure prepared by spending agencies

France Ministry of Finance

Five-year (SGP) MoF: Forecasting Directorate

Government internal; revenue: iteration between various departments in the MoF; expenditure: forecasts made by the MoF’s Budget Directorate in coordination with spending ministries

Italy Ministry of Finance and Economy

Five-year (SGP) … …

New Zealand Treasury Four-year budget forecast

Iterative spreadsheet-based forecast including views of expert panel, business, and senior staff from Treasury

Government internal: two revenue forecasts prepared and published separately by Treasury and revenue administration; based on micro- and macromodels with consistency check with macroeconomic forecasts and assessment against views of practitioners (tax talks); Treasury forecast used in budget; expenditure forecasts prepared by spending agencies

United States Executive (Office of Managementand Budget)

Five-year budget forecast

… President’s forecast assessed by Congressional Budget Office leading to congressional budget resolution that establishes major fiscal aggregates to constrain the decision making of the appropriations, taxing, and authorizing committees

Sources: OECD and World Bank (2003); country authorities; and IMF country desks.Notes: Forecasting horizon includes budget year. MoF = ministry of finance; SGP = Stability and Growth Pact.

©International Monetary Fund. Not for Redistribution

61

also Box 4.2). In most benchmark countries, the agency responsible for the budget develops its economic forecast in-house, using econometric and spreadsheet-based mod-els. These estimates are often supplemented with infor-mation gained from consultations with nongovernmental forecasters or the business sector. In some countries, no outside agency is formally involved and quality control is left to benchmarking against other forecasting agen-cies (for example, Sweden). The main trade-off between the two approaches is that greater involvement of out-side agencies may boost forecast credibility, whereas a broader consultation process may imply the use of less systematic forecasting techniques, which may make it more difficult to pinpoint the cause of forecast errors.

As in most of the other countries, revenue and expen-diture forecasts in Canada are prepared by the ministry of finance. The degree to which the forecasting process is formalized varies quite significantly across countries. Some countries prepare stylized forecasts with some cross-checks against sectoral and revenue experts (Swe-den, Switzerland). Others use detailed model-driven processes and microdata-based models maintained by technical experts (Australia, France, and the United Kingdom). In Canada, there is little direct involvement of outside agencies in preparing revenue and expenditure forecasts for the annual budget. However, projections for the midyear fiscal update are compiled by a small group of private forecasters, providing an independent view of the medium-term implications of current fiscal policies. Other countries have assigned similar tasks to indepen-dent agencies. For example, the U.S. Congressional Bud-

get Office regularly provides 10-year projections of major economic and fiscal variables, based on fiscal policies as legislated by the U.S. Congress. Australia assesses its fiscal forecast through an extensive consultation process with outside experts and the business sector.

The Canadian public has relatively broad access to budgetary information. A comparison of the detail of published fiscal information shows that Canada scores high relative to countries in the benchmark group (see Table 4.6). The primary budget documents available to the public are the annual Budget Plan (usually released in February or March) and the Economic and Fiscal Update prepared midyear. Both provide economic and fiscal forecasts with detailed explanations of antici-pated future developments. The level and detail of pub-lished information is comparatively high.

However, the closed nature of the budget compilation process implies that forecast risks may not be widely understood, limiting public debate on this aspect. As for many other countries, Canada provides relatively little information on the key assumptions and methods underlying the use of macroeconomic assumptions in budget forecasts, making it difficult for outsiders to distinguish between fiscal forecasting performance and errors arising from implicit prudence factors.14 Some

14Beginning with the 2004 Economic and Fiscal Update, the government committed to provide additional information on how national-accounts-based fiscal projections provided by private sec-tor forecasters translate into the accounting framework used in the budget.

Table 4.6. Fiscal Forecasting: Quality Assurance

Availability of Involvement of

Ex Post Assessment of Information on Fiscal

Nongovernmental Agencies1

Forecasting Performance2 Performance3

_______________________________________ _____________________________________

________________________

Macro Revenue Score on detail forecast forecast Self External and regularity

Australia Medium Low Regular Occasional MediumCanada High Medium Regular Occasional HighFrance Medium Low Regular Regular HighGermany Medium High Occasional Occasional LowItaly Low Low . . . No LowNetherlands Medium Medium Regular No LowNew Zealand Medium Medium Regular Occasional HighSweden Low Low Occasional No LowSwitzerland Low Low Occasional Occasional . . .United Kingdom Low Low Regular, legal requirement Regular HighUnited States . . . . . . Regular . . . High

Sources: OECD and World Bank (2003); and data provided by country authorities.1Nongovernmental agencies play an active role (high), are directly consulted (medium), or are not involved (low).2“Self” refers to analysis of forecasting performance in end-of-year reports; “external” refers to reviews by government audit office or other external

agency.3Measures the number of annual and regularly provided central government reports on fiscal forecasting from the list of reporting items based on

OECD Best Practices. Scores of high, medium, and low refer to the country score relative to the group average (medium).

Fiscal Forecasting Practices in International Comparison

©International Monetary Fund. Not for Redistribution

IV BUDGET FORECASTING

62

countries in the benchmark group are more inclusive in this regard. In Germany, for example, tax revenue fore-casts are based on the consensus of a technical expert group with participation of nongovernmental agencies, providing some assurances that fiscal forecasts are untainted by policy objectives. In Australia and New Zealand, the government is legally required to dem-onstrate, at the time the budget is issued, that budget policies are consistent with long-term fiscal objectives, including by establishing a clear link between policy objectives, forecasts, and outcomes. This requirement has led to a greater emphasis on forecast outcomes, with performance assessments being used to gauge the realism of new budget plans.

Unlike most benchmark countries, the Canadian gov-ernment provides regular and detailed analyses of its fiscal forecasting performance, despite the lack of an explicit legal requirement. Only a few countries mandate such reports on an annual basis (Australia, New Zea-land, and the United Kingdom). However, despite the lack of an explicit legal requirement, the government’s Annual Financial Report analyzes fiscal results for the previous fiscal year, including by listing the sources of deviations from initial forecasts. The government also initiated a comprehensive review of its forecasting performance in 1994. A special task force reviewed the accuracy of the Department of Finance’s economic and fiscal forecasts and their role in the budget plan-ning process, initiating changes in the budget process. A more focused review and consultations with a group of private sector economists in 1999 led to a more explicit treatment of the prudence factor and the introduction of five-year fiscal forecasts beginning with the Economic and Fiscal Update in that year (see Box 4.2).

Assessing Forecast Accuracy

Data problems generally limit the analysis of fiscal forecasting performance across countries. A number of studies have compared the accuracy of macroeconomic forecasts by private sector economists and international organizations (Artis, 1996; Artis and Marcellino, 2001; Ash, Smyth, and Heravi, 1998; Batchelor, 2001; Isiklar, Lahiri, and Loungani, 2005; Loungani, 2000; Öller and Barot, 2000). But most analyses of budget projections have focused on a single country, given the difficulties in obtaining a cross-country data set of budget fore-casts. More recently, two studies have analyzed bud-getary forecasts for a group of relatively homogenous countries (the euro area), with one suggesting that the size of forecast errors may depend on structural charac-teristics of a country’s budgetary framework (Strauch, Hallerberg, and von Hagen, 2004), and the other call-ing for independent budget forecasting agencies on the basis of significant forecast biases (Jonung and Larch, 2004).

Information obtained for this study provided suf-ficient detail to compare recent Canadian central gov-ernment budget forecasts with benchmark countries. At a minimum, most budgets provide three to four years of information for key macroeconomic and fiscal variables, including actual or estimated values for the preceding year, an estimate or projection for the cur-rent year, and projections for one or two future years.15

Most budgets are also compiled near the beginning of a new fiscal year, with the result that the values of eco-nomic and fiscal variables reported for the prior year are generally at or close to their final revision. This allows the use of historical data reported in the budget as a basis for comparison against projections contained in earlier budgets.16 A description of available data is contained in Box 4.3.

Budget projections are evaluated against subsequent budget “actuals,” which provides two advantages over using fully revised values. First, data revisions (caused, for example, by changes in the coverage of government accounts) may be retroactively applied to fiscal outcomes but not to past budget projections. Therefore, revised historical data cannot be used to measure the accuracy of projections made before a revision came into force. Under this analysis’s defi-nition of forecast errors, data losses are limited to at most two to three observations around the time a revision is introduced. Moreover, this method is also “fair” in that it focuses on the information that was available to forecasters at the time and that underlay economic agents’ expectations.

On this basis, a comparison of forecast errors shows notable differences between Canada and other bench-mark countries. For example, projections for real GDP growth in Canada appear to have been on the optimistic side in the early 1990s and more cautious during the high-growth phase in the second half of the 1990s.17

A similar pattern can be observed in the United States, whereas German and Swiss budget forecasters appear to have maintained a more optimistic outlook over time. On the other hand, Canadian fiscal forecasts appear to have been consistently one-sided between the mid-1990s and 2004, whereas most other countries have reported two-sided errors. Before proceeding to a more formal evaluation, however, several words of caution are in order.

15Given the small number of countries providing medium-term projections, forecasts for three and more years were not considered. Also, central government forecasts were not available for a number of countries, and so general government forecasts were used.

16For a description of available data and methodological issues, see Mühleisen and others (2005).

17Errors are defined as projected minus actual values. A negative value therefore implies that the outcome has exceeded expectations, and vice versa.

©International Monetary Fund. Not for Redistribution

63

Data Caveats

Reflecting the idiosyncratic nature of every country’s budget process, the empirical analysis remains com-plicated by data limitations. The most important con-straints are the following:• Time series of consistent forecasts and budget out-

comes are relatively short (often with fewer than 10 observations), limiting the power of statistical tests. Many countries updated their budget formats and forecasting methods in the early to mid-1990s. This has generally increased the level of informa-tion provided but also has resulted in structural breaks related to the adoption of new budget con-cepts and coverage. Although the coverage of rev-enue and expenditure data is broadly similar across most countries, there are limits to how closely they can be compared. For example, while tax categories are relatively similar, some countries include social insurance contributions as government revenues. Moreover, sources for nontax revenues (which may include receipts from asset sales, royalties from natu-ral resources, or frequency spectrum fees, to name a few) tend to differ significantly across countries.

• Comparing expenditure subcategories appears to be particularly difficult. For example, the distinc-tion between discretionary and mandatory spending components—each of which poses a different chal-lenge to budget forecasters—is difficult to obtain for most countries or can only be approximated. Simi-larly, data on transfers to other levels of government are not provided on a consistent basis.

• Checks for internal consistency and structural breaks may not have captured all data anomalies. These checks resulted in the rejection of a considerable number of data points. However, given relatively scant institutional knowledge of the information contained in government budgets more than a few years back, only obvious statistical outliers were eliminated.Importantly, revised forecasts published in midyear

budget updates or other publications are also not considered in this study. In many countries, the government provides updated budget projections in the course of the fiscal year—for example, in Canada’s Economic and Fiscal Update or in conver-gence programs provided by countries in the euro area. Other public bodies (such as the U.S. Congres-sional Budget Office) often conduct complemen-tary analyses of fiscal developments. Including such information, however, would have greatly increased the cost of collecting and preparing a consistent dataset.

This may exacerbate problems caused by policy shifts that are implemented midyear. For example, there was a relatively large U.S. fiscal “error,” which underlines the difficulties in limiting the focus of this study to annual bud-get documents. If negotiations over fiscal measures con-clude considerably later than a budget has been published,

there may be a higher likelihood that policy outcomes dif-fer from underlying assumptions, possibly resulting in a significant deviation of fiscal projections from outcomes. However, such deviations would be policy-driven and not the responsibility of budget forecasters.18

Macroeconomic Forecasts

The remainder of this subsection presents a formal comparison of forecast errors since 1995, separated into macroeconomic and fiscal projections. The mean error (ME) and root mean squared error (RMSE) for one-year forecasts of key macroeconomic variables are presented in Table 4.7. The mean error is the simple average of forcast errors over 1995−2003, providing an indication of the direction of forecast errors. The RMSE, defined as the square root of the mean of the errors squared, is independent of the error sign and therefore is a bet-ter measure of the size of forecast errors. Limiting the sample to the years indicated focuses the analysis on the period during which the current Canadian forecast-ing methodology was in force. Moreover, longer time series were not available for many countries, and 2005 budgets were not yet released in most cases.

The evidence suggests that economic growth in Can-ada has on average been ½ percentage point higher than budget projections in recent years. Canadian projections of nominal GDP and real GDP growth show higher RMSEs than for those of most other countries, and Canadian mean errors are at the negative end among the benchmark countries. Decomposing the RMSE into its two components indicates that this result appears to be mostly a function of the large mean error, given that the standard deviation of Canadian forecast errors has not been as high as for many other benchmark countries. This could suggest that Canadian forecasters adopted a relatively consistent forecast bias, whereas the devia-tions for other countries are spread more equally on the positive and the negative side (statistical tests of this hypothesis are presented in the next subsection).

Canadian forecasters also underestimated GDP infla-tion by 0.2 percentage points on average, but short-term unemployment trends were anticipated quite well. Pro-jection errors for increases in the GDP deflator show a distribution similar to the growth forecast, with high RMSEs and a mean at the negative end among the sample countries. By contrast, the one-year forecast of the unemployment rate exhibited a lower RMSE and a lower (positive) mean error than for other countries.

These findings indicate that Canadian budget fore-casters generally adopted a conservative view of mac-roeconomic developments over the past 10 years. Errors made in forecasting major macroeconomic variables are

18Indeed, the consequences of U.S. tax and spending measures were well anticipated at the time of passage.

Assessing Forecast Accuracy

©International Monetary Fund. Not for Redistribution

IV BUDGET FORECASTING

64

Australia

Annual budgets are usually presented in May, two months before the start of the fiscal year in July. Forecast data begins with the 1984/85 budget.• Budgets present activities of the general govern-

ment, which includes central, state/territory, and local governments.

• Beginning in FY1999/2000, Australia moved from a cash to an accrual accounting basis, but subse-quent budgets reported most items on both a cash and an accrual basis. For the sake of consistency, the dataset uses cash forecasts for all fiscal variables, except interest expenses which were only available on an accrual basis from FY1999/2000 to FY2004/05. In FY1999/2000 and FY2000/01, individual, corpo-rate, indirect, and other taxes are omitted from the dataset because they were not reported on a cash basis.

• From FY1984/85 through FY1993/94, there were no revenue projections beyond the budget year—that is, two-year projections were omitted. Projections for real GDP growth and unemployment were also limited to the next fiscal year.

• Final outcomes for FY1996/97 were not reported in the FY1998/99 budget and had to be substituted with estimates reported in the FY1997/98 budget.

Canada

Data were provided in electronic form by the Depart-ment of Finance. Canadian budgets are usually published in February, two months before the start of the fiscal year on April 1.• Projections for FY2000–01 come from the Budget

Update for FY1999–2000, which was published in October 2000.

• Mandatory expenditures include transfer payments; discretionary expenses are defined as program costs.

• Actual outcomes are generally taken from annual financial reports of the government. Annual finan-cial reports are published sufficiently long after the

close of the fiscal year to properly estimate accruals transactions.

France

Data were provided in electronic form by French national authorities. French budgets are usually published in September, with the fiscal year starting on January 1.• Forecast data begins with FY1996. Personal income,

corporate income, and excise and other tax revenue data are not available for FY1996 and FY1997.

Germany

German budgets are published in September for the next fiscal year starting on January 1.

• Forecast data begins with FY1990. Variables directly affected by the 1990 reunification have been omitted.

• Data on mandatory expenditures comprise government wages and salaries and transfer payments. Discretion-ary expenditures include acquisition of goods and ser-vices and capital spending.

Italy

Italian budget proposals are published in the “Docu-mento di Programmazione Economico-Finanziaria” (DPEF) between May and July, a half-year before the start of the next fiscal year in January. Data provided by the national authorities reach back to FY1989.• Personal and corporate income, excise, and other tax

revenue data were not available.• Central government data for FY2000 and FY2001 were

not available.• For FY1990–FY1998, DPEFs did not report final out-

comes for either fiscal or macroeconomic variables, and so estimated outcomes from the previous budget are used as the final outcomes.

Netherlands

Data were provided in electronic form by Dutch national authorities. Dutch budgets are published in September, preceding the fiscal year starting on January 1.

Box 4.3. Data Overview

internally consistent. On average, growth and inflation were stronger than expected, and unemployment rates lower than anticipated. The projection of nominal GDP also suffers from the fact that Canadian forecasters underestimated base-year GDP by about 1 percent on average—the largest negative value in the benchmark group.19 Macroeconomic prudence adjustment through

19For this study, the base year (or “in-year”) is the year preced-ing the budget year (for example, the base year for the FY2004–05 budget is FY2003–04). Although a similarly large base-year error was only found for the United States, cross-country comparisons involving the GDP deflator suffer from the fact that inflation fore-casts were not available for some countries and had to be calculated

the 1998 budget—affecting about half of all sample years for Canada—is estimated to account for 0.1 percentage points of the mean real growth forecast error and for half as much of the mean GDP inflation error.

Fiscal Forecasts

A similarly conservative approach appears to have been applied to Canada’s fiscal projections. An analysis of revenue and expenditure projections generally finds

as the difference between the nominal and real GDP growth rates, with base-year values substituting for actual values.

©International Monetary Fund. Not for Redistribution

65

Canada among the group of countries with relatively weak forecast accuracy (as measured by the RMSE). Moreover, compared to the benchmark group, the aver-age error takes on one of the largest negative values for revenues and one of the largest positive values for expenditures. Taken together, this implies that Canada has the largest negative mean error for the overall defi-cit forecast, even after allowing for economic prudence and contingency factors.20

20Economic prudence and contingency were categorized neither as revenue nor expenditure, with the result that the discrepancy between projected and actual deficits in Canada is larger than the difference between the revenue and expenditure errors. Redefining

On the revenue side, projections of personal income tax revenue and revenue from the Goods and Services Tax (GST) and the Manufacturers’ Sales Tax (MST) contributed most to the overall forecast error. As far as subcomponents of tax revenue are concerned, Canadian RMSEs are generally not as large relative to those of other countries for aggregate revenues. What makes Canada stand out, however, is that the mean error for all subcomponents is negative, compared to at least one positive error for all of the other five countries for

the projected deficit as the difference between revenue and expendi-ture projections corrects for this.

• Forecast data begin with FY1995 and cover general government.

• Most projections were limited to the one-year time frame.

New Zealand

New Zealand publishes its “Budget Economic and Fis-cal Update” (BEFU) in May, prior to the start of the fiscal year on July 1. Growth and unemployment data were pulled directly from BEFU documents; all other observations came from: http://www.treasury.govt.nz/ fiscaldata/default.asp.• Projection data were available for FY1994/95 through

FY2004/05, except for growth and unemployment pro-jections which begin with FY1998/99.

Sweden

Outcome and projection data for Sweden were taken from “Appendix 2: Svensk Economi” of the annual bud-get bill. The bill is published in September, four months prior to the start of the fiscal year on January 1. Data were available for FY1997 through FY2005, with the exception of FY2000.• Revenues and the fiscal balance were provided on a

general government basis. Budgetary expenditure is on a central government basis.

• Data for personal income, corporate income, and excise and other tax revenue were not available for FY1997 and FY1998.

Switzerland

Data were provided in electronic form by Swiss national authorities. Swiss budgets are published in October, with the fiscal year beginning on January 1.• Forecast data begin with FY1990.• Data for personal income, corporate income, and excise

and other tax revenue were not available.

United Kingdom

The U.K. government usually publishes its “Budget Report” in March, shortly before the start of the fiscal

year in April. Data only cover budgets published under the current framework since FY1997/98.• The “Budget Report” refers primarily to the public sec-

tor, although general government aggregates are shown for most years.

• The current U.K. fiscal framework separates the cur-rent and capital budgets. For consistency purposes, current and capital expenditures were consolidated. Total outlays are the sum of current expenditure and net investment.

• The headline balance concept used was “Net bor-rowing” inclusive of net windfall tax receipts and associated spending (WTAS), asset sales, and depreciation.

United States

Federal government data was obtained from “Histori-cal Tables: Budget of the U.S. Government,” which is usually published in February, eight months before the start of the fiscal year on October 1.• Interest expense is recorded on a net basis.• For FY1984/85 through FY1990/91, mandatory spend-

ing was defined as “total, relatively uncontrollable out-lays” and discretionary spending as “total, relatively controllable outlays.”

• Prior to FY1990/91, nominal output is reported as gross national product (GNP). Beginning with FY1990/91, nominal output is reported as gross domestic product (GDP).

Consensus Forecasts

Private sector forecast data for real GDP growth rate (calendar-year basis) and the headline budget deficit value (in local currency) come from Consensus Eco-nomics, Inc. Consensus Economics publishes updated estimates for the current and next calendar/fiscal year every month. The data for this study are drawn from the month in which authorities released their budget documents.

Assessing Forecast Accuracy

©International Monetary Fund. Not for Redistribution

IV BUDGET FORECASTING

66

Table 4.7. Descriptive Statistics of One-Year Budget Forecast Errors, 1995–20031

Nether- New Switzer- United United Australia Canada France Germany Italy lands Zealand Sweden land Kingdom States

Macroeconomic variablesNominal GDP –0.0066 –0.0236 0.0035 0.0244 0.0002 –0.0111 –0.0037 –0.0271 0.0116 –0.0170 –0.0168 0.0290 0.0344 0.0142 0.0292 0.0138 0.0296 0.0209 0.0357 0.0281 0.0199 0.0390 8 9 8 9 9 9 9 4 9 6 9Real GDP growth –0.0500 –0.4750 0.6833 0.9611 0.1395 0.4778 0.0333 0.7250 0.7078 –0.1500 –0.4333 1.1592 1.7211 1.2364 1.3393 1.8577 1.5830 1.6637 1.4679 1.4698 0.9893 1.6809 9 8 6 9 8 9 6 4 9 5 9 GDP deflator 0.2859 –0.1750 0.1833 0.5611 … –0.1611 … 0.2557 0.5762 0.0057 0.1669 1.2536 1.0522 0.3623 0.7792 … 0.6889 … 1.2417 0.9609 0.9038 0.3510 9 8 6 9 0 9 0 4 9 5 9 Unemployment rate 0.4000 0.0875 … … 0.2333 0.3500 0.2000 –0.3500 … … 0.2778 0.6638 0.2834 … … 0.4447 0.6005 0.5797 0.6265 … … 0.8149 9 8 0 0 3 9 6 4 0 0 9

Fiscal variablesGovernment revenue –0.0154 –0.0379 0.0105 0.0155 –0.0180 –0.0278 –0.0175 –0.0329 –0.0209 –0.0085 0.0027 0.0466 0.0620 0.0313 0.0464 0.0280 0.1232 0.0288 0.0351 0.0840 0.0276 0.0921 8 9 6 9 6 6 8 4 9 6 9 Tax revenue –0.0207 –0.0292 0.0086 0.0226 … 0.0024 0.0001 –0.0409 … –0.0055 0.0049 0.0510 0.0569 0.0286 0.0507 … 0.0542 0.0244 0.0430 … 0.0262 0.0993 8 9 6 9 0 9 9 4 0 6 9 Personal income tax 0.0093 –0.0273 … 0.0605 … 0.0199 –0.0063 –0.0257 … –0.0194 –0.0145 0.0234 0.0537 … 0.1032 … 0.0713 0.0215 0.0360 … 0.0435 0.1524 2 9 0 9 0 6 9 4 0 6 9 Corporate income tax –0.0686 –0.0694 … 0.1352 … 0.0388 0.0371 –0.0387 … 0.0065 0.0987 0.1068 0.1652 … 0.4788 … 0.1803 0.1035 0.2194 … 0.1093 0.2340 2 9 0 9 0 6 9 4 0 6 9 Social insurance taxes –0.0885 … … … … … … … … –0.0168 –0.0004 0.1486 … … … … … … … … 0.0234 0.0277 2 0 0 0 0 0 0 0 0 6 9 Indirect taxes –0.0276 –0.0160 … 0.0349 … 0.0087 0.0015 –0.1304 … –0.0017 0.0772 0.0407 0.0603 … 0.0926 … 0.0357 0.0455 0.2043 … 0.0078 0.1039 2 9 0 9 0 6 9 4 0 6 9 Other revenue –0.0434 –0.1861 … –0.0550 … –0.3883 –0.2091 0.0343 … –0.0649 0.0642 0.1244 0.2350 … 0.1589 … 0.5502 0.2592 0.0730 … 0.1695 0.2241 8 9 0 9 0 5 8 4 0 6 9Government expenditure –0.0062 0.0082 –0.0111 –0.0007 0.0076 –0.0172 0.0022 0.0082 0.0110 0.0072 0.0027 0.0288 0.0258 0.0178 0.0234 0.0261 0.0678 0.0092 0.0146 0.0222 0.0100 0.0209 8 9 6 9 6 6 8 7 9 6 9 Mandatory expenditure … –0.0020 … –0.0225 … … … … … … 0.0159 … 0.0435 … 0.0394 … … … … … … 0.0314 0 9 0 9 0 0 0 0 0 0 9 Discretionary expenditure … –0.0051 … 0.0568 … … … … … … –0.0221 … 0.0362 … 0.0715 … … … … … … 0.0340 0 9 0 9 0 0 0 0 0 0 9 Interest expenditure –0.0750 0.0245 … 0.0187 0.0079 –0.0131 –0.0200 0.0364 … 0.0093 0.0295 0.1040 0.0458 … 0.1381 0.0566 0.1260 0.0501 0.1503 … … 0.0816 9 7 0 9 6 9 9 4 0 1 9 Fiscal balance –0.8025 –6.5427 1.9792 1.5599 –2.4218 0.7900 –1.9792 –0.0811 –3.0378 –1.2985 0.0711 5.6913 7.9428 3.6246 5.0669 3.7109 4.2089 3.2954 2.5785 8.7606 3.2585 10.8211 8 9 8 9 6 6 8 4 9 6 9

GDP ratiosGovernment revenue –0.1375 –0.2723 0.1727 –0.0934 –0.3433 –1.5255 –0.5984 –0.3000 –0.3643 0.3264 0.4333 0.6645 0.7071 0.4355 0.4248 0.6721 3.3993 1.0994 1.4663 0.8941 0.8507 1.4397 8 9 6 9 6 5 8 4 9 6 9

Government expenditure –0.0875 0.5204 –0.1671 –0.2888 0.4600 –1.0793 0.0848 0.2883 –0.0076 0.9485 0.3000 1.0256 0.7185 0.3747 0.5317 1.3600 1.9885 0.6233 0.5278 0.4187 1.0896 1.0654 8 9 6 9 6 5 8 7 9 6 9

Fiscal balance –0.1111 –1.1146 0.3625 0.1954 –0.7867 –0.3106 0.1331 –0.0250 –0.3567 –0.5122 –0.0778 1.3950 1.3637 0.6626 0.5926 1.2274 1.9998 0.4821 1.2013 1.0577 1.2705 2.0367 9 9 8 9 6 5 8 4 9 6 9

Source: IMF staff calculations. 1For each variable, rows list mean error, root mean square error, and number of observations. Errors are calculated in percent of actual outcomes,

except for forecasts of GDP growth, GDP inflation, the unemployment rate, and GDP ratios, where simple difference was taken. Error in forecasting fiscal balance is expressed in percent of average of actual revenue and expenditure. Positive error indicates that the forecast was higher than outturn.

©International Monetary Fund. Not for Redistribution

67

which similar data are available. It is the accumulation of small but persistently negative errors, rather than large forecast errors, that make Canadian forecasters appear relatively pessimistic.

Deviations on the expenditure side appear partly driven by smaller-than-expected debt servicing costs. For all countries, expenditure forecasts have been significantly more accurate than revenue forecasts, as evident from substantially lower MEs and RMSEs. Canada has been no exception as far as mandatory and discretionary expendi-ture items are concerned. However, interest payments were on average 2 percent lower than projected, leading to an average forecast error of 0.1 percent of GDP.21

Even when scaled by the size of GDP, Canadian fiscal forecasts appear unusually conservative. When forecast errors are defined as the difference between actual and projected GDP ratios, Canada still has the largest negative ME compared to the benchmark group, although the RMSEs are in a more moderate range. Canada may have been helped by the fact that forecast accuracy improves once revenues are expressed as GDP ratios, given that the elasticity of tax revenues is close to unitary in many countries. On the other hand, projec-tions of expenditure-to-GDP ratios suffer particularly from GDP forecast errors as nominal expenditures tend to be more closely in line with budget targets.

Statistical Analysis of Forecast Outcomes

This subsection uses statistical tests to further explore the forecast characteristics. First, tests are used to check for the presence of a forecast bias and to check whether projections are efficient in the sense that they use all information available at the time of the forecast. Second, budget projections for GDP growth and the fis-cal balance are compared with private sector consensus forecasts. Third, using structural information described previously, country data are pooled to test whether vari-ables describing the forecasting environment have a significant impact on projection outcomes.

Bias and Efficiency Tests

A series of statistical tests confirms a forecasting bias in some components of Canada’s macroeconomic and fiscal forecasts (Table 4.8). The tests suggest that, between 1995 and 2003, the mean and median of the forecasts for nominal GDP as well as for total and non-tax government revenue were significantly different from zero.22 This places Canada in a group with Ger-

21The forecast error for debt service charges also stems partly from a prudence adjustment to the interest rate forecast in the late 1990s, although this effect could not be quantified.

22See Mühleisen and others (2005) for a description of the tests.

many, New Zealand, Sweden, and the United Kingdom, which all exhibit a consistent bias in either the macro forecast or aggregate fiscal revenues or expenditures. By comparison, Australia, France, Italy, the Netherlands, and the United States are largely free of such bias.

The tests also underline that it is the aggregation of small, unidirectional forecast errors that leads to an overall bias in growth and revenue estimates in Canada. For example, both real GDP growth and GDP inflation forecasts have a negative mean error that is not statisti-cally different from zero. However, the hypothesis of a zero nominal GDP error (to which both the growth and inflation error contribute) is clearly rejected. Similarly, the mean errors of individual tax revenue components were not significant at the 10 percent level, unlike the statistically significant aggregate revenue forecast error. Nontax revenues, which account for about 10 percent of total revenues, also appear strongly downward biased.

Errors in the output projection tend to explain a sub-stantial share of revenue errors across most countries, including Canada. In a second battery of mean tests, forecast errors for macroeconomic variables were added to the right-hand side of the test regression. Whereas inflation and unemployment rate forecast errors failed to affect test outcomes, either nominal GDP or real growth errors eliminated much of the apparent bias in revenue forecasts across most countries. In the case of Canada, the null hypothesis of unbiased forecasts was no longer rejected once nominal GDP errors were included, suggesting a close approximation of the country’s tax base.23 Given the typically small share of unemployment assistance and other cyclically sensi-tive components in total government expenditure, it is not surprising that macroeconomic variables appear to have a lesser influence on the outcome of expen-diture projections, with exceptions including Sweden and Switzerland and some spending components in the United States and Germany.

Finally, tests of forecast efficiency suggest that Cana-dian budget forecasters may not have employed all of the information available. Under an “efficient” forecast-ing process, forecasters would update their forecasting models to take into account any source of systematic forecast errors, such as a permanent improvement of a country’s growth prospects. As a result, forecast errors would at least be independently if not normally distributed. This hypothesis is rejected for Canadian growth and revenue estimates, as well as for a number of variables for Germany, the Netherlands, the United Kingdom, and the United States (Table 4.9). Consistent with the results of this test, Canada is also one of the

23Among countries with a significant nominal GDP coefficient, the measured elasticity of revenue errors was between 1¼ and 2, with Canada in the middle (1½) and the United States at the high end.

Statistical Analysis of Forecast Outcomes

©International Monetary Fund. Not for Redistribution

IV

BU

DG

ET

FO

RE

CA

ST

ING

68

Table 4.8. Results of Forecast Error Median and Mean Tests1

Nether- New Switzer- United United Australia Canada France Germany Italy lands Zealand2 Sweden2 land Kingdom States

Nominal GDP – (8) SWVCc (9) – (8) SWVCc (9) – (9) – (9) – (9) C (4) – (9) SWVC (6) – (9)Real GDP growth – (9) – (8) – (6) SWVCc (9) – (8) – (9) – (6) – (4) – (9) – (5) – (9)GDP inflation3 – (9) – (8) – (6) Cc (9) … – (9) … – (4) Cc (9) – (5) c (9)

Unemployment rate3 C (9) – (8) … … – (3) C (9) – (6) – (4) … … – (9)

Government revenue – (8) WVCc (9) – (6) – (9) Cc (6) – (6) WVC (8) VC (4) – (9) – (6) – (9)Tax revenue – (8) – (9) – (6) – (9) … – (9) – (9) VC (4) … – (6) – (9)

of which: Personal income tax3 – (2) – (9) … C (9) … – (6) – (9) C (4) … – (6) – (9) Corporate income tax3 – (2) – (9) … – (9) … – (6) – (9) – (4) … – (6) – (9) Social insurance taxes3 – (2) … … … … … … … … C (6) – (9) Indirect taxes3 – (2) – (9) … – (9) … – (6) – (9) – (4) … – (6) C (9)

Other revenue – (8) SWVCc (9) … – (9) … SWVCc (5) SWVCc (8) – (4) … – (6) – (9)

Government expenditure – (8) – (9) Cc (6) – (9) – (6) – (6) – (8) SWVc (7) SW (9) WVC (6) – (9)Mandatory expenditure … – (9) … C (9) … … … … … … WV (9)Discretionary expenditure … – (9) … SWVCc (9) … … … … … … SWVC (9)Interest expenditure SWVCc (9) – (7) … S (9) – (6) – (9) c (9) – (4) … – (0) – (9)

Fiscal balance3 – (8) C (9) – (8) – (9) C (6) – (6) C (8) – (4) – (9) – (6) – (9)

Mean tests including growth terms4

Government revenue – (8) Gg (9) ng (6) g (9) Nng (6) nGg (6) NG (8) NG (4) Gg (9) n (6) – (9)Tax revenue – (8) g (9) ng (6) – (9) … g (9) – (9) NG (4) … – (6) – (9)Other revenue n (7) Gg (9) … – (9) … NnGg (5) NnGg (8) – (4) … – (6) – (9) Government expenditure – (8) g (9) Nng (6) – (9) – (6) – (6) – (8) N (4) – (9) g (6) – (9) Mandatory expenditure … – (9) … – (9) … … … … … … – (9) Discretionary expenditure … – (9) … g (9) … … … … … … NG (9) Interest expenditure NnGg (8) Nn (7) … – (9) g (6) – (9) n (9) – (4) … … – (9)

Source: IMF staff calculations. See Mühleisen and others (2005) for a description of the underlying methods.1Letters indicate tests that reject a zero median or mean at the 10 percent significance level. (1) Median tests: S = sign test, W = Wilcoxon test, V = van der Maerden test; (2) Mean tests: C = regression

on constant, c = C with AR(1) term. The number of observations is listed in parentheses for each cell.2Test with AR(1) terms and robust residuals were calculated only for variables with more than four observations.3Mean test only.4Letters indicate tests that reject a zero mean at the 10 percent significance level. N = regression on constant and nominal GDP forecast error, n: N with AR(1) term, G = regression on constant and real

GDP growth forecast error, g = G with AR(1) term.

©International Monetary Fund. Not for Redistribution

69

Table 4.9. Results of Efficiency Tests

Nether- New Switzer- United United Australia Canada France Germany Italy lands Zealand2 Sweden2 land Kingdom States

Joint significance tests1 Nominal GDP –(8) FC (9) –(8) FC (9) C (9) C (9) –(9) –(4) –(9) FC (6) –(9)

Real GDP growth –(9) C (8) –(6) FC (9) FC (8) –(9) –(6) –(4) –(9) C (5) FC (9) GDP inflation2 –(9) –(8) FC (6) FC (9) … –(9) … –(4) FC (9) –(5) –(9)

Unemployment rate2 FC (9) –(8) … … C (3) FC (9) –(6) –(4) … … –(9)

Government revenue –(8) C (9) –(6) –(9) C (6) FC (6) –(8) FC (4) –(9) C (6) C (9) Tax revenue –(8) –(9) C (6) –(9) … –(9) –(9) FC (4) … –(6) C (9)

of which: Personal income tax2 … –(9) … FC (9) … –(6) –(9) FC (4) … C (6) –(9) Corporate income tax2 … C (9) … –(9) … FC (6) –(9) –(4) … C (6) FC (9) Social insurance taxes2 … … … … … … … … … FC (6) –(9) Indirect taxes3 … –(9) … –(9) … FC (6) –(9) –(4) … –(6) FC (9)

Other revenue –(8) FC (9) … FC (9) … FC (5) FC (8) –(4) … FC (6) FC (9)

Government expenditure –(8) –(9) –(6) –(9) –(6) –(6) –(8) –(7) –(9) FC (6) C (9) Mandatory expenditure … –(9) … –(9) … … … … … … FC (9) Discretionary expenditure … –(9) … FC (9) … … … … … … FC (9) Interest expenditure FC (9) –(7) … –(9) –(6) FC (9) –(9) C (4) … … –(9)

Error autocorrelation2 Nominal GDP — — — — — 3 — — — — —Real GDP growth — — — — 1 — … — — — —

Government revenue 1 3 — — — — — — — — 2Tax revenue — 2 3 — … — — — … — 3Other revenue — 3 … 1 … — — — … — —

Government expenditure — — — — 2 — — — — — —

Source: IMF staff calculations. See Mühleisen and others (2005) for a description of the underlying methods.1Letters indicate which tests reject the joint hypothesis of a zero constant and unity coefficient in a regression of actual values on a constant and one-year forecasts at the 10 percent significance level.

F = F-test assuming i.i.d. normal residuals. C = chi-square test. The number of observations is listed in parentheses for each cell.2Test reports longest lag for which autocorrelation in error terms was found (with a maximum of 3). This test was run with data going back to 1990.

Sta

tistical A

nalysis o

f Fo

reca

st Ou

tcom

es

©International Monetary Fund. Not for Redistribution

IV BUDGET FORECASTING

70

few countries to exhibit strong autocorrelation in both tax and nontax revenue errors.

Budget versus Private Sector Forecasts

One measure of comparing budget forecasts against each other is to study how they hold up against private sector forecasts in the respective country. For that pur-pose, one-year budget forecasts were compared with consensus projections for growth and the fiscal balance, taken from the month when the corresponding budget was released. Descriptive statistics for consensus pro-jection errors reveal that their magnitude is generally close to those for budget forecast errors and that neither growth nor fiscal forecast errors are consistently larger for public or private forecasters across countries.

Differences in government and private sector forecast errors in Canada are relatively small. Private sector forecasts exhibit a slightly smaller RMSE for growth and fiscal forecasts than those of the govern-ment, similar to the cases of Italy and New Zea-land. Although the difference in the growth forecast appears rather minor—reflecting the fact that budget forecasts are largely based on macroeconomic pro-jections provided by private forecasters—the test of RMSE equality is rejected at relatively high confi-dence levels. As for the fiscal forecast, anecdotal

evidence suggests that the private sector is usually focusing on the underlying budgetary balance (the simple difference between federal revenues and expenditures, excluding the economic prudence and contingency reserve; Figure 4.2). The difference in RMSEs indeed becomes statistically insignificant once that concept is used.

Tests for statistical dominance have also proved inconclusive. While a visual inspection suggests that the difference between the two sets of projections is small relative to the magnitude of the overall error, a formal test can also be used to analyze whether one of the forecasts statistically encompasses the other. These tests often yield inconclusive results—such as when coefficients are estimated with similar magnitude but opposite sign, as in the case of the Canadian growth forecast. The fiscal forecast contained in Canada’s bud-gets appears somewhat weak relative to the consensus, but the only clear-cut cases of statistical dominance relate to fiscal forecasts in Italy and New Zealand, where the private sector appears to have a clear edge over the government, and vice versa in France.

Factors Affecting Forecast ErrorsThis subsection relates forecast performance to

major characteristics of the fiscal environment, as well

Source: IMF staff calculations.

–15

–12

–9

–6

–3

0

3Consensus

Forecast error for the operational balance (excluding prudence and contingency reserves)

Forecast error including prudence and contingency reserve

0302012000999897961995

Figure 4.2. Canada: Fiscal Balance Forecast Errors(Forecast error in percent of size of government)

©International Monetary Fund. Not for Redistribution

71

as to measures of underlying economic volatility. The approach follows Strauch, Hallerberg, and von Hagen (2004), who analyzed whether budget forecasts by euro area countries were influenced by elections or institu-tional factors. Accordingly, some of the information col-lected in subsections of this paper has also been used for empirical testing (a list of variables is contained in Table 4.10).

The paper also tests the hypothesis that strong fluctu-ations in a country’s economy could affect the accuracy of budget forecasts. For example, commodity-exporting countries such as Australia, Canada, and New Zealand could be expected to suffer from larger and more fre-quent exogenous shocks than other countries. Given eco-nomic models’ limitations in predicting turning points, this could make economic projections more difficult.

Indeed, Canada has experienced greater macroeco-nomic volatility than many other countries:

• Overall, Canada registered the third highest output volatility among benchmark countries between 1990 and 2003. Short-term interest rates also fluctuated relatively strongly during that period, but other mac-roeconomic variables, including consumer price infla-tion, business sector wages, and the nominal effective exchange rate, remained comparatively stable.

• However, fiscal aggregates have not been signifi-cantly more volatile in Canada than in other coun-tries. Volatility in Canada’s expenditure-to-GDP ratios was higher than in many benchmark countries. This could partly reflect policy-induced changes in the expenditure ratio, such as cutbacks in spending on economic affairs (subsidies) and social protec-tion related to consolidation in the 1990s, as well as sharp reductions in public debt payments. By contrast, Canada’s revenue volatility (measured rela-tive to the size of GDP) has been lower than in any

Table 4.10. Potential Factors Affecting Forecast Outcomes

Variable Description

Federal structure (dummy variable) Presence of a federal political structure.

Fiscal rule (dummy) Presence of a fiscal rule (see Table 4.4).

Expenditure ceiling (dummy) Presence of format expenditure ceiling.

Deficit ceiling (dummy) Presence of a formal deficit ceiling.

Appropriation Share of budget expenditure subject to appropriation (midpoint of range; see Table 4.1)

Regulatory framework (dummy) Number of aspects regulated by the constitution or by law (see Table 4.1).

Budget reporting Number of OECD Best Practices met (see Table 4.1).

Accountability framework (dummy) Positive response to the question whether a formal comparison is made between the medium-term fiscal policy objectives and the government’s annual budget with explanations given for any deviations.

Performance assessment (dummy) Regular, occasional, or no external ex post assessment of forecasting performance (see Table 4.6).

Budget lead time (dummy) Average number of months between submission of the budget and the budget vote (see Table 4.1).

Prudential framework (dummy) Combination of "Prudential framework 1" and "Prudential framework 2."

Prudential framework 1 (dummy) Positive response to the question whether there is an explicit “prudence” factor built into the economic assumptions which reduces the final economic estimates by a set amount.

Prudential framework 2 (dummy) Positive response to the question whether growth assumption underpinning the medium-term fiscal framework contains a margin of “prudence” vis-à-vis the forecast.

Stable tax revenue Average share of personal income, social security, and indirect tax revenue in total revenue (1991–2002).

Mandatory expenditure Average share of mandatory expenditure in total central government expenditure.

Transfers Share of transfer payments to subnational governments in total central government expenditure (see Table 4.3).

Sources: OECD and World Bank (2003); and IMF staff calculations.

Statistical Analysis of Forecast Outcomes

©International Monetary Fund. Not for Redistribution

IV BUDGET FORECASTING

72

of the other 10 countries—with the exception of cor-porate income tax revenue, which may have been particularly affected by export volatility.24

The results suggest that structural characteristics of the fiscal environment have limited explanatory power for cross-country differences in forecast errors. For the most important variables contained in budget forecasts, a series of simple OLS regressions of MEs and RMSEs on a constant and one of the structural variables yields few significant results.25 The conser-vative stance of Canada’s forecasts is consistent with some of the findings, but there are also counterintui-tive relationships:• There is some evidence that stronger accountability

reduces the RMSE for the growth and tax revenue fore-cast, but that a federal structure has the opposite effect.

• In countries where the budget is presented to the leg-islature early, revenues appear to be harder to fore-cast. However, this result may be influenced by a coincidence with recent policy shifts in the United States, which has the largest budget lead time.

• There is weak evidence that deficit and expenditure ceilings coincide with conservative revenue estimates.

• Fiscal rules are associated with overly optimistic forecasts, although the same applies to countries with a high share of voted appropriations. A higher share of mandatory expenditure is positively correlated with the forecast error for government spending.26

On the other hand, the evidence that forecasts tend to be more conservative in the presence of macroeconomic and fiscal volatility is relatively strong. In particular, a more volatile GDP growth environment pushes growth and, by implication, revenue forecast errors downward while leaving expenditure forecasts unaffected.

In some equations, volatility indicators and institu-tional features were found to be jointly significant. A combination of growth volatility and prudence indi-cators was found to provide the best explanation for fluctuations in MEs and RMSEs across benchmark countries, with volatility being consistently and more strongly significant across the range of regressions car-ried out. Paradoxically, a more formalized accountabil-ity framework and stricter requirements for assessing fiscal policy were found to be associated with overly

24For comparing volatility across countries, fiscal aggregates have been divided by GDP. Sources of volatility include policy changes, such as enhanced public expenditure programs in the United King-dom after 2000, expenditure cuts in Canada or Sweden during the 1990s, or tax cuts in the United States. The results are not corrected for this fact, both because it can be argued that volatility stemming from policy changes also contributes to a more difficult forecasting environment, and because estimates of non-policy-induced volatility are not available for most countries.

25Each of these regressions is run with a maximum sample of only 11 observations, depending on the number of countries for which information was available.

26The results are robust in the sense that they hold even if different countries are removed from the sample.

optimistic expenditure forecasts. This may be due to “adverse selection”—formal accountability may have been strengthened particularly in countries with expen-diture discipline problems.

It remains unclear whether these findings can fully explain the difference between forecast errors in Can-ada and other countries. On the one hand, the existence of a mean error/bias for growth and revenue forecasts in Canada appears to be fully explained by a combina-tion of prudence indicators and macro volatility. For example, the predicted value for the mean error of Can-ada’s nominal GDP forecasts is close to the actual value (Figure 4.3), suggesting that forecasters in other coun-tries would likely arrive at the same outcome if they were operating in Canada’s forecasting environment. On the other hand, the RMSE—which is a better mea-sure for overall forecast quality—appears little affected by macro volatility, and Canada remains the country with the second highest residual in the bottom chart of Figure 4.3. Further research—based on more com-prehensive data and more refined economic models—would be needed to shed greater light on the relationship between the fiscal forecasting environment and forecast accuracy.27

Conclusions

This study suggest that, between the mid-1990s and 2004, Canadian budgets followed a cautious forecasting approach. A descriptive analysis shows Canada with larger and more conservative fiscal forecast errors than most other countries. The study also finds that Cana-da’s aggregate forecast error is composed of small but consistently one-sided errors in fiscal subcomponents, which appears characteristic of a conservative forecasting approach.

A considerable part of this outcome appears to be related to a forecast bias in the macroeconomic com-ponent. This finding may be partly a consequence of Canada’s economic environment at the time, given the link between macroeconomic volatility and pessimistic growth projections. Moreover, Canadian forecasters were not unique in underestimating the global boom of the late 1990s. Although prudence adjustments in budgets of the middle to late 1990s also led to a slight increase in forecast errors, macro projections were likely affected by the fact that Canada unexpectedly outperformed other industrial countries throughout much of the period.

However, other factors are also likely to have played a role. Budget forecasters have had to cope with con-

27Panel estimations are particularly affected by data shortcomings and contribute little additional information. However, time dummies for the late 1990s have generally been significant in regressions cov-ering fiscal variables, suggesting that surprises from a strong global growth environment have not been confined to Canada.

©International Monetary Fund. Not for Redistribution

73

siderable uncertainty relating to the size of provincial transfers and tax-sharing arrangements, which were exacerbated by the relatively large size of provincial budgets relative to the federal government. Moreover, the economic literature suggests that a conservative budgeting approach constitutes a rational response to a regime where the costs of missing a fiscal target are both high and asymmetric, as has been the case in Can-ada over the past 10 years.

Canada could benefit from further improving the transparency of its budgetary forecasts. Given the importance of restoring public confidence in govern-ment finances in the mid-1990s, the consequences of running into deficit were considerably higher than those of achieving a surplus. As Canada’s fiscal situation has improved, it is unclear the extent to which the relative costs of missing budget targets have changed. However,

Canada could benefit from opening up the forecasting process, for example, by involving private forecasters in producing revenue estimates. Equally important, providing more information about critical parts of the forecasting process—in particular, the assumptions and methods used for transforming macroeconomic fore-casts into fiscal projections—would invite greater out-side scrutiny, helping to improve forecast quality and bolster public confidence in budget projections.28

28Examples include Australia and New Zealand, which have adopted transparency legislation to boost public understanding of fiscal forecasts, whereas in other countries—such as Germany and the Netherlands—academic bodies or independent agencies participate in the forecast. IMF (2002) also provides suggestions for expanding the information content of the Economic and Fiscal Update.

Source: IMF staff calculations.1Estimation of the mean error between real GDP growth forecasts and the actual results regressed against the volatility of

real GDP growth and the budget lead time.2Estimation of the root mean squared error between real GDP growth forecasts and the actual results regressed against the

volatility of real GDP growth and prudence indicators.

–3

–2

–1

0

1

2

3

UnitedStates

SwedenNewZealand

NetherlandsItalyUnitedKingdom

FranceGermanyCanadaAustralia

–1.5–1.0–0.50.00.51.01.52.02.53.03.54.04.5

UnitedStates

SwedenNewZealand

NetherlandsItalyUnitedKingdom

FranceGermanyCanadaAustralia

Mean Error

Root Mean Squared Error

Actual

Fitted2

Residual2

Actual

Fitted1

Residual1

Figure 4.3. Impact of GDP Volatility on Forecast Quality(Forecast errors of growth rates in annual percent change)

Conclusions

©International Monetary Fund. Not for Redistribution

74

V The Pension System

Martin Mühleisen

Canada has a comprehensive three-pillared pension system (Box 5.1). The public pillars consist of the

Old-Age Security (OAS) benefit, financed through the federal budget, and the Canada Pension Plan (CPP), which is funded partly by employer/employee contri-butions and partly by the returns on accumulated trust fund assets.1 The private pillar of the system consists of corporate pension plans and individual savings plans.

The Canadian system provides a successful model for pension reform, but challenges remain. As a result of a series of reforms introduced since the mid-1980s, most elderly Canadians at the low- to middle-income level are provided with the means to broadly maintain their living standards in retirement. At the same time, public pension benefits remain relatively modest, providing an incentive for middle- to high-income households to accumulate sufficient assets to fund their retirement. With the retire-ment of the baby boom generation expected to begin at the end of the decade (Figure 5.1), this section examines the state of the Canadian pension system and explores the scope for further policy action. Among other issues, it touches on the role of private saving vehicles, gover-nance of corporate pension plans, incentives for labor market participation of elderly workers, and trends in public pension benefit levels.

The Public Pension System

Old-Age Security

The OAS system is targeted mainly at lower-income seniors. The OAS was initially provided as a universal retirement benefit, with additional benefits provided through the attached Guaranteed Income Supplement (GIS) after 1967 and Spousal Allowance (SPA) after 1975 (see Box 5.1). However, fiscal pressures prompted reforms in the mid-1980s, including the introduction of a more progressive benefit structure. Key measures included freezing the maximum pension levels in real terms after 1984 and introducing means tests or “claw-backs” for OAS benefits for higher-income retirees in 1996, which effectively ended the universality of public

1Residents of the province of Quebec are covered by the separate Quebec Pension Plan.

pension benefits. These measures caused benefits to drop in real terms for all but the poorest seniors and placed government spending on OAS on a downward trend relative to GDP after 1994 (Figure 5.2).2

The means tests have come under criticism, however. In particular, GIS and SPA benefits are tested against personal income only, which raises questions about hor-izontal equity, given the advantage the system provides to retirees with working spouses. The sharp clawback rate of OAS-related benefits and the progressive nature of the old-age income tax credit imply relatively large marginal disincentives to work. Finally, the existence of a large number of tax credits and benefits, with dif-ferent clawback rates, has been viewed as unnecessarily cumbersome and complex.

The last attempt at reforming the OAS system in 1996 failed in the face of widespread political opposition. In 1995 and 1996, the government proposed a “Seniors Ben-efit” that would combine OAS, GIS, and age and pension income tax credits. This benefit would have been tested against full family income, and clawback rates would have been even more progressive than for existing benefits.

2For a discussion, see Hoffman and Dahlby (2001) and Battle (2003).

Sources: World Bank; and OECD.

0.4

0.6

0.8

1.0

1.2

United StatesJapanItalyGermanyCanada

2050204020302020201020001990

Figure 5.1. Old-Age Dependency Ratios in Selected G-7 Countries

©International Monetary Fund. Not for Redistribution

75

Taking into account changes in the tax system that were also envisaged, the plan was not expected to have had a significant immediate impact on retirement incomes but would have produced considerable savings over time. However, as a result of strong political opposition, a less-ening of fiscal pressures owing to improving government finances, and a successful CPP reform, the proposal was withdrawn in 1998 (Battle, 2003).

Canada Pension Plan

Successful implementation of a 1998 reform plan has put the CPP on a sound actuarial footing. Prior to the reforms, it was expected that the combined employer/employee contribution rate would need to rise from 5.6 percent of pensionable earnings in 1996 to more than 14.2 percent by 2030 to preserve the system’s long-term actuarial balance. In order to avoid this sharp increase in contribution rates and the associated inter-generational inequities, the reform package contained the following elements (Battle, 2003):• Benefit cuts: A number of modest changes to the way

benefits are calculated (including with regard to pen-sionable earnings, eligibility for disability benefits, and the amount of death benefits) were estimated to reduce annual benefit payments by 2 percent ini-tially.3 However, these measures were expected to

3CPP premiums are tied to an individual’s pensionable earnings, defined as earnings up to Yearly Maximum Pensionable Earnings (YMPE)—which equals the average industrial wage—less the Year’s Basic Exemption (YBE). Prior to the 1998 reforms, the YBE was set at one-tenth of YMPE. CPP benefits are calculated by the following formula: CPP pension = 0.25 × (average YMPE over the previous 5 years) × (average ratio of pensionable earnings to YMPE over 85 percent of the individual’s working life).

reduce total benefit payments—already low by inter-national standards—by almost 10 percent by 2030.

• Base broadening: For the purpose of calculating pen-sion contributions, the yearly basic exemption (YBE) was frozen at Can$3,500. However, with the maxi-mum amount of yearly maximum pensionable earn-ings (YMPE) continuing to rise with average wages, the amount of earnings subject to premium payments was gradually increased.

• Contribution hike: The CPP contribution rate was gradually raised to a “steady-state” level of 9.9 per-cent, with the increase completed in January 2003. At this level, the CPP was deemed to be in actuarial balance, and no further rate increases were expected pending regular actuarial reviews.

• Market investments: The reforms resulted in a signifi-cant buildup of CPP assets, expected to reach around five years’ worth of benefits by 2010 (Figure 5.3). In addition, prior to 1998, CPP surpluses had been lent to the provinces at a subsidized rate—the market rate paid by the federal government—but the reform also included measures to achieve higher rates of return on plan assets. The provinces’ access to CPP surpluses was gradually eliminated, and the CPP Investment Board (CPPIB) was established and tasked with investing future surpluses in a diversified equity and real estate market portfolio. By 2005, management of the CPP’s fixed-income portfolio was also to have shifted from the Finance Department to the CPPIB, allowing public pen-sion assets to be managed in a manner consistent with employer pension funds in Canada and elsewhere.4

4Provinces have the choice of rolling over nonmarketable bonds placed with the CPP one more time until 2033. OSFI (2001) assumes that the ultimate asset mix will consist of 50 percent bonds and 50 percent equity.

Source: Haver Analytics.

1.25

1.50

1.75

2.00

2.25

2.50

2.75

2001969186817671661961

Figure 5.2. Government Spending on Old-Age Pensions(Percent of GDP)

Source: Office of the Superintendent of Financial Institutions (OSFI), 2001.

1

3

5

7

9

11

Reserve assets (right scale)

Contribution rate (left scale)

20806040202000801960

In p

erce

nt

Res

erve

/exp

endi

ture

rat

io

Figure 5.3. Projected Canada Pension Plan Contribution Rates and Pension Reserves

The Public Pension System

©International Monetary Fund. Not for Redistribution

V THE PENSION SYSTEM

76

The 2002 Actuarial Report confirmed the long-term viability of the CPP through 2075 (Office of the Superintendent of Financial Insitutions (OSFI), 2002). The report found that the steady-state contribution rate required to achieve long-term balance was only 9.8 per-cent as of mid-2002 and that the legislated rate of 9.9 per-cent would result in a larger-than-anticipated buildup of surpluses. Consequently, CPP assets are projected to reach six times annual plan expenditures by 2075 (see Figure 5.3). These results appear to be relatively insensi-tive to changes in underlying assumptions.5

CPP assets have grown rapidly, notwithstanding investment losses early in the decade. As of September 2003, CPP assets amounted to Can$64.4 billion (5 per-cent of GDP), of which the CPPIB held Can$27.4 billion in stocks and real estate (Table 5.1). From the begin-ning of operations in October 1999 through the first half of 2004, the CPPIB registered a cumulative loss of Can$1.2 billion in its active portfolio—mostly a result of heavy stock losses between mid-2002 and early 2003. For the most part, stock losses were offset by gains in the CPP’s bond portfolio, but the CPP as a whole also

5The actuarial estimates assume a real rate of return of 4.5 percent and 5 percent on Canadian and U.S. equities, respectively, and a 3.8 percent real return on bonds. Real average earnings are pro-jected to grow at 1.1 percent over the long run, and the steady-state inflation rate is projected at 3 percent. A 20-basis-point increase in contribution rates would be required if the inflation rate were to equal 2 percent, other things equal. Similarly, the contribution rate would have to increase by 40 basis points if the real rate of return were 1 percentage point lower.

registered a loss of Can$1.1 billion in FY2003. Results improved between March and September 2003, with the CPPIB achieving a 14½ percent rate of return and the CPP’s bond portfolio also gaining 5 percent in value.

Private Pension Schemes

Employer Pension Plans

Accounting for about one-third of total household assets, private pension assets are the second most important source of private wealth next to primary residences. As of 1999, households held an estimated Can$604 billion (50 percent of GDP) worth of assets in corporate registered pension plans (RPPs) and Can$408 billion in individual retirement plans (Sta-tistics Canada, 2001). As of 2003, there were some 15,400 RPPs, covering 5.4 million workers, about one-third of the workforce.6 Included in this number are pension plans for employees of the federal, provincial, and municipal governments, which account for almost half of all workers covered by RPPs. Indeed, some 85 percent of public sector employees are members of an RPP, compared to only 31 percent among full-time private workers.

6These plans are typically registered at the provincial level, except for some 1,200 plans of companies operating in federally regulated areas of employment. These plans cover about 550,000 employees and are registered with and supervised by the OSFI (OSFI, 2003).

• The fi rst tier consists of universal basic pension bene-fi ts fi nanced through the federal budget. They include Old-Age Security (OAS); the Guaranteed Income Supplement (GIS); and Spousal Allowance (SPA) for 60- to 64-year-old spouses (the term OAS is also often used to refer to the three benefi ts together). In total, these benefi ts provide a maximum income of about Can$12,000 per year, which is indexed every three months to the CPI. These benefi ts are taxable, but retirees also receive an age-related tax credit. GIS and SPA benefi ts are tax-free, but are reduced (“clawed back”) by 50 cents and 75 cents for every dollar of other income, respectively, which makes the benefi t structure highly progressive. Finally, retirees receive a tax credit worth around Can$1,000, which is reduced for higher-income seniors, and half of all provinces provide additional income supplements to seniors with low incomes.

• The second tier is formed by the Canada Pension Plan (CPP), a compulsory pension plan covering all

employed and self-employed Canadians. Premiums are split equally between employers and employees (the self-employed pay both parts). The maximum CPP benefi t equals one-quarter of the past fi ve years’ average industrial wage; survivor, disability, and death benefi ts are also provided. Benefi ts are indexed to the CPI and are fully taxable, except for a Can$1,000 pri-vate pension exemption. The CPP is a joint federal- provincial program, with changes requiring the agree-ment of two-thirds of the provinces carrying two-thirds of the population.

• The third tier comprises the private pension system, much of which receives favorable tax treatment. These plans include corporate pension plans (Registered Pen-sion Plans, or RPPs) and individual retirement sav-ings vehicles (Registered Retirement Savings Plans, or RRSPs). Although the income tax system provides tax credits for contributions to RPPs and RRSPs, these plans cover only about a third of the workforce and are con-centrated among those with above-average earnings.

Box 5.1. Canada’s Three-Tiered Pension System

©International Monetary Fund. Not for Redistribution

77

The scope of employer-sponsored pension plans has increased in recent years. In most provinces, vesting in RPPs must now occur after only two years of employ-ment, after which plan members become entitled to have employer contributions made on their behalf. Employers also have to offer coverage for part-time workers, sub-ject to certain conditions, and must allow early retire-ment at actuarially reduced pensions. Indeed, half of all plans are thought to offer unreduced benefits in the case of early retirement once minimum service require-ments are met (OECD, 2001a). Portability provisions were also improved, and plans also have to provide survivor benefits. The FY2004 budget also raised ceil-ings for tax-deductible employer contributions in real terms for the first time in 25 years.7 However, fewer than half of RPP members are in plans that provide some form of benefit indexation, and only one in seven plan members enjoys full CPI indexation of pension benefits (Battle, 2003).

Recent years have also seen an increase in the importance of defined-contribution plans, which shift investment risks to employees (Table 5.2). Defined-contribution plans remain the preferred choice of small companies, in part because individual accounts are easily transferred to individual or group retirement accounts.8 By contrast, defined-benefit plans are gener-ally sponsored by large companies such as banks, rail-

7The budget contained an increase in the amount of contributions to defined-contribution plans that can be deducted from taxable income from Can$14,500 to Can$18,000 as of 2005. The maximum pension benefit permitted under defined-benefit plans was raised to Can$2,000 per year of service as of 2005 from the previous ceiling of Can$1,722. Both limits are indexed to average wage growth for subsequent years.

8Two-thirds of pension plans supervised by OSFI are defined-contribution plans.

roads, and telecommunication organizations. While the majority of private pension plan members still belong to defined-benefit plans, employers are increasingly offering members the option of accruing future benefits on a defined-contribution basis and, in some cases, of converting accrued benefits to cash for transfer into a defined-contribution account. However, almost all pub-lic pension plans have remained on a defined-benefit basis, typically offering 2 percent of salary for each year of service.

The financial position of many defined-benefit pen-sion plans has deteriorated as a result of market losses. Comprehensive information is not available, owing to the fragmented nature of pension fund supervision, but several reports suggest that among larger compa-nies with defined-benefit pension plans, the degree of underfunding is similar or slightly worse than in the United States.9 For those plans supervised by OSFI, a solvency test in June 2003 found that about 210 plans (more than half of all defined-benefit plans under OSFI supervision) were not fully funded—that is, the net present value of future pension obligations exceeded the market value of plan assets—compared with about 180 plans in December 2002 . When assess-ing pension assets at liquidation value—a more strin-gent criterion—the aggregate solvency ratio remained around one, reflecting plan surpluses of some large pension plans. Nevertheless, the number of companies on OSFI’s watchlist increased from 50 to around 90 through the first half of 2003. Since then, pension plans are likely to have benefited from the recovery in equity markets, but relatively low long-term interest rates continue to keep the discounted value of future liabilities at a high level.10

OSFI responded to recent developments by strength-ening its regulatory oversight. OSFI tightened regula-tions on contribution holidays, refined its supervisory tools, and took a more proactive stance in dealing with severely underfunded pension plans. For exam-ple, corporate plans are now subjected to semiannual stress tests, and actuarial reports are requested on an annual basis (instead of every three years) as long as plans are underfunded. Moreover, regulatory changes were considered to limit benefit improvements that would unduly affect the solvency ratio of a plan and to require full funding of a pension plan deficit on plan termination.

9According to a widely quoted UBS study, 49 of the largest 60 companies listed on the Toronto Stock Exchange have defined-benefit plans. These plans had a combined funding shortfall of 3.2 percent of market capitalization at end-2002, compared with only 2.4 percent for S&P 500 companies.

10Members of defined benefit plans are not insured against plan insolvency except in the province of Ontario, where the first Can$1,000 per month of pension benefits is covered by a pension benefits guarantee fund.

Table 5.1. CPPIB Operations

Net Investment Income Net Assets ___________________ Rate of (Can$ billions)_____________

Can$ return Of which:Fiscal Year billions (In percent) CPP CPPIB

1999:H2 0.0 5.0 39.1 0.02000 0.5 40.1 41.3 2.42001 –0.9 –9.4 45.7 7.22002 0.3 3.4 51.9 14.32003 –4.2 –21.1 55.6 17.52004:H1 3.1 14.5 64.4 27.4

Sources: CPP and CPPIB Annual Reports. Notes: CPP = Canada Pension Plan; CPPIB = CPP Investment

Board; H1 = first half-year; H2 = second half-year.

Private Pension Schemes

©International Monetary Fund. Not for Redistribution

V THE PENSION SYSTEM

78

Private Retirement Accounts

Registered Retirement Savings Plans (RRSPs) were introduced in 1957 to provide nonparticipants in employer pension plans, such as the self-employed, with a tax-preferred vehicle for retirement savings. RRSP contributions are deductible from taxable income, and income earned within RRSPs is tax-exempt. Cash withdrawals, however, are treated as ordinary income and are taxable in the year of with-drawal. Contributions to a spouse’s RRSP are also tax-deductible, and since 1991 any unused contribu-tion room can be accumulated and carried forward indefinitely. Workers covered by corporate pen-sion plans are also allowed to contribute to RRSPs, but in recent years the contribution limit was set at Can$14,500 per year (including contributions to RPPs), somewhat lower than for corporate plans. The 2004 budget increased the annual cap to Can$18,000 as of 2006 (equal to the RPP limit) and also provided for the cap’s annual indexation in line with average wage growth.

RRSP contributions have dropped in recent years. The participation rate in the RRSP (defined as the percentage of eligible tax filers making a contribu-tion in a given year) steadily increased from the sys-tem’s inception to reach close to 50 percent in the late 1990s, and the amount of RRSP contributions reached 6½ percent of total wages and salaries in 1998. How-ever, both participation levels and average contribu-tion size declined after about 1997 (OECD, 2003). In 2002, close to six million workers contributed about Can$27 billion to an RRSP, down 7 percent from the peak of Can$29.3 billion in 2000. Tax statistics also indicate that only one-third of all tax filers contributed to an RRSP in 2002, out of about 80 percent of filers who were eligible.

RRSPs are most heavily utilized by high-income groups. Marginal incentives to save for retirement are highest at the very low and middle-to-high ends of the

income scale, owing to the sharp reduction of public pension benefits above the low-income threshold. This is confirmed by a recent Statistics Canada study that found relatively high contribution rates both among low- and high-income wage earners (Palameta, 2003). Nevertheless, a Statistics Canada (2001) survey found that 80 percent of those families without any private pension assets earned less than Can$30,000 per year, and households with more than Can$100,000 in accu-mulated retirement savings accounted for 84 percent of all private pension savings.

Pension Incomes and Retirement Trends

Pension Income Levels and Distribution

Canada’s OSA system has achieved considerable suc-cess in reducing poverty among the elderly. Although public expenditures on income security for seniors have remained modest by international standards—and are projected to peak at levels well below those anticipated by other industrialized countries in the next century—low-income rates among elderly Canadians are now among the lowest in the OECD (Smeeding and Sullivan, 1998; Table 5.3).11 The success in rais-ing retirement incomes for poorer Canadians has also been reflected in a reduction in domestic inequalities, with virtually all of the relative income gains since the early 1980s taking place at the lower end of the income distribution (Table 5.4).

The reduction in poverty among seniors reflects a strong increase in public pension benefits, partly offset by declining labor income. Between 1980 and

11The low-income rate is defined as the share of the popula-tion with disposable income less than half the median of the entire population.

Table 5.2. Membership in Registered Pension Plans(In percent of all employees)

1989 1991 1993 1995 1997 1999 2001

Total 45.7 48.5 48.1 45.7 44.6 43.6 43.5Defined contributions 3.8 4.3 4.7 4.8 5.6 5.9 . . .Defined benefits 41.4 43.6 42.9 40.3 38.3 36.9 . . .Mixed 0.4 0.7 0.6 0.7 0.7 0.8 . . .

Public sector 83.5 89.4 87.4 84.6 83.9 82.2 . . .Private sector 33.6 34.1 33.6 32.1 31.4 31.1 . . .

Source: Statistics Canada.

©International Monetary Fund. Not for Redistribution

79

1999, the combined share of retirement income pro-vided by OAS benefits and the CPP rose from one-third to about 43 percent (Table 5.5). A large part of the increase in public pension payouts is related to the maturation of the CPP, but the relatively con-stant share of OAS income suggests that a substantial increase in basic pension benefits has also allowed poorer retirees to keep up with overall income growth. At the same time, the share of employment income declined to only about 10 percent in 1999, reflecting the trend toward earlier retirement and falling labor participation rates among the elderly. Table 5.3 also illustrates that the share of income from private assets has been essentially stable over the past two decades, except that there has been a major shift of privately invested funds into retire-ment saving schemes.

Participation Rates and Retirement Age

The average retirement age for Canadian workers has fallen sharply over the past 20 years. Household survey data suggest that the labor force participation rate of 55- to 64-year-old males has steadily declined from 86 percent in the early 1960s to 60 percent in the 1990s (Hoffman and Dahlby, 2001).12 As a result, Canada has moved to the middle of the retirement age distribution for major industrial countries. On average, Canadians retire earlier (and stay in retirement longer) than work-ers in the United States or Japan, but later than many of their peers in continental Europe (Table 5.6).

The public pension system contains a number of dis-incentives to work beyond the early retirement age. Gruber (1999) illustrates the choice between continu-ing to work and taking up retirement for Canadian men under a range of circumstances. Although the results are somewhat sensitive to the amount of nonpension income received by a retiree, marginal incentives to work drop sharply beginning at age 55 and especially after age 60, when (reduced) CPP benefits kick in. The steepest disincentives are for workers with little spousal and other income, owing to the sharp clawback of GIS benefits.13 However, Gruber also finds that the actuar-

12The respective drop for males aged 65–69 is from 50 percent to 16 percent. For women, the trend toward earlier retirement is gener-ally dominated by an age cohort effect in favor of higher participa-tion in the labor market (Gruber, 1999).

13For example, a worker in the 10th income percentile (no outside income) achieves a replacement ratio from public pension income of 42 percent if retiring at age 62, but would be subject to an effec-tive 16 percent marginal tax rate if continuing to work. By contrast, the replacement rate for workers in the 90th percentile (with out-side income) is 15 percent and the marginal tax rate 4 percent. At

Table 5.3. Old-Age Income and Labor Market Participation

Low-Income Rate of Elderly Relative Disposable (Percent of the elderly Income of the Elderly Participation Rate, 2001

with income less than (Percent of the Private Pension (In percent) ______________________________

50 percent of median disposable income Funds, 1999 Aged 55–64 _______________ disposable income) of all individuals) (Percent of GDP) Aged over 65 Male Female

Australia 16.1 67.6 63.8 6.0 60.0 36.9Canada 2.5 97.4 45.7 6.0 61.3 41.7France 10.7 89.7 6.3 1.2 43.8 34.1Germany 10.4 85.6 3.2 3.0 50.6 32.4Italy 15.3 84.1 3.0 3.4 57.8 26.6Japan . . . . . . 18.7 21.8 83.4 49.2Netherlands 1.9 86.3 119.3 3.1 52 26.9Spain 11.3 . . . 2.3 1.6 61.4 23.6Sweden 3.0 89.2 . . . 9.4 73.5 67.4United Kingdom 11.6 77.8 84.1 4.8 64.4 44.6United States 20.3 91.7 74.4 13.1 68.1 53.0

Source: OECD (2003).

Table 5.4. Distribution of the Elderly by Population Income Quintile, 1980–95(In percent)

ChangeQuintile 1980 1990 1995 1980–95

Bottom 39.7 25.2 17.5 –22.22nd 22.1 29.7 32.5 10.53rd 12.2 16.2 20.0 7.84th 13.3 14.9 16.0 2.7Top 12.8 13.9 14.0 1.2

Source: Myles (2000).

Pension Incomes and Retirement Trends

©International Monetary Fund. Not for Redistribution

V THE PENSION SYSTEM

80

ial adjustment of CPP benefits for workers retiring after the statutory retirement age of 65 is insufficient.14

Are Canadians Saving Enough for Retirement?

Lise (2003) suggests that Canadian pension arrange-ments will remain adequate for the foreseeable future, with even lower-income Canadians able to meet basic needs. On the basis of cohort-specific income and con-sumption data, Lise finds that elderly Canadians expe-rience no sudden decline in consumption as they enter retirement, implying that existing pension arrange-ments are sufficient to maintain living standards in old age. These findings, which are robust across different income quartiles, are supported by evidence that sav-ing rates are also broadly maintained and that even low-income households continue to make gifts as they age. Lise (2003) also observes that the saving behavior of those expected to reach retirement during the next two decades has so far been similar to that of previous generations. Therefore, he concludes, future pensioners should also be able to maintain consumption and living standards in retirement.

Nonetheless, another study suggests that a large share of Canadians have failed to amass sufficient retire-ment savings to avoid a dependence on public pension schemes. A survey by Statistics Canada (2001) indicates that about 3½ million family units (close to one-third of all families) had no private retirement savings at all. Moreover, even among families with the household head approaching retirement (aged 45 and above), one-third

age 65, the low-income (high-income) case receives a pension of 124 (32) percent of income at a 64 (18) percent marginal tax rate.

14CPP benefits are reduced by 0.5 percent for each month that they are received before age 65 and increased by 0.5 percent for each month that retirement is postponed after age 65.

may not have saved enough to replace two-thirds of their earnings or secure an income above the poverty line. To be sure, this group includes a significant number of high-income households that may experience lower income after retirement but are still able to live rela-tively comfortably. Yet about one-quarter of households with income between Can$20,000 and Can$40,000 were also found to have insufficient savings and thus will likely remain dependent on retirement income from public sources. These findings raise concern, even though this study did not consider nonfinancial retire-ment assets or the continued accrual of benefits under defined-benefit pension schemes, which may alleviate the financial situation of many households.

Other factors suggest a risk that relative poverty lev-els among the elderly could increase again.15 Seniors depending on public pension benefits are likely to see their relative incomes shrink, vis-à-vis both other retir-ees and the general working population, because• The income replacement value of OAS benefits—

which are adjusted to cost-of-living increases rather than wages—can be expected to decline over time.

• Similarly, CPP benefits are wage-based at the point of retirement, but subsequent adjustments are also linked to the price level, implying a stronger decline

15Most of the reductions in old-age poverty took place during the 1980s, owing to factors that are expected to weaken in the future (Myles, 2000). Chiefly among those is that public pension schemes reached maturity while real incomes of the working population stagnated during most of the 1980s and early 1990s, raising relative incomes of the elderly. The income distribution among seniors also became more even as the decline in elderly employment reduced a source of income concentration, whereas growing public retirement benefits were of a more equitable nature.

Table 5.6. Withdrawal from Labor Force and Retirement Duration, 1999

Age of Male Labor Market Duration Withdrawal (In years) of Male _______________________ First Third Retirement quartile Median quartile (In months)

Italy 54.5 58.8 63.4 20.7Finland 56.0 59.6 63.0 18.9Germany 57.4 60.3 63.9 18.8Netherlands 57.8 60.4 64.1 18.2Canada 57.8 62.4 66.5 18.2United Kingdom 57.8 62.6 66.5 16.8Sweden 59.9 63.7 66.7 17.5United States 59.4 64.6 71.4 16.3Japan 62.7 68.5 77.7 14.9Average 58.1 62.3 67.0 17.8

Source: OECD (2001).

Table 5.5. Sources of Pretax Income for the Elderly(Shares of total, in percent)

1980 1985 1990 1995 1999

Employment income 26.4 20.1 17.3 16.7 10.0Investment income 23.4 22.6 21.2 14.9 13.1Retirement income 11.6 13.0 16.2 21.0 28.5C/QPP benefits 7.7 10.9 14.3 17.3 19.0OAS/GIS/SPA 25.9 28.0 25.1 23.6 24.0Other 4.9 5.5 6.0 6.4 5.3

Sources: Statistics Canada, Income Trends in Canada, 1980–1999.Notes: C/QPP = Canada/Quebec Pension Plan; OAS = Old-Age

Security; GIS = Guaranteed Income Supplement; SPA = Spousal Allowance.

©International Monetary Fund. Not for Redistribution

81

in relative incomes as the duration of retirement increases with life expectancy.

• Moreover, if industrial wage incomes were to fall relative to other incomes, for example, in the ser-vice sector, an increasing share of CPP beneficiaries could see effective replacement ratios shrink.

Policy Issues

Although the Canadian pension system appears sound, there may be scope for further reforms ahead of the retirement of the baby boom generation. Con-sideration could be given to strengthening the private pension pillar, including by reviewing the structure of retirement saving vehicles and strengthening gover-nance of corporate pension schemes. Amendments to the public pension system could also assist in foster-ing labor market participation among older workers, boosting pension saving, and relieving impending fis-cal pressures. Moreover, over the longer term, care will be needed to ensure that the support provided by the basic public pension system is kept at adequate levels to avoid a rise in old-age poverty.

Reviewing Personal Saving Vehicles

Tax incentives appear to have had only a marginal impact on private saving. For example, Burbidge, Fretz, and Veall (1998) suggest that the tax treatment of RPPs and RRSPs has mainly led to a reallocation of savings into these vehicles. This view is supported by Veall (1999), who found that a substantial change in mar-ginal income tax rates in 1988 had a negligible impact on RRSP contributions. Similarly, Milligan (2002) found that a 10 percentage point increase in marginal tax rates would increase the probability of RRSP par-ticipation by only 8 percent. This result implies that tax increases between 1982 and 1996 explain only 5 per-cent of the increase in RRSP contributions during the same period.

Nevertheless, the recent focus on increasing RRSP contribution limits appears appropriate. Against the background of increased labor market mobility and a trend away from lifetime attachment to a single employer, gradual increases in RRSP contribution limits, at least in line with incomes and other pen-sion parameters, provide room for personal saving to increase, especially for those not covered by RPPs. The immediate fiscal implications of rising contribu-tion limits would likely be minimal and would likely be roughly matched in the long run by the tax paid on withdrawals.16 Moreover, these instruments also tend

16Milligan (2003) found that the option to accumulate unused contribution room has a negative short-term impact on RRSP con-

to improve the broader efficiency of the tax system by reducing the extent to which corporate income is double taxed and by increasing the extent to which tax is paid on consumption rather than income.

Tax-prepaid savings plans (TPSPs) could usefully supplement existing private savings plans. In contrast to RRSPs, contributions to TPSPs would not provide an immediate deduction from income tax, but would enable tax-free accumulation of capital gains and ben-efit payments as well as tax-exempt withdrawals. As detailed in Kesselman and Poschmann (2001), TPSPs would have the advantage of allowing households to optimize their retirement asset allocation free from tax considerations. The authors suggest that overall economic efficiency could be increased as households shift their investments away from (tax-preferred) life insurance policies or real estate holdings. Moreover, TPSPs could provide a saving incentive for low-income households, particularly if TPSP earnings do not cause a clawing back of GIS and SPA benefits.

Strengthening Governance of Corporate Pension Plans

Safeguarding the soundness of corporate pension plans remains a key priority. Recent changes in the fed-eral supervisory regime—partly prompted by financial weaknesses among defined-benefit plans—have already provided OSFI with the means to focus its surveillance more tightly on weaker pension funds. However, the extent to which supervisory practices at the provincial level have strengthened remains unclear. For exam-ple, most provinces have yet to follow OSFI in deal-ing with underfunded plans, and transparency about the state of provincial corporate pension funds could also be improved. These differences suggest that there exists a potential for closer harmonization of federal-provincial regulation and supervision, aimed at ensuring uniform adoption of best practices, improving system-wide transparency, and ensuring a more efficient use of supervisory resources.

For defined-benefit plans, relaxing limitations on con-tributions to fully funded pension plans could strengthen their financial position over the cycle. Although most defined-benefit plans would be expected to return to being fully funded in the course of the next economic upswing, recent experience illustrates that placing a ceiling on plan assets at a cyclical peak can leave assets at an insufficient level to cope with the ensuing down-turn. Therefore, an increase in the permissible surplus of pension funds (currently two years’ worth of pension contributions, or 20 percent of liabilities) could help encourage full funding over the cycle.

tributions, as individuals attempt to smooth contributions over the life cycle.

Policy Issues

©International Monetary Fund. Not for Redistribution

V THE PENSION SYSTEM

82

Encouraging Labor Market Participation among the Elderly

Unlike in many other countries the Canadian pen-sion system does not seem to significantly affect labor market participation among the elderly. Studies sug-gest that the generosity of public pension benefits has a sizable effect on retirement decisions in many countries, in many cases encouraging early retirement (Baker, Gruber, and Milligan, 2003; Gruber and Wise, 1999; Johnson, 2000). In Canada, however, the sys-tem appears to be relatively neutral, with some esti-mates suggesting that only about 20 percent of the trend decline in average retirement age is explained by public pension incentives (Baker and Benjamin, 1999; OECD, 2003).17 Indeed, while the CPP allows retirement from the age of 60, the OAS system does not provide an early retirement option at all. This contrasts with corporate pension plans, which have often been used to finance early retirement of employees in recent years, in an effort to mitigate the effect of corporate downsizing and restructuring programs.

Nevertheless, there appears scope to further reduce incentives for early retirement in the CPP. For example, the upward adjustment to CPP benefits for retirement before the statutory pension age of 65 is somewhat too low relative to the adjustment that is made for early

17Gruber (1999) and Tompa (1999) also found many early retirees to be already semi-detached from the labor market as a result of unemployment, part-time or low-income employment, or disability.

retirement.18 Similarly, although eligibility require-ments have been stiffened in recent years, disability benefits still seem to have dampened old-age labor participation rates (Campolieti, 2001). OECD (2003) research also suggests that disability insurance is being used as an exit route from the labor market by a signifi-cant share of male workers. Finally, recent proposals for reforming Quebec province’s pension plan could also be reviewed for their applicability at the national level (Box 5.2).

Especially in view of projected increases in longev-ity, there would seem to be scope for increases in the statutory retirement age.19 Assuming that the increase would apply to both OAS and CPP and that the entry age for early retirement would also be shifted upward, potential advantages include the following:• Direct fiscal savings would accrue from a shorter

period over which basic pension benefits are paid. Moreover, additional tax revenues would be gener-ated as working lives are extended. However, increas-ing the retirement age would also imply that workers for whom early retirement has been primarily a means to exit from unemployment would need to be supported longer through the social safety net.

18To achieve actuarial fairness, the pension benefit would need to be raised from 0.5 percent to 0.7 percent for every month of retire-ment after reaching the age of 65.

19To ensure political acceptance and mitigate the impact on work-ers close to retirement age, any increase in the retirement age would need to be gradual and phased in over a longer time frame.

The QPP—an independent public pension plan for the residents of Quebec province—is designed to be very similar to the CPP. Both systems share the same basic parameters, including contribution rates, statutory retire-ment age, and maximum contribution and benefit levels. Minor differences include the design of disability and survivor benefits, which are slightly more generous in Quebec.

The Quebec government is considering more signifi-cant changes to the QPP, motivated by its less favor-able actuarial position relative to the CPP. Owing to higher dependency ratios and lower immigration rates for Quebec, the actuarially neutral contribution rate of the QPP has increased to around 10.1 percent, compared to 9.8 percent for the CPP. The relatively low level of projected reserves and the growing gap between the CPP and QPP’s underlying contribution rates have led to calls for reforms.

The proposals aim to reduce the incidence of early retirement by providing commensurate financial incen-tives and allowing for a more flexible transition into retirement. Suggested measures include

• Adjusting the formula for calculating benefi ts at retire-ment by increasing the number of years in employ-ment required to receive full pension benefi ts;

• Providing an actuarially fair adjustment of pension benefi ts if retirement begins after the statutory age of 65; and

• Allowing benefi ciaries to continue to work without fi nancial penalty and crediting additional pension con-tributions toward the benefi ciary’s account (allowing pension benefi ts to increase until the maximum benefi t level is reached).The reform proposals also take into account the grow-

ing incidence of divorce and the large number of children living with only one parent. Proposed measures would focus survivor benefits to a larger extent on surviving chil-dren and no longer provide lifetime benefits for younger spouses. By also introducing elements of income testing, effective survivor benefit levels would decline slightly. Additional measures would reduce disability pensions in some cases, to bring them more closely in line with pen-sion benefits.

Box 5.2. Reform Proposals for the Quebec Pension Plan

©International Monetary Fund. Not for Redistribution

83

• Increasing the retirement age could leave room for reducing CPP contribution rates. This would reduce the wage wedge and could have positive effects on labor market participation across all age groups, as well as facilitating a more efficient allocation of labor resources.

• The accumulation of additional wealth would relieve some of the pressure on the two public pension pillars. The resources accumulated as a result of longer life-time employment could boost private pension assets and increase the share of retirement income from private sources after several decades of decline.

Maintaining Adequate Basic Public Pension Benefits

Basic pension benefits are projected to decline sig-nificantly in real terms in coming decades.20 Assuming continued price indexation, the average basic pen-

20With population aging, voting power is shifting toward the elderly generation, possibly precipitating a reallocation of public spending toward meeting old-age needs.

sion benefit could decline from around 20 percent of average incomes in 2001 to 14 percent by 2030, with lower-income groups facing a still larger decline in percentage points (OECD, 2003). Even on this basis, however, public pension expenditure is projected to increase by 2–3 percent of GDP through 2050—higher than in some other G-7 countries, including Japan and the United Kingdom (Table 5.7). More generous benefit provisions would push spending considerably higher—a shift to wage indexation, for example, could result in an increase of 5 percent of GDP, comparable to some of the large continental European economies.

It therefore remains important to explore options for simplifying the existing system and targeting benefits more directly at the neediest elderly. Despite failed attempts to reform the basic pension system in the late 1990s, there appears considerable scope for simplify-ing and unifying benefits that are currently provided through different programs (including GIS, SPA, and the tax code). Merging these programs and testing benefits for family income instead of personal income would contribute to increasing intragenerational equity and make the benefit structure more transparent.

Table 5.7. Projected Increases in Old-Age Pension and Social Spending

Change in Public Pension Old-Age Social Change in Old-Age Social Expenditure, 2000–501 Spending, 19952 Spending, 2000–502

(In percent of GDP) (In percent of GDP) (In percent of GDP)

Australia 1.6 3.1 3.3Canada 2.0 4.6 2.9France 3.9 12.3 2.7Germany 5.0 11.1 4.0Italy –0.3 13.0 4.8Japan 0.6 6.1 4.3Netherlands 4.8 7.3 3.5Sweden 1.6 8.8 1.3United Kingdom –0.7 8.1 2.7United States 1.8 6.2 1.6

1Source: Gruber and Wise (2001).2OECD (2003).

Policy Issues

©International Monetary Fund. Not for Redistribution

84

VI Fiscal Prudence and Tax Reduction Opportunities

The fiscal framework adopted by Canada in the mid-1990s led to a rapid reduction in government

debt, and several consecutive years of federal surpluses and falling government debt created a strong social consensus for prudent fiscal policies. The federal objec-tive of “budget balance or better” resulted in by far the largest reduction in debt among the G-7 countries. The authorities also established the objective of reduc-ing the federal debt-to-GDP ratio below 25 percent by FY2013–14. Most provinces also adopted balanced budget rules, and the aggregate provincial debt ratio also fell.

Fiscal prudence, together with reforms to the public pension system that put it on an actuarially sound basis, leave Canada relatively well prepared to cope with the fiscal pressures of population aging. This opens the opportunity to reduce the tax burden while maintaining fiscal prudence. In particular, tax cuts pledged by the conservative government that took power in February 2006 included a 2 percentage point reduction in the goods and services tax, of which the first percentage point has already been implemented, as well as cuts in a range of taxes on business before 2011. This has been accompanied by a framework that envisages pay-ing down federal debt by Can$3 billion (¼ percent of GDP) a year.

This section examines two issues related to the potential tax reductions: the first subsection examines the implications of delaying tax cuts in order to further reduce government debt, and the second assesses the efficiency gains of reducing the GST versus the level of personal income taxation.

A. Jam Today or More Jam Tomorrow? The Timing of Tax Cuts

Tamim Bayoumi and Dennis Botman

This analysis uses the IMF’s Global Fiscal Model (GFM) to both assess the long-term benefits of reduc-ing government debt by delaying tax cuts and examine issues of tax spillovers. The simulations here examine the consequences of forgoing immediate tax cuts in response to reductions in government spending so that

government debt can fall further, allowing larger tax cuts in the future. In addition, they examine the impact on Canada of tax rate changes elsewhere in the world.

The Model and Calibration

The GFM model is part of a suite of models with similar underlying structures adapted to look at dif-ferent macroeconomic issues.1 These models share important characteristics, including a firm grounding in microeconomic theory—consumers maximize util-ity and firms do the same with profits—which ensures that the long-run properties of the simulations conform to those predicted by theory. Because the underlying parameters correspond to assumptions about underly-ing behavior (such as the elasticity of substitution for hours worked with respect to real wages or consump-tion with respect to the real interest rate), these models are well designed to analyze how simulations depend on key behavioral assumptions, while real and nominal rigidities generate realistic dynamic responses. Finally, these models accommodate more than one country and hence can examine international linkages.

These features of GFM make it particularly well suited to examine fiscal issues:• The private sector is impatient. More specifically, the

discount rate used by the private sector is higher than the real rate of interest. In the absence of such impa-tience, the private sector fully anticipates the future costs of tax changes, leading to the Ricardian result that movements in aggregate demand from changes in taxes or transfers have no impact on spending. Addi-tionally, the model assumes that a certain proportion of wages accrue to “rule-of-thumb” individuals who vary their consumption one-for-one with their post-tax income. Finally, tax rates also distort relative prices and hence the allocation of resources.2

• Markets are not fully competitive. Firms and workers have some monopolistic power, implying that prices

1See Bayoumi and others (2004) for a discussion of the overall modeling effort and Botman, Laxton, Muir, and Romanov (2006) for a description of GFM.

2The model is not useful, however, for analyzing issues of inter-generational equality.

©International Monetary Fund. Not for Redistribution

85

are above their perfectly competitive levels. One con-sequence of this is that a corporate income tax affects not only the return to capital, but also the economic rent firms are able to extract from their monopolistic power. As a result, corporate income taxes are less distortionary than when these rents do not exist.The remainder of the model can be briefly sum-

marized as follows. Consumption and production are characterized by constant elasticity of substitution util-ity/production functions. There are two factors of pro-duction, labor and capital, which can be moved across sectors to produce traded or nontraded goods. Invest-ment is driven by a Tobin’s Q-relationship, in which firms respond sluggishly to differences between the future discounted value of their profits and the market value of their capital stock. Perfect capital mobility implies that real interest rates in countries are equalized over time.3 Wages and prices are assumed to be per-fectly flexible, which reduces the short-term aggregate demand impact of fiscal policies. Accordingly, the dis-cussion here focuses on medium- and long-term results. As noted, the model accommodates multiple countries; this analysis uses a two-country version of the model.

The impact of a tax cut on real activity depends on the response of aggregate supply and demand. The supply-side effects come through the increase in equi-librium hours worked (as a drop in the wage tax rate raises take-home wages) or the capital stock (as a cut in the corporate income tax rate increases post-tax rates of return). The increase in aggregate demand depends on the extent to which individuals view a larger fis-cal deficit as an increase in their permanent income, which, in turn, depends on the level of impatience and the proportion of rule-of-thumb consumers. Over the longer term, these effects spill over to other countries as the global real interest rate rises to re-equilibrate aggregate demand and supply and through international trade channels.

The model was parameterized to reflect the macro-economic features of Canada and the rest of the world. The latter is based on the United States, Canada’s main trading partner. Canada is about one-tenth the size of the rest of the world, and hence its policies have only a limited impact on the global rate of interest. The ratios relative to GDP of consumption, investment, govern-ment spending, wage income, and income from capital correspond to those in Canada and the United States. Canadian exports and imports as a ratio to GDP are set at current values. Tax rates on capital, labor, and con-sumption have been calibrated to reflect current yields across the two economies.4

3Lower levels of international capital mobility would raise the beneficial effects of debt reduction for the Canadian economy, while lowering spillover from tax policy in the rest of the world.

4Rather than try to model the complexities of actual tax systems, it is assumed that taxes are levied on the relevant base as a single

A number of other key behavioral parameters are set equal across the two economies.5 In addition to those characterizing real rigidities in investment and the markups of firms, simulations examine the impact of changing the values of the following key parameters:• The Frisch elasticity of labor supply, which measures

the sensitivity of labor supply to real wages: In the baseline, this is set at 0.04, in the midrange of val-ues produced by microeconomic studies. Alternative simulations assume values of 0.08 and 0.01, around the upper and lower limits of these estimates.

• The elasticity of substitution between labor and capi-tal in the production function: In the baseline, this is set at −0.8, while alternative simulations use values of −0.6 and −1 (the latter is the familiar Cobb- Douglas case, which implies constant shares of income accru-ing to labor and capital).

• The intertemporal elasticity of substitution of con-sumption, which measures the sensitivity of con-sumption to changes in the real interest rate: This is set at −0.33 in the base case, with a lower and upper level of −0.25 and −0.5, again broadly covering the range of microeconomic estimates.

• The impatience of forward-looking consumers: This parameter has not been subjected to significant microeconomic analysis. One approach is to consider the gap between interest rates charged to government debt and those charged to consumers on credit card debt, the main source of unsecured loans in which the lender takes the full risk that consumers may be unable to repay. Given the substantial margins seen in this comparison, the private sector discount rate was set some 10 percentage points above the real rate of interest, while simulations are also reported with wedges of 5 percentage points and 15 percent-age points.

• The proportion of wages associated with rule-of-thumb consumers: In the base case, this parameter was set at 10 percent, being raised to 20 percent and set to zero in alternative simulations. At the macro economic level, consumption is known to be relatively sensitive to changes in disposable income, although some of this comes from the impatience of forward-looking consumers discussed previously. In the base parameterization, the assumptions about impatience and rule-of-thumb consumers imply a multiplier of around one-fourth for temporary tax cuts or changes in income, broadly in line with other empirical work.6

marginal rate, so that there is no difference between average and marginal tax rates. Were taxes assumed to be progressive, this would affect tax distortions.

5See Laxton and Pesenti (2003) for a more detailed discussion of evidence on parameter values.

6For a discussion of current evidence on fiscal multipliers, see IMF (2004b).

A. Jam Today or More Jam Tomorrow? The Timing of Tax Cuts

©International Monetary Fund. Not for Redistribution

VI FISCAL PRUDENCE AND TAX REDUCTION OPPORTUNITIES

86

Implications of Cutting Transfers and Lowering Taxes

This subsection compares the consequences of match-ing a cut in transfers with an immediate tax cut versus making a larger tax cut later. The simulations assume that room for tax cuts is provided by a permanent cut in lump-sum transfer payments of 1 percentage point of GDP.7 The results compare the following two policy responses: (1) immediately implementing a permanent cut in tax rates so as to reduce tax revenues by the same amount as the cut in transfer payments; and (2) leaving tax rates unchanged for 10 years, followed by a larger permanent cut in tax rates made possible by the lower level of interest costs due to the intervening decline in the government debt ratio. In the second scenario, the government ends up with permanently lower tax rates and permanently lower levels of government debt, but at the cost of not offsetting the negative short-term impact of the cut in transfers on aggregate demand. While such scenarios are clearly stylized—in practice, the main reason for reducing government debt at pres-ent is to prepare for the future pressures on government spending from population aging—they help illustrate the effects of choosing to cut taxes or reduce debt in a simple and intuitive manner.8

Results for the base parameterization suggest that there are significant long-term benefits to delaying a cut in wage taxes, but that there are also costs to not offsetting the fall in transfers in the short term (Figure 6.1). Immediately replacing a reduction in lump sum transfers of 1 percentage point of GDP with a cut in wage taxes leads to a 0.3 percent increase in real GDP. About two-thirds of this boost occurs relatively rapidly, as the reduction in taxes leads to a boost in hours worked, while the remainder accumulates more slowly, as the capital stock rises. Delaying the cut in wage taxes by 10 years results in a small fall in real GDP after 5 years, as the impact on aggregate demand of the reduction in transfer payments is not offset, but also leads to an eventual tax rate reduction that is one and a half times larger than when taxes are cut imme-diately. Once implemented, the larger tax cut leads to real GDP gains that rise gradually from double to triple those generated by an immediate tax cut.

Cuts in corporate income taxes produce larger bene-fits that accumulate more slowly over time, while delay-ing cuts produce a similar ratio of losses and benefits (Figure 6.2). Replacing an immediate cut in wage taxes with a cut in corporate income taxes leads to a similar gain in real GDP after five years and a much larger

7Lump-sum transfers have no impact on incentives and hence allow a focus on the distortions caused by tax rates.

8The baseline also does not take account of future fiscal pressures from population aging. However, the qualitative results would not be significantly altered if the baseline were changed.

gain over time as capital is accumulated. The larger long-term benefits from a cut in corporate income taxes compared to wage taxes reflects the fact that the corpo-rate tax is more distortionary, as the supply of capital (which can be easily reproduced) is more elastic than that of labor—a standard result in the literature.9 Delay-ing the cut in taxes for 10 years again implies forgoing significant benefits to real GDP over the first 5 years in return for reaping benefits that are of the order of double the base case beyond 10 years.

A key advantage of a model such as GFM with well-defined microeconomic foundations is that the impli-cations of alternative behavioral assumptions can be examined. Figure 6.3 reports the change in real GDP after 5 years, 15 years, and 40 years when wage taxes are cut immediately, after 10 years, and the difference between these two values for a range of parameter val-ues. The results after 5 years can be interpreted as the medium-term effect of the cut in spending and (if implemented) the tax cut; those after 15 years represent the medium-term impact of a delayed tax cut; while the

9For a Canadian application, see Department of Finance Canada (2004b).

Source: IMF staff calculations.

–0.5

0

0.5

1.010-year delay

Immediate tax cut

4035302520151050

-0.5

0.0

0.5

1.0

1.5Difference 10-year delay Immediate tax cut

50 years15 years5 years

Figure 6.1. Effects on Real GDP of Immediate versus Delayed Wage Tax Cuts(Percent deviation from baseline)

©International Monetary Fund. Not for Redistribution

87

results after 40 years represent the long-term impact of alternative scenarios.

The results indicate that the Frisch elasticity of labor supply largely determines the overall size of the wage tax distortion. The benefits from immediate cuts in wage taxes vary approximately proportionally with the value of the Frisch elasticity of labor supply, for exam-ple, approximately doubling when it is raised from 0.04 to 0.08. This is because the elasticity determines the response of hours worked to changes in post-tax real incomes and hence the distortion to labor supply. No other parameter has a significant effect on the impact of an immediate cut in wage taxes.

The Frisch elasticity also has a proportional impact on the dynamic benefits of delaying wage tax cuts, while the ratio of short-term losses to long-term ben-efits rises as consumption becomes less sensitive to the real interest rate. The difference in the changes in real GDP implied by the different timing of tax cuts again varies approximately proportionately with the size of the Frisch elasticity. In addition, the ratio of longer-term benefits from delayed tax cuts to short-term losses rises

when consumption is less sensitive to real interest rates as it induces a larger fall in the real interest rate and a greater rise in the capital stock. This occurs when the intertemporal elasticity of consumption is reduced and when forward-looking consumers are made more impa-tient. Increasing the elasticity of substitution between labor and capital also raises this ratio as it increases the response of the capital stock to changes in labor supply. Finally, changing the proportion of rule-of-thumb con-sumers had little impact on the path of real GDP (and is not reported in the figure).

The key parameter for a corporate income tax is the elasticity of substitution between capital and labor, and changes in the sensitivity of consumption to real interest rates again matter for the dynamic responses (Figure 6.4). The higher the elasticity of substitution between labor and capital, the greater the incentive for firms to respond to a tax cut by raising the capital stock, and hence the larger the benefits of a delayed tax cut. In addition, the increased impact on the real interest rate resulting from the reduced sensi-tivity of consumption to the real interest rate again raises the dynamic benefits of a tax cut. By contrast, changes in the Frisch elasticity and the proportion of rule-of-thumb consumers have little impact on the simulations.

Fiscal Spillovers

The model also can be used to examine issues of fis-cal spillovers across countries. As in the earlier subsec-tion, the analysis is based on a highly stylized scenario designed to illustrate the impact of tax competition on the Canadian economy. In particular, it is assumed that there is a wage or corporate income tax rate cut in the rest of the world that lowers revenues by 1 percentage point of GDP and leads to larger fiscal deficits. After five years, this tax cut is rescinded and replaced by a permanent tax rate increase that generates sufficient revenues to cover the additional interest costs incurred by the intervening rise in government debt. The simu-lations examine first the impact on the Canadian econ-omy assuming no response by the tax authorities, and then the results if the Canadian authorities follow the rest of the world in cutting, and then later raising, the same tax rate.

The results suggest that fiscal spillovers from tax cuts in the rest of the world can be significant and rise over time (Figure 6.5). A temporary cut in wage taxes in the rest of the world equivalent to 1 percentage point of their GDP is followed by a subsequent increase on the order of one-quarter of this amount. The loss to Canadian and global GDP is on the order of three-quar-ters of a percent after 15 years. The losses accumulate gradually over time, as global real interest rates rise by around a half percentage point and crowd out invest-ment in response to the 8 percentage points of GDP

Source: IMF staff calculations.

–0.5

0.0

0.5

1.0

1.5

2.0

2.5

10-year delay

Immediate tax cut

4035302520151050

–0.5

0.0

0.5

1.0

1.5

2.0

2.5

50 years15 years5 years

Difference10-year delayImmediate tax cut

Figure 6.2. Effects on Real GDP of Immediate versus Delayed Corporate Income Tax Cuts(Percent deviation from baseline)

A. Jam Today or More Jam Tomorrow? The Timing of Tax Cuts

©International Monetary Fund. Not for Redistribution

VI FISCAL PRUDENCE AND TAX REDUCTION OPPORTUNITIES

88

Higher Frisch Elasticity of Labor Supply(Elasticity = 0.08)

–0.5

0

0.5

1.0

1.5

Difference10-year delayImmediate tax cut

40155

Lower Absolute Frisch Elasticity of Labor Supply(Elasticity = 0.01)

–0.5

0

0.5

1.0

1.5

40155Years after reduction in transfers Years after reduction in transfers

Higher Absolute Elasticity of Substitution betweenLabor and Capital (Elasticity = –1)

Lower Absolute Elasticity of Substitution betweenLabor and Capital (Elasticity = –0.6)

Years after reduction in transfers Years after reduction in transfers

Higher Absolute Intertemporal Elasticity of Substitutionof Consumption (Elasticity = –0.5)

Lower Absolute Intertemporal Elasticity of Substitutionof Consumption (Elasticity = –0.25)

Years after reduction in transfers Years after reduction in transfers

–0.5

0

0.5

1.0

1.5

40155

–0.5

0

0.5

1.0

1.5

40155

More Impatient Consumers (Wedge is 15 percentage points)

Less Impatient Consumers(Wedge is 5 percentage points)

Years after reduction in transfers Years after reduction in transfers

–0.5

0

0.5

1.0

1.5

40155

–0.5

0

0.5

1.0

1.5

40155

–0.5

0

0.5

1.0

1.5

40155

–0.5

0

0.5

1.0

1.5

40155

Source: IMF staff calculations.

Figure 6.3. Effects of Alternative Parameterizations on Impact of a Cut in Transfers and in the Wage Tax Rate on Real GDP(Percent deviation from baseline)

©International Monetary Fund. Not for Redistribution

89

Higher Frisch Elasticity of Labor Supply(Elasticity = 0.08)

–0.5

0

0.5

1.0

1.5

2.0

2.5

40155

Lower Absolute Elasticity of Labor Supply(Elasticity = 0.01)

Years after reduction in transfers Years after reduction in transfers

Higher Absolute Elasticity of Substitution betweenLabor and Capital (Elasticity = –1)

Lower Absolute Elasticity of Substitution betweenLabor and Capital (Elasticity = –0.6)

Years after reduction in transfers Years after reduction in transfers

Higher Absolute Intertemporal Elasticity of Substitutionof Consumption (Elasticity = –0.5)

Lower Absolute Intertemporal Elasticity of Substitutionof Consumption (Elasticity = –0.25)

Years after reduction in transfers Years after reduction in transfers

–0.5

0

0.5

1.0

1.5

2.0

2.5

40155

–0.5

0

0.5

1.0

1.5

2.0

2.5

40155

More Impatient Consumers (Wedge is 15 percentage points)

Less Impatient Consumers(Wedge is 5 percentage points)

Years after reduction in transfers Years after reduction in transfers

–0.5

0

0.5

1.0

1.5

2.0

2.5

40155

Source: IMF staff calculations.

Difference10-year delayImmediate tax cut

–0.5

0

0.5

1.0

1.5

2.0

2.5

40155

–0.5

0

0.5

1.0

1.5

2.0

2.5

40155

–0.5

0

0.5

1.0

1.5

2.0

2.5

40155

–0.5

0

0.5

1.0

1.5

2.0

2.5

40155

Figure 6.4. Effects of Alternative Parameterizations on Impact of a Cut in Transfers and in the Corporate Income Tax Rate on Real GDP(Percent deviation from baseline)

A. Jam Today or More Jam Tomorrow? The Timing of Tax Cuts

©International Monetary Fund. Not for Redistribution

VI FISCAL PRUDENCE AND TAX REDUCTION OPPORTUNITIES

90

increase in rest of the world government debt.10 The resulting losses to real GDP for equivalent changes in corporate income taxes are somewhat larger, reflecting the fact that corporate tax rate increases are more dis-tortionary, but the basic mechanisms are similar.

The Canadian government can mitigate the medium-term effects of fiscal spillovers by matching foreign tax cuts, but only at the cost of larger long-term costs to real GDP. These results broadly mirror those found pre-viously about the benefits and costs of immediate and delayed tax cuts. However, because the main effect on Canada occurs through the global rate of interest, rather than through differences in tax rates across countries,

10This is broadly consistent with results reported in IMF (2004b), which finds that a 1 percentage point rise in the U.S. fiscal deficit raises interest rates by up to ½ percentage point (recalling that the United States represents about one-third of global GDP at market prices), as well as Elmendorf and Mankiw (1999) who find that as a general rule, a rise in government debt lowers the capital stock by an approximately equal amount.

Canada’s own tax policy has only a limited impact on the long-term losses in output. This observation applies more to wage tax cuts than to corporate income tax cuts, as the latter involve larger domestic distortions.

These results reflect a range of assumptions about the structure and behavior of the global economy. It should be emphasized that the impact of fiscal policies in individual countries would be smaller for the rest of the world, as they would have a lesser impact on global debt and real interest rates. In addition, the size of fiscal spillovers is an area of considerable controversy, and these simulations are only one approach to addressing this question.11

Conclusions

The conclusions of this analysis can be summarized as follows:• There are significant potential benefits to be gained

by reducing government debt and delaying tax cuts. In the base case, delaying tax reductions by 10 years at the cost of short-term losses in output doubles the medium-term gains to real GDP of the eventual tax cut, and the long-term benefits can be even larger.

• A corporate income tax cut provides significantly greater benefits over time than a wage tax cut. This is because capital is a more mobile factor of pro-duction and hence is more responsive to changes in incentives.

• The key parameters that determine the benefits of delaying tax cuts are the Frisch elasticity of labor supply for a wage tax and the elasticity of substitution between labor and capital for a corporate tax cut. In addition, the ratio of losses and benefits to real GDP rise as the response of consumption to real interest rates is reduced or consumer impatience rises, as this boosts the long-term impact on the capital stock.

• International fiscal spillovers can be significant, with much of the impact coming through the global rate of interest rather than through differences in tax rates across countries. This limits the effectiveness of Canadian policies in reducing these spillovers, particularly for wage taxes.

B. Comparing Efficiency Gains from Reducing the GST and the Personal Income Tax

Dennis Botman

This subsection uses the GFM to assess the effi-ciency gains from reducing the Goods and Services Tax and the personal income tax.12 For these purposes,

11See IMF (2004b) for a discussion of these issues.12GFM is discussed in more detail in Botman and others (2006).

–2

–1.5

–1

–0.5

0

0.5

Canada matches tax cut

Canadian taxes remain unchanged

403530252015105

Temporary Cut in Wage Tax in the Rest of the World

–3

–2.5

–2

–1.5

–1

–0.5

0

0.5

Canada matches tax cut

Canadian taxes remain unchanged

403530252015105

Temporary Cut in Corporate Income Tax in the Rest of the World

Source: IMF staff calculations.

Figure 6.5. Impact of Tax Spillovers from the Rest of the World(Canadian GDP, percent deviation from baseline)

©International Monetary Fund. Not for Redistribution

91

the model has been extended to include a sales tax, and the model was then used to consider the following questions:• What are the macroeconomic effects of reducing the

GST versus lowering the personal income tax?• How do different taxes rank in terms of efficiency in

Canada according to GFM?• To what extent are the results consistent with other

evidence for Canada and robust to alternative model specifications?

Comparing Wage, Consumption, and Income Taxation

In a simple model, without initial wealth, wage and consumption taxation are equivalent. To illustrate this point, from the two-period budget constraint it follows that a wage tax affects the consumption-leisure choice as follows:

1C1 + ––––– C2 = (1 – tw)wL. (1) (1 + r)

In other words, lifetime consumption is equal to after-tax wage income. Similarly, a consumption tax affects the consumption-leisure choice in the following manner,

(1 + tc)(1 – tc)C1 + ––––– C2 = wL (1 + r)

1 1→ C1 + ––––– C2 = –––––– wL. (2) (1 + r) (1 + tc)

As a result, consumption and payroll taxation are equiv-alent in terms of their impact on consumption and lei-sure.13 Taxation of consumption and taxation of wage income both reduce the cost of leisure. An equivalent analysis for an income tax results in

C2 = [1 + r(1 – ty)] × [w(1 – ty)L – C1]

1→ C1 + –––––––––– C2 = w(1 – ty)L. (3) 1 + r(1 – ty)

An income tax has a broader tax base than both labor income and a sales tax and includes the return to sav-ings. As such, income taxation affects not only the con-sumption-leisure decision, but also the intertemporal choice between consumption today and consumption in the future.

However, with initial assets, a consumption tax base becomes broader than a wage tax base. Denoting initial assets by A, the incidence of labor income and con-sumption taxes discussed previously will change to, respectively,

1C1 + ––––– C2 = (1 – tw)wL + A (1 + r)

13Which is the case for tc = tw/(1 – tw).

(1 + tc)(1 – tc)C1 + ––––– C2 = wL + A (1 + r)

1 1→ C1 + ––––– C2 = –––––– (wL + A). (4) (1 + r) (1 + tc)

The wage base is smaller than the consumption tax base by the amount of initial assets in the economy. As a result, the economic distortions caused by these two types of taxes will differ. In addition, the inci-dence of the tax will differ as well, with consump-tion taxes also applying to the nonworking population with asset holdings (retirees, for example), which has potentially important implications for intra- and inter-generational equity, especially in an aging society such as Canada’s.

GFM allows a more rigorous assessment of alterna-tive types of taxation than the simple framework above, but the basic intuition remains the same. Specifically:• GFM features two types of economic agents: “opti-

mizing consumers” who have access to financial markets and own wealth, and “rule-of-thumb con-sumers” who do not have access to financial mar-kets and do not own any wealth. A consumption and payroll tax imply different effects in GFM owing to the presence of the former group of consumers. Optimizing agents smooth their consumption over their lifetimes—extending the two-period exam-ples above—but discount the future based on their remaining life expectancy (which does not depend on age), as in the overlapping-generations framework in Blanchard (1985) and Weil (1989). Rule-of-thumb consumers are not directly affected by corporate and personal income taxation as they own no assets (domestically or foreign-issued government bonds or corporate equity), although over time changes in the capital-labor ratio and real wages imply that consum-ers that rely relatively more on labor income will be affected—tax incidence is determined endogenously in GFM.

• The presence of monopolistically competitive firms implies that a personal or corporate income tax applies not only to the rate of return of capital but also to monopolistic rents distributed as dividends. In addition, the fact that firms are able to set their prices above marginal costs also implies that, in the short term, output is partly demand determined. As a result, alternative types of taxation can have differ-ent effects on output in the short term insofar as their effect on aggregate demand varies.

• Labor supply is endogenous. As a result, the extent to which the labor supply declines after higher payroll or consumption taxation depends critically on the sensitivity of workers to changes in the real wage—the elasticity of labor supply—and this sensitivity is assumed to be relatively low in advanced countries, especially for males (see Evers, de Mooij, and van Vuuren, 2005).

B. Comparing Efficiency Gains from Reducing the GST and the Personal Income Tax

©International Monetary Fund. Not for Redistribution

VI FISCAL PRUDENCE AND TAX REDUCTION OPPORTUNITIES

92

• Taxation of capital and excess profits affects incen-tives to save and invest, because capital accumulation is endogenous in the model. As a result, personal or corporate income taxation affects the supply side of the economy through investment, which is subject to adjustment costs.

Ranking Tax Distortions in Canada

The model is used to analyze the distortions caused by alternative types of taxation by assuming that Can-ada has fiscal space to reduce taxes because of a reduc-tion in lump-sum transfers from the government to households. In this case, any macroeconomic effects of tax reform follow directly from the resulting substitu-tion effects. After all, a commensurate reduction in a nondistortionary tax would have no macroeconomic effects, as the respective income effects would offset each other. The fiscal space is assumed to allow for a

reduction in the effective GST by 1 percentage point.14

With the exception of the fraction of rule-of-thumb consumers—set equal to 25 percent—the calibration is identical to that used by Bayoumi and Botman in the first part of this section to analyze the macroeconomic effects of an immediate versus a delay in tax cuts in order to further reduce government debt.

A reduction in the effective GST stimulates con-sumption and output (Figure 6.6). A 1 percentage point reduction in the effective GST rate reduces revenue by about 0.6 percentage point of GDP. In the short term, the effects of a GST reduction impact immediately on consumption and GDP. This generates only modest gains in efficiency and potential output, however, as a

14A reduction in the effective GST by 1 percentage point implies a larger reduction in the statutory rate if GST exemptions are pres-ent in practice.

Government Accounts Government Debt and Net Foreign Assets

Source: IMF, Global Fiscal Model (GFM) simulations.Note: Fiscal space results from a permanent reduction in lump-sum transfers. The GST is reduced in a

debt-neutral manner.

–1.2–1

–0.8–0.6–0.4–0.2

–00.20.40.60.8

Fiscal spaceGST (percentage points)Revenue

4036322824201612840

Real GDP and Consumption Capital Stock and Labor Effort

–0.6

–0.4

–0.2

0

0.2

0.4

Total consumption (percent)Real GDP (percent)

4036322824201612840

–0.20

0.20.40.60.81.01.21.4

Net foreign assetsGovernment debt

4036322824201612840

–0.10

0.10.20.30.40.50.60.70.8

Labor effort (percent)Capital stock (percent)

4036322824201612840

Figure 6.6. Macroeconomic Effects of a Reduction in the Goods and Services Tax(Deviation from initial steady state in percent of GDP unless otherwise noted)

©International Monetary Fund. Not for Redistribution

93

cut in the GST only directly affects the supply side of the economy through incentives for labor supply.

On the other hand, a reduction in the personal income tax rate has a stronger effect on potential output by stimulating incentives to invest (Figure 6.7). In the short term, the gain in output (real GDP) takes longer to materialize compared to reducing the GST, given that investment is subject to adjustment costs. In the medium term, the gains in potential output, however, are approximately three times larger for a reduction in the personal income tax relative to an equivalent reduc-tion in the GST. This reflects two considerations. First, by also taxing accumulated assets, a cut in the GST implies a windfall gain to asset holders, which does not contribute to economic efficiency. Second, by reducing taxation of a reproducible factor of production (capital), which is relatively more sensitive to taxation than labor supply, the personal income tax reduction has a stron-ger effect on output.

In general, the results confirm the view that the GST is a relatively efficient form of taxation. In net pres-

ent value terms, the increase in potential output could be substantially larger if other taxes were reduced (Figure 6.8):• A cut in the payroll tax—similar to the GST—also

affects the consumption-leisure decision. Labor sup-ply will increase by more, however, as no windfall gain is provided to asset holders.

• Personal income taxes are, in turn, more distortion-ary than payroll taxes, because the base includes dividend income in addition to wage and interest income and transfers.

• Finally, corporate income taxation is the least effi-cient form of taxation, although the presence of monopolistic competition in GFM implies that part of the tax burden falls on rents rather than on the return to capital. The latter implies that corporate income taxation in GFM is less distortionary than in models that assume firms are perfectly competitive.This order of efficiency is consistent with evidence

from various international studies. Baylor (2005) sur-veys dynamic computable general equilibrium analyses

Government Accounts Government Debt and Net Foreign Assets

Source: IMF, Global Fiscal Model (GFM) simulations.Note: Fiscal space results from a permanent reduction in lump-sum transfers. Personal income taxation is

reduced in a debt-neutral manner.

–1.2–1

–0.8–0.6–0.4–0.2

00.20.40.60.8

Fiscal spacePersonal income tax (percentage points)Revenue

4036322824201612840

Real GDP and Consumption Capital Stock and Labor Effort

–0.6

–0.4

–0.2

0

0.2

0.4

4036322824201612840

–0.2-0

0.20.40.60.81.01.21.4

4036322824201612840

–0.10

0.10.20.30.40.50.60.70.8

4036322824201612840

Net foreign assetsGovernment debt

Total consumption (percent)Real GDP (percent) Labor effort (percent)

Capital stock (percent)

Figure 6.7. Macroeconomic Effects of a Reduction in Personal Income Taxation(Deviation from initial steady state in percent of GDP unless otherwise noted)

B. Comparing Efficiency Gains from Reducing the GST and the Personal Income Tax

©International Monetary Fund. Not for Redistribution

VI FISCAL PRUDENCE AND TAX REDUCTION OPPORTUNITIES

94

of tax distortions, focusing on the issue of tax rank-ing and reviewing results from standard neoclassical growth models and human-capital-driven endogenous growth models. In general, the ranking based on results from neoclassical growth models indicates that capital taxes are the most distortionary, followed by labor and then consumption taxes. Tax rankings based on results from endogenous growth models, on the other hand, are more heterogeneous and vary across framework, settings, and ranking criteria. Nonetheless, results from these models indicate that both capital and labor tax reductions tend to have greater economic impact than consumption tax reductions.

The ranking of taxation found in GFM is also con-sistent with results of general equilibrium models for the Canadian economy (Department of Finance Can-ada, 2004b).15 Matier and Wu (2000) and Baylor and Beauséjour (2004) both construct representative-agent dynamic general equilibrium models to analyze tax dis-tortions. They simulate the long-run impact of deficit-neutral tax reductions equivalent to 1 percent of GDP. Consumption taxes are found to be least distortionary in Canada, followed by wage, personal, and corporate income taxation. Consistent with our results, they also find that, over the long run, consumption expands more if the personal income tax is reduced than when the general sales tax is reduced.

These findings are also robust to variations in the under-lying assumptions (Figure 6.9). The strong microfounda-tions in GFM allow for exploration of the sensitivity of the

15See Baylor and Beauséjour (2004) for a detailed description of the model and a demonstration that the conclusion is robust under alternative values for important model parameters.

macroeconomic effects of fiscal policy and tax reform to uncertainty about the fundamental determinants of con-sumer and producer optimization. In this case, the model is used to analyze the sensitivity of the finding that a per-sonal income tax cut is more efficient than GST reduction to uncertainty about the planning horizon of consumers, access to financial markets, the elasticity of labor supply, the characteristics of utility and production functions, and the degree of competition:• A longer planning horizon of consumers, a reduction in

the share of rule-of-thumb consumers, or higher substi-tutability between capital and labor all imply somewhat larger incentives to save and invest and therefore stron-ger gains in potential output from reducing either the GST or personal income taxation.

• More elastic labor supply implies that reducing either the GST or the personal income tax causes larger efficiency gains.

• If consumption is less sensitive to changes in the real interest rate, efficiency gains from reducing the per-sonal income tax are larger. Higher national saving

–0.2–0.10.00.10.20.30.40.50.6

GST

Payroll tax levied on workers

Corporate income tax

Personal income tax

4035302520151050

Source: IMF, Global Fiscal Model (GFM) simulations.Note: Fiscal space results from a permanent reduction in

lump-sum transfers. Taxes are reduced in a debt-neutral manner.

Effect on Real GDP

Figure 6.8. Efficiency Gains from Reducing Alternative Types of Taxation(Deviation from initial steady state, in percent)

Source: IMF, Global Fiscal Model (GFM) simulations.1In the baseline, the planning horizon is 10 years, the fraction

of rule-of-thumb consumers is 25 percent, labor supply elasticity parameter equals 0.96 (moderately elastic; higher values correspond to a less elastic supply), the elasticity of substitution between capital and labor in the production function is equal to 0.80, the intertemporal elasticity of substitution is 0.33, and the price markups are equal to 20 and 40 percent in the traded and nontraded goods sectors, respectively.

2Planning horizon is equal to 100 years.3Labor supply elasticity parameter is equal to 0.9.4Elasticity of substitution between capital and labor is equal to 1

(Cobb-Douglas case).5Intertemporal elasticity of substitution is equal to 0.2.6Price markups in the traded and nontraded goods sectors are

9 and 17 percent, respectively.

0

2

4

6

8

10

12DifferencePersonal income taxGST

Lower price

markups6

Lower intertemporal elasticity of substitution5

Higherfactor

substitut-ability4

More elasticlaborsupply3

Norule-of-thumb

consumers

Longer planning horizon2

Baseline

Figure 6.9. Efficiency Gains from Reducing the GST versus Personal Income Taxation: Sensitivity Analysis1

(Deviation in percent of initial steady-state GDP)

©International Monetary Fund. Not for Redistribution

95

following a reduction in the personal income tax is followed by higher consumption in the future. The required reduction in real interest rates is larger if consumers have a lower intertemporal elasticity of substitution, leading to larger incentives to invest.

• Lower price markups imply larger efficiency gains from reducing taxation of dividends because a larger share of the tax burden falls on the return to capital rather than on excess profits.

Conclusions

From an efficiency point of view, Canada would be best served if tax reductions took the form of lower

taxes on saving and investment rather than reduced general sales taxes. This conclusion is consistent both with the general theoretical evidence about the ranking of taxation and with specific evidence for Canada. The conclusion that the GST is a relatively efficient tax in Canada is also robust to alternative assumptions about the structure of the economy and uncertainty about the fundamental determinants of consumer and producer behavior. It is also noteworthy that the consumption tax rate in Canada is low by international standards and that maintaining its current level also helps to protect the tax base as the retiree population rises. On the other hand, taxation of saving and investment is relatively high in Canada.

Conclusions

©International Monetary Fund. Not for Redistribution

This page intentionally left blank

©International Monetary Fund. Not for Redistribution

Part III

Monetary and Financial Policies

©International Monetary Fund. Not for Redistribution

This page intentionally left blank

©International Monetary Fund. Not for Redistribution

99

VII Disintermediation and Monetary Transmission

Jorge E. Roldos

The financial system links monetary policy and the real economy. Thus, events or trends that affect the

financial system also can change the monetary trans-mission mechanism. Over the last two decades, Cana-da’s financial system (like those in other industrialized countries) has been transformed by several rounds of financial deregulation, innovation, and disintermedia-tion. This transformation has likely affected the trans-mission of monetary policy.

Monetary transmission occurs through the tradi-tional channel (the impact of interest rates on compo-nents of aggregate demand) and through the so-called credit channel, including constraints in the availability of loanable funds to banks and corporates. Those who stress the predominance of the credit channel argue that the traditional channel cannot explain the magni-tude, timing, and compositional effects of the monetary policy actions in the U.S. economy (see Bernanke and Gertler, 1995). Credit market frictions give rise to a “financial accelerator” mechanism that amplifies and makes more persistent the impact of a monetary policy action on the economy. These credit market frictions depend on several features of the financial system, in particular the degree to which the financial system is securitized (broadly understood as the development of securities markets versus banks) and the size and the state of the banking system (Cecchetti, 1999).

The credit channel is more an enhancement mecha-nism than a truly independent or parallel channel of monetary transmission (Bernanke and Gertler, 1995). Earlier literature on the credit channel focused on the bank lending channel and argued that monetary policy operated through the lack of substitutes in both the liability and asset sides of banks’ balance sheets. In particular, it was argued, a monetary tightening would reduce excess reserves and contract the supply of loans, and this would have a direct negative impact on output. The evidence on this mechanism is somewhat mixed (see, for instance, Kashyap and Stein, 1994), and it is likely that, with financial deregulation and innova-tion, the importance of the bank lending channel has diminished. More recent literature has focused on a broader balance sheet channel and argues that a mon-etary contraction weakens firms’ financial positions—either directly through reduced cash flows or indirectly

through the declining value of assets and/or collat-eral. The deterioration in firms’ financial condition increases the external finance premium, and the associ-ated increase in the cost of capital operates through the financial accelerator mechanism (Bernanke, Gertler, and Gilchrist, 1999). Recent evidence on the macro-economic significance of these financial frictions is provided in Levin, Natalucci, and Zakrajsek (2004).

This section uses two methodologies to study changes in Canada’s monetary policy transmission associated with important changes in the financial structure dur-ing the 1990s. First, a series of vector autoregression (VAR) models is used to characterize the dynamics of output and prices after a monetary shock. Standard VAR models used to study monetary transmission of the type surveyed in Christiano, Eichenbaum, and Evans (1999) are extended with the inclusion of financial vari-ables emphasized in the credit channel literature and are tested for the existence of structural breaks or parameter instability around the dates of major changes in financial sector regulation. Second, this section studies the impact of financial variables on structural econometric models of aggregate demand, which are increasingly used by central banks to conduct monetary policy analysis. In particular, it gauges the impact of the ratio of direct (or market) to indirect (or intermediated) finance—a sum-mary measure of disintermediation—on the interest rate sensitivity of aggregate demand.

Financial Deregulation and Disintermediation

The Canadian financial system underwent two important changes over the last two decades as a result of deregulation and financial innovation.1 First, it became more market based, as corporates increased the use of direct (or market) financing compared to indi-rect (or intermediated) financing. Second, there was an increase in households’ access to credit, reflected in the rise in consumer and mortgage credit.

1For a thorough account of these and other trends, see Freedman and Engert (2003) and Calmes (2004).

©International Monetary Fund. Not for Redistribution

VII DISINTERMEDIATION AND MONETARY TRANSMISSION

100

The first trend is apparent in Figure 7.1. During the 1970s, the rise in inflation and nominal interest rate uncertainty led the corporate sector to rely increas-ingly on short-term bank loans and much less on issu-ing securities. However, this trend was reversed in the 1980s, and bank lending declined as a source of fund-ing for the corporate sector from approximately 60 per-cent in 1980 to less than 40 percent in the 2000s. This was accompanied by an increase in the issuance of corporate bonds and commercial paper (Figure 7.2).

This process of disintermediation is likely to have reduced both the bank dependency of borrowers and the constraints on the availability of loanable funds,2which together would indicate a decline in the rela-tive importance of the bank lending channel. How-ever, they do not necessarily imply a reduction in the importance of the credit channel because the cost of external finance, summarized in the “external finance premium,” may still operate as a key mechanism for the transmission of monetary policy.

Changes in the regulatory framework, together with changes in global and Canadian financial market con-ditions, were major drivers of these structural changes. Calmes (2004) notes that significant amendments to the Bank Act in 1980, 1987, 1992, and 1997 contributed to (and were in part driven by) the sharp change in cor-porates’ funding mix. Figure 7.3 shows the increase in the ratio of direct to indirect lending beginning in the

2See Kashyap and Stein (1994) and Bernanke and Gertler (1995).

late 1980s and lasting through the 1990s, following a decade of stagnation and a decline during the 1970s.

This pattern of disintermediation, which parallels changes in other countries, was accompanied by a switch in banks’ activities, from lending to large cor-porates to underwriting and other securities-related activities. In particular, the 1987 amendments to the Bank Act allowed banks to conduct brokerage activities (Calmes, 2004), and banks made substantial invest-ments in this area between 1987 and 1989. Nonethe-less, banks continue to be the main providers of loans to small and medium-size enterprises. Also, several indicators show that the Canadian banking system has

Source: Haver Analytics.

0

20

40

60

80

100

Loans

Equities Other

Bonds, debentures,and short-term paper

Source: Bank of Canada.

1971 75 79 83 87 91 95 99 2003

Figure 7.1. Nonfinancial Corporate Financing Sources, 1971–2005(Stocks outstanding, in percent of total)

Source: Bank of Canada.

Credit Flows, 1971–2005

Credit Flows, 1991–2005

–10

–5

0

5

10

15

Commercial paper

Commercial paper

Bonds and debentures

Bonds and debentures

Loans

Loans

2003999591878379751971

–10

–5

0

5

10

15

0503200197 9995931991

Figure 7.2. Credit Flows, 1971–2005(In billions of Canadian dollars)

©International Monetary Fund. Not for Redistribution

101

continued to grow despite banks’ declining share of lending, especially when off-balance-sheet activities are included (see Calmes, 2004).

The second trend is the growth in credit to the house-hold sector (broadly defined as persons and unincorpo-rated businesses), which rose from just under 50 percent of GDP in the early 1980s to almost 70 percent of GDP in 2001 (see Freedman and Engert, 2003). This steady increase was driven by the growth in consumer and mortgage credit, associated with the decline in infla-tion and the escalation of housing prices (Figure 7.4). Indeed, consumer credit as a share of the total (business and consumer credit), increased from 42 percent in 1982 to 55 percent in 2005.

The growth in credit to households, which was also shared by other industrialized nations, raised the impor-tance of the household sector and opened the potential for restrengthening the bank lending channel and/or the financial accelerator (which would work more directly through housing prices and collateral). Although the securitization of mortgages has had a profound impact on monetary transmission in the United States (Estrella, 2002), mortgage securitization has advanced at a much slower pace in Canada (Freedman and Engert, 2003). The following subsections explore how financial variables may have changed the monetary transmission mechanism, first, by looking at the impact of financial variables in tra-ditional VAR models and, second, by testing the impact of such variables in structural econometric models.

Evidence from VAR Models Changes in monetary transmission associated with

these changes in financial structure can be analyzed with VAR models, where a minimum structure is imposed and monetary shocks are identified as innova-

tions to the interest rate equation. Standard VAR mod-els used in studies of monetary transmission include a set of endogenous variables (output and prices) and an interest rate equation that captures a general monetary policy rule that affects the endogenous variables with a lag.3 Given Canada’s small open economy characteris-tics, the exchange rate is included in the model after the interest rate equation, together with a set of exogenous variables X.4

The benchmark VAR model has the representation

Yt = A(L)Yt–1 + B(L)Xt + ut, (1)

where

Yt = [yt, pt , Rt, St ] (2)

and y is quarterly GDP, p is the GDP deflator, R is the three-month T-bill rate,5 and S is the exchange rate. The vector of exogenous variables is given by

Xt = [ytUS, Rt

US, cpt ], (3)

where yUS is U.S. GDP, RUS is the federal funds rate, and cp is an index of world commodity prices.

This benchmark VAR model is then extended with a block of financial variables, F, suggested by the credit channel literature. This includes quantity variables, aimed at capturing the role/existence of alternative

3This is the standard recursivity assumption; see Christiano, Eichenbaum, and Evans (1999) and Favero (2001).

4The exogenous variables included are U.S. GDP, U.S. federal funds rate, and an index of world commodity prices.

5Although the overnight rate is the Bank of Canada monetary policy instrument, which is the best variable to summarize the mon-etary policy stance (Fung and Yuan, 1999), it is rather unstable in the first part of the sample. Thus, we use the T-bill rate, which is highly correlated with the overnight rate and is also the variable used in several European Union studies that will serve as compari-sons to the Canada case.

0

0.2

0.4

0.6

0.8

1.0

Source: Bank of Canada.

2003999591878379751971

Figure 7.3. Ratio of Direct to Indirect Private Lending, 1971–2005

0

100

200

300

400

500

600

700

Consumer credit

Business loans

Household mortgages

2003999591878379751971

Source: Bank of Canada.

Figure 7.4. Outstanding Credit, 1971–2005(In billions of Canadian dollars)

Evidence from VAR Models

©International Monetary Fund. Not for Redistribution

VII DISINTERMEDIATION AND MONETARY TRANSMISSION

102

sources of funding for the corporate and household sectors, as well as asset price variables, aimed to cap-ture the role of collateral and the “external finance premium.”6

The VAR models are estimated in levels using quarterly data starting from 1971, the beginning of the floating exchange rate period, and ending in 2005. As in most VAR models of the monetary transmission mechanism, there is no explicit analysis of the long-run behavior of the economy. Doing the analysis in levels allows for implicit cointegrating relationships in the data, but cointegration is not explicitly imposed. Imposing cointegrating restrictions on a VAR in levels could increase efficiency in the estimation, but this would be at the cost of potential inconsistencies if the incorrect identifying restrictions are imposed. Since the monetary transmission mechanism is a short-run phenomenon, most researchers prefer to employ unre-stricted VARs in levels to evaluate impulse responses over the short to medium run (Favero, 2001).

The dynamics of Canada’s output and prices are broadly similar to those of the United States, the Euro-pean Union, the United Kingdom, and Japan, using sim-ilar benchmark VAR models.7 The impulse responses for the benchmark VAR for Canada for the full sample following a monetary shock are presented in the left-hand column of Figure 7.5. The decline in output after a contractionary monetary policy shock is somewhat smoother and more persistent in Canada, while the slug-gish decline in prices (with an initial spike consistent with the so-called price puzzle) is similar to that of most other industrialized countries. The exchange rate appre-ciates for a period of about three years, but the results are not statistically significant for the full sample.

The responses to a monetary shock during the full sample period are consistent with those of other indus-trialized countries, but the lack of significance and the fact that there were important regime changes8 suggest the possibility of a structural break in the statistical model in the late 1980s or early 1990s. Statistical tests for structural breaks at the dates when key changes to the Bank Act were implemented (1988:Q1 and 1993:Q2, as suggested by Calmes, 2004) confirm such breaks for both the benchmark and financial VARs.9

6The variables included were total loans to businesses and house-holds; securities (bond, equities, and commercial paper); and ratios that micro studies have found relevant for the credit channel (such as the ratio of commercial paper to business loans; see Kashyap and Stein, 1994). The price variables included spreads on loans, com-mercial paper, and bonds, as well as stock and housing prices (and an aggregate asset price index, with equal weights of both of them).

7See, respectively, Christiano, Eichenbaum, and Evans (1999); Peersman and Smets (2001); Bean, Larsen, and Nikolov (2002); and Morsink and Bayoumi (2001).

8Changes occurred in both financial regulation and in monetary policy; on the latter, see Atoyan (2004).

9See the last line in Tables 1 and 2 in Roldos (2006). The relatively small sample does not provide the degrees of freedom necessary to

The instability of the VAR coefficients is confirmed with likelihood-ratio tests for all VAR models at the 5 percent confidence levels.

The second and third columns of Figure 7.5 show that the impulse responses in the first subsample (1971:Q1–1991:Q1) are more tightly estimated and show a sta-tistically significant decline in output between the fifth and tenth quarters. Prices fall more gradually and per-sistently, while the exchange rate appreciates sharply on impact and returns to its baseline level afterward. How-ever, the impulse responses for the second subsample (1991:Q1–2005:Q2) show a very different shape and are statistically insignificant.10

The extended VARs provide a tighter and improved characterization of the dynamics of macro variables after a monetary shock, especially for the first subsam-ple. The variables added included both asset prices and quantities, and the ones that contributed to an improved characterization of the transmission of interest shocks to output and prices were business loans (see Figure 7.5) and asset prices (Figures 7.6 and 7.7). In particular, when each of these variables is added to the benchmark VAR, a statistically significant recession and a more tightly estimated response of the price level decline emerges for the first subsample, and the appreciation of the exchange rate after the increase in interest rates becomes statistically significant (Figure 7.8).

More important, the contribution of the interest rate shock to the variance of output fluctuations (as mea-sured by the variance decompositions) increases mark-edly in the models with financial variables, suggesting that their addition contributes to a richer set of dynamic interactions and an improved characterization of the transmission mechanism. Monetary shocks explain close to 35–40 percent of the variance of output in the financial models, while they explain around 27 percent in the baseline model (see Roldos, 2006, Table 4).

The dynamics of output following a monetary shock seem to have changed earlier in the sample period than the dynamics of prices. The F-tests for individual equations in each VAR show a structural break for the output equations in all the models by 1988, while the price equation changes are more robust in the early 1990s (Roldos, 2006, Tables 1–3). This is consistent with changes in output being associated with the ear-lier changes in financial structure and with changes in

perform continuous break tests and let the data show the true break point. Tests for an intermediate date, 1991:Q1, reported in Table 3 in Roldos (2006), yield results that are qualitatively similar to those of the second break test.

10A similar pattern arises for the different components of aggre-gate demand: private consumption falls smoothly but persistently between the fourth and tenth quarters; investment falls three times as much as consumption; and residential investment falls more than total investment. In the second subsample, residential investment falls significantly during the first year only (despite the mild response in GDP), and several of the results are statistically insignificant.

©International Monetary Fund. Not for Redistribution

103

Source: IMF staff calculations.Note: Solid black line shows impulse response. Other lines (dashed) show a two-standard-deviation confidence band.

Full Sample

Response of GDP_R to TBILL3MO Response of GDP_R to TBILL3MO Response of GDP_R to TBILL3MO

1971:Q1–1991:Q1 1991:Q1–2005:Q1

–0.004–0.003–0.002–0.001

00.0010.0020.0030.004

2018161412108642–0.004–0.003–0.002–0.001

00.0010.0020.0030.004

2018161412108642–0.002

–0.001

0

0.001

0.002

0.003

2018161412108642

Response of GDP_P to TBILL3MO Response of GDP_P to TBILL3MO Response of GDP_P to TBILL3MO

Response of TBILL3MO to TBILL3MO Response of TBILL3MO to TBILL3MO Response of TBILL3MO to TBILL3MO

–0.003

–0.002

–0.001

0.000

0.001

0.002

0.003

–0.003

–0.002

–0.001

0

0.001

0.002

0.003

2018161412108642 2018161412108642–0.0010

–0.0005

0

0.0005

0.0010

0.0015

0.0020

2018161412108642

–0.6–0.4–0.2

00.20.40.60.8

2018161412108642–0.6–0.4–0.2

00.20.40.60.8

2018161412108642–0.6–0.4–0.2

00.20.40.60.8

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–2

–1

0

1

2

3

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–2

–1

0

1

2

3

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–2

–1

0

1

2

3

2018161412108642

Figure 7.5. Impulse-Response Functions from Benchmark VAR

Evidence from VAR Models

©International Monetary Fund. Not for Redistribution

VII DISINTERMEDIATION AND MONETARY TRANSMISSION

104

Source: IMF staff calculations.Note: Solid black line shows impulse response. Other lines (dashed) show a two-standard-deviation confidence band.

Full Sample

Response of GDP_R to TBILL3MO Response of GDP_R to TBILL3MO Response of GDP_R to TBILL3MO

1971:Q1–1991:Q1 1991:Q1–2005:Q1

–0.004–0.003–0.002–0.001

00.0010.0020.0030.004

2018161412108642–0.004–0.003–0.002–0.001

00.0010.0020.0030.004

2018161412108642–0.002

–0.001

0

0.001

0.002

0.003

2018161412108642

Response of GDP_P to TBILL3MO Response of GDP_P to TBILL3MO Response of GDP_P to TBILL3MO

Response of TBILL3MO to TBILL3MO Response of TBILL3MO to TBILL3MO Response of TBILL3MO to TBILL3MO

–0.003

–0.002

–0.001

0

0.001

0.002

0.003

2018161412108642–0.003

–0.002

–0.001

0.000

0.001

0.002

0.003

2018161412108642–0.0010

–0.0005

0

0.0005

0.0010

0.0015

0.0020

2018161412108642

–0.6–0.4–0.2

00.20.40.60.8

2018161412108642–0.6–0.4–0.20.00.20.40.60.8

2018161412108642–0.6–0.4–0.2–0.00.20.40.60.8

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–0.8–0.4

00.40.81.21.62.02.42.8

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–1.2–0.8–0.4

00.40.81.21.62.02.4

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–2

–1

0

1

2

2018161412108642

Figure 7.6. Impulse-Response Functions from VAR, Including Business Loans

©International Monetary Fund. Not for Redistribution

105

Source: IMF staff calculations.Note: Solid black line shows impulse response. Other lines (dashed) show a two-standard-deviation confidence band.

Full Sample

Response of GDP_R to TBILL3MO Response of GDP_R to TBILL3MO Response of GDP_R to TBILL3MO

1971:Q1–1991:Q1 1991:Q1–2005:Q1

–0.004–0.003–0.002–0.001

00.0010.0020.0030.004

2018161412108642–0.003–0.002–0.001

00.0010.0020.0030.004

2018161412108642–0.002

–0.001

0.000

0.001

0.002

2018161412108642

Response of GDP_P to TBILL3MO Response of GDP_P to TBILL3MO Response of GDP_P to TBILL3MO

Response of TBILL3MO to TBILL3MO Response of TBILL3MO to TBILL3MO Response of TBILL3MO to TBILL3MO

–0.003

–0.002

–0.001

0

0.001

0.002

0.003

2018161412108642–0.003

–0.002

–0.001

0.000

0.001

0.002

2018161412108642–0.0010

–0.0005

0

0.0005

0.0010

0.0015

0.0020

2018161412108642

–0.6–0.4–0.2

00.20.40.60.8

2018161412108642–0.6–0.4–0.2–0.00.20.40.60.8

2018161412108642–0.6–0.4–0.2–0.00.20.40.60.8

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–1.0–0.5

00.51.01.52.02.5

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–2

–1

0

1

2

3

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–2

–1

0

1

2

2018161412108642

Figure 7.7. Impulse-Response Functions from VAR, Including Asset Prices

Evidence from VAR Models

©International Monetary Fund. Not for Redistribution

VII DISINTERMEDIATION AND MONETARY TRANSMISSION

106

Source: IMF staff calculations.Note: Solid black line shows impulse response. Other lines (dashed) show a two-standard-deviation confidence band.

Full Sample

Response of GDP_R to TBILL3MO Response of GDP_R to TBILL3MO Response of GDP_R to TBILL3MO

1971:Q1–1991:Q1 1991:Q1–2005:Q1

–0.003–0.002–0.001

00.0010.0020.0030.004

2018161412108642–0.003–0.002–0.001

00.0010.0020.0030.004

2018161412108642–0.0015–0.0010–0.0005

00.00050.00100.00150.0020

2018161412108642

Response of GDP_P to TBILL3MO Response of GDP_P to TBILL3MO Response of GDP_P to TBILL3MO

Response of TBILL3MO to TBILL3MO Response of TBILL3MO to TBILL3MO Response of TBILL3MO to TBILL3MO

–0.003

–0.002

–0.001

0

0.001

0.002

0.003

2018161412108642–0.003

–0.002

–0.001

0

0.001

0.002

0.003

2018161412108642–0.0015–0.0010–0.0005

00.00050.00100.00150.0020

2018161412108642

–0.6–0.4–0.2

00.20.40.60.8

2018161412108642–0.6–0.4–0.2

00.20.40.60.8

2018161412108642–0.6–0.4–0.2

00.20.40.60.8

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–1.0–0.5

00.51.01.52.02.5

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–2

–1

0

1

2

3

2018161412108642

Response of XR_INDEX_LEV to TBILL3MO

–2

–1

0

1

2

2018161412108642

Figure 7.8. Impulse-Response Functions from VAR, Including Asset Prices and Business Loans

©International Monetary Fund. Not for Redistribution

107

prices being more closely associated with the changes in the monetary regime in the early 1990s.

The fact that responses to a monetary shock became weaker in the more recent subsample can be interpreted to reflect the fact that the systemic component of monetary policy (that is, the one captured by the monetary policy rule) has become more important with the increased transparency and other institutional improvements asso-ciated with the adoption of inflation targeting. This is consistent with a decline in the importance of monetary shocks—that is, the unexpected part of monetary policy captured by the VAR shocks. Indeed, the standard devia-tion of the monetary shocks in the second subsample is systematically smaller than in the first subsample. Similar conclusions can be drawn for the United States, in terms of both the instability of monetary VAR models and the increasing role of systematic monetary policy.11

In sum, both the benchmark and the VARs, which include financial variables, show a clear change in mon-etary transmission in the late 1980s or early 1990s. Since VARs are largely unrestricted, we cannot test specific hypothesis of what key factors are behind the changes in monetary transmission within this framework. For this, we now turn to specific tests of the role of financial variables in structural econometric models.

Evidence from Structural Models

Structural macroeconometric models used to ana-lyze monetary policy issues generally do not explicitly incorporate financial sector features. However, Bean, Larsen, and Nikolov (2002) argue that a reduction in financial frictions associated with financial deepening and/or disintermediation would lower the persistence of moves in the output gap and would reduce the elastic-ity of the gap to the real interest rate (consistent with a reduction in the external finance premium and lower amplification of the initial monetary shock). Alterna-tively, it has been argued that bank lending is more relationship based and therefore less interest rate sensi-tive than market financing, suggesting that disinter-mediation would increase the elasticity of the gap to the interest rate.12 Thus, in the absence of clear-cut, testable implications from micro-founded models, this analysis estimates closed and open economy versions of an aggregate demand or IS equation and tests the sensitivity of the main coefficients to the evolution of financial variables.13

11See Bernanke, Gertler, and Watson (1997) and Boivin and Giannoni (2002).

12Bank loans are implicit contracts that allow for more flexibility in renegotiation and risk sharing, features that are not necessarily reflected in market prices or interest rates (Allen and Gale, 2000).

13Typical New Keynesian models of monetary policy also include a Phillips curve (or aggregate supply) and interest rate equations (Clarida, Galí, and Gertler, 1999).

One simple closed-economy IS equation is that pro-posed by Rudebusch and Svensson (1999):

yt = α0 + α1 yt−1 + α2yt−2 + α3(Rt−1 − πt−1) +εt, (4)

where y is the output gap and π is the annualized quar-terly rate of inflation.14 Following Estrella (2002), the coefficients of the lagged gap and the real interest rate are allowed to vary with measures of financial inno-vation or disintermediation, and the significance of these variables is tested. Since the focus here is on the increase in the share of securities markets fund-ing relative to bank lending, the coefficient α3 is made a function of the ratio of direct to indirect finance, DIF.15 To capture the two trends described previously (the decline in the share of bank lending to corporates and the growth of credit to households), two broad measures of financial disintermediation are considered, where DIF1 is the ratio of securities to business loans and DIF2 is the ratio of securities to total loans—to capture the increased lending to the household sector. The equations to be estimated are of the form

yt = α0 + α1 yt−1 + α2 yt−2

+ (α31 + α32 DIFt )(Rt−1 − πt−1) + εt. (5)

The estimation results confirm the changes in mone-tary transmission in the early 1990s, but do not seem to support the hypothesis that disintermediation is behind such changes. The estimates show that the interest rate elasticity of the output gap was significantly differ-ent from zero in the full sample, but also show that the significance is only for the more recent subsample (see Roldos, 2006, Tables 6 and 7, respectively). This stands in contrast to the results from the VAR models, which seem to suggest a loss in the effectiveness of monetary policy in the more recent sample period, and confirms the interpretation of a more important role for the systematic effects of monetary policy—given that this specification captures both the systematic and unexpected effects of interest rate changes.

There is no clear evidence from the closed-economy IS equation that the interest rate elasticity of the output gap has been affected by the shift toward market financ-ing. The interest rate elasticity does not seem to depend on either measure of financial disintermediation.16

14The output gap is defined as the ratio of actual GDP to a Hodrick-Prescott-filtered version of the same series; the inflation rate is the same period change in the GDP deflator.

15Estrella (2002) makes the coefficient α3 a function of the degree of securitization of mortgage loans only. For the United States, he finds that the elasticity of the gap to the real interest rate falls as the degree of mortgage securitization increases, and he interprets the result as a decline in the efficacy of monetary policy. The coef-ficients on the output lags do not seem to be affected by the disin-termediation variables.

16For the average values of the variables DIF1 and DIF2, the interest elasticity coefficients become −0.33 and −0.46, respectively. However, for their current values, the coefficients are not very dif-ferent from zero. See Roldos (2006), Table 6.

Evidence from Structural Models

©International Monetary Fund. Not for Redistribution

VII DISINTERMEDIATION AND MONETARY TRANSMISSION

108

Although equation (4) is a useful simple benchmark, other more recent specifications of an aggregate IS equation may be more appropriate for Canada. These specifications are derived from micro-founded theoret-ical models and stress both forward-looking and open-economy aspects of aggregate demand.

Monetary policy models in the “New Keynesian” tradition are grounded in dynamic general equilibrium theory and capture the forward-looking behavior of optimizing firms and consumers. However, the empiri-cal performance of these models is not satisfactory, and backward-looking elements have to be added to achieve a reasonable fit. Following Clarida, Galí, and Gertler (1999), a typical forward-looking IS equation with endogenous persistence can be specified as

yt = α0 + α1Et+1 + α2 yt−1 + α3(Rt − πt) + εt, (6)

where E is the expectations operator and the lagged real interest rate has been replaced by the current rate. However, the estimation results are not satisfactory when the expected output gap is replaced by the actual future gap. Although the future gap enters with a large and highly significant coefficient, the real interest rate loses significance and a high DW statistic suggest that

an important degree of autocorrelation is not captured by this specification. The coefficients incorporating the ratio of direct to indirect finance DIF are also not statistically significant.17

Monetary policy models for the small open economy are isomorphic to the ones just discussed for the closed economy, except for the size of the parameters of the aggregate demand function. Clarida, Galí, and Gertler (2001) show that the introduction of a foreign country and foreign goods does not change the shape of the IS equation, except that the interest rate sensitivity of aggregate demand—the parameter α2 in the previous equations—depends now on the degree of openness or the share of foreign goods in total consumption, a slow-moving variable.

Despite the fact that the real exchange rate does not appear as an independent variable in the micro-funded models of aggregate demand, several analysts and central banks include this variable in the IS equation. Moreover, popular open economy models, such as the one in McCallum and Nelson (1999a), include shocks to foreign output in the IS equation. Thus, an open economy IS equation of this form

yt = α0 + α1Et+1 + α2 yt−1 + α3(Rt − πt−1)

+ α4 zt−1 + α5 y*1 + εt (7)

was estimated, where z is the real exchange rate gap and y* is the U.S. GDP gap.18 The real exchange rate gap is the difference between the actual real exchange rate and a Hodrick-Prescott-filtered version of the same variable. The resulting real exchange rate gap measure is qualitatively similar to the measure presented in Figure 2 in Berg, Laxton, and Karam (2006), which imposes uncovered interest rate parity and uses a flexible combination of backward- and forward-looking elements to calibrate an equation similar to (7).

Estimates of the open economy IS equation pro-vide evidence that disintermediation has contributed to the changes in monetary transmission experienced in Canada. In particular, the interest rate elasticity of aggregate demand is not statistically different from zero for the full sample period under study (1971–2005, see Table 7.1), but the elasticity becomes sig-nificant when it is estimated as a linear function of the ratio of direct to indirect finance as in equation (5).19 This suggests that the trends of intermediation

17Results are available upon request.18Since the forward-looking component of the output gap gener-

ates a high degree of unexplained autocorrelation and a nonsignifi-cant interest rate elasticity, a version with two lags of the Canadian output gap was estimated. The U.S. GDP measure of potential output is a Hodrick-Prescott filter of GDP until 1980, and staff estimates afterwards.

19When calculated using the average sample values of DIF1 and DIF2, both elasticities are, respectively, −0.15 and −0.09.

Table 7.1. Estimates of Open Economy IS Equation (1971:Q1–2005:Q2)

Open Economy Model Model IS Equation with DIF1 with DIF2

α0 –0.096 0.137 –0.054 (0.837) (0.768) (0.906)

α1 1.021 0.972 0.979 (0.000) (0.000) (0.000)

α2 –0.252 –0.229 –0.229 (0.001) (0.003) (0.003)

α30 –0.072 0.874 0.658 (0.312) (0.022) (0.031)

α31 . . . –0.505 –1.191 . . . (0.012) (0.014)

α4 –0.027 –0.018 –0.019 (0.628) (0.747) (0.737)

α5 0.440 0.628 0.602 (0.000) (0.000) (0.000)

R2 0.838 0.845 0.845

DW 1.79 1.76 1.78

Source: Author’s calculations.Notes: DIF1 is the ratio of direct to indirect finance for the

corporate sector; DIF2 is the ratio of direct finance to total lending—including the household sector.

p-values are in parentheses.

©International Monetary Fund. Not for Redistribution

109

(in the 1970s) and disintermediation (in the 1990s) might have cancelled out when they were omitted in the initial specification. Moreover, the coefficients on DIF1 and DIF2 become significant in the second half of the sample and yield relatively larger estimates of the interest elasticity (respectively, −0.46 and −0.48, for the average sample values of DIF1 and DIF2; see Table 7.2). This suggests that the responsiveness of aggregate demand to monetary policy actions has become stronger with the process of financial disin-termediation during the 1990s.

Conclusions and Policy Implications

This analysis shows that monetary transmission in Canada has changed markedly since the late 1980s and also provides evidence that financial disintermedia-tion has contributed to these changes. Estimated VAR models show a clear break in monetary transmission beginning in 1988, after changes in financial sector regulation initiated the process of financial disinterme-diation. Although the inclusion of financial variables in VAR models improves the characterization of monetary transmission in the period before the structural break, the models suggest a decrease in the effectiveness of monetary policy during the 1990s. However, there is

evidence suggesting that this may be related to the fact that VAR models capture the impact of unexpected monetary shocks, while the systemic component of monetary policy—which is not captured in the impulse responses—has likely become more important with the improvements in the monetary framework during the 1990s. As in other industrialized countries, changes in monetary policy, the nature of global shocks, and other structural changes probably also played a significant role in monetary transmission.20

The process of disintermediation has contributed to changes in the sensitivity of aggregate demand to real interest rates, a key parameter in the transmission of monetary policy to the real economy. Estimates of the interest rate elasticity of aggregate demand in IS equations increase during the 1990s, confirming the interpretation that the systemic component of monetary policy has become more relevant recently. Moreover, changes in the ratio of direct to indirect finance, a mea-sure of disintermediation, help explain the changes in the interest rate elasticity and suggest an increase in the effectiveness of monetary policy associated with a larger use of market-based sources of finance during the

20Stock and Watson (2003) estimate that for the United States, changes in policies account for around one-quarter of the reduction in volatility of the major macroeconomic aggregates.

Conclusions and Policy Implications

Table 7.2. Estimates of Open Economy IS Equation (Sample Break in 1988:Q1)

Open Economy IS Equation Model with D1F1 Model with D1F2 ________________________________ ________________________________ ________________________________ (1971:1–1988:1) (1988:1–2005:2) (1971:1–1988:1) (1988:1–2005:2) (1971:1–1988:1) (1988:1–2005:2)

α0 –1.567 0.337 –1.495 1.209 –1.292 1.346 (0.058) (0.428) (0.072) (0.043) (0.142) (0.025)

α1 0.747 1.303 0.740 1.172 0.734 1.143 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

α2 –0.204 –0.427 –0.187 –0.333 –0.192 –0.313 (0.057) (0.000) (0.087) (0.005) (0.075) (0.008)

α30 0.201 –0.17 –0.978 1.672 –0.579 2.13 (0.115) (0.029) (0.475) (0.070) (0.519) (0.030)

α31 . . . . . . 0.653 –0.941 1.333 –3.571 . . . . . . (0.387) (0.046) (0.381) (0.019)

α4 0.026 –0.025 0.032 –0.104 0.039 –0.112 (0.791) (0.697) (0.738) (0.165) (0.687) (0.125)

α5 1.025 0.375 0.956 0.594 0.999 0.626 (0.000) (0.017) (0.000) (0.002) (0.000) (0.001)

R2 0.771 0.943 0.773 0.947 0.773 0.948

DW 1.80 1.93 1.82 1.82 1.80 1.80

Source: Author’s calculations. Notes: DIF1 is the ratio of direct to indirect finance for the corporate sector; DIF2 is the ratio of direct finance to total lending—including the

household sector. p-values are in parentheses.

©International Monetary Fund. Not for Redistribution

VII DISINTERMEDIATION AND MONETARY TRANSMISSION

110

1990s. This could be attributed to the lower interest rate sensitivity of relationship-based bank lending compared to the more price-sensitive direct or market funding.

Monetary policy appears to have become more effective during the 1990s, when measured through the average impact of interest rate changes on the out-put gap or, alternatively, on aggregate demand. How-ever, this increased effectiveness may be undermined by more recent increases in household borrowing and

the relative decline in the issuance of corporate secu-rities, the ratio of direct to indirect finance, which has been lowering since 2002. Although the increas-ing role of the household sector may change mon-etary transmission in ways not captured in the simple models considered here, the analysis of monetary policy could benefit from a more explicit consider-ation of the evolving role of financial markets and intermediaries.

©International Monetary Fund. Not for Redistribution

111

VIII Inflation Targeting and Macroeconomic Volatility

Tamim Bayoumi and Vladimir Klyuev

Canada was one of the pioneers of inflation target-ing (IT). The IT regime that was introduced in

February 1991 was the second in the world, with New Zealand having adopted a similar framework in 1989. Despite its novelty, the IT regime was a great success from an early stage and has been widely credited as a key element in reducing inflation, anchoring private sector inflationary expectations, and reducing macro-economic volatility (Ragan, 2005).

The IT framework has also been allowed to evolve over time, particularly around the renewals of the IT agreement between the Bank of Canada and the federal government that have occurred about every five years. Early in the process, these changes included lowering the inflation target, but this has been fixed at 2 per-cent since December 1994. On occasion, they involved fine-tuning the way core inflation is measured. More radical changes, such as lengthening the time hori-zon over which the inflation target should be achieved in response to certain types of shocks, have at least been contemplated and discussed within the Bank of Canada. Analyzing the link between IT and macroeco-nomic volatility is an important ingredient in determin-ing the role of the framework in shaping the Canadian economy and in evaluating the potential benefits and costs of proposed regime changes.

This section uses a small dynamic monetary model to assess the link between IT and macroeconomic volatility. The notion that the monetary policy frame-work itself and the credibility it enjoys are impor-tant for macroeconomic stability is a common one among policymakers.1 This analysis, which fol-lows the approach in Bayoumi and Sgherri (2004a, 2004b), suggests that the most important link goes through the impact of monetary credibility on the forward- lookingness of private sector expectations and presents evidence that IT lowered macroeco-nomic volatility through these “credibility” effects. This implies that changes to the framework should be approached in a careful manner so as to avoid eroding the credibility of the overall regime.

1See, for example, Dodge (2005).

The Framework

Analysts often use small models to assess the impact of monetary policy on the economy. A typical closed economy model involves three equations of the follow-ing type:

it = ρit–1 + (1–ρ)(θ0 + θ1πet+1 + θ2yt–1) + εi

t (1)

yt = αyet+1 + (0.99 – α)yt–1 + βrt–1 + εy

t (2)

πt = λπet+1 + (1 – λ)πt–1 + φyt–1 + επ

t , (3)

where y is the output gap (the difference between actual and potential output), r is the real interest rate (current nominal rate less expected inflation), π is annualized inflation, i is the nominal interest rate, εs are error terms, superscript e represents expectations, and other Greek letters reflect parameters. This type of model highlights two main monetary transmission channels: the real interest rate channel, through which the central bank affects the spending decisions of the private sec-tor; and an expectation channel, where the monetary authority influences the private sector’s expectations by conveying information about the future course of monetary policy.2

The monetary reaction function equation (1) follows the approach of Clarida, Galí, and Gertler (1999). It comprises a Taylor rule in which the nominal inter-est rate responds to expected future inflation and the lagged output gap (assuming the current gap is not observable), augmented by a lagged dependent variable to reflect interest rate smoothing, while the unobserved inflation target and natural rate of interest are sub-sumed in the constant term.

Aggregate demand is determined in equation (2), which links the output gap to the short-term real inter-est rate, as well as forward- and backward-looking weights on the output gap itself. The weights add up to the subjective discount rate, taken to be 0.99 in quar-terly data. This type of relationship comes from a Euler

2Applications of such models include Taylor (1999); Clarida, Galí, and Gertler (1998); Rudebusch and Svensson (1999); Svensson (2000); King (2000); and McCallum and Nelson (1999b).

©International Monetary Fund. Not for Redistribution

VIII INFLATION TARGETING AND MACROECONOMIC VOLATILITY

112

equation for consumption augmented to allow for habit persistence.

The expectation-augmented Phillips curve given in equation (3) describes the model’s supply side. It relates current inflation to a weighted average of future and past inflation, with the weights adding up to one. This type of specification is typical in empirical work,3although its theoretical justification has been a source of contention.

Modern inflation theory suggests that more pre-dictable monetary policy can improve macroeco-nomic stability by making the Phillips curve more forward-looking. Recent theoretical work has estab-lished that the inflationary expectations become more forward- looking as uncertainty about future demand is reduced.4 In particular, modern versions of the Lucas “islands” model of nominal rigidities that take into account uncertainties about how people assume oth-ers will react to a given event imply that the degree to which expectations in the Phillips curve are backward-looking should rise with uncertainty about aggregate demand. As discussed in Bayoumi and Sgherri (2004a), this implies that the level of nominal inertia should be negatively related to monetary credibility, as this is an important determinant of uncertainty about aggregate demand. That paper finds a strong, statistically signifi-cant and correctly signed cointegrating relationship in the U.S. data between the uncertainty about real inter-est rates (the proxy for credibility in this model) and the weight on backward-looking expectations in the Phillips curve, with the former Granger causing the lat-ter and no evidence of reverse feedback. This research motivates us to investigate whether the introduction of the IT regime, by lowering uncertainty about monetary policy, has made the inflation process more forward-looking, thereby producing a more flexible economy and reducing macroeconomic instability.

It should also be recognized that policymakers in Canada were among the first to recognize this potential link and relied on it as a rationale for introducing infla-tion targets. For example, in March 1995, Gordon Thies-sen, then Governor of the Bank of Canada, observed that “[b]y making its inflation-control objectives more explicit, the Bank hoped not only to influence infla-tion expectations, but also reduce uncertainty in the economy and the financial markets” (Thiessen, 1995). He went on to say that “with credible targets, inflation expectations, and therefore inflation, are less likely to react to temporary demand and supply shocks.” In addi-tion, inflation targets impose discipline on the Bank, which “makes monetary actions more predictable and

3See, for example, Clarida, Galí, and Gertler (1999); King and Wolman (1999); and Levin, Wieland, and Williams (1999).

4Mankiw and Reis (2001); Woodford (2001); Amato and Shin (2003), based on the original insights of Lucas (1972) and Phelps (1983).

less a source of uncertainty for others as they make economic decisions.”

Empirical Results

Empirical estimates suggest that the introduction of IT may indeed have made the Canadian Phillips curve more forward-looking. The model was estimated using quarterly data over 1982–89 (the period before the introduction of inflation targeting), and the IT period of 1992–2005—the years 1990–91, being a period of transition, were eliminated from the sample.5 Estima-tion used the generalized method of moments (GMM) with the first four lags of inflation, the output gap, and interest rates to instrument the forward-looking variables in the equation. The results produce little evi-dence of instability over time in the IS curve or the coefficient on the output gap in the Phillips curve, but do show a statistically significant shift in the coef-ficient on forward-looking inflation in the Phillips curve after the introduction of IT in addition to more aggressive response to inflation and greater interest rate smoothing in the monetary reaction function. In particular, the preferred specification—reported in Table 8.1—involves a rise in the coefficient on forward-looking inflation of almost one-third, from 0.54 to 0.71, as IT reduced uncertainty about the monetary reaction function.

More forward-looking inflation expectations enhanced the expectations channel, reducing the need for aggressive responses to macroeconomic shocks. Figure 8.1 reports impulse-response functions from the estimated model with regard to disturbances to the output gap, inflation, and interest rates over the two samples.6 As shown in the first two columns of the figure, after the introduction of IT, more limited mone-tary responses were needed to stabilize the economy in response to macroeconomic shocks, largely reflecting the speedier impact of interest rate changes on private sector behavior. In addition, the increased importance of the expectation channel implied by more forward-looking inflation expectations also allowed monetary policymakers to respond to shocks in a more gradual manner.

The results suggest that the increase in Canada’s macroeconomic stability since 1991 largely reflects a

5In the implementation, π was annualized quarterly change in the CPI, in percent; i was the target for overnight rate set by the Bank of Canada, in percent; and y was the output gap calculated by the Bank of Canada as percentage of potential real GDP.

6The coefficient below one on inflation in the pre-IT monetary policy rule violates the Taylor principle and makes the model dynamically unstable. In simulations, a value of 1.2 was imposed on that coefficient. The negative estimated coefficient on the output gap in the IT monetary policy rule was insignificantly different from zero and was replaced with zero in simulations.

©International Monetary Fund. Not for Redistribution

113

decline in inflation inertia. To illustrate this further, the first four rows in Table 8.2 report the standard deviation of inflation and the output gap and the inter-est rate before and after the introduction of IT and compares them with the volatility implied by our esti-mated monetary model. While the model does not very accurately replicate the standard deviations observed historically, it does capture the reduction in macroeco-nomic volatility after the introduction of IT, including in the interest rate as IT made policy responses more predictable.

Further analysis suggests that the change in the behavior of the central bank and the private sector clearly contributed to the decline in macroeconomic variability in Canada since 1991, but that the difference in the magnitude or nature of shocks had an ambiguous effect. We explore the reasons behind the fall in model volatility, distinguishing between the impact of changes in the shocks, the monetary policy reaction function, and private sector behavior. Row E in Table 8.2 shows the implied level of macroeconomic volatility if the economy having the 1990s structure had been subject to shocks that prevailed in the 1980s, while the next row reverses the experiment, looking at the volatility of an economy with a 1980s’ structure buffeted by 1990s-like shocks. The results suggest that the shocks in the 1990s imposed less output variability but more infla-tion and interest rate volatility. On the other hand, the change in the structure of the economy helped reduce the variability of inflation and nominal interest rates without a noticeable effect on output volatility.

More forward-looking inflation expectations played a key role in reducing macroeconomic variability. Simply replacing the pre-IT monetary reaction function with the IT-period rule without any attendant change in the forward-looking nature of the Phillips curve (row G) produces only a small reduction in inflation and inter-est rate volatility at a cost of higher output gap variabil-ity. By contrast, the rise in the expectations component of the Phillips curve (row H) explains almost all of the implied reduction in macroeconomic volatility.

The model helps evaluate the impact of possible changes to the IT framework. One particular modifica-tion currently under discussion is an extension of the time over which the central bank aims to achieve its tar-get. In our framework, such an extension is equivalent to reducing the coefficient on inflation in the monetary reaction function. As reported in Table 8.3, assuming no change in the Phillips curve, lowering the coefficient on inflation in the monetary response function from 1.5 to 1.2 would slightly reduce the volatility of interest rates and increase the volatility of the output gap and inflation by negligible amounts.

A key issue is whether a modification of the frame-work, such as delaying or muting response to shocks, might degrade the regime’s credibility. Should the mar-kets perceive the change as a weakening of the Bank’s commitment to the inflation target, they would adjust their expectation-formation process. That adjustment would likely result in an increase in inflation inertia, as expectations became less forward-looking. As a con-sequence, the reduction in macroeconomic volatility

Table 8.1. Estimates of Monetary Model1

1982–89 Both Periods 1992–2005

IS curveExpected output gap 0.507 (.006)* Lagged real interest rate −0.008 (.001)* Standard error of regression 0.492 0.226

Phillips curveExpected inflation 0.54 (.130)* 0.71 (.039)*Lagged output gap 0.030 (.024)Standard error of regression 1.487 2.110

Monetary reaction functionLagged interest rate 0.699 (.024)* 0.833 (.026)*Constant term 6.965 (.480)* 0.986 (.711)Long-run coefficient on inflation 0.813 (.088)* 1.532 (.317)*Long-run coefficient on output gap 0.610 (.074)* −0.230 (.279)

Standard error of regression 1.016 0.851

Source: IMF staff calculations.1Estimated using systems generalized method of moments with a constant term, the first four lags of inflation,

the output gap, and nominal interest rates as instruments. Standard errors, reported in parentheses, are robust to heteroscedasticity and to a first order moving average process. An asterisk indicates that the coefficient is different from zero at the 1 percent significance level.

Empirical Results

©International Monetary Fund. Not for Redistribution

VIII INFLATION TARGETING AND MACROECONOMIC VOLATILITY

114

over the past 15 years might be partially reversed, as the private sector would respond more to demand and supply shocks and less to policy action. As the last row of Table 8.3 demonstrates, a reduction in the coefficient on expected inflation in the Phillips curve as small as one-fifth of the earlier gain would completely negate the benefits to interest rate volatility and increase per-ceptibly the volatility of inflation.

Conclusions and Policy Implications

This analysis suggests that the reduction in mac-roeconomic volatility in Canada after the introduc-

tion of inflation targeting is largely attributable to the reaction of the private sector to the establishment of a credible monetary policy framework. With greater confidence in the central bank’s commitment to price stability, the private sector started forming infla-tion expectations in a more forward-looking manner, thereby reducing considerably the degree of nominal inertia in the Phillips curve. This has attenuated pri-vate sector reaction to demand and supply shocks, thus lessening the volatility of inflation and muting the business cycle.

An implication of this analysis is that further refine-ments of the IT framework should be carried out in a gradual manner that does not compromise monetary

Source: IMF staff calculations.

A. Demand Shock

Inflation Inflation Inflation

B. Supply Shock C. Policy Shock

Output gap Output gap Output gap

Interest rate Interest rate Interest rate

–0.10

0.10.20.30.40.50.60.70.80.9

1

4037343128252219161310741

Pre-IT

IT

–0.20

0.20.40.60.81.01.21.41.61.8

4037343128252219161310741–0.16–0.14–0.12–0.10–0.08–0.06–0.04–0.02

0

4037343128252219161310741

–0.20

0.20.40.60.81.01.21.41.61.8

4037343128252219161310741–0.08–0.07–0.06–0.05–0.04–0.03–0.02–0.01

00.010.02

4037343128252219161310741–0.16–0.14–0.12–0.10–0.08–0.06–0.04–0.02

00.02

4037343128252219161310741

–0.20

0.20.40.60.81.01.21.41.6

4037343128252219161310741–0.1

00.10.20.30.40.50.60.70.80.9

4037343128252219161310741–0.2

0

0.2

0.4

0.6

0.8

1.0

4037343128252219161310741

Figure 8.1. Impulse-Response Functions

©International Monetary Fund. Not for Redistribution

115

credibility. The private sector’s expectation-formation process and reaction to shocks are crucial determi-nants of macroeconomic stability. Potential benefits from adjustments in the regime should be weighed carefully against the possibility that a loss of cred-ibility from a perceived waning of commitment to the inflation target would worsen the macroeconomic environment. This also underscores the importance of communicating carefully the rationale for any pro-posed changes.

Table 8.2. Standard Deviation of Inflation, Output Gap, and Interest Rate before and after Introduction of Inflation Targeting (IT)(In percent)

Output Interest Equation Shocks Inflation Gap Rate

A Actual Pre-IT 2.1 2.2 2.0B Actual IT 1.8 1.5 1.6C Pre-IT Pre-IT 4.8 1.5 4.1D IT IT 3.2 0.8 1.6E IT Pre-IT 2.4 1.6 1.8F Pre-IT IT 6.5 0.8 4.8G IT MR; Pre-IT PC IT 6.1 1.2 4.5H IT PC; Pre-IT MR IT 3.2 0.7 1.8

Source: IMF staff calculations.Note: Rows A and B: actual standard deviations in pre-IT and IT

periods; C and D: asymptotic standard deviations based on esti-mated model (1) for the two periods; E and F: mixing estimated model parameters from one period with estimated standard devi-ations of shocks from the other; G and H: model-based asymp-totic standard deviations assuming the structure of the economy and the shocks are as estimated for the IT period, except that row G uses the pre-IT period Phillips curve (PC) and row H uses the pre-IT period monetary reaction (MR) function.

Table 8.3. Standard Deviation of Inflation, Output Gap, and Interest Rate under Hypothetical Regimes

Output InterestEquation Parameters Inflation Gap Rate

(In percent)IT λ=.71; θ1=1.5 3.24 0.81 1.58Longer reaction λ=.71; θ1=1.2 3.25 0.83 1.52Longer reaction;

less credibility λ=.68; θ1=1.2 3.54 0.84 1.62

Source: IMF staff calculations.

Conclusions and Policy Implications

©International Monetary Fund. Not for Redistribution

116

IX Competition in the Banking SystemIryna V. Ivaschenko1

There is ongoing discussion on the merits of allowing mergers between Canada’s large financial institu-

tions. Some analysts argue that a lack of mergers may constrain banks’ ability to compete internationally, given the comparatively small size of the Canadian market.2Others caution that mergers would increase the already high concentration in Canada’s banking sector, adversely affecting competition and thus consumer interests.3 Some observers are particularly concerned about the prospect of mergers among the six largest banks, which account for about 90 percent of all banking assets.

This section uses an industrial organization (IO) approach to measure the degree of competition among the largest Canadian banks compared with other coun-tries. In particular, it analyzes competition in banking systems that have more banks than Canada, such as the United States, which has a large and fragmented banking system; the United Kingdom, which has fewer banks; and continental Europe.

Optimal Level of Bank Competition: A Review of the Literature

The literature is inconclusive on the relative merits of a highly competitive banking system versus a structure that retains some degree of monopolistic power.4 This applies to the literature studying the consequences of competitive structures for allocative efficiency and productivity, as well as to work looking into the effects on banking sys-tem stability. Theoretical results largely depend on which particular model is chosen, and empirical approaches have also failed to support any firm conclusion.

1This section benefited from comments provided by the Bank of Canada and the Canadian Department of Finance.

2In an interview on September 3, 2004, Industry Minister David L. Emerson said that Canadian banks risk becoming low-level play-ers on global lending markets if Ottawa does not allow them to merge. See also Bond (2003).

3For example, a majority of respondents to a recent survey con-ducted among members of the Canadian Federation of Independent Business and other small and medium-sized enterprises agreed that competition in the financial services sector should increase before additional bank mergers were approved.

4See Northcott (2004) for a comprehensive review of the literature.

Standard IO methods applied to the banking indus-try suggest that perfect competition fosters allocative efficiency by channeling credit to its most productive use. However, other theoretical approaches that take into account some specific aspects of the banking busi-ness—such the presence of information asymmetry and the effect of a bank’s net worth on the quantity of credit supplied—suggest that a banking system with some market power may provide more and higher-quality credit. This result rests on the argument that banks with some monopoly power are more prepared to engage in costly activities that mitigate information asymmetries, such as relationship lending and screening (Petersen and Rajan, 1995; Cetorelli and Peretto, 2000).5 Empiri-cal studies also have failed to find convincing evidence that market power is detrimental to credit allocation (Northcott, 2004, and references therein). Earlier stud-ies found a negative relationship between the degree of competition and bank profits, but the results were not robust across time, products, or profit specification. More recent research has found that this relationship is weakened or eliminated if differences in banks’ pro-ductive efficiency are taken into account (see Berger, 1995; and Punt and van Rooij, 2001).

Standard IO methods also suggest that perfect competition achieves productive efficiency because it maximizes the quantity of credit supplied at the low-est interest rate. However, the results are not clear-cut once economies of scale—which usually exist in banking—are taken into account. Berger and Mester (1997) review studies that highlight some empirical evidence pointing at inefficiencies in banking, but it is not clear whether these arise from a lack of competition or unrealized scale economies.

There is no consensus in the literature on what mar-ket structure may promote prudent behavior, which benefits the stability of the financial system. Some studies suggest that market power may encourage pru-dent risk-taking and screening of borrowers, improving loan quality (Keeley, 1990; Salas and Saurina, 2003). However, other research finds that strong regulations

5In addition, Petersen and Rajan (1995), using U.S. data, also found that supply of credit to young firms is greater in the system with market power, which should encourage innovation and produc-tivity growth.

©International Monetary Fund. Not for Redistribution

117

and disclosure requirements can mitigate risk-taking and promote screening regardless of the competitive environment (Cordella and Levy Yeyati, 2002).

Methodology and Data

This paper uses a standard IO approach to assess the competitiveness of large Canadian banking groups. The Panzar-Rosse approach measures market power by the extent to which changes in factor prices are reflected in revenues. Under perfect competition, an increase in factor prices induces a proportional change in revenues because firms face perfectly elastic demand for their products. Conversely, under monopolistic competition, revenues change less than proportionally in response to changes in factor prices. In the case of a perfect monop-oly, there may be no response or even a negative response of revenues to changes in input costs. To measure the degree of competition in a particular market, Panzar and Rosse (1987) proposed the so-called H-statistic, which is computed as the sum of elasticities of revenues to unit factor costs in a reduced form revenue equation.

Although not initially intended to be applied to the financial system, Panzar and Rosse’s original method-ology has since been adapted to investigate the compet-itive structure of banks.6 Applications in the literature include the cases of Germany and several European countries (De Bandt and Davis, 2000; and IMF, 2003); the United States (Shaffer, 1982); and Canada (Nathan and Neave, 1989). Canada’s broad financial system was found not to exhibit monopoly power at the time, but the consolidation that has taken place in recent years suggests a need to revisit this result.

Following the approach of Nathan and Neave (1989), the following base model was estimated:

log INCNET = constant + α log PFUND + β log PCAP + λ log PLAB

+ δ log AAST, (1)

where INCNET is net income, PFUND is the unit price of funds, PCAP is the unit price of capital, PLAB is the unit price of labor, and ASST total assets. Total assets are included to identify possible scale economies, given the wide range of asset sizes across countries. For this model, H = α + β + λ. In a perfect competition case H = 1. A positive value of H, which is below unity, indicates monopolistic competition, with higher val-ues of H corresponding to a more competitive indus-try. Negative values of H could indicate either that the banking system is perfectly monopolistic or that the

6The extension of the Panzar-Rosse methodology to banking requires banks to be treated as single-product firms, consistent with the intermediation approach to banking in which banks are viewed as financial intermediaries (see Colwell and Davis, 1992, for details).

market is not in equilibrium (for example, because of structural change), in which case the H-statistic could not be applied.7

This analysis uses annual BankScope data for the larg-est banks in Canada, the United States, Germany, France, Italy, and the United Kingdom from 1999 to 2003. The unit price of funds was calculated as interest expenses over total deposits, the unit price of labor as personnel expenses over total liabilities, and the unit price of capital as other expenses divided by total liabilities.8 The sam-ple contains 35 banks in Canada, 27 of the largest U.S. banks, 290 French banks, 266 German banks, 127 Italian banks, and 200 banks in the United Kingdom.

Results

The empirical evidence supports the view that Cana-dian banks have grown more slowly than their major foreign competitors (Table 9.1). Although the Canadian Big Six have maintained profit growth, they have lagged

7The Panzar and Rosse (1987) model is based on the premise that the competitive structures under analysis are in a long-term equilib-rium. Adjustments to shocks or structural change could affect the way changes in factor prices translate into revenue changes, render-ing the H-statistic less useful.

8The cost per square foot of premises would be a better mea-sure of the cost of physical capital, but these data are not currently available.

Table 9.1. Size and Profitability Indicators of 25 Largest Banks

Total Assets1 Return on Return onCountry (billion US$) Assets Equity

1999Canada 37.4 0.6 8.6

Big Six 137.0 0.8 15.9France 113.0 0.5 2.5Germany 75.7 0.2 5.0Italy 55.9 0.9 10.8Spain 29.8 1.0 11.0United Kingdom 151.0 0.8 18.2United States 143.0 1.6 18.1

2003Canada 52.4 0.8 9.4

Big Six 195.0 0.8 14.7France 154.0 0.3 4.3Germany 140.0 0.1 1.9Italy 68.5 0.3 6.0Spain 45.8 0.9 10.0United Kingdom 325.0 0.6 12.3United States 217.0 1.6 16.7

Source: BankScope.1Unweighted average.

Results

©International Monetary Fund. Not for Redistribution

IX COMPETITION IN THE BANKING SYSTEM

118

their U.S. counterparts in profitability and, particularly, in balance sheet growth. The average size of the Big Six was comparable to banks in the United States and the United Kingdom in 1999. By 2003, however, large Canadian banks were on average 12 percent smaller than U.S. banks and more than 60 percent smaller than banks in the United Kingdom.

The Panzar-Rosse competitiveness measure provides a mixed assessment of the competitiveness of the Cana-dian banking system. The statistic was calculated for all banks in the sample, based on a fixed-effect panel esti-mation.9 The results indicate that the banking system in Canada is slightly more competitive than those in con-tinental Europe, the United Kingdom, and the 27 larg-est national banks in the United States (Table 9.2). At the same time, the H-statistics for the Big Six banks is negative, which could suggest the presence of some monopolistic power. Canada is clearly not a case with large banks operating as a perfect monopoly, given that the H-statistic is not significantly different from zero, but alternative tests confirm that the Panzar-Rosse sta-tistic is valid and not tainted by structural change.10

The analysis indicates that Canada is not the only country where large banks seem to enjoy some pricing power. The level of competition among large institu-

9The fixed-effect approach was suggested by a Hausman test.10As discussed above, the negative value of the H-statistic could

also indicate that the Canadian banking system goes through a period of structural change, in which case the Panzar-Rosse approach would not be valid. However, Shaffer (1982) argued that the return on assets or on equity (ROA or ROE) should not be correlated with input prices in the absence of structural change. Therefore, equa-tion (1) was re-estimated with log ROA on the left-hand side. The hypothesis H=0 could not be rejected, suggesting that the Panzar-Rosse statistic is valid in Canada’s case.

tions in other countries varies considerably and—as in Canada—differs from the competitiveness measure of the broader system for some countries. For example, when comparing the 25 largest banks by asset value across countries, the results suggest that only U.K. and Spanish large banks operate in a fully competitive envi-ronment (Table 9.3). All other countries are similar to Canada in that large banks enjoy some degree of monopolistic power.

The results also reveal that the number of large banks in a country is not as important for the level of compe-tition as their combined market share. For example, the Panzar-Rosse measure calculated for banks that hold 95 percent of a country’s total consolidated bank assets—equal to the market share of the Canadian Big Six—again suggests that banking systems in the United Kingdom and Spain are particularly exposed to compe-tition (Table 9.3, upper panel).11 However, competition increases across countries if the analysis is limited to banks that only account for 90 percent of total assets (Table 9.3, lower panel). At the same time, the number of banks accounting for a given level of bank assets varies greatly across countries and is not correlated with the Panzar-Rosse statistic (Table 9.4).

Conclusions

This section analyzes the level of competition in the Canadian banking system, with a particular focus on the six large banks, and compares it with banks in other

11For Canada, the market share of the Big Six only takes into account domestic banking assets, excluding foreign subsidiaries and credit unions.

Table 9.2. H-Statistics, by Country and Region

Region H-Statistics1 Standard Error P-Value

All countries 0.509** 0.084 0.000

United States2 0.283 0.182 0.119United Kingdom 0.581** 0.087 0.000Continental Europe 0.321** 0.192 0.095

France 1.150 4.200 0.784Germany 0.482** 0.134 0.000Italy 0.471** 0.127 0.000Spain 0.283** 0.122 0.020

Canada, all banks 0.698** 0.178 0.000Canada, Big Six3 –0.389 0.391 0.319

Source: IMF staff calculations.1Asterisks signify that coefficients are significant at the 5 percent level.2Only the 27 largest national banks are included.3The Big Six banks include Canadian Imperial Bank of Commerce (CIBC), Bank of Montreal, National Bank of

Canada, Toronto Dominion, Royal Bank of Canada, and Scotiabank.

©International Monetary Fund. Not for Redistribution

119

industrial countries. Using an approach from the IO lit-erature, the analysis does not reject the hypothesis that the Big Six may enjoy some degree of market power, although the broad banking system in Canada is found to be strongly competitive. The analysis of other coun-tries reveals similar differences between large banks

and the total banking system, with the United Kingdom and Spain being the only countries where large banks appear to operate in a fully competitive environment. The analysis also finds that the number of large banks in a country is not as important for the level of competition as their combined market share.

Table 9.3. H-Statistics for Largest Banks, by Country and Region

Region H-Statistics1 Standard Error P-Value

Panel A. Banks accounting for 95 percent of total consolidated assets

All countries 0.180 0.112 0.108

United States 0.261 0.185 0.158United Kingdom 0.528** 0.166 0.001France 0.629 0.385 0.102Germany –0.947 2.036 0.642Italy 0.025 0.170 0.884Spain 0.421** 0.157 0.007

Canada –0.389 0.391 0.319

Panel B. Banks accounting for 90 percent of total nonconsolidated assets

All countries 0.416** 0.123 0.001

United States 0.261 0.185 0.158United Kingdom 0.509** 0.145 0.000France 0.584** 0.286 0.041Germany –0.021 0.609 0.972Italy 0.295 0.389 0.447Spain 0.520** 0.157 0.001

Canada –0.249 0.531 0.639

Source: IMF staff calculations.1Asterisks signify that coefficients are significant at 5 percent level. Specification includes total assets.

Table 9.4. Size and Concentration of Banking Systems across Regions, 2003

Number of Banks Holding Number of Banks Holding Size 95 Percent of Assets 90 Percent of Assets (billion US$) (Consolidated) (Nonconsolidated)

Canada 1,274 6 6France 3,764 12 17Germany 2,264 4 14Italy 2,783 87 22Spain 1,493 15 11United Kingdom 7,226 12 17United States1 7,809 25 25

Sources: Statistics Canada; U.S. Federal Reserve; Primark DataStream; and BankScope.1National commercial banks only, accounting for 70 percent of consolidated assets.

Conclusions

©International Monetary Fund. Not for Redistribution

120

X Large Banking Groups and Financial System Soundness

Gianni De Nicolò, Alexander Tieman, and Robert Corker

This section1 uses market-based soundness indica-tors to assess the stability implications of the rapid

growth and shifting business strategies of Canada’s large banking groups. Helped by changes in the regula-tory framework, a process of mergers and acquisitions led to the emergence of six large banking groups that account for a large share of the Canadian financial system. Over the past 20 years, these banks diversi-fied their income and balance sheets by reducing the share of traditional deposit and lending activities and taking on more exposure to domestic and international financial markets. The soundness measures presented here—based on “distance-to-default” (DD) models—suggest that the rapid growth of the large banking groups during the second half of the 1990s was asso-ciated with a significant increase in their overall risk profile. More recently, however, their risk-adjusted returns appear to have improved, and their soundness was further underpinned by ongoing increases in capi-tal adequacy ratios.1

The strong performance of the six large banking groups speaks to a high degree of resilience of the Cana-dian financial system. In 2001, financial market weak-ness following the demise of the high-tech stock-market bubble and the global slowdown affected the banks’ performance. However, bank profitability recovered, and other financial soundness indicators continued to strengthen from already comfortable levels. Moreover, the relatively modest rise in default risk during the slowdown, compared with that during the Russian, Bra-zilian, and Long Term Capital Management (LTCM) crises of the late 1990s, suggests a strengthening of risk management practices on the part of the banks.

Banking Sector Trends

The Canadian banking sector is highly concentrated. The six large banking groups account for around 90 percent of Canadian deposits and banking assets. Based on mid-2004 balance-sheet data, the largest of these accounts for almost 25 percent of total banking

1Prepared with research assistance from Marianne El-Khoury.

assets, followed by four institutions each holding close to 15 percent of total assets. The sixth bank accounts for about 5 percent of total assets.

These banks went through a rapid growth spell in the second half of the 1990s, partly in response to legislative changes (Box 10.1). From the end of 1996 to the end of 2001, their deposit bases and total assets expanded by more than 50 percent. On the asset side, investments were channeled in particular into secu-rities and mortgages. After 2001, however, balance sheet expansion slowed dramatically, in part for cycli-cal reasons.

The investment behavior of these banks since the mid-1990s is the continuation of a longer trend reflect-ing the expansion of their fee-earning businesses. As part of their growth and diversification strategy, they aquired mortgage loan companies, securities busi-nesses, and trust companies (Calmes, 2004). Through this process, they expanded their links to financial mar-kets and, in particular, their exposure to equity mar-kets. In mid-2004, securities accounted for 27 percent of total assets, compared with 19 percent at the end of 1996 (Table 10.1).2

The six large banking groups also acquired substan-tial foreign investments. Consistent with many analysts’ view that the Canadian market does not offer suffi-cient banking economies of scale, these banks sought to expand their business abroad. During the 10 years ending in 2004, their foreign securities holdings grew at roughly three times the pace of domestic securities. While accumulating large U.S. dollar exposures in gen-eral, they also acquired extensive direct investments in the United States and the Caribbean.

The counterpart to this portfolio shift was a sharp decline in the relative importance of the banks’ tradi-tional lending business, excluding mortgages. Increases in nonmortgage lending barely kept pace with inflation after the mid-1990s, notwithstanding rapid growth in the banks’ deposit base. This was only partly offset by mortgage lending that expanded in line with, or even faster than, the total asset base. As a result, in 2004 the

2This reflects an increase in the share of assets devoted to securi-ties for all except one bank.

©International Monetary Fund. Not for Redistribution

121

banks’ net interest income accounted for less than half of total net income, compared with about 60 percent in the mid-1990s.

These trends and strategies are not unique to large banking groups in Canada. Most of the large finan-cial institutions in the United States and elsewhere also expanded their securities and mortgage busi-nesses and consequently derive a greater share of their

income from fees and other non-interest sources (IMF, 2004c).

Foreign and financial market exposures contributed to significant strains in the banking sector after the stock market bubble burst in the early 2000s. Although the downturn of the Canadian economy was relatively mild, non-interest income growth dropped from over 25 percent a year between 1996 and 2000 to virtually

The rapid growth of the large banking groups has been facilitated by changes to Canada’s financial sector legislation:• Amendments to the regulatory framework in 1987 and

1992 removed legal barriers separating the activities of various types of financial institutions, allowing Cana-dian financial institutions to develop into financial con-glomerates (Freedman, 1998).

• In 2001, limitations on investment in nonfinancial busi-ness were relaxed, and a holding company regime was introduced. At the same time, a new merger review policy was introduced, mainly in response to the government’s 1998 decision not to allow two mergers involving four of Canada’s largest banks (Ministry of Finance, 1998; Group of Ten, 2001).

• The new policy raised the ownership limit to 20 percent for voting shares and 30 percent for nonvoting shares and loosened the requirement that large banks be widely held, while retaining the requirement that no investor hold a majority share in the bank (Daniel, 2002).The Canadian regulatory and supervisory regime is

subject to regular reviews in order to keep pace with the changing technological and market environment. Can-ada’s financial legislation contains sunset clauses that prescribe periodic reassessments and updating of the regulatory framework that governs the financial system. In the past, the review used to take place every 10 years. In 1992, the review period was shortened to five years and extended to the legislation governing all federal financial institutions.

Box 10.1. Changes in Canada’s Financial Sector Legislation

Table 10.1. Balance Sheet Items of Six Large Banking Groups(In percent, unless otherwise noted)1

1996 2000 2001 2002 2003 2004

Total assetsIn billions of Canadian dollars 1,015 1,438 1,587 1,628 1,627 1,671Annual percent change2 9.1 10.4 2.6 –0.1 2.7As percent of GDP 121.3 133.5 143.2 140.6 133.5 129.4

Aggregate balance sheet shares Assets

Securities 19.0 23.4 25.2 24.7 27.4 26.8Non-mortgage loans 41.0 37.2 36.0 34.6 31.0 32.7Mortgage loans 19.9 19.3 19.3 20.0 21.2 21.4Other 20.1 20.1 19.5 20.7 20.4 19.1

LiabilitiesDeposits 68.2 67.6 66.6 66.1 65.8 67.1Other 31.8 32.4 33.4 33.9 34.2 32.9

Net income before tax3 1.0 1.0 0.8 0.6 0.9 1.2

Net interest income/total net income 60.6 39.5 40.9 43.6 47.9 47.2

Capital adequacy ratio4 9.2 11.7 12.3 12.3 13.3 13.4

Source: Office of the Superintendent of Financial Institutions.1Figures as of year-end, except for 2004, for which figures are for June.2Figure for 2000 is the average during 1996–2000.3As percent of total assets.4In percent of risk-weighted assets.

Banking Sector Trends

©International Monetary Fund. Not for Redistribution

X LARGE BANKING GROUPS AND FINANCIAL SYSTEM SOUNDNESS

122

zero between 2001 and 2003. With impaired assets and loan loss provisions up sharply, return on equity for the banking system as a whole fell from 15.8 percent in 1999 to 9.4 percent in 2002, but has since rebounded (Table 10.2).

Nevertheless, throughout both the growth spell and subsequent slowdown, Canada’s large banking groups continued to build up capital relative to risk-weighted assets. Their total capital increased by about 70 percent between 1996 and 2001, translating into a rise in the average capital ratio from 9.2 percent to 12.3 percent. Following a drop in both capital and risk-weighted assets in 2002, the average capital ratio increased to 13.3 percent in 2003, boosted by increases in capital and a further decline in risk-weighted assets. Through-out the period, all banks remained substantially above the minimum capital requirements set in the 1988 Basel capital accords.

The banks have also reduced their off-balance-sheet activities. Prior to 1998, they took on increasing expo-sures in the derivatives markets and also boosted other off-balance-sheet exposures. These were scaled back in subsequent years, following the Russia and Brazil cri-ses and the LTCM collapse. With one exception, these banks have not expanded—and in some cases made further sizable contractions in—their off-balance-sheet activity since 1999.

Despite similar features, the business strategies of the six large banking groups have not been identical and financial performance has been varied. Two of these, the largest and one of the mid-tier banks, deviated from the other institutions in that they built up their secu-

rities portfolio share more rapidly and reduced their share of mortgage loans. The financial performance of these two institutions has been quite different. The largest bank appeared to weather the post-2000 slow-down relatively comfortably, whereas profitability in the other bank—which had the fastest growing balance sheet and the most rapid buildup of securities invest-ments of all six large banks—was quite volatile, culmi-nating in pretax losses in 2002. One other bank—the one with the fastest growing exposure to the mortgage market—also saw its profits fall to a very low level in 2002, whereas the other three institutions experienced only mild dips in profitability.

Market-Based Financial Soundness Indicators

The large banking groups’ business strategies affect the financial soundness of both these individual insti-tutions and the financial system as a whole. Increased diversification, both internationally and across differ-ent business lines, should in principle yield better risk profiles for financial institutions. But when diversifica-tion takes place at the expense of either lower capi-tal or profitability, on one hand, or increased earnings volatility, on the other, financial soundness may not necessarily improve—as reflected in the performance of individual banks. Moreover, were all the large banks to diversify in the same direction, the system’s overall vulnerability to large, common shocks (that is, the risk of several banks experiencing distress at the same time)

Table 10.2. Vulnerability Indicators of the Banking System1

1997 1998 1999 2000 2001 2002 2003

Balance sheet Total loans to assets (percent) 62.8 58.4 58.7 57.7 55.8 55.6 53.8Total loans to deposits (percent) 92.5 90.2 86.8 84.2 84.5 83.1 81.3Impaired assets/total assets 0.68 0.66 0.59 0.60 0.84 0.90 0.64Loan loss provision (in percent of total assets) 0.18 0.20 0.23 0.27 0.37 0.57 0.23Total foreign currency assets/total assets2 41.5 46.4 40.2 40.5 42.7 41.4 36.2Total foreign currency liabilities/total assets2 43.5 47.7 42.4 42.2 44.7 43.8 37.7Total foreign currency deposits/total assets2 30.7 32.2 31.3 29.5 30.4 28.9 25.6

Profitability Return on total shareholders’ equity (percent) 16.4 13.4 15.8 15.3 13.9 9.4 14.7Return on average assets 0.71 0.57 0.71 0.72 0.66 0.44 0.69Average intermediation spread 2.9 2.6 2.7 2.9 3.0 3.0 3.1Net interest income

(in percent of average total assets) 2.1 1.8 1.8 1.8 1.9 2.0 1.9

Sources: Bloomberg; Canadian Bankers Association; Haver Analytics; and Office of the Superintendent of Financial Institutions.1Unless otherwise indicated, based on data reported by the six largest chartered Canadian banks, which account for more than 90 percent of the

total market share.2All chartered banks.

©International Monetary Fund. Not for Redistribution

123

could increase even if each bank were individually bet-ter hedged against risks.

This subsection evaluates trends in the vulner-ability of Canada’s financial system using a market-based soundness indicator. The indicator, which was estimated for the period 1991–2003, is based on DD models commonly used in the finance literature and increasingly reported in central bank financial stabil-ity reports. DD is a composite measure computed as the sum of the return on the estimated market value of assets and the capital-to-assets ratio at market prices, divided by the volatility of assets.3 It thus combines measures of profitability, balance sheet strength, and market uncertainty. A higher DD indicates an improve-ment in financial soundness at the company level, for example, because of improved profitability, a higher capital ratio, or reduced volatility—or a combination of all three.4 Although DD measures are sensitive to vari-ations in the underlying assumptions, they have been shown to predict supervisory ratings, bond spreads, and rating agencies’ downgrades.5

The analysis provides a number of conclusions regarding the impact of evolving business strategies on large banking groups’ financial soundness:• There appears to be no trend movement in the aver-

age DD for the six large banking groups (LBGs) since 1990 (Figure 10.1). Although the DD narrowed sharply in the late 1990s, as a substantial deterio-ration in the risk-adjusted return outweighed rising capital ratios, the average risk profile improved sig-nificantly in subsequent years. Overall, this suggests that shifting business strategies and changing bal-ance sheet structures have not led to a noticeable increase or decrease in average default probabilities for the large banking groups.

• With one notable exception, there has been no ten-dency for risk to concentrate in the largest banks. Risk concentration can be measured as the differ-ence between the weighted and unweighted average of DDs over all six institutions, with weights given by each bank’s share of total market valuation. A

3Estimates of the market value of assets are based on the structural valuation model of Black and Scholes (1973) and Merton (1974) and were computed using the estimation procedure described in Vas-salou and Xing (2004) using daily market and annual accounting data.

4Two caveats apply. First, the DD as employed here does not take into account the stochastic interest rate risk stemming from the correlation between the risk-free rate and the value of a company’s assets. As this is potentially another important source of risk for banks, risks might be underestimated in this analysis. See Liu, Papakirykos, and Yuan (2004) for an extension incorporating inter-est rate risk. Second, as the DD is based on market data, the DDs for large banking groups can be subject to large fluctuations, which tend to be associated with the business cycle and “expectation cycles” regarding future earnings prospects.

5See Krainer and Lopez (2001); Gropp, Vesala, and Vulpes (2002); and Chan-Lau, Jobert, and Kong (2004).

decline in this measure indicates a concentration of risk in the largest banks, and vice versa. For most of the period observed, this measure of concentration was unchanged. However, it dropped in 2002, owing to a substantial reduction in the market valuation of two of the largest LBGs.6 The subsequent spike high-lighted risks in two of Canada’s important banks, even though average soundness measures for the sec-tor remained satisfactory (see Figure 10.5).

• Risk profiles of the large banking groups do not appear to have converged. The standard deviation of DD indicators for the six banks does not exhibit a trend over the past 15 years. In fact, a rising stan-dard deviation suggests that banks’ risk profiles have diverged somewhat in the last few years.

• The large banking groups appear resilient to com-mon shocks. A “system” DD can be computed using market measures of return and volatility for the aggregate subsector and its capital-asset ratio. As such, it measures the default risk of an entity whose balance sheet is the aggregate of all the large bank-ing groups’ balance sheets. The system DD is almost always substantially higher than the average DD, indicating that correlations between the portfolios of the different banks are low, and hence that the group of large banks as a whole remains well diversi-fied (Figure 10.2). The exception was around 2000, when a relatively high correlation among individual DD measures—reflecting the broad-based decline in equity prices—implied that the system as a whole fared no better than individual large banks. How-ever, the difference between the system and average

6If balance sheet valuations of assets are used to weight the DDs, the downward spike largely disappears. However, the use of market valuations seems preferable for diagnostic purposes because the objective is to capture the market’s assessment of risk.

Source: IMF staff calculations.

0

3

6

9

12.5

0320019919971995931991

Figure 10.1 Average Distance to Default (DD) of Large Banking Groups

Market-Based Financial Soundness Indicators

©International Monetary Fund. Not for Redistribution

X LARGE BANKING GROUPS AND FINANCIAL SYSTEM SOUNDNESS

124

DD subsequently rebounded to above its long-term average.

• However, high correlation of DD measures across sectors suggests that the broader financial system typically has a high degree of common risk expo-sure. In particular, the correlation between system DDs of the banking and insurance sectors is mostly around unity, suggesting that risks across the finan-cial sector are closely aligned (Figure 10.3).7 In effect, most of the time, the fortunes of the large banking groups and other financial institutions are expected to follow similar paths. Risks in the bank-ing sector and in the nonfinancial sector also tend to be positively correlated—with the notable excep-tion of the period between 1999 and 2001, when market sentiment turned against financial institu-tions well ahead of the general fall in the stock market. The relatively strong performance of financial

soundness indicators for large Canadian banks likely reflects improvements in risk management. Perhaps surprisingly at first sight, given the banks’ increased exposure to stock markets, the deterioration of DD indicators after 2000 was relatively modest. In par-ticular, the decline in the average DD—signaling a rise in vulnerability—was much less than the drop follow-ing the Russian default, Brazil crisis, or LTCM col-lapse in 1998 and 1999, when market volatility of bank valuations increased substantially. This ob servation is consistent with anecdotal evidence that large Cana-dian banks, like other large and internationally active banks, have improved their risk-management capa-

7Correlations are computed using daily data and a one-year roll-ing window.

bilities in recent years, strengthening their ability to absorb the impact of shocks on their balance sheets (see, for example, Bank of International Settlements, 2002).

Soundness Indicators in Canada and the United States

The broad trends in financial soundness indicators in Canada have much in common with those in the United States. Large banking groups in both countries have undergone rapid growth and similar structural shifts in their balance sheets. During the second half of the 1990s, this expansion was accompanied by almost identical declines in average DD measures for U.S. and Canadian large banking groups (Figure 10.4).8Since 2000, DDs in both countries have been volatile, although on average they have increased by a greater amount in the Canadian case—which may reflect a somewhat stronger build-up of capital ratios. Nonethe-less, differences between the two countries are small, and large banking groups in both countries have come through the recent testing period of economic down-turn and financial market distress with strengthened soundness indicators.

In contrast to the United States, structural change does not appear to have led to an increase in risk con-centration in the Canadian banking system. In Can-ada, the risk-concentration measure (weighted minus unweighted average DD) has been relatively unchanged since the beginning of the 1990s (Figure 10.5). In

8For more analysis of DD measures in the United States, see IMF (2004c).

Source: IMF staff calculations.Note: DD = distance to default.

–0.5

0

0.5

1.0

1.5

2.0

2.5

032001999795931991

Figure 10.2. System DD Minus Average DD of Large Banking Groups

–1

–0.5

0

0.5

1

032001999795931991

Source: IMF staff calculations.Note: DD = distance to default.

Figure 10.3. Correlation between Banking and Insurance Sector DDs

©International Monetary Fund. Not for Redistribution

125

essence, the Canadian system has remained dominated by the same number of players, broadly equal in size and with relatively independent risk profiles. By con-trast, risk concentration in the United States increased markedly in the second half of the 1990s, owing in part to consolidation—on the same scale, the temporary increase in risk concentration in the Canadian system in 2002 noted above is much smaller.

Conclusions

This analysis suggests that substantial growth and structural change in the Canadian banking sector have not added to systemic risk over time. Default risk did

rise significantly when balance sheets expanded rapidly in the second half of the 1990s. However, helped by the ongoing buildup of additional capital and seemingly improved risk management practices, risk profiles have subsequently recovered. Moreover, the large banking groups’ individual risk profiles have remained rela-tively uncorrelated, notwithstanding that their business strategies have shared many common features, and risk concentration has not on average increased. However, with a relatively small number of players, each individ-ual large bank is of systemic consequence. A close cor-relation of risk profiles between banks and insurance companies implies that the financial sector as a whole remains exposed to common sources of risks.

Source: IMF staff calculations.Note: DD = distance to default.

0

2

4

6

8

10

12

United States

Canada

032001999795931991

Figure 10.4. Average DD in Canada and the United States

Source: IMF staff calculations.Note: DD = distance to default.

–0.50

–0.25

0

0.25

0.50

Canada (right axis)

032001999795931991–1

–0.50

0.51.01.52.02.53.0

United States (left axis)

Figure 10.5. Risk Concentration in Canada and the United States(Difference between weighted and unweighted average of DDs)

Conclusions

©International Monetary Fund. Not for Redistribution

126

References

Acharya, R.C., P. Sharma, and S. Rao, 2003, “Canada-U.S. Trade and FDI Patterns,” in North American Linkages: Opportunities and Challenges for Canada, Industry Canada Research Series, Vol. 11, ed. by R. Harris (Cal-gary, Alberta: University of Calgary Press).

Alberta Employment, Immigration and Industry, 2006, Oil Sands Industry Update. Available via the Internet: http://www.alberta-canada.com/oandg/files/pdf/oil-SandsUpdate_December2006.pdf.

Allen, Franklin, and Douglas Gale, 2000, Comparing Finan-cial Systems (Cambridge, Massachusetts: MIT Press).

Alt, R., 1993, “Revenue Forecasting and Estimation: How It’s Done, State by State,” State Tax Notes, Vol. 4, pp. 1038–51.

Amano, R., and S. van Norden, 1993, “A Forecasting Equa-tion for the Canada–U.S. Dollar Exchange Rate,” in The Exchange Rate and the Economy (Ottawa: Bank of Canada).

Amato, J.D., and H.S. Shin, 2003, “Public and Private Infor-mation in Monetary Policy Models,” BIS Working Paper No. 138 (Basel: Bank for International Settlements). Available via the Internet: http://www.bis.org/publ/work138.pdf.

Armstrong, P., T. Harchaoui, C. Jackson, and F. Tarkhani, 2002, “A Comparison of Canada-U.S. Economic Growth in the Information Age, 1981–2000: The Importance of Investment in Information and Communication Tech-nologies,” Statistics Canada Research Paper No. 001 (Ottawa: Statistics Canada). Available via the Internet: http://www.statcan.ca/english/research/11F0027MIE/11F0027MIE2002001.pdf.

Artis, M.J., 1996, “How Accurate Are the IMF’s Short-Term Forecasts? Another Examination of the World Eco-nomic Outlook,” IMF Working Paper 96/89 (Washing-ton: International Monetary Fund).

———, and M. Marcellino, 2001, “Fiscal Forecasting: The Track Record of the IMF, OECD, and EC,” Economet-rics Journal, Vol. 4 (June), pp. S20–36.

Ash, J.C.K., D.J. Smyth, and S.M. Heravi, 1998, “Are OECD Forecasts Rational and Useful? A Directional Analysis,” International Journal of Forecasting, Vol. 14 (Septem-ber), pp. 381–91.

Ashley, R., C.W.J. Granger, and R. Schmalensee, 1980, “Advertising and Aggregate Consumption: An Anal-ysis of Causality,” Econometrica, Vol. 48 (July), pp. 1149–67.

Atoyan, Ruben, 2004, “Econometric Investigation of Policy Preference Evolution: The Bank of Canada Case” (unpub-lished; Washington: International Monetary Fund).

Bailliu, J., and M.R. King, 2005, “What Drives Movements in Exchange Rates?” Bank of Canada Review (Autumn), pp. 27–39.

Baker, M., and D. Benjamin, 1999, “Early Retirement Pro-visions and the Labor Force Behavior of Older Men: Evidence from Canada,” Journal of Labor Economics,Vol. 17, Part 1 (October), pp. 724–56.

Baker, M., J. Gruber, and K. Milligan, 2003, “The Retire-ment Incentive Effects of Canada’s Income Security Programs,” Canadian Journal of Economics, Vol. 36 (May), pp. 261–90.

Baldwin, J., and J. Chowhan, 2003, “The Impact of Self-Employment on Productivity Growth in Canada and the U.S.,” Canadian Economic Observer, Vol. 16 (August), pp. 21–30.

Baldwin, J., and J. Tang, 2003, “The Contribution of Small and Medium Sized Producers to the Productivity Gap Between the Manufacturing Sectors of Canada and the United States,” paper presented at the Annual Meet-ing of the Canadian Economics Association, Carleton University, Ottawa, May 30–June 1. Available via the Internet: http://www.csls.ca/events/cea2003/baldwin-tang-cea2003.pdf.

Bank for International Settlements (BIS), 2002, BIS 72nd Annual Report (Basel). Available via the Internet: http://www.bis.org/publ/arpdf/ar2002e.pdf.

Basu S., J. Fernald, N. Oulton, and S. Srinivasan, 2003, “The Case of the Missing Productivity Growth: Or, Does Information Technology Explain Why Productivity Accelerated in the US but not the UK?” NBER Working Paper No. 10010 (Cambridge, Massachusetts: National Bureau of Economic Research).

Batchelor, R.A., 2001, “How Useful Are the Forecasts of Intergovernmental Agencies? The IMF and OECD Ver-sus the Consensus,” Applied Economics, Vol. 33 (Febru-ary), pp. 225–35.

Battle, K., 2003, “Sustaining Public Pensions in Canada: A Tale of Two Reforms,” in Taste of Pie: Searching for Better Pension Provisions in Developed Countries, ed. by N. Takayama (Tokyo: Maruzen).

Baylor, M., 2005, “Ranking Tax Distortions in Dynamic General Equilibrium Models: A Survey,” Working Paper 2005-06 (Ottawa: Department of Finance).

———, and L. Beauséjour, 2004, “Taxation and Economic Efficiency: Results from a Canadian CGE Model,” Work-ing Paper 2004-10 (Ottawa: Department of Finance).

Bayoumi, T., and D. Botman, 2005, “Jam Today or More Jam Tomorrow?” in Canada: Selected Issues, IMF Country

©International Monetary Fund. Not for Redistribution

References

127

Report No. 05/116 (Washington: International Mon-etary Fund).

Bayoumi, T., and R. Cardarelli, 2005, “How Flexible Is the Canadian Economy?” in Canada: Selected Issues, IMF Country Report No. 05/116 (Washington: International Monetary Fund).

Bayoumi, T., H. Faruqee, and J. Lee, 2005, “A Fair Exchange? Theory and Practice of Calculating Equi-librium Exchange Rates,” IMF Working Paper 05/229 (Washington: International Monetary Fund).

Bayoumi, T., and V. Klyuev, 2006, “Canadian Inflation Tar-geting and Macroeconomic Volatility in Retrospect and Prospect,” in Canada: Selected Issues, IMF Country Report No. 06/229 (Washington: International Mon-etary Fund).

Bayoumi, T., and M. Mühleisen, 2006, “Energy, the Exchange Rate, and the Economy: Macroeconomic Benefits of Canada’s Oil Sands Production,” IMF Working Paper 06/70 (Washington: International Monetary Fund).

Bayoumi, T., and others, 2004, GEM: A New International Macroeconomic Model, IMF Occasional Paper No. 239 (Washington: International Monetary Fund).

Bayoumi, T., and S. Sgherri, 2004a, “Deconstructing the Art of Central Banking,” IMF Working Paper 04/195 (Washington: International Monetary Fund).

———, 2004b, “Monetary Magic? How the Fed Improved the Flexibility of the Economy,” IMF Working Paper 04/24 (Washington: International Monetary Fund).

Bayoumi, T., B. Sutton, and A. Swiston, 2006, “Shocking Aspects of Canadian Labor Markets,” IMF Working Paper 06/83 (Washington: International Monetary Fund).

Bean, Charles, Jens Larsen, and Kalin Nikolov, 2002, “Finan-cial Frictions and the Monetary Transmission Mecha-nism: Theory, Evidence, and Policy Implications,” ECB Working Paper No. 113 (Frankfurt: European Central Bank).

Berg, Andrew, Philippe Karam, and Douglas Laxton, 2006, “A Practical Model-Based Approach to Monetary Pol-icy Analysis—Overview,” IMF Working Paper 06/80 (Washington: International Monetary Fund).

Berg, A., D. Laxton, and P. Karam, 2006, “Practical Model-Based Monetary Policy Analysis—A How-To Guide,” IMF Working Paper 06/81 (Washington: International Monetary Fund).

Berger, A., 1995, “The Profit-Structure Relationship in Bank-ing—Tests of Market-Power and Efficient-Structure Hypotheses,” Journal of Money, Credit and Banking,Vol. 27 (May), pp. 404–31.

———, and L. Mester, 1997, “Inside the Black Box: What Explains Differences in the Efficiencies of Financial Institutions,” Journal of Banking and Finance, Vol. 21, No. 7, pp. 895–947.

Bernanke, Ben, and Mark Gertler, 1995, “Inside the Black Box: The Credit Channel of Monetary Policy Trans-mission,” Journal of Economic Perspectives, Vol. 9 (Autumn), pp. 27–48.

———, and Simon Gilchrist, 1999, “The Financial Accel-erator in a Quantitative Business Cycle Framework,” in Handbook of Macroeconomics, Vol. 1C, ed. by J.B. Taylor and M. Woodford (Amsterdam: North-Holland).

Bernanke, Ben, Mark Gertler, and Mark Watson, 1997, “Sys-temic Monetary Policy and the Effects of Oil Price Shocks,” Brookings Papers on Economic Activity: 1,Brookings Institution, pp. 91–142.

Birol, F., and W. Davie, 2001, “Oil Supply Costs and Enhanced Productivity,” Energy Prices and Taxes(Fourth Quarter), pp. 17–23. Available via the Internet: http://data.iea.org/ieastore/assets/products/eptnotes/feature/4Q2001B.pdf.

Bjørnland, Hilde C., 1998, “The Economic Effects of North Sea Oil on the Manufacturing Sector,” Scottish Journal of Political Economy, Vol. 45, No. 5, pp. 552–85.

Black, F., and M. Scholes, 1973, “The Pricing of Options and Corporate Liabilities,” Journal of Political Economy,Vol. 81, No. 3, pp. 637–54.

Blanchard, Olivier J., 1985. “Debt, Deficits, and Finite Horizons,” Journal of Political Economy, Vol. 93, pp. 223–47.

Blöndal, J.R., and J.K. Kristensen, 2002, “Budgeting in the Netherlands,” OECD Journal of Budgeting, Vol. 1, No. 3, pp. 43–80.

Boivin, Jean, and Marc Giannoni, 2002, “Assessing Changes in the Monetary Transmission Mechanism: A VAR Approach,” Federal Reserve Bank of New York Eco-nomic Policy Review, Vol. 8 (May), pp. 97–111.

Bond, D., 2003, “Bank Mergers: Why We Need Them, How to Get Them,” Backgrounder No. 74 (Toronto: C.D. Howe Institute). Available via the Internet: http://www. cdhowe.org/pdf/backgrounder_74.pdf.

Botman, D., 2006, “Efficiency Gains from Reducing the GST Versus Personal Income Taxation in Canada,” in Can-ada: Selected Issues, IMF Country Report No. 06/229 (Washington: International Monetary Fund).

———, and D. Laxton, 2004, “The Effects of Tax Cuts in a Global Fiscal Model,” Box 2.2 in World Economic Outlook, April (Washington: International Monetary Fund).

———, D. Laxton, D. Muir, and A. Romanov, 2006, “A New-Open-Economy Macro Model for Fiscal Policy Evaluation,” IMF Working Paper 06/45 (Washington: International Monetary Fund).

Bowden, R., and V. Martin, 1996, “International Business Cycles and Financial Integration,” Review of Economics and Statistics, Vol. 77 (May), pp. 305–20.

Burbidge, J., D. Fretz, and M.R. Veall, 1998, “Canadian and American Saving Rates and the Role of RRSPs,” Cana-dian Public Policy, Vol. 24 (June), pp. 259–63.

Burbidge, J., and A. Harrison, 1985, “(Innovation) Account-ing for the Impact of Fluctuations in U.S. Variables on the Canadian Economy,” The Canadian Journal of Economics, Vol. 18 (November), pp. 784–98.

Calmes, Christian, 2004, “Regulatory Changes and Finan-cial Structure: The Case of Canada,” Working Paper No. 2004–26 (Ottawa: Bank of Canada). Avail-able via the Internet: http://www.bankofcanada.ca/en/res/2004/wp04-26.htm.

Campolieti, M., 2001, “The Canada/Quebec Pension Plan Disability Program and the Labor Force Participation of Older Men,” Economic Letters, Vol. 70 (March), pp. 421–26.

Canadian Association of Petroleum Producers (CAPP), 2005, “Canadian Crude Oil Production and Supply Forecast, 2005–

©International Monetary Fund. Not for Redistribution

REFERENCES

128

2015” (Calgary). Available via the Internet: http://www.capp.ca/raw.asp?x=1&dt=NTV&e=PDF&dn=89357.

Canadian Energy Research Institute (CERI), 2004, Oil Sands Supply Outlook: Potential Supply and Costs of Crude Bitumen and Synthetic Crude Oil in Canada, 2003–2017 (Calgary).

Cardarelli, R., 2003, “Assessing the Long-Term Fiscal Posi-tion of Canada,” in Canada: Selected Issues, IMF Country Report No. 03/34 (Washington: International Monetary Fund).

———, and M.A. Kose, 2004, “Economic Integration, Busi-ness Cycle, and Productivity in North America,” IMF Working Paper 04/138 (Washington: International Mon-etary Fund). Also in Canada in the Global Economy,proceedings of a conference held at the Bank of Canada, November 2004 (Ottawa: Bank of Canada). Available via the Internet: http://www.bankofcanada.ca/en/con-ference/2004/integration.pdf.

Cecchetti, Stephen G., 1999, “Legal Structure, Financial Structure, and the Monetary Policy Transmission Mech-anism,” Federal Reserve Bank of New York Economic Policy Review, Vol. 5 (July), pp. 19–28.

Cetorelli, N., and P. Peretto, 2000, “Oligopoly Banking and Capital Accumulation,” Working Paper No. 12 (Chi-cago: Federal Reserve Bank of Chicago).

Chan-Lau, J., A. Jobert, and J. Kong, 2004, “An Option-Based Approach to Bank Vulnerabilities in Emerging Markets,” IMF Working Paper 04/33 (Washington: International Monetary Fund).

Christiano, Lawrence, Martin Eichenbaum, and Charles Evans, 1999, “Monetary Policy Shocks: What Have We Learned and to What End?” in Handbook of Macroeco-nomics, Vol. 1A, ed. by J.B. Taylor and M. Woodford (Amsterdam: North-Holland).

Clarida, Richard, Jordi Galí, and Mark Gertler, 1998, “Mon-etary Policy Rules in Practice: Some International Evi-dence,” European Economic Review, Vol. 42 (June), pp. 1033–67.

———, 1999, “The Science of Monetary Policy,” Jour-nal of Economic Literature, Vol. 37 (December), pp. 1661–707.

———, 2001, “Optimal Monetary Policy in Open versus Closed Economies: An Integrated Approach,” American Economic Review, Vol. 91 (May), pp. 248–52.

Coats, W., D. Laxton, and D. Rose, eds., 2003, The Czech National Bank’s Forecasting and Policy Analy-sis System (Prague: Czech National Bank). Avail-able via the Internet: http://www.cnb.cz/www.cnb.cz/en/publications/download/forecasting_2003_1-4.pdf.

Colwell, R., and E.P. Davis, 1992, “Output and Productiv-ity in Banking,” Scandinavian Journal of Economics,Vol. 94 (Supplement), pp. S111–29.

Cordella, T., and E. Levy Yeyati, 2002, “Financial Open-ing, Deposit Insurance, and Risk in a Model of Bank-ing Competition,” European Economic Review, Vol. 46 (March), pp. 471–85.

Corden, W.M., 1981, “The Exchange Rate, Monetary Pol-icy and North Sea Oil: The Economic Theory of the Squeeze on Tradeables,” Oxford Economic Papers,Vol. 33 (Supplement, July), pp. 23–46.

Crawford, A., 2002, “Trends in Productivity Growth in Can-ada,” Bank of Canada Review (Spring), pp. 19–32.

Cushman, D., and T. Zha, 1997, “Identifying Monetary Pol-icy in a Small Open Economy Under Flexible Exchange Rates,” Journal of Monetary Economics, Vol. 39 (August), pp. 433–48.

Dabán, T., E. Detragiache, G. di Bella, G.M. Milesi-Ferretti, and S. Symansky, 2003, Rules-Based Fiscal Policy in France, Germany, Italy, and Spain, IMF Occasional Paper No. 225 (Washington: International Monetary Fund).

Daniel, F., 2002, “Recent Changes to Canada’s Financial Sector Legislation,” Bank of Canada Review (Winter 2002–2003). Available via the Internet: http://www.bankofcanada.ca/en/review/2002/daniele.pdf.

De Bandt, O., and E.P. Davis, 2000, “Competition, Con-testability and Market Structure in European Banking Sectors on the Eve of EMU,” Journal of Banking and Finance, Vol. 24 (June), pp. 1045–66.

De Masi, P., and I. Ivaschenko, 2003, “Corporate Gover-nance in Canada: Coping with Financial and Regulatory Spillovers from the United States,” in Canada: Selected Issues, IMF Country Report No. 03/34 (Washington: International Monetary Fund).

De Masi, P., and C. Towe, 2003, “The Canadian Health Care System: Cross-Country Comparisons and Options for Reform,” in Canada: Selected Issues, IMF Country Report No. 03/34 (Washington: International Monetary Fund).

Demers, F., 2003, “The Canadian Phillips Curve and Regime Shifting,” Working Paper 2003−32 (Ottawa: Bank of Canada).

De Nicolò, G., P. Hayward, and A.V. Bhatia, 2004, “U.S. Large Complex Banking Groups: Business Strate-gies, Risks, and Surveillance Issues,” in United States: Selected Issues, IMF Country Report No. 04/228 (Washington: International Monetary Fund). Available via the Internet: http://www.imf.org/external/pubs/ft/scr/2004/cr04228.pdf.

De Nicolò, G., A. Tieman, and R. Corker, 2005, “Large Bank-ing Groups and Financial System Soundness,” in Can-ada: Selected Issues, IMF Country Report No. 05/116 (Washington: International Monetary Fund).

Department of Finance Canada, 2004a, The Economic and Fiscal Update: Annexes to the Presentation (Ottawa). Available via the Internet: http://www.fin.gc.ca/ec2004/ec04e.pdf.

———, 2004b, “Taxation and Economic Efficiency: Results from a General Equilibrium Model,” in Tax Expenditures and Evaluations 2004 (Ottawa). Available via the Internet: http://www.fin.gc.ca/taxexp/2004/TaxExp04_e.pdf.

———, 2004c, Tax Expenditures and Evaluations 2004 (Ottawa). Available via the Internet: http://www.fin.gc.ca/taxexp/2004/TaxExp04_e.pdf.

Dion, R., 1999, “Trends in Canada’s Merchandise Trade,” Bank of Canada Review (Winter 1999–2000). Avail-able via the Internet: http://www.bankofcanada.ca/en/review/2000/r001c-e.pdf.

Dodge, D., 2005, “Our Approach to Monetary Policy: Inflation Targeting,” remarks to the Regina Chamber of Commerce, Regina, Saskatchewan, December 12. Available via the Internet: http://www.bank-banque-canada.ca/en/speeches/2005/sp05-18.html.

©International Monetary Fund. Not for Redistribution

References

129

Elmendorf, D.W., and N.G. Mankiw, 1999, “Government Debt,” in Handbook of Macroeconomics, Vol. 1C, edited by J.B. Taylor and M. Woodford (Amsterdam: North-Holland).

Estrella, Arturo, 2001, “Financial Innovation and the Mon-etary Transmission Mechanism” (unpublished; New York: Federal Reserve Bank of New York). Available via the Internet: http://www.newyorkfed.org/research/economists/estrella/estrellapaper.pdf.

———, 2002, “Securitization and the Efficacy of Monetary Policy,” Federal Reserve Bank of New York Economic Policy Review, Vol. 8 (May), pp. 271–84.

Evers, M., R. de Mooij, and D. van Vuuren, 2005, “What Explains the Variation in Estimates of Labour Supply Elasticities?” CESifo Working Paper No. 1633 (Munich: Center for Economic Studies). Available via the Inter-net: http://www.cesifo.de/wp.

Fair, R.C., and R.J. Shiller, 1990, “Comparing Information in Forecasts from Econometric Models,” American Eco-nomic Review, Vol. 80 (June), pp. 375–89.

Faruqui, U., W. Gu, M. Kaci, M. Laroche, and J.P. Maynard, 2003, “Differences in Productivity Growth: Canadian-U.S. Business Sectors, 1987–2000,” Monthly Labor Review, Vol. 126 (April), pp. 16–29.

Faulkner-MacDonagh, M., 2004, “Canadian Household Sav-ing—Developments and Risk,” in Canada: Selected Issues, IMF Country Report No. 04/60 (Washington: International Monetary Fund).

Favero, Carlo, 2001, Applied Macroeconometrics (Oxford: Oxford University Press).

Forni, M., M. Hallin, M. Lippi, and L. Reichlin, 2001, “The Generalized Dynamic-Factor Model: Identification and Estimation,” Review of Economics and Statistics,Vol. 82 (November), pp. 540–54.

Freedman, Charles, 1998, “The Canadian Banking System,” Technical Report No. 81 (Ottawa: Bank of Canada). Available via the Internet: http://www.bank-banque-canada.ca/en/res/tr/1998/tr81.pdf.

———, and Walter Engert, 2003, “Financial Developments in Canada: Past Trends and Future Challenges,” Bank of Canada Review (Summer), pp. 3–16. Available via the Internet: http://www.bankofcanada.ca/en/review/sum-mer03/engerte.pdf.

Fung, Ben S.C., and Mingwei Yuan, 1999, “Measuring the Stance of Monetary Policy,” in Money, Monetary Pol-icy and Transmission Mechanisms (Ottawa: Bank of Canada). Available via the Internet: http://www.bankof-canada.ca/en/res/wp/1999/fungyuan-final.pdf.

Geweke, J., 1999, “Using Simulation Methods for Bayes-ian Econometric Models: Inference, Development, and Communication,” Econometric Reviews, Vol. 18, No. 1, pp. 1–76.

Gordon, R., 2003, “Hi-Tech Innovation and Productivity Growth: Does Supply Create Its Own Demand?” NBER Working Paper No. 9437 (Cambridge, Massachusetts: National Bureau of Economic Research).

Gregory, A., A. Head, and J. Raynauld, 1997, “Measur-ing World Business Cycles,” International Economic Review, Vol. 38 (August), pp. 677–701.

Gropp, R.E., J. Vesala, and G. Vulpes, 2002, “Equity and Bond Market Signals as Leading Indicators of Bank Fragility,” ECB Working Paper No. 150 (Frankfurt:

European Central Bank). Available via the Internet: http://www.ecb.int/pub/pdf/scpwps/ecbwp150.pdf.

Group of Ten, 2001, Report on Consolidation in the Finan-cial Sector (Basel, Switzerland: Bank for International Settlements). Available via the Internet: http://www.bis.org/publ/gten05.pdf.

Gruber, J., 1999, “Social Security and Retirement in Canada,” in Social Security and Retirement Around the World,ed. by J. Gruber and D.A. Wise (Chicago: University of Chicago Press).

———, and D.A. Wise, eds., 1999, Social Security and Retirement Around the World (Chicago: University of Chicago Press).

———, 2001, “An International Perspective on Policies for an Aging Society,” NBER Working Paper No. 8103 (Cambridge, Massachusetts: National Bureau of Eco-nomic Research).

Gust, C., and J. Marquez, 2004, “International Comparison of Productivity Growth: The Role of Information Tech-nology and Regulatory Practices,” Labour Economics,Vol. 11 (February), pp. 33–58.

Harchaoui, T., and F. Tarkhani, 2002, “Information Technol-ogy and Productivity Growth in Canada and the United States: A Comprehensive Examination” (unpublished; Ottawa: Statistics Canada).

Helbling, T., and T. Bayoumi, 2003, “Are They All in the Same Boat? The 2000–2001 Growth Slowdown and the G-7 Business Cycle Linkages,” IMF Working Paper 03/46 (Washington: International Monetary Fund).

Helliwell, J., R. Issa, R. Lafrance, and Q. Zhang, 2005, “NEMO: An Equation for the Canadian Dollar,” in Canada in the Global Economy, proceedings of a con-ference held at the Bank of Canada, November 2004 (Ottawa: Bank of Canada). Available via the Internet: http://www.bankofcanada.ca/en/conference/2004/nemo.pdf.

Ho, M., S. Rao, and J. Tang, 2003, “Sources of Output Growth in Canadian and U.S. Industries in the Information Age” (unpublished; Ottawa: Industry Canada).

Hoffman, M., and B. Dahlby, 2001, “Pension Provision in Canada,” in Pension Systems and Retirement Incomes Across OECD Countries, ed. by R. Disney and P. John-son (Cheltenham: Edward Elgar).

Hutchison, M.M., 1994, “Manufacturing Sector Resiliency to Energy Booms: Empirical Evidence from Norway, the Netherlands, and the United Kingdom,” Oxford Eco-nomic Papers, Vol. 46, No. 2, pp. 311–29.

International Monetary Fund, 2001, Manual on Fiscal Trans-parency (Washington).

———, 2002, Canada: Report on the Observance of Stan-dards and Codes—Fiscal Transparency Module, IMF Country Report No. 02/51 (Washington).

———, 2003, Germany: Selected Issues, IMF Country Report No. 03/342 (Washington).

———, 2004a, Canada: Selected Issues, IMF Staff Country Report 04/60 (Washington).

———, 2004b, “The Global Implications of the U.S. Fiscal Deficit and of China’s Growth,” in World Economic Outlook, April (Washington).

———, 2004c, United States: Selected Issues, IMF Staff Country Report 04/228.

©International Monetary Fund. Not for Redistribution

REFERENCES

130

Isiklar, G., K. Lahiri, and P. Loungani, 2005, “How Quickly Do Forecasters Incorporate News? Evidence from Cross-Country Surveys” (unpublished). Available via the Internet: http://www.albany.edu/econ/Research/2005/HowQuickly_jul20_2005.pdf.

Issa, R., R. Lafrance, and J. Murray, 2005, “The Turning Black Tide: Energy Prices and the Canadian Dollar,” Working Paper No. 2006-29 (Ottawa: Bank of Canada).

Ivaschenko, I., 2005, “Competition in Canada’s Banking System,” in Canada: Selected Issues, IMF Country Report No. 05/116 (Washington: International Mon-etary Fund).

———, and A. Swiston, 2005, “The Effects of U.S. Shocks on the Canadian Economy: Results from a Two-Coun-try Model,” in Canada: Selected Issues, IMF Country Report No. 05/116 (Washington: International Mon-etary Fund).

Jackson, A., 2003, “Why the ‘Big Idea’ Is a Bad Idea: A Critical Perspective on Deeper Economic Integration with the United States,” Research Paper No. 20 (Ottawa: Canadian Labour Congress). Available via the Internet: http://canadianlabour.ca/updir/paper20.pdf.

Johnson, R., 2000, “The Effect of Old-Age Insurance on Male Retirement: Evidence from Historical Cross-Country Data,” Research Working Paper No. 00-09 (Kansas City: Federal Reserve Bank of Kansas City). Available via the Internet: http://www.kc.frb.org/Publicat/Reswkpap/PDF/rwp00-09.pdf.

Jonung, L., and M. Larch, 2004, “Improving Fiscal Policy in the EU: The Case for Independent Forecasts,” Economic Paper No. 210 (Brussels: Directorate-General for Eco-nomic and Financial Affairs, European Commission). Available via the Internet: http://ec.europa.eu/econ-omy_finance/publications/economic_papers/2004/ecp210en.pdf.

Jorgenson, D., M. Ho, and K. Stiroh, 2002, “Information Technology, Education, and the Sources of Economic Growth Across U.S. Industries,” paper presented at the National Bureau of Economic Research Confer-ence on Research in Income and Wealth, “Measur-ing Capital in the New Economy,” Washington, April 26–27. Available via the Internet: http://www.nber.org/~confer/2002/criws02/ho.pdf.

Joumard, Isabelle, and Per Mathis Kongsrud, 2003, “Fiscal Relations across Government Levels,” OECD Economic Studies, Vol. 36, No. 1, pp. 156–229.

Justiniano, A., 2004, “Estimation and Model Selection in Dynamic Factor Analysis” (unpublished; Washington: International Monetary Fund).

———, 2005, “Factoring in Canadian Cycles,” in Canada: Selected Issues, IMF Country Report No. 05/116 (Wash-ington: International Monetary Fund).

Kashyap, Anil K., and Jeremy C. Stein, 1994, “Monetary Policy and Bank Lending,” Chapter 7 in Monetary Pol-icy, NBER Studies in Business Cycles, Vol. 29, ed. by N. Gregory Mankiw (Chicago: University of Chicago Press).

Kass, R., and A. Raftery, 1995, “Bayes Factors,” Journal of the American Statistical Association, Vol. 90 (June), pp. 773–95.

Kaufman, M., P. Swagel, and S. Dunaway, 2003, “Regional Convergence and the Role of Federal Transfers in Can-

ada,” IMF Working Paper 03/97 (Washington: Interna-tional Monetary Fund).

Kaufmann, S., 2000, “Measuring Business Cycles with a Dynamic Markov Switching Factor Model: An Assess-ment Using Bayesian Simulation Methods,” Economet-rics Journal, Vol. 3 (June), pp. 39−65.

Keeley, M., 1990, “Deposit Insurance, Risk, and Market Power in Banking,” American Economic Review, Vol. 80 (December), pp. 1183–200.

Kesselman, J.R., and F. Poschmann, 2001, “Expanding the Recognition of Personal Savings in the Canadian Tax System,” Canadian Tax Journal, Vol. 49, No. 1, pp. 40–101.

Khan, H., and M. Santos, 2002, “Contribution of ICT Use to Output and Labour-Productivity Growth in Canada,” Working Paper No. 2002-7 (Ottawa: Bank of Canada). Available via the Internet: http://www.bankofcanada.ca/en/res/wp/2002/wp02-7.pdf.

Kim, C.-J., and C. Nelson, 1999, State Space Models with Regime Switching: Classical and Gibbs-Sampling Approaches with Applications (Cambridge, Massachu-setts: MIT Press).

King, Robert, 2000, “The New IS-LM Model: Language, Logic, and Limits,” Economic Quarterly (Federal Reserve Bank of Richmond), Vol. 86, pp. 45–103.

———, and Alexander Wolman, 1999, “What Should the Monetary Authority Do When Prices Are Sticky?” in Monetary Policy Rules, ed. by John B. Taylor (Chicago: University of Chicago Press).

Klyuev, V., and R. Luzio, 2006, “Regional Dimensions of the Canadian Economy,” in Canada: Selected Issues, IMF Country Report No. 06/229 (Washington: International Monetary Fund).

Klyuev, V., and P. Mills, 2006, “Is Housing Wealth an ‘ATM’? The Relationship Between Household Wealth, Home Equity Withdrawal, and Saving Rates,” IMF Working Paper 06/162 (Washington: International Mon-etary Fund).

Kopits, G., and S. Symansky, 1998, Fiscal Policy Rules, IMF Occasional Paper No. 162 (Washington: International Monetary Fund).

Kose, M.A., 2004, “Canada-U.S. Economic Integration: Developments and Prospects,” in Canada: Selected Issues, IMF Country Report No. 04/60 (Washington: International Monetary Fund).

———, C. Otrok, and C. Whiteman, 2003, “International Business Cycles: World, Region, and Country Specific Factors,” American Economic Review, Vol. 93 (Septem-ber), pp. 1216–39.

———, 2005, “Understanding the Evolution of World Busi-ness Cycles,” IMF Working Paper 05/211 (Washington: International Monetary Fund).

Krainer, J., and J.A. Lopez, 2001, “Incorporating Equity Market Information into Supervisory Monitoring Mod-els,” Working Paper No. 2001-14 (San Francisco: Fed-eral Reserve Bank of San Francisco). Available via the Internet: http://www.frbsf.org/publications/economics/papers/2001/wp01-14bk.pdf.

Krelove, R., J.G. Stotsky, and C.L. Vehorn, 1997, “Canada,” in Fiscal Federalism in Theory and in Practice, ed. by T. Ter-Minassian (Washington: International Monetary Fund).

©International Monetary Fund. Not for Redistribution

References

131

Lane, P., 1999, “The New Open Economy Macroeconomics: A Survey,” Trinity Economic Paper No. 3 (Dublin: Trin-ity College). Available via the Internet: http://www.tcd.ie/Economics/TEP/1999_papers/TEPNo3PL99.pdf.

Laxton, D., and P. Pesenti, 2003, “Monetary Rules for Small, Open, Emerging Economies,” Journal of Monetary Economics, Vol. 50 (July), pp. 1109–46.

Lee, J., and M. Mühleisen, 2004, “The Canadian Dollar: Back to Fundamentals?” in Canada: Selected Issues,IMF Country Report No. 04/60 (Washington: Interna-tional Monetary Fund).

Leung, D., 2004, “The Effect of Adjustment Costs and Organizational Change on Productivity in Can-ada: Evidence from Aggregate Data,” Working Paper No. 2004-1 (Ottawa: Bank of Canada). Avail-able via the Internet: http://www.bankofcanada.ca/en/res/wp/2004/wp04-1.pdf.

Levin, Andrew T., Fabio Natalucci, and Egon Zakrajsek, 2004, “The Magnitude and Cyclical Behavior of Financial Market Frictions,” Finance and Economic Discussion Series Working Paper No. 2004-70 (Washington: Fed-eral Reserve Board). Available via the Internet: http://www.federalreserve.gov/Pubs/FEDS/2004/200470/200470pap.pdf.

———, Volcker Wieland, and John C. Williams, 1999, “Robustness of Simple Monetary Policy Rules under Monetary Uncertainty,” in Monetary Policy Rules, ed. by John B. Taylor (Chicago: University of Chicago Press).

Lise, J., 2003, “Is Canada’s Retirement Income System Work-ing?” Working Paper No. 2003-02 (Ottawa: Department of Finance Canada).

Liu, Y., E. Papakirykos, and M. Yuan, 2004, “Market Valu-ation and Risk Assessment of Canadian Banks,” Work-ing Paper No. 2004-34 (Ottawa: Bank of Canada). Available via the Internet: http://www.bankofcanada.ca/en/res/wp/2004/wp04-34.pdf.

Lopes, H.F., and M. West, 2004, “Bayesian Model Assess-ment in Factor Analysis,” Statistica Sinica, Vol. 14, pp. 41–67. Available via the Internet: http://www3.stat.sinica.edu.tw/statistica/oldpdf/A14n12.pdf.

Loungani, P., 2000, “How Accurate Are Private Sector Fore-casts? Cross-Country Evidence from Consensus Fore-casts of Output Growth,” IMF Working Paper 00/77 (Washington: International Monetary Fund).

Lucas, R.E., 1972, “Expectations and the Neutrality of Money,” Journal of Economic Theory, Vol. 4 (April), pp. 103–24.

Lumsdaine, R., and E.S. Prasad, 2003, “Identifying the Com-mon Component of International Economic Fluctua-tions: A New Approach,” Economic Journal, Vol. 113, No. 484, pp. 101–27.

Luzio, R., 2004, “The Information Content of Real Return Bonds,” in Canada: Selected Issues, IMF Country Report No. 04/60 (Washington: International Monetary Fund).

Macklem, T., 2003, “Future Productivity Growth in Canada: Comparing to the United States,” International Produc-tivity Monitor, Vol. 7 (Fall), pp. 50–57.

Mankiw, G., and R. Reis, 2001, “Sticky Information Versus Sticky Prices: A Proposal to Replace the New Keynes-ian Phillips Curve,” NBER Working Paper No. 8290

(Cambridge, Massachusetts: National Bureau of Eco-nomic Research).

Matier, C., and L. Wu, 2000, “CGE Model-Based Esti-mates of the Macroeconomic Impacts of Tax Reduc-tions,” Department of Finance Analytical Note (Ottawa: Department of Finance Canada).

McCallum, Bennett, and Edward Nelson, 1999a, “Nomi-nal Income Targeting in an Open-Economy Optimiz-ing Model,” Journal of Monetary Economics, Vol. 43 (June), pp. 553–78.

———, 1999b, “Performance of Operational Policy Rules in an Estimated Semiclassical Structural Model,” in Monetary Policy Rules, National Bureau of Economic Research Studies in Business Cycles, ed. by J.B. Taylor (Chicago: University of Chicago Press).

Melek Mansour, J., 1999, “Do National Business Cycles Have an International Origin?” (unpublished; Brussels: Uni-versité Libre de Bruxelles).

Merton, R.C., 1974, “On the Pricing of Corporate Debt: The Risk Structure of Interest Rates,” Journal of Finance,Vol. 29 (May), pp. 449–70.

Milligan, K., 2002, “Tax-Preferred Savings Accounts and Marginal Tax Rates: Evidence on RRSP Participation,” Canadian Journal of Economics, Vol. 35 (August), pp. 436–56.

———, 2003, “How Do Contribution Limits Affect Contri-butions to Tax-Preferred Savings Accounts?” Journal of Public Economics, Vol. 87 (February), pp. 253–81.

Ministry of Finance, 1998, “Minister of Finance Announces Decision on Bank Merger Proposals,” Press Release 1998-123, December 14 (Ottawa). Available via the Internet: http://www.fin.gc.ca/news98/98-123e.html.

Morsink, James, and Tamim Bayoumi, 2001, “A Peek Inside the Black Box: The Monetary Transmission Mecha-nism in Japan,” IMF Staff Papers, Vol. 48 (October), pp. 22–57.

Mühleisen, M., 2004, “Canada’s Pension System: Status and Reform Options,” in Canada: Selected Issues, IMF Country Report No. 04/60 (Washington: International Monetary Fund).

———, S. Danninger, D. Hauner, K. Krajnyák, and B. Sut-ton, 2005, “How Do Canadian Budget Forecasts Com-pare with Those of Other Industrial Countries?” IMF Working Paper 05/66 (Washington: International Mon-etary Fund).

Myles, J., 2000, “Incomes of Seniors,” Perspectives on Labour and Income, Vol. 12 (Winter), pp. 23–32.

Nathan, A., and E. Neave, 1989, “Competition and Con-testability in Canada’s Financial System: Empirical Results,” Canadian Journal of Economics, Vol. 22 (August), pp. 576–94.

National Energy Board (NEB), 2003, Canada’s Energy Future: Scenarios for Supply and Demand to 2025(Calgary). Available via the Internet: http://www.neb-one.gc.ca/energy/SupplyDemand/2003/SupplyDe-mand2003_e.pdf.

———, 2005, Canada’s Oil Sands: Opportunities and Chal-lenges to 2015, Energy Market Assessment (Calgary).

Nordhaus, W.D., 1987, “Forecasting Efficiency: Concepts and Applications,” Review of Economics and Statistics,Vol. 69 (November), pp. 667–74.

©International Monetary Fund. Not for Redistribution

REFERENCES

132

Northcott, C.A., 2004, “Competition in Banking: A Review of Literature,” Working Paper No. 2004-24 (Ottawa: Bank of Canada).

Obstfeld, M., and K. Rogoff, 2000, “The Six Major Puzzles in International Macroeconomics: Is There a Com-mon Cause?” NBER Working Paper No. 7777 (Cam-bridge, Massachusetts: National Bureau of Economic Research).

Office of the Superintendent of Financial Institutions (OSFI), 2001, Actuarial Report (18th) on the Canada Pension Plan: As at 31 December 2000 (Ottawa). Available via the Internet: http://www.osfi-bsif.gc.ca/app/DocReposi-tory/1/eng/oca/reports/CPP/CPP1801_e.pdf.

———, 2002, Actuarial Report (19th) Supplementing the Actuarial Report on the Canada Pension Plan: As at 31 December 2000 (Ottawa). Available via the Internet: http://www.osfi-bsif.gc.ca/app/DocRepository/1/eng/oca/reports/CPP/CPP1902_e.pdf.

———, 2003, Annual Report 2002–03 (Ottawa). Available via the Internet: http://www.osfi-bsif.gc.ca/app/DocRe-pository/1/eng/reports/osfi/osfi2003_e.pdf.

Oliner, S.D., and D.E. Sichel, 2002, “Information Technology and Productivity: Where Are We Now and Where Are We Going?” Federal Reserve Bank of Atlanta Economic Review, Vol. 87 (Third Quarter), pp. 15–44.

Öller, L.-E., and B. Barot, 2000, “The Accuracy of European Growth and Inflation Forecasts,” International Journal of Forecasting, Vol. 16, No. 3, pp. 293–315.

Organization for Economic Cooperation and Develop-ment (OECD), 2001a, Ageing and Income: Finan-cial Resources and Retirement in 9 OECD Countries(Paris).

———, 2001b, Measuring Productivity—OECD Manual: Measurement of Aggregate and Industry-Level Produc-tivity Growth (Paris).

———, 2002, “The OECD Budgeting Database,” OECD Journal of Budgeting, Vol. 1, No. 3, pp. 155–70.

———, 2003, Canada, OECD Economic Surveys, Vol. 14 (Paris).

———, and World Bank, 2003, Results of the Survey on Budget Practices and Procedures (Paris). Available via the Internet: http://ocde.dyndns.org.

Palameta, B., 2003, “Profiling RRSP Contributors,” Per-spectives on Labour and Income, Vol. 4 (January), pp. 29–35.

Panzar, J., and J. Rosse, 1987, “Testing for ‘Monopoly’ Equi-librium,” Journal of Industrial Economics, Vol. 35 (June), pp. 443–56.

Peersman, Gert, and Frank Smets, 2001, “The Monetary Transmission Mechanism in the Euro Area: More Evi-dence from VAR Analysis,” ECB Working Paper No. 91 (Frankfurt: European Central Bank).

Petersen, M., and R.G. Rajan, 1995, “The Effect of Credit Mar-ket Competition on Lending Relationships,” Quarterly Journal of Economics, Vol. 110 (May), pp. 407–43.

Phelps, E.S., 1983, “The Trouble with ‘Rational Expectations’ and the Problem of Inflation Stabilization,” in Individ-ual Forecasting and Aggregate Outcomes: “Rational Expectations” Examined, ed. by R. Frydman and E.S. Phelps (Cambridge, United Kingdom: Cambridge Uni-versity Press).

Punt, L., and M. van Rooij, 2001, “The Profit-Structure Rela-tionship and Mergers in the European Banking Industry: An Empirical Assessment,” DNB Staff Report No. 58 (Amsterdam: De Nederlandsche Bank). Available via the Internet: http://www.dnb.nl/dnb/home/file/sr058_tcm13-36663.pdf.

Radler, M., 2002, “Worldwide Reserves Increase as Pro-duction Holds Steady,” Oil & Gas Journal, Vol. 100, No. 52, pp. 113–15.

Ragan, C., 2005, “The Road Ahead for Canadian Inflation Targeting,” in Issues in Inflation Targeting, proceed-ings of a conference held at the Bank of Canada in April (Ottawa: Bank of Canada). Available via the Inter-net: http://www.bankofcanada.ca/en/conference/2005/ragan.pdf.

Reynolds, D.B., 2005, “The Economics of Oil Definitions: The Case of Canada’s Oil Sands,” OPEC Review, Vol. 29 (March), pp. 51–73.

Robidoux, B., 2003, “Future Productivity Growth in Canada: The Role of the Service Sector,” International Produc-tivity Monitor, Vol. 7 (Fall), pp. 58–65.

Roldos, J., 2006, “Disintermediation and Monetary Trans-mission in Canada,” IMF Working Paper 06/84 (Wash-ington: International Monetary Fund).

Rudebusch, Glenn D., and Lars E.O. Svensson, 1999, “Pol-icy Rules for Inflation Targeting,” in Monetary Policy Rules, ed. by John B. Taylor (Chicago: University of Chicago Press).

Salas, V., and J. Saurina, 2003, “Deregulation, Market Power and Risk Behaviour in Spanish Banks,” European Eco-nomic Review, Vol. 47 (December), pp. 1061–75.

Sargent, T., and C. Sims, 1977, “Business Cycle Modeling Without Pretending to Have Too Much A Priori Economic Theory,” in New Methods in Business Cycle Research: Proceedings from a Conference, ed. by C. Sims (Min-neapolis: Federal Reserve Bank of Minneapolis).

Schmitt-Grohé, S., 1998, “The International Transmission of Economic Fluctuations: Effects of U.S. Business Cycles on the Canadian Economy,” Journal of International Economics, Vol. 44 (April), pp. 257–87.

Schroeder, W., E. Mui, and B. Maxwell, 2005, “The Cana-dian Oil Sands: One of the World’s Greatest Sources of Oil—But Not Without Risk,” Energy Industry Study (Toronto: Dominion Bond Rating Service).

Shaffer, S., 1982, “A Non-Structural Test for Competition in Financial Markets,” in Federal Reserve Bank of Chi-cago, Bank Structure and Competition: Conference Proceedings (Chicago).

Smeeding, Timothy M., and Dennis H. Sullivan, 1998, “Gen-erations and the Distribution of Economic Well-Being: A Cross-National View,” American Economic Review, Vol. 88, No. 2, pp. 254–58.

Söderbergh, B., 2005, “Canada’s Oil Sands Resources and Its Future Impact on Global Oil Supply” (unpublished; Uppsala, Sweden: Uppsala Hydrocarbon Depletion Study Group, Uppsala University). Available via the Internet: http://www.peakoil.net/uhdsg/OilSandCanada.pdf.

Spatafora, Nikola, and Andrew Warner, 1999, “Macroeco-nomic and Sectoral Effects of Terms-of-Trade Shocks—The Experience of the Oil-Exporting Developing Countries,” IMF Working Paper No. 99/134 (Washing-ton: International Monetary Fund).

©International Monetary Fund. Not for Redistribution

References

133

Statistics Canada, 2001, The Assets and Debts of Cana-dians: Focus on Private Pension Savings (Ottawa). Available via the Internet: http://www.statcan.ca/bsolc/english/bsolc?catno=13-596-X&CHROPG=1.

Stevens, Paul, 2003, “Resource Impact: Curse or Blessing? A Literature Survey,” Journal of Energy Literature, Vol. 9, No. 1, pp. 3–42.

Jean-Philippe Stijns, 2003, “An Empirical Test of the Dutch Disease Hypothesis Using a Gravity Model of Trade,” paper presented at the 2003 EEA Congress, Stockholm.

Stock, James H., and Mark W. Watson, 1998, “Diffusion Indexes,” NBER Working Paper No. 6702 (Cam-bridge, Massachusetts: National Bureau of Economic Research).

———, 2002, “Macroeconomic Forecasting Using Diffusion Indexes,” Journal of Business and Economic Statistics,Vol. 20 (April), pp. 147–62.

———, 2003, “Has the Business Cycle Changed, and Why?” in NBER Macroeconomics Annual 2002, ed. by M. Gertler and K. Rogoff (Cambridge, Massachusetts: MIT Press).

———, 2005, “Understanding Changes in International Busi-ness Cycle Dynamics,” Journal of the European Eco-nomic Association, Vol. 3 (September), pp. 966–1006.

Strauch, R., M. Hallerberg, and J. von Hagen, 2004, “Budget-ary Forecasts in Europe: The Track Record of Stability and Convergence Programmes,” ECB Working Paper No. 307 (Frankfurt: European Central Bank).

Svensson, Lars, 2000, “Open-Economy Inflation Targeting,” Journal of International Economics, Vol. 50 (Febru-ary), pp. 155–83.

Taylor, John B., 1999, “Monetary Policy Guidelines for Employment and Inflation Stability,” in Inflation, Unemployment, and Monetary Policy, ed. by Robert M. Solow and John B. Taylor (Cambridge, Massachusetts: MIT Press), pp. 29–54.

Thiessen, G.G., 1995, “Uncertainty and the Transmission of Monetary Policy in Canada,” The HERMES-Glendon

Lecture, York University–Glendon College, Toronto, March 30.

Timilsina, G.R., N. LeBlanc, and T. Walden, 2005, Economic Impacts of Alberta’s Oil Sands: Final Report (Calgary: Canadian Energy Research Institute). Available via the Internet: http://www.ceri.ca/Publications/documents/OilSandsReport-Final.PDF.

Timilsina, G.R., J.P. Prince, D. Czamanski, and N. LeBlanc, 2005, “Impacts of Crude Oil Production from Alberta Oil Sands on the Canadian Economy,” paper presented at EcoMod 2005 International Conference on Policy Modeling, Istanbul, June 29–July 1. Available via the Internet: http://www.ecomod.net/conferences/eco-mod2005/ecomod2005_papers/Timilsina.pdf.

Tompa, E., 1999, “Transitions to Retirement: Determinants of Age of Social Security Take Up,” Social and Eco-nomic Dimensions of an Aging Population Research Paper No. 6 (Hamilton, Ontario: McMaster University).

Tsounta, E., 2006, “Why Are Women Working So Much More in Canada? An International Perspective,” IMF Working Paper 06/92 (Washington: International Mon-etary Fund).

Vassalou, M., and Y. Xing, 2004, “Default Risk in Equity Returns,” Journal of Finance, Vol. 59 (April), pp. 831–68.

Veall, M.R., 1999, “Did Tax Flattening Affect RRSP Contri-butions?” Quantitative Studies in Economics and Pop-ulation Research Report No. 342 (Hamilton, Ontario: McMaster University).

Weil, Philippe, 1989, “Overlapping Families of Infinitely-Lived Agents,” Journal of Public Economics, Vol. 38, pp. 183–98.

Woodford, M., 2001, “Imperfect Common Knowledge and the Effects of Monetary Policy,” NBER Working Paper No. 8673 (Cambridge, Massachusetts: National Bureau of Economic Research).

Zellner, A., 1986, “Biased Predictors, Rationality and the Evaluation of Forecasts,” Economic Letters, Vol. 21, No. 1, pp. 45–48.

©International Monetary Fund. Not for Redistribution

134

Recent Occasional Papers of the International Monetary Fund

258. Northern Star: Canada’s Path to Economic Prosperity, edited by Tamim Bayoumi, Vladimir Klyuev, and Martin Mühleisen. 2007.

257. Economic Growth and Integration in Central America, edited by Dominique Desruelle and Alfred Schipke. 2007.

256. Moving to Greater Exchange Rate Flexibility: Operational Aspects Based on Lessons from Detailed Country Experiences, by Inci Ötker-Robe and David Vávra, and a team of IMF economists. 2007.

255. Sovereign Debt Restructuring and Debt Sustainability: An Analysis of Recent Cross-Country Experience, by Harald Finger and Mauro Mecagni. 2007.

254. Country Insurance: The Role of Domestic Policies, by Törbjörn Becker, Olivier Jeanne, Paolo Mauro,Jonathan D. Ostry, and Romain Rancière. 2007.

253. The Macroeconomics of Scaling Up Aid: Lessons from Recent Experience, by Andrew Berg, Shekhar Aiyar, Mumtaz Hussain, Shaun Roache, Tokhir Mirzoev, and Amber Mahone. 2007.

252. Growth in the Central and Eastern European Countries of the European Union, by Susan Schadler, Ashoka Mody, Abdul Abiad, and Daniel Leigh. 2006.

251. The Design and Implementation of Deposit Insurance Systems, by David S. Hoelscher, Michael Taylor, and Ulrich H. Klueh. 2006.

250. Designing Monetary and Fiscal Policy in Low-Income Countries, by Abebe Aemro Selassie, Benedict Clements, Shamsuddin Tareq, Jan Kees Martijn, and Gabriel Di B ella. 2006.

249. Official Foreign Exchange Intervention, by Shogo Ishi, Jorge Iván Canales-Kriljenko, Roberto Guimarães, and Cem Karacadag. 2006.

248. Labor Market Performance in Transition: The Experience of Central and Eastern European Countries, by Jerald Schiff, Philippe Egoumé-Bossogo, Miho Ihara, Tetsuya Konuki, and Kornélia Krajnyák. 2006.

247. Rebuilding Fiscal Institutions in Post-Conflict Countries, by Sanjeev Gupta, Shamsuddin Tareq, Benedict Clements, Alex Segura-Ubiergo, Rina Bhattacharya, and Todd Mattina. 2005.

246. Experience with Large Fiscal Adjustments, by George C. Tsibouris, Mark A. Horton, Mark J. Flanagan, and Wojciech S. Maliszewski. 2005.

245. Budget System Reform in Emerging Economies: The Challenges and the Reform Agenda, by Jack Diamond. 2005.

244. Monetary Policy Implementation at Different Stages of Market Development, by a staff team led by Bernard J. Laurens. 2005.

243. Central America: Global Integration and Regional Cooperation, edited by Markus Rodlauer and Alfred Schipke. 2005.

242. Turkey at the Crossroads: From Crisis Resolution to EU Accession, by a staff team led by Reza Moghadam. 2005.

241. The Design of IMF-Supported Programs, by Atish Ghosh, Charis Christofides, Jun Kim, Laura Papi, UmaRamakrishnan, Alun Thomas, and Juan Zalduendo. 2005.

240. Debt-Related Vulnerabilities and Financial Crises: An Application of the Balance Sheet Approach to Emerging Market Countries, by Christoph Rosenberg, Ioannis Halikias, Brett House, Christian Keller, Jens Nystedt, Alexander Pitt, and Brad Setser. 2005.

239. GEM: A New International Macroeconomic Model, by Tamim Bayoumi, with assistance from Douglas Laxton, Hamid Faruqee, Benjamin Hunt, Philippe Karam, Jaewoo Lee, Alessandro Rebucci, and Ivan Tchakarov. 2004.

238. Stabilization and Reforms in Latin America: A Macroeconomic Perspective on the Experience Since the Early 1990s, by Anoop Singh, Agnès Belaisch, Charles Collyns, Paula De Masi, Reva Krieger, Guy Meredith, and Robert Rennhack. 2005.

237. Sovereign Debt Structure for Crisis Prevention, by Eduardo Borensztein, Marcos Chamon, Olivier Jeanne, Paolo Mauro, and Jeromin Zettelmeyer. 2004.

236. Lessons from the Crisis in Argentina, by Christina Daseking, Atish R. Ghosh, Alun Thomas, and Timothy Lane. 2004.

OCCASIONAL PAPERS

©International Monetary Fund. Not for Redistribution

135

235. A New Look at Exchange Rate Volatility and Trade Flows, by Peter B. Clark, Natalia Tamirisa, and Shang-Jin Wei, with Azim Sadikov and Li Zeng. 2004.

234. Adopting the Euro in Central Europe: Challenges of the Next Step in European Integration, by Susan M. Schadler, Paulo F. Drummond, Louis Kuijs, Zuzana Murgasova, and Rachel N. van Elkan. 2004.

233. Germany’s Three-Pillar Banking System: Cross-Country Perspectives in Europe, by Allan Brunner, Jörg Decressin, Daniel Hardy, and Beata Kudela. 2004.

232. China’s Growth and Integration into the World Economy: Prospects and Challenges, edited by Eswar Prasad. 2004.

231. Chile: Policies and Institutions Underpinning Stability and Growth, by Eliot Kalter, Steven Phillips, Marco A. Espinosa-Vega, Rodolfo Luzio, Mauricio Villafuerte, and Manmohan Singh. 2004.

230. Financial Stability in Dollarized Countries, by Anne-Marie Gulde, David Hoelscher, Alain Ize, David Marston, and Gianni De Nicolò. 2004.

229. Evolution and Performance of Exchange Rate Regimes, by Kenneth S. Rogoff, Aasim M. Husain, Ashoka Mody, Robin Brooks, and Nienke Oomes. 2004.

228. Capital Markets and Financial Intermediation in The Baltics, by Alfred Schipke, Christian Beddies, Susan M. George, and Niamh Sheridan. 2004.

227. U.S. Fiscal Policies and Priorities for Long-Run Sustainability, edited by Martin Mühleisen and Christopher Towe. 2004.

226. Hong Kong SAR: Meeting the Challenges of Integration with the Mainland, edited by Eswar Prasad, with contributions from Jorge Chan-Lau, Dora Iakova, William Lee, Hong Liang, Ida Liu, Papa N’Diaye, and Tao Wang. 2004.

225. Rules-Based Fiscal Policy in France, Germany, Italy, and Spain, by Teresa Dában, Enrica Detragiache, Gabriel di Bella, Gian Maria Milesi-Ferretti, and Steven Symansky. 2003.

224. Managing Systemic Banking Crises, by a staff team led by David S. Hoelscher and Marc Quintyn. 2003.223. Monetary Union Among Member Countries of the Gulf Cooperation Council, by a staff team led by Ugo

Fasano. 2003.222. Informal Funds Transfer Systems: An Analysis of the Informal Hawala System, by Mohammed El Qorchi,

Samuel Munzele Maimbo, and John F. Wilson. 2003.221. Deflation: Determinants, Risks, and Policy Options, by Manmohan S. Kumar. 2003.220. Effects of Financial Globalization on Developing Countries: Some Empirical Evidence, by Eswar S. Prasad,

Kenneth Rogoff, Shang-Jin Wei, and Ayhan Kose. 2003.219. Economic Policy in a Highly Dollarized Economy: The Case of Cambodia, by Mario de Zamaroczy and

Sopanha Sa. 2003.218. Fiscal Vulnerability and Financial Crises in Emerging Market Economies, by Richard Hemming, Michael Kell,

and Axel Schimmelpfennig. 2003.217. Managing Financial Crises: Recent Experience and Lessons for Latin America, edited by Charles Collyns and

G. Russell Kincaid. 2003.216. Is the PRGF Living Up to Expectations? An Assessment of Program Design, by Sanjeev Gupta, Mark Plant,

Benedict Clements, Thomas Dorsey, Emanuele Baldacci, Gabriela Inchauste, Shamsuddin Tareq, and Nita Thacker. 2002.

215. Improving Large Taxpayers’ Compliance: A Review of Country Experience, by Katherine Baer. 2002.214. Advanced Country Experiences with Capital Account Liberalization, by Age Bakker and Bryan Chapple.

2002.213. The Baltic Countries: Medium-Term Fiscal Issues Related to EU and NATO Accession, by Johannes Mueller,

Christian Beddies, Robert Burgess, Vitali Kramarenko, and Joannes Mongardini. 2002.212. Financial Soundness Indicators: Analytical Aspects and Country Practices, by V. Sundararajan, Charles Enoch,

Armida San José, Paul Hilbers, Russell Krueger, Marina Moretti, and Graham Slack. 2002.211. Capital Account Liberalization and Financial Sector Stability, by a staff team led by Shogo Ishii and Karl

Habermeier. 2002.

Note: For information on the titles and availability of Occasional Papers not listed, please consult the IMF’s Publications Catalog or contact IMF Publication Services.

Occasional Papers

©International Monetary Fund. Not for Redistribution