23
COLM KEARNEY MEHDI MONADJEMI University of New South Wales Kensington, New South Wales, Australia Fiscal Policy and Current Account Performance: International Evidence on the Twin Deficits* Divergent predictions about the twin deficits relationship, which links the stance of fiscal policy to the current account balance, continue to emanate from variously specified theoretical models of the open economy. This paper utilizes the vector autoregressive (VAR) technique to examine the international evidence from eight countries on quarterly data over the recent period of floating exchange rates from 1972:i-1987:iv. The evidence is consistent with a temporary twin deficits relation- ship that is not invariant to the government’s financing decision and does not persist over time. 1. Introduction The extent to which variations in the stance of fiscal policy can lead to predictable developments in an open economy’s per- formance on current account of the balance of payments remains a controversial issue. An important aspect of this issue concerns the so-called twin deficits analysis, according to which fiscal deficits and current account balances are very closely related so that reductions in the former are both necessary and sufficient to obtain improved performance in the latter. While the applicability of this analysis is clearly limited in a world of international macroeconomic policy harmonization, its relevance in the present context of substantial international economic interdependence coupled with imperfect policy harmornzation is more clear-cut. Theoretical work on the relationship that exists between vari- ations in the stance of fiscal policy and the current account balance has been based upon two types of models. The traditional approach has employed macroeconomic models that are constructed from *This paper was written while the first author was on leave at the Central Bank of Ireland. The views expressed are not necessarily those of the Bank. We are grateful to Kenneth Clements, Gabriel Fagan, John Frain, Michael Moore, and Tom O’Connell for helpful discussions. Remaining weaknesses are the authors’ re- sponsibility. Journul of Macroeconomics, Spring 1999, Vol. 12, No. 2, pp. 197-219 197 Copyright 0 1999 by Louisiana State University Press OR%-0794/90/$1.50

Fiscal policy and current account performance: International evidence on the twin deficits

Embed Size (px)

Citation preview

Page 1: Fiscal policy and current account performance: International evidence on the twin deficits

COLM KEARNEY MEHDI MONADJEMI

University of New South Wales Kensington, New South Wales, Australia

Fiscal Policy and Current Account Performance: International Evidence on the Twin Deficits*

Divergent predictions about the twin deficits relationship, which links the stance of fiscal policy to the current account balance, continue to emanate from variously specified theoretical models of the open economy. This paper utilizes the vector autoregressive (VAR) technique to examine the international evidence from eight countries on quarterly data over the recent period of floating exchange rates from 1972:i-1987:iv. The evidence is consistent with a temporary twin deficits relation- ship that is not invariant to the government’s financing decision and does not persist over time.

1. Introduction The extent to which variations in the stance of fiscal policy

can lead to predictable developments in an open economy’s per- formance on current account of the balance of payments remains a controversial issue. An important aspect of this issue concerns the so-called twin deficits analysis, according to which fiscal deficits and current account balances are very closely related so that reductions in the former are both necessary and sufficient to obtain improved performance in the latter. While the applicability of this analysis is clearly limited in a world of international macroeconomic policy harmonization, its relevance in the present context of substantial international economic interdependence coupled with imperfect policy harmornzation is more clear-cut.

Theoretical work on the relationship that exists between vari- ations in the stance of fiscal policy and the current account balance has been based upon two types of models. The traditional approach has employed macroeconomic models that are constructed from

*This paper was written while the first author was on leave at the Central Bank of Ireland. The views expressed are not necessarily those of the Bank. We are grateful to Kenneth Clements, Gabriel Fagan, John Frain, Michael Moore, and Tom O’Connell for helpful discussions. Remaining weaknesses are the authors’ re- sponsibility.

Journul of Macroeconomics, Spring 1999, Vol. 12, No. 2, pp. 197-219 197 Copyright 0 1999 by Louisiana State University Press OR%-0794/90/$1.50

Page 2: Fiscal policy and current account performance: International evidence on the twin deficits

Cohn Kearney and Mehdi Monadjemi

postulated behavioral relationships that purport to describe how the economy works in aggregate without explaining the behavior of the agents who make up the economy. Examples of this approach are to be found in the work of Mundell (1963), Branson (1976), Dorn- busch (1976), Kawai (1985), and Marston (1985). The second type of model, which has become more popular in recent years, derives the important macroeconomic relationships from the microeconomic foundations of individual optimizing behavior. Examples of this ap- proach have been reported by Dixit (1978), Neary (1980), Obstfeld (1981), Persson (1982), Kimbrough (1985), Frenkel and Razin (1986), Cuddington and Vinals (1986a, 198613) and Moore (1989). Both of these approaches to explaining the current account response to vari- ations in the stance of fiscal policy have yielded divergent results. The growth in empirical evidence, however, has not kept pace with these theoretical advancements. The purpose of this paper is to contribute toward redressing this imbalance.

In view of the considerable uncertainty that pervades appro- priate specification of theoretical models of open economy stabili- zation policy, the appropriate modeling strategy that is adopted in this paper utilizes the vector autoregressive (VAR) technique. The suitability of this approach in the current context stems from the fact that a VAR model constitutes an unrestricted reduced form of some unknown structural system of equations. Therefore, as Zellner and Palm (1974), and Zellner (1979) have demonstrated, any struc- tural econometric model amounts to a restricted VAR model. In the present context, this procedure facilitates examination of the extent to which the emergence of a twin deficits relationship depends im- portantly upon the government’s financing decision, as well as upon the implications of the latter for the transmission mechanisms that operate through the exchange rate.

The remainder of the paper is organized as follows. Section 2 summarizes the current state of theoretical knowledge about the relationship that exists between the government’s fiscal deficit and the current account of the balance of payments. It demonstrates the extent to which the twin deficits relationship is exceedingly complex while being influenced by factors that are not exclusively under the control of those who formulate and implement macroeconomic pol- icy. Section 3 presents econometric analysis of the time series data for eight OECD countries (Australia, Britain, Canada, France, Ger- many, Ireland, Italy and the United States) over the period of float- ing exchange rates from 1972:i to 1987:iv. Section 4 calculates the

198

Page 3: Fiscal policy and current account performance: International evidence on the twin deficits

Fiscal Policy and Current Account Performance

impulse response functions from the VAR models in order to ex- amine the adjustment dynamics of balance of payments on current account to innovations in government expenditures that are ii- nanced alternatively by higher taxes, debt creation, and by mone- tary expansion. The final section summarizes the paper before con- cluding that the existence of strong twin deficits relationships is not corroborated by the international evidence.

2. The Theory We begin by outlining the conventional accounting relation-

ship that exists between domestic output and the components of aggregate demand. Equation (1) indicates how domestic output (Y) is divided between private consumption (C,), private investment (I,), government expenditure for consumption and investment pur- poses (C, and I, respectively), exports and imports of goods and services (EX and ZM respectively), and the interest that is payable on the country’s net ownership of foreign assets (F).

Y=C,+Z,+C,+Z,+EX-ZM+F. (1)

We can rewrite this identity by defining the government’s fiscal deficit (DEF) as the sum of its expenditures less tax revenues (T), and the current account deficit of the balance of payments (CUR), together with private savings (S,) as follows:

DEF=C,+Z,-T, (2)

CUR = -(EX - ZM + F), (3)

S, = Y - T - C, . (4)

Substituting these definitions into the first identity yields the fa- miliar twin deficits relationship (5), which describes the current ac- count deficit in terms of the difference between private investment and savings (I, - S,) plus the government’s fiscal deficit.

CUR = (I, - S,) + DEF . (5)

The relationship that exists between the variables in identity (5) is obviously crucial for any analysis of the twin deficits relationship. It occupies a pivotal position in open economy models of stabili-

199

Page 4: Fiscal policy and current account performance: International evidence on the twin deficits

Colm Kearney and Mehdi Monadjemi

zation policy. We shall have more to say about this in the context of both postulated and optimizing models of the economy. For ex- ample, proponents of the Ricardian equivalence theorem, as res- tated by Barro (1974), argue that increases in (G - T) will be offset by increases in S, insofar as the private sector fully discounts the future tax liabilities associated with financing the fiscal deficit. If foreign residents purchase the government’s debt at unchanged in- terest rates, Dwyer’s (1985) capital inflow hypothesis will yield a close association between the current account and fiscal deficits. Proponents of the recently popular twin deficits analysis, such as Laney (1984), argue that the private sector tends to run a stable (I - S) relationship so that current account deficits tend to mirror fiscal deficits.

Models Based on Postulated Behavioral Relationships The classic analysis of how fiscal policy works in an open econ-

omy under fixed and flexible exchange rates is presented by Mun- dell (1963). He demonstrates how fiscal expansion can raise do- mestic output under fixed rates, but not under flexible exchange rates. Subsequent work by Dornbusch (1976) extends this earlier analysis while endowing agents in the economy with rational ex- pectations before supporting the earlier conclusions about the ef- fectiveness of fiscal policy under floating exchange rates. Both Mun- dell (1963) and Dornbusch (1976) d escribe a world of perfect capital mobility in which flows of international capital are unrestricted and where assets that are denominated in domestic and foreign curren- cies are perfectly substitutable. If a small open economy’s exchange rate is market-determined in this world, a policy of fiscal expansion will cause the domestic currency to appreciate until the trade bal- ance deteriorates by an amount that exactly offsets the fiscal stim- ulus. In short, the entire fiscal expansion is transmitted abroad as the deficit on current account “twins” the government’s fiscal def- icit. More recent developments in the study of open economy sta- bilization policy using postulated models, however, have demon- strated the extent to which these earlier results are not robust to changes in model specification. These developments are readily in- terpretable as stemming from recognition of three main weaknesses in the original Mundell-Fleming framework; namely, an inappro- priate stock-flow specification of the capital account of the balance of payments, no consideration of wealth effects, and an exclusive focus on the demand side of the macroeconomy with fixed wages and prices.

200

Page 5: Fiscal policy and current account performance: International evidence on the twin deficits

Fiscal Policy and Current Account Performance

Concerning the first of these weaknesses, the Mundell-Fleming model specifies capital flows rather than stocks as depending upon the level of interest rates, with the result that changes in the latter will induce permanent flows of capital across political frontiers. As Branson (1976), amongst others, has pointed out, this feature of the model is not consistent with portfolio balance because it allows the current account to be permanently nonzero while implying that do- mestic residents are willing to allow non-residents to accumulate indefinite amounts of claims and/or liabilities on them. By making all asset stocks dependent upon interest rate levels, the portfolio balance approach overcomes this weakness, while emphasizing the degree of substitutability between domestic and foreign currency denominated assets, along with wealth effects, as important deter- minants of the potency of stabilization policy. When domestic and foreign currency denominated assets are perfectly substitutable, the portfolio balance model predicts the same short-run effects of sta- bilization policy, as does the Mundell-Fleming model with perfect capital mobility. As Rodriguez (1979) and Sachs (1980) demonstrate, however, these effects can be reversed over time in the portfolio balance model if expansionary fiscal policy reduces the net foreign asset position of the economy due to current account deficits. In this case, the achievement of long-run equilibrium may necessitate lower domestic output in order to generate trade account surpluses to match deteriorating debt service obligations. This result ob- viously conflicts with the twin deficits case described above. When the assumption of perfect substitutability of domestic and foreign currency denominated assets is relaxed, however, the effects of sta- bilization policy will differ from the Mundell-Fleming predictions in the short run as well as in the long run. Expansionary fiscal policy will generally raise domestic interest rates relative to their foreign counterparts, but the exchange rate effects will be ambig- uous depending upon the relative responses of the current and cap- ital accounts of the balance of payments. This situation has been studied by Greenwood (1983) and Kawai (1985), who demonstrate that no unique relationship exists between the government’s macro- economic policy mix, the exchange rate, and the current account of the balance of payments.

Discussion of the role of wealth prompts consideration of the application of Ricardian equivalence, as restated by Barro (1974), to open economies. The conditions under which this equivalence theo- rem will operate in an open economy context will match the closed economy case if the foreign interest rate, which is faced by the

201

Page 6: Fiscal policy and current account performance: International evidence on the twin deficits

Calm Kearney and Mehdi Monadjemi

domestic private and public sectors, is the same. In this case, as Leiderman and Blejer (1987) point out, the domestic private sector internalizes the government’s debt regardless of the extent to which it is held by domestic or foreign residents. There exists a well-doc- umented set of conditions under which Ricardian equivalence will not hold: Stiglitz (1983) provides examples of market imperfections such as price rigidities and incomplete markets for intergenerational risk sharing; Barro (1979) and Razin and Svensson (1983) illustrate the effects of distortionary taxes; Greenwood and Kimbrough (1985) illustrate the effects of capital controls; Blanchard (1985) illustrates how finite horizons drive a wedge between private and public dis- count rates; Helpman and Razin (1987) employ Blanchard’s (1985) insight to provide an important example of Ricardian inequivalence in a flexible exchange rate open economy.

When wage and price flexibility is added to the existence of wealth effects in the portfolio balance model, the Mundell-Fleming conclusions about the operation of fiscal policy require further mod- ification. Marston (1985) provides a good account of how this comes about. The primary effect of fiscal expansion in an economy with internationally mobile capital is to raise the real exchange rate. With fixed nominal exchange rates, this will occur wholly through higher domestic prices, and the resulting drop in real wages will ensure that extra output is generated. When exchange rates are market- determined, however, the higher real exchange rate will partly con- sist of nominal appreciation, and the resulting wealth effects will ensure that output expands more than under fixed rates. There is no presumption of a twin deficits relationship in this extension of the Mundell-Fleming model unless a fixed real exchange rate (due to perfectly substitutable domestic and foreign goods) prevents do- mestic output from responding to the fiscal policy initiative.

Models Based on Optimizing Behavior Macroeconomists since Keynes have become aware of the need

to derive their behavioral relationships from the microeconomic first principles of individual optimizing behavior. A renewed sense of urgency has been injected into this endeavor since the realization of Lucas (1976) that perceived changes in the policy regime may well cause individuals to alter their maximizing behavior in ways that modify many previously postulated macroeconomic relation- ships. The significance of this development for the present discus- sion lies in the fact that a number of recent studies have analyzed the relationship that exists between fiscal policy and the balance of

Page 7: Fiscal policy and current account performance: International evidence on the twin deficits

Fiscal Policy and Current Account Performance

payments using models that are based on microeconomic optimizing behavior. These models fall conspicuously into a twofold classifica- tion, namely, those that result in Walrasian market-clearing macro- economic frameworks and those that allow for the existence of mar- ket disequilibrium. Obstfeld (1981), Sachs (1980), and Frenkel and Razin (1986) are amongst those who have constructed the former type of model, while Dixit (1978), Near-y (1980), Persson (1982), Cuddington and Vinals (1986a, 1986b) and Moore (1989) amongst others have constructed models that allow for the existence of var- ious types of market disequilibria.

Obstfeld (1981) develops an equilibrium model of the deter- mination of the exchange rate and current account of a small open economy that is inhabited by utility-maximizing households with in- finite planning horizons, who consume a single good and hold only domestic money, although they have access to world credit mar- kets. A tax-financed increase in government expenditure does not lead to higher private consumption in this model unless the au- thorities purchase public goods that enter the private sector’s utility function. In Obstfeld’s general case, consumption expenditures de- cline by an amount that is greater than the fiscal expansion because the private sector saves more in order to attain the higher level of national income at its initial level of consumption. The increased savings generate a current account surplus, which raises national income through the additional interest payments from foreign res- idents. This analysis has been extended by Sachs (1982) and by Frenkel and Razin (1986) who investigate the potency of fiscal pol- icy in a general equilibrium, two-country world that is inhabited by overlapping generations of inter-temporal utility maximizers. The latter model permits investigation of the effects of both anticipated and previously unanticipated policy initiatives, which are either tem- porary or permanent in nature. Amongst the major conclusions that emerge from this work are that temporary tax-financed fiscal ex- pansions are associated with current account deficits, while per- manent policy initiatives have ambiguous effects.

The early work on optimizing disequilibrium open economy models by Dixit (1978) and Near-y (1980) assumes fixed exchange rate regimes, while the former study also assumes that the economy produces and consumes a single tradeable commodity. More recent work by Cuddington and Vinals (1986a, 198613) generalizes the ear- lier work in this regard and also allows for the existence of both classical and/or Keynesian unemployment, while Moore (1989) al- lows for one of the goods to be invested for inventory purposes.

203

Page 8: Fiscal policy and current account performance: International evidence on the twin deficits

Colm Kearney and Mehdi Monadjemi

Intertemporal utility maximizers in the economy analyzed by Cuddington and Vinals can consume traded and non-traded goods, and they form rational expectations about the effects of current and future probable policy initiatives of the authorities. Consider, for example, a temporary tax-financed expansion in government ex- penditure on non-traded goods in an economy that suffers from Keynesian unemployment. The resulting excess demand for non- tradeables is eliminated by exchange rate appreciation that raises the relative price and production of these goods while reducing the value of tradeables, which generates a current account deficit. If the fiscal expansion is anticipated to occur in the future, higher expected future non-tradeables prices will induce a shift towards non-tradeables today, and this will generate a current account sur- plus. The effects of a permanent fiscal expansion are the sum of the above two policies (that is, temporary and future), and its current account effects are consequently indeterminate.

The conclusion that emerges from this overview of the rele- vant theoretical literature is that the relationship that exists be- tween the government’s fiscal stance and the economy’s perfor- mance on current account of the balance of payments is a complex one that is influenced by many factors, not all of which are under the control of those who formulate and implement macroeconomic policy. The early vintage theoretical models that yielded strong twin deficit relationships did so by abstracting from important complex- ities, which, as Kawai (1985) remarks, explains why they have not performed well empirically. It is with interest, therefore, that we now turn to examine the international evidence utilizing the VAR technique.

3. Empirical Analysis Theoretical work on open economy stabilization policy points

to the importance of the government’s financing decision as well as the transmission mechanisms that operate through the exchange rate in determining the extent of a twin deficits relationship. The mod- eling strategy adopted in this section explicitly incorporates these factors in the empirical tests. More specifically, the following VAR model is suggested.

A (=Q it = ut > (6)

Page 9: Fiscal policy and current account performance: International evidence on the twin deficits

Fiscal Policy and Current Account Performance

with

A (L) = Z - A,L - AzL2 - . . . - A,LP ,

E(u,) = 0, E(t&) = z, E(u,u,‘) = 0 for t # s ,

E(ytd) = 0 for t # s ,

This is a standard VAR representation in which y is a (1Xn) vector of variables, A is an (nxn) matrix of coefficients, u is an (nxl) vector of white noise disturbance terms, and L denotes the lag operator (for example, L’y, = y,-,). The variables that appear in the y vector are government expenditures (g), tax revenues (t), monetary crea- tion (m), the exchange rate (X), and the current account of the balance of payments (c). The dataset consists of quarterly observa- tions on these five variables for eight countries (Australia, Britain, Canada, France, Germany, Ireland, Italy and the United States) over the period 1972:i-1987:iu. All data comes from various issues of Znternational Financial Statistics. The choice of countries reflects the inclusion of the G7 economies (except Japan, which was ex- cluded due to unavailability of some quarterly series) plus two small open economies. All series (except X, which is the real effective exchange rate index) are seasonally adjusted and expressed in nom- inal terms as a ratio of nominal GDP.

A convenient feature of the VAR model of Equation (6) is that it can be estimated by ordinary least squares that will yield con- sistent and asymptotically efficient estimates of the A matrix be- cause the right hand side variables are predetermined and are the same in each equation. The first step in estimating the VAR model is to decide upon the appropriate lag length (p). Following Hakkio and Morris (1984), this was accomplished by setting the maximum lag length at 10 and using likelihood ratio tests to examine each restriction against all other possibilities. This procedure yielded lag lengths of between four and eight quarters for the countries stud- ied. The second column of Table 2 provides the details. Exami- nation of the Ljung-Box Q-statistics provided support for this choice of lag length by failing to detect the existence of residual autocor- relation in any of the equations.

205

Page 10: Fiscal policy and current account performance: International evidence on the twin deficits

Colm Kearney and Mehdi Monadjemi

Table 1 provides a summary description of the performance of the model by presenting the F-statistics (with their associated marginal significance levels) of the joint significance of all lags of the relevant variables in the equations for government expenditure, tax revenues, and the current account balance. These results cast light upon the direction of causality that exists between the twin deficits.

Looking first at the government expenditure and tax equa- tions, there is evidence of significant feedback to the government’s fiscal deficit from macroeconomic variables such as the exchange rate and the current account balance. Although the latter is con- sistent with the existence of a twin deficits relationship, it implies the existence of reverse directional causality, which has been re- ported for the United States by Darrat (1988). Inspection of the current account equation suggests that this feedback effect exists internationally and may well dominate the conventional view of fis- cal policy as the control variable that causes current account per- formance. Indeed, the performance of the current account equation for each of the eight countries included in this study indicates that this variable is exceedingly difficult to model and may well ap- proximate a random walk.

Table 2 sheds further light on this issue by presenting the sums of lag coefficients for each of the explanatory variables in the equations for the current account balance. This is a convenient pre- sentational device because the individual coefficient estimates of the A matrix in Equation (6) do not convey much information and be- cause it provides a good indication of the simulation properties of the VAR model. Examination of the diagnostics in the last three columns of the table indicates that the equations are well specified with no evidence of residual autocorrelation. A number of inter- esting points emerge from inspection of the estimates. First, for 6 of the 8 countries the lagged values of the dependent variable pro- vide the best explanation of current balance behavior with cumu- lative coefficients that are close to unity. This finding corroborates that reported in Table 2. Second, in spite of this, there is evidence of other sum coefficients being different from zero at the 5% and 10% levels of statistical significance, There does not, however, emerge any internationally unified set of results concerning the qualitative effects of government expenditure and its financing on the current account balance. Third, with the exception of one country, the evi- dence presented in the table is consistent with the Marshall-Leaner

206

Page 11: Fiscal policy and current account performance: International evidence on the twin deficits

TABL

E 1.

]o

int

Sign

ifica

nce

Test

s fo

r th

e Tw

in

Def

icits

Eq

uatio

ns

of

the

VAR

Mod

el

Equa

tion

Austr

alia

Brita

in Ca

nada

Fr

ance

Ge

rman

y Ire

land

Italy

USA

(1)

Gov

ernm

ent

Expe

nditu

re

(G)

G

T M

c X

(2)

Tax

Rec

eipt

s (T

) G

T M

c X

(3)

Cur

rent

Ac

coun

t (C

) G

T M

C

X

76.8

(0

.00)

0.

9 (0

.48)

3.

4 (o

.ol)*

1.

7 (0

.15)

4.

7 (o

.oo)

*

2.0

(0.0

9)

3.9

(0.0

1)

0.3

(0.9

1)

2.8

(0.0

3)*

1.6

(0.1

8)

1.2

(0.3

3)

1.8

(0.1

3)

0.6

(0.6

9)

29.8

(0

.00)

1.

7 (0

.15)

29.6

(0

.00)

1.

8 (0

.13)

1.

2 (0

.35)

2.

3 (0

.06)

1.

6 (0

.18)

0.9

(0.5

2)

5.4

(0.0

0)

1.9

(0.1

1)

2.2

(0.0

7)

0.5

(0.8

1)

0.6

(0.7

3)

0.7

(0.6

4)

1.3

(0.2

8)

44.1

(0

.00)

0.

4 (0

.88)

25.2

(0

.00)

0.

4 (0

.89)

0.

4 (0

.90)

1.

0 (0

.49)

0.

9 (0

.51)

0.8

(0.6

1)

2.8

(0.0

2)

0.7

(0.6

7)

0.9

(0.5

2)

1.5

(0.2

1)

2.9

(0.0

3)*

2.1

(0.0

9)

2.0

(0.1

0)

33.6

(0

.00)

1.

6 (0

.18)

3.8

(0.0

1)

2.3

(0.0

6)

0.7

(0.6

2)

3.1

(0.0

2)*

2.5

(0.0

5)

0.9

(0.0

5)*

2.6

(0.0

4)

0.5

(0.7

4)

1.5

(0.2

1)

2.6

(0.0

4)*

1.3

(0.2

9)

2.1

(0.0

9)

1.0

(0.4

2)

63.3

(0

.00)

0.

8 (0

.53)

87.2

(0

.00)

3.

1 (0

.03)

* 1.

2 (0

.32)

1.

3 (0

.28)

2.

0 (0

.11)

1.3

(0.3

0)

18.5

(0

.00)

0.

6 (0

.69)

3.

9 (o

.ol)*

0.

3 (0

.87)

1.0

(0.4

4)

2.5

(0.0

6)

1.0

(0.4

4)

206

(0.0

0)

6.8

(o.o

o)*

44.0

(0.

00)

1.9

(0.1

2)

2.2

(0.0

8)

3.2

(0.0

2)*

1.8

(0.1

4)

3.5

(o.o

l)*

2.2

(0.0

7)

3.6

(O.O

l)*

2.0

(0.1

1)

1.3

(0.2

7)

1.3

(0.2

7)

1.2

(0.3

2)

1.1

(0.3

6)

110

(0.0

0)

1.8

(0.1

3)

2.3

(0.0

6)

43.6

(0

.00)

3.

1 (0

.02)

* 1.

1 (0

.36)

0.

4 (0

.89)

2.

15

(0.0

9)

3.1

(0.0

2)*

3.0

(0.0

2)*

2.2

(0.0

7)

2.1

(0.0

9)

2.9

(0.0

3)*

0.5

(0.7

5)

6.0

(0.0

1)

46.2

(0

.00)

1.

7 (0

.16)

0.

5 (0

.76)

7.

7 (O

.OO)

* 0.

5 (0

.76)

2.

9 (0

.03)

* 1.

6 (0

.20)

2.0

(0.1

0)

5.5

(o.o

o)*

0.9

(0.5

0)

4.3

(o.o

o)*

1.8

(0.1

3)

3.8

(O.O

l)*

17.7

(0

.00)

71

.0

(0.0

0)

2.4

(0.0

6)

2.6

(0.0

4)*

NOTE

: Th

e va

riabl

es

C,

T,

M,

C an

d X

are

as d

escr

ibed

in

the

text.

Th

e fig

ures

re

porte

d ar

e th

e F-

statis

tics

that

te

st th

e jo

int

sign&

cam

e of

al

l la

gs

of

the

rele

vant

va

riabl

e in

ea

ch

equa

tion.

Th

e fig

ures

in

pa

rent

hese

s ar

e m

argi

nal

signi

fican

ce

leve

ls.

Aste

risks

de

note

st

atist

ical

sign&

canc

e at

th

e 5%

le

vel.

For

clarit

y, as

teris

ks

do

not

appe

ar

in

“own

” va

riabl

es

in

each

eq

uatio

n.

Page 12: Fiscal policy and current account performance: International evidence on the twin deficits

TABL

E 2.

Su

ms

of

Coef

ficie

nts

in th

e Cu

rrent

Ac

coun

t Eq

uatio

ns

of

the

VAR

Mod

el

Sum

of

Coe

fficie

nts

lag

Leng

th

Cons

tant

Tr

end

G T

M

c X

R-2

SER

0

Austr

alia

6 0.

100*

-0

.001

**

-0.0

41

-0.3

54*

-0.0

70

0.39

8**

-0.0

10

0.98

0.

003

18.4

5 (2

.16)

(1

.W

(1.0

3)

(1.5

8)

(0.2

5)

(1.9

5)

(1.2

4)

(0.6

2)

Brita

in 6

-0.0

28

-0.0

22

0.12

1 -0

.348

* 0.

932*

* -0

.006

0.

94

0.00

4 24

.50

(0.8

2)

(0”: y

(0

.16)

(0

.68)

(1

.48)

(1

4.51

) (0

.39)

(0

.30)

Ca

nada

8

-0.0

05

-0.0

001

0.31

2*

0.00

4 0.

143*

1.

008*

* -0

.007

0.

97

0.00

05

24.0

4 (0

.29)

(0

.W

(1.5

9)

(0.0

1)

(1.3

6)

(13.

04)

(1.1

6)

(0.2

9)

Fran

ce

5 0.

015

-0.0

001

0.06

3 -0

.098

-0

.019

0.

909*

* -0

.007

0.

95

0.00

2 17

.48

(0.2

3)

(1.0

2)

(0.9

7)

(0.5

2)

ww

(9.9

2)

(0.4

7)

(0.6

8)

Germ

any

4 0.

016*

* 0.

0001

**

-0.0

95**

-0

.259

**

-0.1

13*

0.96

1**

-0.0

07**

0.

98

0.00

1 15

.30

(2.9

4)

(4.1

6)

(1.7

0)

(1.8

4 (1

.61)

(2

4.76

) (3

.53)

(0

.81)

Zr

eland

5

-0.1

99**

0.

0000

0.

318*

* -0

.084

-0

.491

* 1.

032*

* 0.

085*

* 0.

98

0.01

0 18

.02

(2.5

7)

(0.1

2)

(2.0

3)

(0.4

1)

(1.5

0)

(19.

60)

(2.3

2)

(0.6

9 Ita

ly 8

0.15

3**

-0.0

012*

* -0

.179

**

0.16

8*

0.41

2**

0.10

9 -0

.069

**

0.97

0.

004

17.5

2 (3

.12)

(2

.22)

(2

.15)

(1

.67)

(2

.57)

(0

.42)

(3

.21)

lo.

@)

Unite

d St

ates

5

(0”:;

(0

0:;

-0.0

91**

0.

130*

* 0.

093*

* 0.

852*

* -0

.009

E-2*

* 0.

99

0.00

2 21

.26

(2.4

7)

(3.5

8)

(1.9

9 (1

5.67

) (1

.99)

to

.44

NOTE

: Th

e va

riabl

es

G,

T,

M,

C an

d X

are

defin

ed

in

the

text.

I?

, SE

R,

and

Q de

note

re

spec

tively

th

e ad

just

ed

corre

latio

n co

effic

ient

, th

e re

gres

sion

stand

ard

erro

r, an

d th

e Lj

ung-

Box

Q-sta

tistic

wi

th

its

mar

gina

l sig

nific

ance

le

vel

in

brac

kets.

Th

e ot

her

figur

es

in

pare

nthe

ses

are

t-sta

tistic

s wi

th

one

and

two

aste

risks

de

notin

g sig

nific

ance

at

th

e 10

%

and

5%

leve

ls.

The

mar

gina

l sig

niBc

ance

le

vel

of

thes

e sta

tistic

s dif

fers

ac

ross

eq

uatio

ns

due

to

vary

ing

degr

ees

of

freed

om

that

re

sult

from

th

e d&

rent

la

g le

ngth

s.

The

F-sta

tistic

s in

Ta

ble

1 te

st wh

ethe

r al

l la

gs

of

a va

riabl

e ca

n be

ex

clude

d fro

m

an

equa

tion,

wh

erea

s th

e t-s

tatis

tics

repo

rted

in

this

tabl

e te

st wh

ethe

r th

e su

m

of

the

lag

coe&

ient

s is

stat

istica

lly

diffe

rent

fro

m

zero

.

Page 13: Fiscal policy and current account performance: International evidence on the twin deficits

Fiscal Policy and Current Account Performance

conditions whereby an exchange rate depreciation will result in im- proved performance on the current account balance.

These results may, of course, conceal important short-term ef- fects as the current account adjusts dynamically in response to vari- ations in government spending, its financing and to the exchange rate. Since these short-term responses are of concern to those who design and implement macroeconomic policy, we now turn to ex- amine the simulation properties of the model in order to cast light upon the dynamics of the twin deficits relationship.

4. Policy Simulation Results Our purpose in this section is to investigate the dynamic prop-

erties of the twin deficits relationship by examining the impulse response functions of the VAR model as presented in Equation (6). The latter correspond to the vector moving average (VMA) repre- sentation of the VAR model, which exists when the conditions of the multivariate Wold decomposition theorem obtain. Appropriate construction of the impulse response functions, however,. necessi- tates orthogonalization of the errors by ordering the variables in our VAR model with the most important variables being placed first and the least important variables being placed last in the y vector. Con- veniently, the VMA representation allows us to order the variables in Equation (6) according to their exogeneity by examining the vari- ance decomposition of the k-step-ahead prediction of yt, conditional on all information available at time t.

Table 3 provides the results of this exercise for the eight coun- tries in our sample with the most important variables appearing first. A number of interesting observations emerge. First, no variable in this study is completely exogenous insofar as a maximum of 60% of the 40-step-ahead prediction error of any variable is explained by innovations in itself. Second, the ranking of variables by degree of importance (that is, exogeneity) is different in each country. For example, both tax revenues (t) and the exchange rate (X) range from being the most exogenous variables in some countries (Aus- tralia and Germany, respectively) to being the least exogenous vari- ables in others (Britain and France together with Italy, respec- tively). This variation reflects ditYerent market and political structures and partly explains why different countries react differently to sim- ilar economic disturbances and why the achievement of interna- tional macroeconomic policy coordination is not easily achievable. Third, in terms of the twin deficits relationship, the ranking of fiscal

209

Page 14: Fiscal policy and current account performance: International evidence on the twin deficits

TABLE 3. Variance Decomposition of the VAR Model

Percentage of 40-step-ahead Dependent prediction error in y which is

Variable explained by innovations in Country (Y) (GTMCX)

Australia:

T X C G M

Britain:

G C M X T

Canada:

G 60 30 5 3 2 X 67 22 7 2 3 T 53 23 18 2 4 C 73 5 8 10 4 M 40 31 18 8 3

C 50 22 12 8 8 G 35 32 16 9 8 T 18 29 28 10 15 M 9 15 16 55 5 X 33 17 25 6 19

Germany:

X 58 21 13 3 5 G 24 46 14 8 8 T 41 19 27 4 9 M 25 9 5 49 12 C 49 20 12 3 15

T X C G M

54 13 22 6 5 24 46 7 19 5 37 10 44 4 5 20 38 4 28 10 14 9 37 8 32

G C M X T

61 25 5 6 3 25 45 16 11 2 34 20 29 6 11 47 32 7 7 3 49 28 4 14 18

G X T C M

C G T M X

X G T M C

Page 15: Fiscal policy and current account performance: International evidence on the twin deficits

Fiscal Policy and Current Account Performance

TABLE 3. Variance Decomposition qf the VAR Model (cont’d)

Country

Ireland:

Percentage of 4O-step-ahead Dependent prediction error in y which is

Variable explained by innovations in

(Y) (GTMCX)

G T M X C

Italy:

United States: G T X C M

G 55 15 10 10 10 T 52 21 10 9 8 M 33 19 25 3 19 X 45 14 9 23 8 c 44 18 10 14 14

G M C T X

G 48 23 16 9 4 M 36 26 19 6 3 C 36 26 16 18 4 T 42 17 17 18 6 X 34 36 8 15 7

G 39 25 14 17 6 T 30 32 21 14 3 X 33 19 28 16 4 C 36 21 20 18 4 M 27 17 13 9 34

policy variables (that is, g and t) in order of importance exceeds the current account balance (that is, c) in only half the countries (Canada, Germany, Ireland and the United States). As alluded to earlier, however, there is evidence of considerable reverse causa- tion from c to g and t even in these countries.

Having appropriately orthogonalized the errors in our VAR model, we can now proceed to examine the impulse response func- tions in order to cast light upon the dynamics of the twin deficits relationship. Borrowing previous notation allows us to write the government’s budget constraint as

G=T+AB+AM, (7)

211

Page 16: Fiscal policy and current account performance: International evidence on the twin deficits

Colm Kearney and Mehdi Monadjemi

where B denotes outstanding government debt, M denotes the stock of money, and G = C, + Z, as in Equation (2). Equation (7) states that the government can finance its expenditures by raising tax rev- enues, by issuing debt or by monetary expansion. Figures I-3 de- pict the impulse response functions for the current account balance to the following innovations:

(i) an increase in government expenditure, which is financed by issuing debt,

(ii) an increase in government expenditure, which is financed by raising taxes (that is, a balanced budget fiscal expan- sion), and

(iii) an increase in government expenditure, which is financed by additional monetary growth.

In each case a one standard deviation innovation in the relevant variable was normalized to unity in order to ensure that the budget remains in balance in simulation (ii) and that the additional mon- etary expansion in simulation (iii) is equal to the increase in gov- ernment expenditure. Simulation (i) was performed by noting from the government’s budget constraint, Equation (7), that additional expenditure which is not financed by taxes or by monetary expan- sion must be financed residually by issuing more debt. Finally, all variables continue to be expressed as a ratio of nominal GDP in order to facilitate international comparison of the magnitude of re- sponses to the fiscal policy innovations.

Figure 1 presents the results of simulation (i). Perhaps the most obvious conclusion that follows from casual inspection of the Figure concerns the richness of the adjustment dynamics across countries. The current account performance in six of the eight countries studied here (the exceptions being Australia and France) experiences initial deterioration that varies in duration and magni- tude. The adverse current account performance lasts approximately one year in Ireland, two years in Britain and Canada, three years in Italy and the United States, and five years in Germany. Con- cerning the magnitude of responses, half of the countries (Canada, Germany, Ireland and the United States) exhibit short-term mul- tipliers of unity or greater, while the rest exhibit short-term mul- tipliers that are considerably less than unity. Finally, all countries eventually experience prolonged periods of enhanced current ac- count performance as the systems adjust dynamically toward their stationary long-run equilibria.

Simulations (ii) and (iii) provide interesting comparisons with the results presented above. Casual inspection of Figures 2 and 3

212

Page 17: Fiscal policy and current account performance: International evidence on the twin deficits

0.3

0

-0.2

2.0 12 24 36 48 . 1

France 0 - n /\e

4 \/ \ / \/ -0.34 v v -

12 24 36 48 1.6, -

1.8 12 24 36 48 I /- 1

0 - -1.6 ’

Ireland

n 3 12 24 J

36 48

Figure 1. Simulation 1: The Current Account Effects of an Unanticipated Bond-Financed

Increase in Government Expenditure

Page 18: Fiscal policy and current account performance: International evidence on the twin deficits

-:;;;I if:: -;:;I] -::;;\ -;1:3/\::““1 I:;;]

0.8 12 24 36 48

0

-1.2 J 12 24 36 4%

Figure 2. Simulation 2: The Current Account Effects of an Unanticipated Balanced Budget

Increase in Government Expenditure

Page 19: Fiscal policy and current account performance: International evidence on the twin deficits

Australia A- \

12 24 36 48

0.3-

0- A - I

Italy

Figure 3.

Simulation 3: The Current Account Effects of an Unanticipated Money-Financed Increase in Government Expenditure

Page 20: Fiscal policy and current account performance: International evidence on the twin deficits

Colm Kearrey and Mehdi Monadjemi

suggests that the pattern of the current account’s dynamic response to innovations in the stance of fiscal policy is independent of the government’s financing decision. Although this is obviously the case in long-run equilibrium, it does not constitute evidence in favor of Ricardian equivalence. Inspection of Figures 1 and 2, for example, reveals that a substitution of taxes for debt has important implica- tions for the short-term dynamic response of the current account balance to innovations in government spending. These effects, how- ever, are not uniform across countries. A swap in government fi- nancing of this type will exacerbate Australia’s current account po- sition while improving Germany’s by reducing the duration of its deficit from approximately five to three years. It is intuitively plau- sible that this result stems from the different net external asset po- sitions of these countries.

The findings that emerge from our empirical analysis of eight countries can be summarized as indicating the existence of a tem- porary twin deficits relationship between the stance of fiscal policy and performance on the current account of the balance of pay- ments, which does not persist over time. Examination of the im- pulse response functions confirms that fiscal expansions will lead to prolonged periods of improved current account performance as the economy adjusts towards its long-run equilibrium. The twin deficits relationship varies internationally in magnitude and duration, and it is not independent of the government’s financing decision. Fi- nally, the evidence presented here suggests the existence of strong feedback effects from current account performance to the stance of fiscal policy.

5. Concluding Comments Theoretical macroeconomic models of recent vintage yield di-

vergent predictions about the relationship that exists between the stance of fiscal policy and the economy’s performance on current account of the balance of payments. In view of the considerable uncertainty that continues to pervade appropriate specification of open economy macromodels, the empirical analysis of this paper utilizes the VAR technique to investigate the extent of international evidence in support of the twin deficits proposition. This technique is particularly suited to this problem because a VAR model consti- tutes an unrestricted reduced form of some unknown structural model. Using quarterly data from eight countries over the recent

216

Page 21: Fiscal policy and current account performance: International evidence on the twin deficits

Fiscal Policy and Current Account Performance

period of floating exchange rates from 1972:i-1987:iv, the existence of a temporary twin deficits relationship that does not persist over time was demonstrated. In addition, substantial international evi- dence of reverse causation is uncovered, along with the existence of complex short-term adjustment dynamics that are not invariant to the government’s financing decision, and are capable of inverting the twin deficits relationship for substantial periods of time. The policy conclusion that emerges from this study is that tight fiscal policy should not be relied upon in isolation to deliver sustained improvement in current account performance.

Received: November 1987 Final version: August 1989

References Barro, Robert J. “Are Government Bonds Net Wealth?” Journal of

Political Economy 82 (1974): 1095-117. -. “On the Determination of Public Debt.” Journal of Polit-

ical Economy 87 (1979): 940-71. Blanchard, Olivier J. “Debt, Deficits and Finite Horizons.” Journal

of Political Economy 93 (1985): 223-47. Branson, William H. “Asset Markets and Relative Prices in Ex-

change Rate Determination.” Socialwissenschaf lithe Annalen 1 (1976): 69-89.

Cuddington, John T., and Jose M. Vinals. “Budget Deficits and the Current Account in the Presence of Classical Unemployment.” Economic Journal 96 (1986): 101-19.

-. “Budget Deficits and the Current Account: An Inter- temporal Disequilibrium Approach.” Journal of International Economics 21 (1986): l-24.

Darrat, Ali F. “Have Large Budget Deficits Caused Rising Trade Deficits?’ Southern Economic Journal 56 (1988): 879-87.

Dixit, Avinesh K. “The Balance of Trade in a Model of Temporary Equilibrium with Rationing.” Review of Economic Studies 45 (1978): 393-404.

Dornbusch, Rudiger. “Expectations and Exchange Rate Dynamics.” Journal of Political Economy 84 (1976): 101-19.

Dwyer, Gerald P. “Federal Deficits, Interest Rates and Monetary Policy.” Journal of Money, Credit, and Banking 17 (1985): 655- 81.

217

Page 22: Fiscal policy and current account performance: International evidence on the twin deficits

Colm Kearney and Mehdi Monadjemi

Frenkel, Jacob A., and Assaf Razin. “Fiscal Policies in the World Economy.” Journal of Political Economy 94 (1986): 564-94.

Greenwood, Jeremy. “Expectations, the Exchange Rate and the Current Account.” Journal of Monetary Economics 12 (1983): 543- 69.

Greenwood, Jeremy, and Kent P. Kimbrough. “Capital Controls and Fiscal Policy in the World Economy.” Canadian Journal of Eco- nomics 18 (1985): 743-65.

Hakkio, Craig S., and Charles S. Morris. “Vector Autoregressions: A User’s Guide.” Discussion Paper No. RWP84-10, Federal Re- serve Bank of Kansas City, 1984.

Helpman, Elhanan, and Assaf Razin. “Exchange Rate Management: Intertemporal Tradeoffs.” American Economic Review 77 (1987): 107-23.

Kawai, Masahiro. “Exchange Rates, the Current Account and Mon- etary-Fiscal Policies in the Short Run and in the Long Run.” Oxford Economic Papers 37 (1985): 391-425.

Kimbrough, Kent P. “An Examination of the Effects of Government Purchases in an Open Economy.” Journal of International Money and Finance 1 (1985): 113-33.

Laney, Leroy 0. “The Strong Dollar, the Current Account and Federal Deficits: Cause and Effect.” Federal Reserve Bank of Dallas Economic Review (January 1984): 1-14.

Leiderman, Leonardo, and Mario I. Blejer. “Modelling and Testing Ricardian Equivalence.” Znternational Monetary Fund Staff Pa- pers 34 (1987): l-35.

Lucas, Robert E., Jr. “Econometric Policy Evaluation: A Critique.” In The Phillips Curve and Labour Markets, edited by Karl Brun- ner and Alan Meltzer, 19-46. Carnegie-Rochester Conference Series on Public Policy No. 1. Amsterdam: North-Holland, 1976.

Marston, Richard C. “Stabilization Policy in Open Economies.” In vol. 2 of Handbook of Znternational Economics, edited by Ron- ald W. Jones and Peter B. Kenen, 859-916. Amsterdam: Elsev- ier Science Publishers B.V., 1985.

Moore, Michael J. “Inventories in the Open Economy Macromodel: A Disequilibrium Analysis. ” Review of Economic Studies 56 (1989): forthcoming.

Mundell, Robert A. “Capital Mobility and Stabilization Policy un- der Fixed and Flexible Exchange Rates.” Canadian Journal of Economics and Political Science 29 (1963): 475-85.

Neary, J. Peter. “Nontraded Goods and the Balance of Trade in a

218

Page 23: Fiscal policy and current account performance: International evidence on the twin deficits

Fiscal Policy and Current Account Performance

Neo Keynesian Temporary Equilibrium.” Quarterly Journal of Economics 95 (1980): 403-29.

Obstfeld, Maurice. “Macroeconomic Policy Exchange Rate Dynam- ics and Optimal Asset Accumulation.” Journal of Political Econ- omy 89 (1981): 1142-61.

Persson, Torsten. “Global Effects of National Stabilization Policies under Fixed and Floating Exchange Rates.” Scandinavian Jour- nal of Economics 84 (1982): 165-92.

Razin, Assaf, and Lars E.O. Svensson. “The Current Account and the Optimal Government Debt.” Journal of Znternational Money and Finance 2 (1983): 215-24.

Rodriguez, Carlos A. “Short and Long Run Effects of Monetary and Fiscal Policy Under Flexible Exchange Rates with Perfect Capital Mobility.” American Economic Review 69 (1979): 176-82.

Sachs, Jeffrey D. “Wage Indexation, Flexible Exchange Rates and Macroeconomic Policy.” Quarterly Journal of Economics 94 (1989): 731-47.

Stiglitz, George E. “On the Relevance or Irrelevance of Public Fi- nancial Policy: Indexation, Price Rigidities and Optimal Mone- tary Policy.” In Znflation Debt and Zndexation Policy, edited by Rudiger Dornbusch and Mark H. Simonsen. Cambridge: MIT Press, 1983.

Zellner, Arnold. “Causality and Econometrics.” In Three Aspects of Policy and Policymaking: Knowledge, Data and Institutions, ed- ited by Karl Bruner and Alan Meltzer, 9-54. Carnegie-Rochester Conference Series on Public Policy No. 10. Amsterdam: North- Holland, 1979.

Zellner, Arnold, and Franz Palm. “Time Series Analysis and Si- multaneous Equation Econometric Models.” Journal of Econo- metrics 8 (1974): 17-54.

219