Regional – Local Dimension of Russia’s Fiscal … · Regional-Local Dimension of Russia’s Fiscal Equalization 5 may also be driven by political goals unrelated to economic objectives

International Studies ProgramWorking Paper 06-16 (Updated)November 2006

Regional – Local Dimension of Russia’s Fiscal Equalization

Jorge Martinez-Vazquez Andrey Timofeev

International Studies Program Andrew Young School of Policy Studies Georgia State University Atlanta, Georgia 30303 United States of America Phone: (404) 651-1144 Fax: (404) 651-4449 Email: [email protected] Internet: http://isp-aysps.gsu.edu Copyright 2006, the Andrew Young School of Policy Studies, Georgia State University. No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means without prior written permission from the copyright owner.

International Studies Program Working Paper 06-16 (Updated)

Regional – Local Dimension of Russia’s Fiscal Equalization Jorge Martinez-Vazquez Andrey Timofeev November 2006

International Studies Program Andrew Young School of Policy Studies The Andrew Young School of Policy Studies was established at Georgia State University with the objective of promoting excellence in the design, implementation, and evaluation of public policy. In addition to two academic departments (economics and public administration), the Andrew Young School houses seven leading research centers and policy programs, including the International Studies Program. The mission of the International Studies Program is to provide academic and professional training, applied research, and technical assistance in support of sound public policy and sustainable economic growth in developing and transitional economies. The International Studies Program at the Andrew Young School of Policy Studies is recognized worldwide for its efforts in support of economic and public policy reforms through technical assistance and training around the world. This reputation has been built serving a diverse client base, including the World Bank, the U.S. Agency for International Development (USAID), the United Nations Development Programme (UNDP), finance ministries, government organizations, legislative bodies and private sector institutions. The success of the International Studies Program reflects the breadth and depth of the in-house technical expertise that the International Studies Program can draw upon. The Andrew Young School's faculty are leading experts in economics and public policy and have authored books, published in major academic and technical journals, and have extensive experience in designing and implementing technical assistance and training programs. Andrew Young School faculty have been active in policy reform in over 40countries around the world. Our technical assistance strategy is not to merely provide technical prescriptions for policy reform, but to engage in a collaborative effort with the host government and donor agency to identify and analyze the issues at hand, arrive at policy solutions and implement reforms. The International Studies Program specializes in four broad policy areas: Fiscal policy, including tax reforms, public expenditure reviews, tax administration reform Fiscal decentralization, including fiscal decentralization reforms, design of intergovernmental

transfer systems, urban government finance Budgeting and fiscal management, including local government budgeting, performance-

based budgeting, capital budgeting, multi-year budgeting Economic analysis and revenue forecasting, including micro-simulation, time series

forecasting, For more information about our technical assistance activities and training programs, please visit our website at http://isp-aysps.gsu.edu or contact us by email at [email protected].

1

Regional-Local Dimension of Russia’s Fiscal Equalization

Jorge Martinez-Vazquez and Andrey Timofeev* Andrew Young School of Policy Studies, Georgia State University

Abstract

National equalization policies often are concerned with the quality and accessibility of services delivered at the lowest (or local) level of government. When these policies are implemented in a hierarchical fashion through the intermediate level governments, national equalization policies need to take into account possible offsetting effects originating in regional or provincial government policies. In this paper, we examine recent fiscal equalization outcomes for about 2,000 Russian local governments to assess and explain the extent of equalization differences between and within regions. In particular, we examine the claim that intraregional policies should be blamed for rising disparities in local fiscal outcomes despite Russia’s federal government efforts to equalize sub-national government finances. Keywords: Russia; Fiscal Disparities; Tax-sharing; Equalization Grants; Local Government JEL Classification: H73, H77, P35

* Corresponding author: Andrew Young School of Policy Studies, Georgia State University, P.O. Box 3992, Atlanta, GA

30302-3992, tel: 404-651-0421, fax: 404-651-4449, e-mail: [email protected].

nternational Studies Program Working Paper Series

2

1. Introduction

Most of the fiscal federalism literature has been traditionally centered on central-

regional relations.1 At the same time, in many countries the bottom tier of government is

responsible for many social services the quality and accessibility of which is often

targeted by national equalization policies.2 Thus, when designing their equalization

policies central governments need to take into account possible offsetting effects of

policies implemented by intermediate level governments.3 In those countries where the

central government does not deal directly with the bottom tier but has to rely instead on

the liaison provided by regional governments, the central authorities need to embed

incentive mechanisms for aligning the preferences of the regional governments with

national equalization goals. The importance of regional-local relations for national

equalization policies has been identified in the literature but it remains relatively little

studied and understood (Bahl and Wallace, 2003). In this paper, we examine fiscal

outcomes in about 2,000 local governments belonging to 72 Russian regions to assess and

explain the extent of equalization between and within regions in 2001.

In the paper we will not attempt to assess the efficiency of the actual equalization

outcomes against any normative criteria. In fact, the first-best theory of fiscal federalism

provides no rationale for equalization grants. Local governments are prescribed to engage

1 In this paper we use the term “regions” as a general description of the intermediate level government, which receives

other names in different country settings, such as provinces, states, and so on. 2 One review found that of 165 countries with available written constitutions, 116 made reference to a right to education

and 73 to a right to health care (Gauri 2004). At the same time in many countries, local governments are to some extent engaged in the provision of these two social services.

3 This is especially important in hierarchical organizations of government structures where central governments rely on regional governments to implement their policies, including equalization, toward local governments. Alternative government structures, such as bi-furcated systems where the central government directly implements its policies with local governments and there is little hierarchical dependency between regional and local governments, are less prone to this type of problems. Regional government policies vis-à-vis equalization may differ form those of their central governments for a variety of reasons; for example, regional governments, especially poorer ones, may weigh economic development objectives more heavily than equalization objectives.


3

only in the provision of local public goods financed with benefit taxation. When the

benefits of local public goods spill over the jurisdiction border, the central government is

supposed to internalize these benefits through conditional matching grants to local

governments. However, in the second-best world it might be unfeasible for local

government to rely exclusively on benefit taxation.4 At the same time non-benefit

taxation necessary to make up the difference would be associated with a larger

deadweight loss when undertaken at the local level due to mobility of economic agents

across local jurisdictions. These are a few examples when resource allocation by local

governments might require unconditional grants from a higher-level government.

In practice, unconditional grants constitute a prevalent form of intergovernmental

transfers in many countries. Moreover, the notion of “fiscal equity” appears to be the

prime goal for most grant schemes (Le Grand 1975). The common equalization targets

include equal service for a given level of tax effort or equal tax effort for a given level of

service provision. Easing the burden of providing the standard quality of public services

requires larger assistance in the form of unconditional grants to localities with smaller

revenue capacity and larger expenditure need.

In this paper we accept the declared equalization policy of Russia’s government

and assess the actual outcome against this declared goal. A number of federal acts declare

equalization of local fiscal outcomes as a national policy goal in Russia. For example, the

Budget Code of 1998 requires both the federal and regional governments to establish

equalization schemes for jurisdictions at the level immediately below.5 In another

4 For, example if the production of local public goods exhibits decreasing costs, revenue generated through benefit taxation

would not be sufficient to cover total costs with marginal costs being below average costs. See, for example, the discussion in Tresch (2002, p. 844).

5 In particular, these goals appear in articles 131 and 137-138 of the Budget Code.

International Studies Program Working Paper Series

4

example, the national reform agenda for 2001-2005 stated among the goals of the system

of intergovernmental relations equal access of citizens to basic public services and social

security regardless of where they reside.6 Because many of those services are provided at

the local level, the fulfillment of this goal would require some equalization of fiscal

resources available to local governments. As part of this reform agenda, the 2001 reforms

of revenue sharing aimed at making it more transparent and to narrow existing fiscal

disparities.

Nevertheless, local government budget data for 2001 show a wide variation in

local governments’ per capita revenue from their own taxes and from federal and regional

tax sharing sources. Because in Russia local governments do not receive direct aid from

the federal government, the only vehicle for their equalization is through the allocation of

shared tax revenue and grants by the regional authorities.7 Indeed, as we document

further below, regional-local revenue sharing and grants on average considerably

decrease total variation in fiscal outcomes (by comparison to the distribution of own

revenues.) However, comparing the extent of inequality before and after “equalization”

(effected through shared tax revenues and grants) across regions shows a dramatic

decrease in local fiscal disparities in some regions while almost no equalization seems to

be achieved in others.

There are multiple reasons for these outcomes. Firstly, some regional

governments may value more heavily other economic objectives, such as economic

development. Also in reality the allocation of regional-local intergovernmental grants

6 Program of developing Fiscal Federalism in the Russian Federation for 2001-2005 (Resolution No. 584 of the RF

Government of August 15, 2001). 7 This study excludes so-called ‘restricted access’ territories (‘zakrytye administrativno-territorialnye obrazovanya’ or

ZATOs), which have no fiscal relations with the regional government. These are about 40 localities that enjoy a special status of a federal territory due to presence of sensitive military or nuclear installations.


5

may also be driven by political goals unrelated to economic objectives. For example, a

regional government might use its control over equalization grants to reward local allies

or punish local authorities representing a rival political party. As a result, the political

choice for the desired level of equalization is hard to trace from publicly stated policy

documents.

Our approach in this paper will be to assume that the desired degree of

equalization is implicitly revealed by the local fiscal disparities that exist before and after

equalization. The empirical results will shed some light on whether regional policies

offset or reinforce the central government’s equalization efforts. In addition, in this paper

we attempt to explain the existing differences in intra-regional equalization by looking at

three sets of possible causes grouped as political, economic, and structural (demographic

and geographic) factors.

The remainder of the paper is organized as follows: In the next section we

provide a brief description of local governments in Russia and their revenue sources.

Then we briefly review the legal framework for intergovernmental transfers in Russia.

After that we turn to the assessment of fiscal outcomes at the local government level

resulting from the assignment of tax revenues sources and allocation of

intergovernmental transfers. Next, we attempt to explain why the extent of intra-regional

equalization varies from region to region. Our conclusions follow.

2. Local Government Revenue in Russia

The Russian Federation has a population of 144 million, out of which 73 percent

reside in urban areas. Before the recent wave of government amalgamations, the


6

Federation was comprised of 89 subjects: ethnic republics, krais, okrugs, oblasts, and

autonomous areas – hereafter referred to as “regions.” Below the regional level, there are

one or two levels of local government, depending on the region. The prevailing one-tier

system has local government established at the level of cities and rayons (an equivalent

of US county) with an average size of 57.9 thousand. However, in about fifteen regions,

local government is also established at the level of sub-rayon townships and rural

districts, with an average size of 5.4 thousand. In this latter case, at the rayon level we

have either an overlapping local government or a deconcentrated arm of the regional

government. All formally established governments have political autonomy secured in

the constitution. However, as discussed below, this autonomy is constrained with the de

facto monopolization of tax policy by the federal government.

Revenue assignments in the Russian Federation have been evolving on an

ongoing basis since the beginning of the transition. The current federal legislation does

not allow any level of government to introduce taxes beyond those enumerated in the Tax

Code.8. The list of permitted taxes is broken into three categories: “federal” revenue

sources, “regional” revenue sources, and “local” revenue sources.

Classification of a tax into a particular category does not always determine the

level of government that receives the proceeds from this tax. For some taxes (designated

as “assigned”), revenue sharing is fixed in legislation. For others (designated as

“regulated”), revenue sharing is determined by the higher-level government as part of the

annual budget process.

8 In the course of 2000-05, gradual enactment of the Tax Code chapters replaced respective provisions of the previous Law

on the Basic Principles of Taxation.


7

Parts of Personal Income Tax (PIT) collection and excises on goods (and until

2001, also the VAT) are shared by the federal government with regional governments

based on the point of collection for these taxes. In turn, regional governments can

reallocate a portion of those revenues to their local government. Similarly, royalties on

natural-resources extraction are in part assigned to regional governments (and until year

2002, also to localities) at the point of collection. In addition, regional governments can

determine a surtax on the federal Corporate Income Tax (CIT) base of up to 17.5 percent

(and also 2 percent by local governments before 2005).

For the purpose of this paper, we classify local government revenue sources into

four broad categories ranked by the decreasing degree of control that sub-national

governments can exercise: own-source revenue, assigned revenue, regulated revenue, and

intergovernmental grants (see table 1).

The category of “own-source” revenue encompasses all revenue sources whose

yield can be affected at the margin by local governments. This ability can stem from the

local governments’ discretion to determine taxable bases, determine rates, introduce the

tax, or any combination of these three. The category of “assigned sources” of revenue

refers to legislated long-term entitlements to (a share of) the yield from tax instruments

over which local officials have no discretion.9 The category of “regulated revenue” refers

to tax-revenue sharing determined by the higher-level government as part of its annual

budget process. The final category in our taxonomy, intergovernmental grants, compiles

all discretionary transfers to local governments whether formula-based or completely

discretionary.

9 These statements would need to be qualified if the local authorities can affect the behavior and diligence of

tax administration officials. The assumption here is that at the present time the dual subordination of tax officials in Russia does not present a problem any longer.


8

Table 1. Average Structure of Local Government Revenue

1996 1997 1998 1999 2000 2001Own-source revenue 8.38%

(0.67) 9.06% (0.62)

10.83% (0.63)

11.72% (0.66)

12.95% (0.70)

13.96% (0.84)

Local CIT surtax n.a. n.a. n.a. n.a. n.a. 6.86% (1.30)

Housing maintenance tax 3.69% (0.90)

3.75% (0.87)

4.76% (0.89)

6.13% (0.94)

7.88% (0.97)

2.01% (1.14)

Land tax and lease 1.41% (0.88)

1.95% (0.84)

2.73% (0.90)

3.18% (0.85)

2.91% (0.89)

2.56% (1.02)

Earmarked duties 0.97% (0.98)

0.96% (0.94)

0.41% (2.99)

0.29% (1.44)

0.19% (0.90)

0.16% (1.77)

Non-tax 0.71% (1.48)

0.63% (1.19)

0.64% (3.12)

1.20% (1.09)

1.61% (1.15)

1.66% (1.13)

Assigned revenue 7.30% (1.05)

8.31% (1.11)

8.27% (0.99)

9.91% (0.89)

10.25% (0.95)

8.50% (1.13)

Sales tax n.a. n.a. 0.13% (3.95)

2.74% (1.01)

3.20% (0.91)

2.68% (1.57)

Enterprise assets tax 5.86% (1.09)

6.20% (1.14)

6.51% (0.97)

5.16% (1.12)

4.31% (1.13)

3.56% (1.29)

Subsoil fees 0.99% (3.62)

1.72% (3.16)

1.28% (3.52)

1.67% (3.52)

2.43% (3.38)

1.95% (3.49)

Regulated revenue 28.57% (0.56)

26.87% (0.58)

28.08% (0.59)

29.16% (0.61)

28.16% (0.63)

22.79% (0.67)

VAT 5.28% (1.00)

5.09% (1.02)

5.69% (0.96)

4.78% (0.94)

3.31% (1.07)

—

PIT 13.02% (0.57)

13.29%(0.54)

13.55% (0.61)

12.50%(0.59)

13.73%(0.60)

15.39% (0.65)

CIT 8.13% (0.87)

5.73% (1.17)

5.32% (1.22)

7.79% (1.19)

7.19% (1.21)

2.64% (2.28)

Excises 0.55% (4.84)

0.98% (3.76)

1.11% (3.60)

1.65% (3.57)

1.15% (4.33)

1.37% (4.20)

Grants 55.76% (0.42)

55.71% (0.43)

52.83% (0.47)

49.22% (0.52)

48.64% (0.54)

54.75% (0.47)

Notes: - Coefficients of variation are provided in parentheses. Averaging over all localities in our sample. - n.a. Not applicable Source: Martinez-Vazquez, Timofeev, Boex (2006).

Table 1 shows the composition of local government revenue according to our

taxonomy. In addition, the table’s descriptive statistics note not only the mean share of

each revenue categories, but also the extent of variations in the composition of local

revenues. The relative importance of particular sources of marginal revenue in local

governments’ total revenue is a result of both the productivity of authorized tax


9

instruments and the political capacity to tap particular tax bases. Apparently, out of the

two dozen "local" own taxes (over which local governments had had some form of

discretion up to 2001), the most utilized are those on enterprise turnover, profits, and land

use—all of them essentially business taxes.

The table clearly shows the increasing importance of own sources and assigned

sources of revenue over time. The combined share of these two categories in local

governments’ total revenue increased from 16 percent in 1996 to 22 percent in 2001. The

share of own-source revenue alone increased from 8.4 percent in 1996 to 14 percent in

2001. Even so, the importance of own-source revenue for local governments in Russia is

still well below the levels observed in most OECD countries. In 2001, with the goal of

eliminating the most distortionary taxes, the federal government repealed the local

governments’ right to levy the Housing Maintenance Tax (with a rate of up to 1.5 percent

of enterprise turnover, and which accounted for 14 percent of local pre-transfer revenues

in 1999). As compensation, local governments gained the right to introduce a “piggy-

back” corporate income tax of up to a maximum of 5 percent.

Another development revealed in the table is a sharp drop in the share of

regulated tax revenue that was almost entirely offset with an increase in the share of

grants in 2001. At the same time, the mean share of regulated revenues in regional

budgets decreased from 20.7 percent in 1999 to 18.1 percent in 2001 and the share of

federal grants increased from 44.4 percent in 1999 to 50 percent in 2001. The mean ratio

of federal transfers to consolidated regional-local expenditures increased from 25 percent

in 1999 to 33 percent in 2001.


10

This reshuffling is a part of the fundamental reforms in the system of revenue

sharing aimed at making it more transparent and targeted. Thus, the federal government

introduced new earmarked grants in substitution for the former partial retention of VAT

revenues by regional and local governments at the point of collection.10 The reduction of

the sub-national share in VAT collections from 25 percent in 1998 to 15 percent in 1999–

2000 to zero in 2001 resulted in a significant loss to sub-national pre-transfer revenues, of

which about fifteen percent was accounted for by VAT in 1998. However, the centralized

share of VAT collections is transferred back to regions in the form of subventions

earmarked for the fulfillment of two major federal mandates.

The cumulative impact of the post-1998 tax reforms and the reassignment of

revenue sources on the vertical fiscal balance is reflected by the centralization of tax

collections to the federal budget. Thus, the federal government’s share of overall tax

collections increased from 51 percent in 1999 to 60 percent by 2001. Potentially, this

should have enabled the federal government to narrow fiscal disparities across localities.

Indeed, if the yield of the centralized taxes is distributed more unevenly than that for

taxes still retained by local government, we should observe a decrease in disparities both

within and between regions. In addition, centralization of tax revenue gives the federal

government more resources to perform inter-regional equalization and, if federal grants

are passed on by regional government to the local level in an equalizing manner, this

would also decrease within-region disparities.

10 Centralization of VAT collections was partially compensated by increasing the retention rate for PIT collections from 84

to 99 percent.


11

Figure 1. Lorenz curves for total local revenues in 1999 and 2001

However, according to Lorenz curves on figure 1 disparity in local fiscal

outcomes increased between 1999 and 2001. There can be several reasons for that

development. First, an economic recovery occurring since 1999 might have widened

fundamental economic disparities across localities. Indeed, according to Lorenz curves on

figure 2, an increase in disparity is more pronounced for own-source and assigned

revenue than for total revenues. However, national tax reforms appear to have contributed

to this process too. Thus, according to figure 3 the local turnover tax (Housing

maintenance tax) eliminated in 2001 had a more evenly distributed yield than that of the

CIT, which was provided to localities as a surtax in place of the turnover tax. Similarly

the yield from the VAT that was centralized in 2001 was more evenly distributed than

that of CIT.


12

Figure 2. Lorenz curves for own-source and assigned local revenues in 1999 and 2001

Figure 3. Lorenz curves for the yield of selected taxes in 1999


13

Therefore, the intergovernmental reforms of 2001 were not designed to reduce

fiscal disparities via tax revenue assignments. However, there was still potential to reduce

disparities by distributing the centralized tax resources via intergovernmental grants. In

the rest of this paper we attempt to explain why larger intergovernmental grants actually

did not lead to lesser disparities in fiscal outcomes. The main issue to uncover is whether

the national equalization policy was blocked at the regional level or whether federal

grants were not adequate to mitigate increased disparities arising from own-source and

assigned revenue to sub-national governments

2.1 The federal transfer system and regional-local fiscal transfers

An appropriate starting point for the discussion on local fiscal equalization in the

Russian Federation is the country’s federal-regional transfer system. Not only does this

system determine the level and scope of resources available for subsequent regional-local

distribution, but it also serves as an important model for formulating regional-local

transfer systems. In 1994, rates for sharing tax revenue between the federal government

and regional governments were in effect standardized across regions (while varying by

tax) and since that year have been annually set in the federal budget law virtually

unchanged.11 The implication was that all regions, be they rich or poor, had to be treated

the same. Therefore, the attainment of more efficient and transparent revenue

assignments brought about potentially larger fiscal disparities in the Russian Federation.

11 Whereas revenue-sharing arrangements had been relatively stable from 1994 through 1997, several substantial changes

were made in revenue-sharing arrangements before and after the economic crisis of August 1998 in the direction of recentralizing tax revenues. Changes in the assignment of revenues, especially periodic centralization of a share of PIT collections, reflected a tug of war over resources between the federal and regional governments. Fiscal policy reforms in recent years have aimed at reassigning tax revenues away from subnational governments and toward the federal government. Thus the reassignment of the VAT to the central government most significantly contributed to the centralization of public-sector resources since 2000. The federal government’s share of overall tax collections reached 65 percent in 2002 from a low of 42.5 percent in 1997.


14

The federal government had to rely on a new system of equalization grants to offset

potential increase in disparities in the distribution of fiscal resources.12 In addition, the

federal government maintained an ad hoc mechanism of “mutual settlements,” which

were mostly non explicitly budgeted transfers granted to some regions during the process

of budget execution.

Equalization grants (under the title of Federal Fund for the Financial Support of

the Regions) represent the most dominant form of federal-regional transfers. The size of

the federal equalization grant pool is set prior to the beginning of the fiscal year in the

annual federal budget law.13 Although the shares of individual regions in the equalization

fund are drafted based on the need/capacity formula, the legislated amounts may end up

considerably different as a result of political bargaining in the parliament, in particular

over specific weights to be used in the formula.14

At the regional-local level there has been much less progress. Although there have

been some reform efforts by a group of regions, most Russian regional governments

continue to employ budgetary practices inherited from the Soviet system, such as

“regulation” of taxes and simple “gap filling” transfers, as the main methodology for

closing horizontal fiscal imbalances among local governments. The “regulation”

approach consists of lowering revenue-sharing rates for one or more taxes in richer

12 Apparently, unification of sharing rates across regions was associated with reduced between-region variation in per-

capita pre-transfer revenues. At the same time, the variation in per-capita expenditures increased. We can attribute this evidence to greater bargaining power of wealthier regions with respect to differentiated sharing rates. When the sharing rates were standardized, this bargaining power appears to have been redirected toward the distribution of intergovernmental grants.

13 The size of the equalization pool was initially set at 15% of the federal government’s internal revenue. However, the fixed percentage was reduced to 13 percent by 1999 and eventually fixed as the previous year nominal amount annually adjusted for inflation.

14 The transfer formula undergoes some changes every year. Thus, for the 2005 transfers, the Index of Budget Expenditures was computed for the composite subnational expenditures rather than 15 separate expenditure categories as in previous years. However, the set of factors used in the formula is quite stable and traditionally has included variables accounting for differences in wage costs, consumer price level, utility costs, accessibility of the region’s localities by transport, and demographic composition of the population.


15

localities and increasing them in poorer localities. It was only in the late 1990s that

federal legislation attempted to bring some structure to the regional-local fiscal relations.

The federal law of 1997 required all regions to allocate a portion of shared tax revenue

using a stable formula and provided a (in some ways too specific) list of parameters to be

used. 15

This piece of federal legislation attempted also to secure the vertical fiscal

balance at the local level by requiring that for major taxes, regional governments should,

on average, meet certain minimum sharing rates with local governments at the point of

collection. This was considered overall a toothless provision because it included the

phrase “on average” and the vast majority of regions had been already in compliance

with it. The same law also required regions to set and secure a minimum expenditure

level for local governments.

2.2 The extent of fiscal equalization at the sub-national level

Actual per capita public expenditures in a locality are essentially a product of

three components: the scope of service responsibilities assigned to the local government

by its regional authorities, the quality (or level of provision) of public services selected

by the local authorities, and per capita costs of public services in that particular locality

(per unit of service quality). Thus, the local government expenditures are determined by

both, the factors under local government control (quality of services and efficiency of

provision) and those determined exogenously (the extent of assigned responsibilities and

15 See provisions of the federal Law on the Financial Foundations of Local Self-Government in the Russian Federation

(1997) concerning the “Fund for the Financial Support of Municipal Settlements”


16

the unit cost of service provision at an average efficiency level).16 Sound

intergovernmental schemes aim at equalizing fiscal disparities that are due to the factors

lying outside the local government control. To assess the effectiveness of Russia’s

equalization we analyze disparities in local governments’ per capita resources while

controlling for differences in the costs of service provision, the scope of assigned

responsibilities, and revenue-raising effort.

Because of the differences in input prices and socio-economic environment, local

governments differ in the amount of money they must spend to achieve a given quality of

public services. For example, in the Northern territories, inputs to public provision are

likely to be more expensive due to high transportation costs. In addition, the harsh

climate might require more inputs (e.g., heating fuel) to achieve the same outcome of

public provision. To control for differences in public service costs across localities, we

adjust nominal fiscal resources by the region-wide index of subsistence level.

Alternatively, we use the Index of Budget Expenditures used in the allocation of federal

grants in 2001.17 In addition to unit costs this index also captures demand for public

services that is due to exogenous factors such as the client-group size (e.g., school-age

children).

Secondly, we control for region-wide decisions about the division of service

responsibilities between the levels of government. Because a local government’s role in

the service provision affects how much the locality must spend to ensure a certain quality

of this service, a local government’s share in service responsibility is a key component of

assessing disparities in local government resources.

16 If the local government operates inefficiently, for example using the wrong combination of inputs of production such as

labor and capital, then the costs per unit of service will be higher than they need to be. 17 See footnote 14.


17

There is an argument that centralized provision is characterized by more

uniformity in the level and quality of public services across the national territory This

argument finds some support in empirical studies showing that centralized taxes and

expenditures achieve redistribution of resources from richer areas to poorer ones

(Prud'homme 1995). By applying this argument to Russia’s setup, one can hypothesize

that regional government spending uniformly eases the burden of local governments in

the provision of public services. Even if regional government service facilities are not

evenly distributed among localities, one can argue that, regardless of their residence,

people can use transport to access those facilities.

To allow for such impact, one of our measures adjusts local fiscal resources for

the extent of regional service co-responsibility by adding per capita regional spending to

per capita local spending. This adjustment would obviously narrow measured intra-

regional disparities in local fiscal outcomes. At the same time, anecdotal evidence

suggests that even in the most centralized countries we do not always find uniformity in

service provision so that the capital city often has higher levels of public infrastructure

and better access to public services. Therefore, we also include in our presentation

unadjusted local spending for a reference.

The adjustments for differences in the costs of service provision and the scope of

assigned responsibilities allow the local budget data to reveal disparities in the level of

public services in Russia’s localities. The next question we would like to explore is

whether these disparities result from the failure of the higher-level government to assign

adequate sources of revenue to the local level or the failure of local governments to fully

utilize those resources. To answer this question we will transform actual tax collections


18

into the revenue capacity by adjusting for the region-wide level of tax effort.18 Fiscal

capacity concerns the ability of the jurisdiction to raise revenue for public spending given

the level of economic activity within the boundaries of the jurisdiction and the devolved

authority to translate this economic activity into public revenues. We rely on the

information on regions’ tax capacity taken from the 2001 calculations of federal grants.

3. How Big are Fiscal Disparities at the Sub-national Level?

Because localities do not receive direct aid from the federal government, the only

vehicle for inter-municipal equalization is through tax revenue sharing and grants

allocated by regional authorities (who in turn receive grants from the federal

government). Therefore, we can attempt to assess the extent of equalization achieved

with federal-regional and regional-local grants and compare relative contribution of the

two instruments to narrowing the ultimate disparity in fiscal outcomes. In order to do

that, we examine the variation in per capita fiscal resources of local government resulting

from different stages of the equalization process.19

We use two indexes of inequality belonging to the class of general entropy

measures: square coefficient of variation (I2) and mean log deviation (I0), also known as

Theil's second measure. These measures have a number of properties instrumental for this

study, as explained further in the appendix. Firstly, these indexes allow us to quantify

relative contributions of between-region and within-region disparities to the nation-wide

inequality (Shorrocks, 1984). Secondly, we can derive quantitative measures for relative

contributions of different revenue sources to the ultimate inequality (Shorrocks, 1982). 18 We are indebted to Vladimir Popov for this point. 19 To capture persistent disparities we focus on local budget revenues. In our sample, local budget revenues and

expenditures differ on average by 3 percent due to sales of assets, borrowing, and other below-the-line receipts.


19

Thirdly, we can decompose relative contributions of each revenue sources into within-

region and between region components (Tsui, 1998). And finally, the two measures that

we chose out of the broad class of general entropy measures nicely complement each

other as I0 gives more weight to disparities in the lower tail and I2 gives more weight to

disparities in the upper tail of the distribution. In addition, the I2 measure is equivalent to

the coefficient of variation, which is commonly reported among descriptive statistics.

Table 2. Disparities in Per Capita Revenue of Local Governments, 2001

Own -source Plus assigned

Plus regulated

Plus grants

Grand mean, RUR 727 1226 2 423 4 923Grand median, RUR 337 529 1 269 3 389Grand maximum, RUR 32 968 56 633 120 566 129 081Grand minimum, RUR -137 0 4 17

Coefficient of Variation 1.69 1.93 1.69 1.32 Note: Negative amounts stem from refunds for past collections. Source: Authors’ calculation based on data from the Center for Fiscal Policy, Moscow.

Table 2 shows the differences in per capita fiscal resources (before and after

cumulatively adding assigned revenues, regulated revenues, and intergovernmental

transfers) among local governments of the 72 regions (1,963 localities in total) of the

Russian Federation for 2001.20 In that year, the per capita amount of own-source

revenues plus assigned revenue ranged from zero rubles in Sunzhensk Rayon (Ingush

Republic) to 56,633 rubles per person in Yak-Bodinsk Rayon (Udmurt Republic), with

the median amount of 529 rubles per person. The coefficient of variation in own-source

and assigned revenue stands at a rather high level of 1.93.21 This disparity is somewhat

20 The data set, collated by the Center for Fiscal Policy (Moscow), covers 72 out of a total 89 regions of Russia, with about

1,900 out of total 2,500 district level governments. In regions with two-tier local governments, data are aggregated at the top tier

21 For a somewhat smaller sample for which data are available, the coefficient of variation in the average wage among localities is only 0.79 compared to the coefficient of variation of 2.07 computed for own-source and assigned revenue in


20

lessened after the allocation of regulated tax revenues among localities, so that the

coefficient of variation decreases from 1.93 to 1.69. However, considerable equalization

is achieved only after the distribution of regional-local grants, after which the coefficient

of variation drops to 1.32.

Table 3. Disparities in Per Capita Revenue of Local Governments within and between

Regions, 2001

Own -source Plus assigned

Plus regulated

Plus grants

Plus regional

expenditures Unadjusted fiscal resources

Square Coefficient of Variation (I2) 2.8468 3.7067 2.8410 1.7410 1.0448 Within regions 1.7290 1.9526 1.3740 0.7872 0.4423 Between regions 1.1177 1.7541 1.4670 0.9538 0.6025 Mean log deviation (I0) 0.2433 0.2417 0.2034 0.1187 0.0982 Within regions 0.1502 0.1382 0.1153 0.0552 0.0319 Between regions 0.0931 0.1036 0.0882 0.0635 0.0662

Fiscal resources adjusted for the price level Square Coefficient of Variation (I2) 2.4258 2.7765 2.2029 1.3561 0.9282 Within regions 1.7504 1.7565 1.3701 0.8240 0.292 Between regions 0.6754 1.0200 0.8328 0.5321 0.6362 Mean log deviation (I0) 0.2250 0.2191 0.1805 0.092 0.0662 Within regions 0.1502 0.1382 0.1153 0.0552 0.0219 Between regions 0.0748 0.0809 0.0652 0.0368 0.0443

Fiscal resources adjusted for the tax effort and expenditure need Square Coefficient of Variation (I2) 1.8033 1.7942 1.5803 0.8929 0.4366 Within regions 1.4603 1.3104 1.1096 0.6450 0.2200 Between regions 0.3430 0.4838 0.4707 0.2479 0.2166 Mean log deviation (I0) 0.2096 0.2022 0.1746 0.0863 0.0509 Within regions 0.1502 0.1382 0.1153 0.0549 0.0215 Between regions 0.0594 0.0640 0.0594 0.0314 0.0293 Source: Authors’ calculation based on data from the Center for Fiscal Policy, Moscow.

the same sample. This suggest that tax instruments assigned as own revenue sources of local governments have more uneven distribution than personal income. This is not surprising given that in 2001 more than half of own-sources revenue is accounted for by the local surtax on enterprise profits.


21

The true extent of cross-jurisdictional equalization achieved with the allocation of

regulated tax revenue and grants is somewhat disguised in the descriptive statistics

presented in table 2—because we pooled localities from different regions. Decomposition

of the square coefficient of variation reveals that the between-region disparity in local

fiscal outcomes is only slightly smaller than it was for the initial between-region disparity

in own-source revenue (see the top panel in table 3). The other way to look at the cross-

regional outcome of equalization is to note that it was just enough to offset an increase in

between-region disparity arising from the allocation of assigned tax revenue. For both I0

and I2 measures most of the reduction in total variation should be attributed to narrowing

disparities within regions. This is evident from the reduced share of the within-region

inequality in the total inequality compared to the respective share before equalization.

Figure 4. Relative factor components for inequality in own-source and assigned local revenues


22

The same story holds if we adjust per capita resources of local governments for the

region-wide index of living costs (see the middle panel of table 3). Accounting for inter-regional

cost differences reduces total variation, which nevertheless undergoes similar transformation at

each stage of resource allocation. Because the cost index is region-wide, cost adjustments leave

the within-region variation unchanged so that it now accounts for almost two-thirds of total

variation in own-source and assigned revenue. Using the index of expenditure need and tax effort

reduces the between-region variation even more dramatically but still following the same pattern

at each stage of resource allocation (see the bottom panel of table 3).22

Examining relative contributions of different taxes to the ultimate inequality, shows that

within-region disparity is dominant for each of own and assigned sources of revenue except

subsoil fees (Figure 4). While subsoil fees account for about nine percent of own-source and

assigned revenue, they contribute more than a quarter of the measured inequality. The CIT surtax

contributes about one third of total inequality, which is proportional to its share in own-source

and assigned revenue.

22 Ideally we would like to adjust locality-level data for price differences and tax effort. However, this information is not available for

the local level. Some comfort is provided by the fact that the adjustment using region-level indexes, while showing less between-region inequality at each cumulative stage of revenue allocation, nevertheless does not change dramatically how the extent of disparity is transformed been stages.


23

Table 4. Coefficients of Variation of Local Revenues in 72 Regions, 2001

Oblast code

Subject of RF Own-source + Assigned + Regulated + Grants After assigned / After grants

1 Republic of Bashkortostan 0.62 0.49 0.46 0.21 2.282 Republic of Buriatia 0.82 1.01 0.83 0.38 2.663 Republic of Dagestan 1.59 1.50 1.28 0.34 4.384 Kabarda-Balkar Republic 0.45 0.43 1.29 0.61 0.715 Republic of Kalmykia 2.13 2.58 2.62 1.54 1.686 Republic of Karelia 1.41 1.37 0.90 0.64 2.137 Republic of Komi 0.70 0.96 0.49 0.28 3.438 Republic of Mari El 0.60 0.53 0.68 0.46 1.169 Republic of Mordovia 0.48 0.51 0.42 0.22 2.36

10 Republic of North Osetia 0.57 0.50 0.92 0.42 1.1811 Republic of Tatarstan 1.52 1.35 1.03 0.69 1.9412 Republic of Tyva 0.97 1.08 1.40 0.64 1.6813 Udmurt Republic 5.03 4.28 4.52 3.67 1.1714 Ingush Republic 1.40 1.74 1.76 1.46 1.1915 Chuvash Republic 0.65 0.58 0.60 0.15 3.9018 Krasnodar Krai 0.54 0.56 0.50 0.38 1.4719 Krasnoyarsk Krai 1.49 1.62 1.27 0.79 2.0420 Primorski Krai 0.69 0.58 0.49 0.16 3.6322 Khabarovsk Krai 0.86 1.00 0.66 0.61 1.6323 Amur Oblast 1.32 1.12 0.56 0.45 2.5124 Arkhangelsk Oblast 0.47 0.45 0.34 0.15 2.9025 Astrakhan Oblast 1.21 1.33 0.54 0.27 4.8827 Bryansk Oblast 0.64 0.58 0.57 0.30 1.9728 Vladimir Oblast 0.45 0.43 0.42 0.22 1.9029 Volgograd Oblast 0.46 0.55 0.45 0.15 3.6830 Vologda Oblast 0.45 0.45 0.40 0.14 3.1931 Voronezh Oblast 0.51 0.50 0.48 0.22 2.3132 Nizhny Novgorod Oblast 0.62 0.54 0.45 0.27 2.0034 Irkutsk Oblast 0.49 0.50 0.36 0.39 1.3035 Kaliningrad Oblast 0.62 0.81 0.66 0.42 1.9336 Tver Oblast 0.63 0.58 0.44 0.63 0.9237 Kaluga Oblast 0.51 0.48 0.45 0.26 1.8138 Kamchatka Oblast 0.36 0.30 0.23 0.44 0.6939 Kemerovo Oblast 4.59 4.00 3.95 3.03 1.3240 Kirov Oblast 0.63 0.54 0.35 0.14 3.7741 Kostroma Oblast 3.15 2.89 2.82 2.11 1.3742 Samara Oblast 0.63 0.48 0.67 0.60 0.8043 Kurgan Oblast 0.41 0.53 0.60 0.12 4.5945 Leningrad Oblast 0.78 0.80 0.57 0.33 2.4046 Lipetsk Oblast 0.45 0.42 0.40 0.19 2.2547 Magadan Oblast 0.22 0.45 0.28 0.22 2.0648 Moscow Oblast 0.73 0.64 0.41 0.29 2.2249 Murmansk Oblast 0.35 0.34 0.19 0.26 1.3050 Novgorod Oblast 1.12 1.21 1.26 1.08 1.1151 Novosibirsk Oblast 0.45 0.46 0.42 0.15 3.0252 Omsk Oblast 0.57 0.55 0.54 0.21 2.68


24

Table 4. Coefficients of Variation of Local Revenues in 72 Regions, 2001 (continued)

Oblast code

Subject of RF Own-source + Assigned + Regulated + Grants After assigned / After grants

53 Orenburg Oblast 0.54 0.68 0.47 0.36 1.8954 Oryol Oblast 0.72 0.61 0.41 0.15 4.1255 Penza Oblast 0.65 0.63 0.53 0.14 4.3756 Perm Oblast 0.70 0.70 0.39 0.20 3.4957 Pskov Oblast 0.50 0.43 0.42 0.16 2.6358 Rostov Oblast 0.57 0.52 0.47 0.18 2.8559 Ryazan Oblast 0.90 0.83 0.48 0.19 4.3560 Saratov Oblast 0.58 0.50 0.45 0.25 1.9961 Sakhalin Oblast 0.91 0.74 0.51 0.36 2.0262 Sverdlovsk Oblast 0.52 0.48 0.36 0.23 2.0763 Smolensk Oblast 0.54 0.66 0.60 0.41 1.6264 Tambov Oblast 0.46 0.50 0.52 0.18 2.8865 Tomsk Oblast 1.12 1.45 0.77 0.39 3.6966 Tula Oblast 0.56 0.55 0.52 0.20 2.7267 Tumen Oblast 0.63 0.62 0.53 0.26 2.4068 Ulianovsk Oblast 0.53 0.48 0.43 0.21 2.3170 Chita Oblast 1.65 1.07 0.95 0.45 2.3872 City of St. Petersburg 0.64 0.54 0.52 0.20 2.7176 Republic of Adygeya 0.70 0.59 0.38 0.16 3.6778 Jewish AO 0.93 0.86 0.78 0.78 1.0979 Karachayevo-Circassian Republic 0.84 1.10 0.85 0.40 2.7880 Republic of Khakasia 0.73 0.67 0.65 0.18 3.6882 Komi-Perm AO 0.42 0.86 0.76 0.40 2.1283 Koryak AO 0.43 0.49 0.45 0.39 1.2487 Khanty-Mansi AO 0.55 1.01 0.66 0.42 2.4289 Evenk AO 0.59 0.90 0.51 0.38 2.3590 Yamal-Nenets AO 0.54 0.68 0.47 0.36 1.89

Mean 0.88 0.88 0.77 0.48 2.38 Median 0.63 0.59 0.52 0.29 2.26 Minimum 0.22 0.30 0.19 0.12 0.69 Maximum 5.03 4.28 4.52 3.67 4.88

AO. Autonomous oblast. Source: Authors’ calculation based on data from the Center for Fiscal Policy, Moscow.


25

Table 4 reports coefficients of variation for per capita local resource allocations

within individual regions for cumulative steps in the local resource allocation process.

Similar to the trend revealed by the pooled (inter-regional) set, consideration of the mean

and median values in table 4 suggests that consecutive allocation of regulated revenue

and grants results in a more equal distribution of local resources within each region. As a

result of equalization, in a median region per capita fiscal resources of local governments

differ by no more than 30 percent, while for the pooled sample this variation is 132

percent.

At the same time, between regions, the average per capita fiscal resources of local

governments vary by almost 100 percent of the mean—only slightly less than the

disparity in own-source revenue (see the top panel of table 3).23 If we account for inter-

regional cost differences, the average per capita fiscal resources of local governments

vary between regions by about 70 percent of the mean compared to 82 percent variation

for the own-source revenue (50 and 60 percent respectively if adjusted for tax effort and

expenditure need). Thus, the impact of equalization on the between-region variation is

somewhat more pronounced if we account for differences in costs and tax capacity. But

even after accounting for these differences, cross-regional disparity in fiscal outcomes is

still high especially when compared to the extent of within-region variation of fiscal

outcomes. Significant reduction of within-region disparities resulting from the allocation

of regional-local grants is possible due to the dominance of the latter revenue source in

total local revenues (over fifty percent of the total compared to only a quarter of

consolidated regional-local revenues accounted for by federal grants).

23 To allow decomposition, table 3 reports square coefficients of variation while the discussion of disparities refers to the

square root of the reported values.


26

It is also interesting to track how the extent of disparity is transformed after each

cumulative stage of revenue allocation by observing the extent of inequality for own-

source revenues, and then noting changes in inequality measures after consecutively

adding assigned tax revenues, regulated revenue, and intergovernmental grants. Assigned

tax revenues (from corporate assets, retail sales, and natural resources) considerably

increase the disparity between regions (table 3). This effect is quite intuitive given the

uneven distribution of natural resources (and to a lesser extent, corporate assets) across

regions (Figure 4).

However, at the lower tail of distribution within-region disparities are slightly

lessened by the allocation of assigned revenue, as revealed by the I0 measure in table 3.

Moreover, the reduction in within-region disparities drives a decrease in the I0 measure

for the total variation in the lower tail. This equalizing effect (albeit only very small) of

the assigned revenue on the within-region disparities is hard to explain, as the

aforementioned taxable bases are not evenly distributed among localities of one region

(recall Figure 4). One possible explanation is that poorer localities might have more

corporate assets than corporate profits, which dominated the own-source revenue in 2001.

Another possibility is that some unaccounted regulation might be going on through the

allocation of the corporate assets tax in excess of the mandatory 50 percent retention

rate.24

The cumulative allocation of regulated tax revenues to local governments

(including the personal income tax, corporate income tax, and excises) lessens the local

fiscal disparities both between and within regions (table 3). The between-region

24 A review of 1999 budget laws for 72 regions of Russia shows that 21 of these regions set differentiated

rates for sharing with constituent localities revenues from the corporate assets tax.


27

equalization through tax regulation is rather unexpected. The consolidated regional-local

share of regulated taxes should essentially act as an assigned-revenue source because of

the uniform remittance rates to the federal level used after 1994. Thus, there are no clear

reasons why regulated taxes should bring about any equalization between regions. One

possibility is that tax bases for regulated taxes are more evenly distributed between

regions than that for assigned taxes, especially those on natural resources. Another

explanation is that regional governments in poorer regions allow local governments to

retain a larger share of the regional-local tax collections since these are negligible

compared to the amount of federal grants received by such regions (Timofeev 2003).

The allocation of grants has a much stronger equalizing effect than that of the

regulated tax revenue. At the lower tail, a weaker within-region equalization effect of

regulated tax revenue might be because for poorer localities retention rates are already set

at 100 percent and thus cannot be increased any further. Weaker within-region

equalization in the upper tail might be explained by the fact that in many regions this

source of revenue might, in effect, be treated more like an assigned tax than an

intergovernmental transfer or grant. 25 One reason why regions might be treating

regulated taxes more or less like assigned taxes is that federal legislation requires regions

to meet some minimum sharing rates for this type of revenue with local governments,

even if only on average. 26 In addition, while nominal amounts of grants become eroded

with annual inflation, the last year's retention rates present a more stable benchmark for a

local government bargaining with regional authorities. The share of grants in total 25 Bahl and Wallace (forthcoming) show that income-elasticity of regulated revenue is only slightly less than

that of own-source and assigned revenue in the Leningrad region of Russia (1.30 vs. 1.38). They conclude that “variable sharing rates do not significantly reduce disparities in fiscal capacity within the region.”

26 The lack of data on total collections of regulated tax revenue in a local territory precludes us from exploring some obvious interesting questions such as what the extent of variation in sharing rates is for major taxes, or what determines these sharing rates, and whether these sharing rates are negatively related to the size of the tax base or tax effort.


28

revenue of local government being twice the share of regulated taxes might also explain

why most of the within-region equalization occurs only through the allocation of grants

(Table 1).

We can consider one additional stage of resource allocation, which is constituted

by direct expenditures undertaken by regional governments. Assuming that expenditures

from the regional budget uniformly benefit residents of all constituent localities, we can

add per capita amount of regional expenditures to the per capita amount of local

expenditures.27 In particular, it will account for different extent of local governments’

involvement in the provision of public services in different regions.

Inclusion of the regional expenditures into the consideration of local fiscal

outcomes considerably decreases within-region disparities as shown in the last column of

Table 3. However, this equalizing effect hinges on our assumption of uniform distribution

of benefits from regional expenditures. At the same time, between-region disparities in

price-adjusted resources increase with the inclusion of regional expenditures. This

suggests that between-region disparity in local fiscal resources is mostly determined by

the total tax yield in the region rather than the local share of that yield proportional to the

scope of expenditure responsibilities devolved to the local level. Note however a decrease

in between-region disparity in resources adjusted for tax effort, which suggests higher

regional expenditures in regions with lower tax capacity and higher expenditure needs.

After all these regions receive more federal grants, which tend to “stick” at the regional

level.

27 Accounting for regional government spending can be crucial for the assessment of between-region disparities. Thus,

Tsui (2005) shows that in China between-province disparity in per-capita local expenditures actually increases after the allocation of intergovernmental grants in 1994-2000. At the same time, Bahl and Wallace (2003) find that for Russia inter-provincial distribution of per capita provincial-local expenditures was more equitable than distribution of per capita provincial-local revenue in 1995. Similarly, Hofman and Guerra (2004) also find that the allocation of grants reduces disparity in provincial-local per capita expenditures in China.


29

The preceding analysis of descriptive statistics for within-region equalization still

does not show us the entire picture. In fact, comparing the inequality measures before

and after “equalization” (effected through regulated tax revenues and grants) for

individual regions shows a dramatic decrease in local fiscal disparities in some regions

while almost no equalization is achieved in others (Table 4). We can measure

equalization effort of regional governments with the ratio of inequality measures before

and after equalization correspondingly. Thus, in all but five regions the I2 equalization

effort is higher than that for the between-region equalization (in all but 15 regions for the

I0 equalization effort).

Correlation analysis suggests that larger inequality in own-source and assigned

revenue does not lead to a larger extent of equalization undertaken by regional

authorities. In fact, equalization effort is negatively related to the initial extent of intra-

regional disparity even if only weakly (the correlation coefficient between the initial

inequality and the equalization effort is –0.17 for the I2 measure and –0.14 for the I0

measure).

4. Why does Equalization Effort Differ Across Regions?

In this section, we attempt to explain differences in equalization efforts of

regional governments with a set of regional characteristics capturing political, economic,

demographic, and geographic differences. As we argued above, regional governments

determine the ultimate fiscal outcomes in individual localities through the allocation of

shared tax revenue and grants. We are interested, therefore, in finding out what set of

regional characteristics can potentially explain the differences in decisions made by


30

regional governments about allocation of grants and regulated revenue that impact local

governments’ per capita resources.

First, initial inequality in own-source and assigned revenue should potentially

determine both the need for and political feasibility of an equalization policy. A second

variable that a priori may be expected to have an impact on disparities in local

government revenues is the relative level of devolution of expenditure responsibilities.

When local governments are assigned a de facto larger share of responsibility for sub-

national public services, other things being equal, local fiscal outcomes have a greater

impact on the well-being of the local population, which in turn affects the regional

government. Thus, we use the local government's share in the regional-local expenditures

to account for varying concern of regional governments with local fiscal outcomes. Third,

we use the local government's share in the regional-local tax collections to account for

regional government leaning toward financing localities through tax revenue retention as

apposed to using cash grants.

Fourth, an important potential determinant of the extent of intra-regional

equalization is the relative dependence of the regional government on federal transfers.

As local governments account for a considerable portion of the regional-local

expenditures, a share of federal grants is likely to be passed on to the local level and thus

dilute disparities in own-source revenue of local governments. Empirical studies (e.g.,

Plekhanov and Freinkman, 2005, p.34) find that federal transfers “tend to be much less

shared with municipal governments than more conventional regional revenues.”

Moreover, there are indications that regional governments in poorer regions allow local

governments to retain a larger share of the regional-local tax collections since these are


31

negligible compared to the amount of federal grants received by such regions (Timofeev

2003).This fly-paper effect can potentially lead to a situation when regional-local

transfers fail to off-set disparities arising from larger retention of tax revenue at the point

of collection. However, empirical evidence suggests that the local tax yield in poor

regions is so small that even a smaller share of federal transfers passed on to localities

accounts for a larger share of local budgets in those regions and thus effectively dilutes

disparities in own-source revenue of local governments (Martinez-Vazquez et al, 2006, p.

103) .

Fifth, the relative availability of natural resources should also affect inequalities in

local government revenue and resources available for redistribution by the regional

government. We employ the volumes of oil and gas extraction to capture the importance

of the subsoil fee as an assigned source of revenue. Before the 2001 amendments to the

Law on Payment for Natural Resources, 30 percent of the subsoil fee revenues were

allocated to the municipality where extraction took place. 28

Sixth, we employ the gross regional product (GRP) to approximate the elasticity

of local tax instruments with respect to the level of economic activity. In addition, GRP

should capture the income elasticity of regional redistribution policies. Seventh, we use

poverty headcount to capture the importance for fiscal equalization for poverty

alleviation.

We also make an attempt to take into account the role of politics at the regional

level. Identifying political trends and issues within the regions is difficult because of the

lack of available data. During all transition years, there has been a big divide between

28 Federal Law No. 2395-1 of Feb. 21, 1992, Article 42, decrees revenues from all hydrocarbon resources (oil,

gas, and coal) to be split among federal, regional, and local governments, with these levels receiving 40, 30, and 30 percent respectively.


32

reformist regions and traditional or conservative regions in Russia. To sort regional

governments into those two camps we employ the number of votes cast for the

Communist Party candidate in the 1996 presidential elections. We expect that the higher

the percentage for this variable the more conservative the regional governments would

be, because the candidate's supporters would lean toward the old-style equalizing system

of governance. Two alternatives proxies for reform-mindedness of regions are the extent

on privatization in the 1990s and costs-recovery in housing and utilities relative to

household income. Another variable reflecting political landscape is educational

attainments of the population. Finally, we employ the urbanization rate of regions to

capture urban-rural cleavages in regional politics.

Table 5 reveals strong correlation among some of the proposed explanatory

variables. Therefore, when these variables jointly enter the regression equation, our

estimates can suffer from large standard errors and low significance levels due to multi-

colinearity. In general, the significance level is inversely related to one minus the

proportion of the independent variable variation explained by all other regressors (Greene

2000, p. 256). For example, the local share of expenditures explains only 29 percent (that

is 0.542) of variation in tax revenue retention at the local level and therefore both

variables have significant effects in our regression. By contrast, oil production explains

88 percent (that is 0.942) of variation in gross regional product. In this latter case, while

including in the regression equation only those among the related variables that have the

highest statistical significance, we should remember that these estimates also capture the


33

Table 5. Pair-wise correlation coefficients for selected variables

Local spending

Local tax retention

Grant dependence

Oil Gas Real GRP

Poverty Urban Privatization

Local spending 1.000 Local tax retention 0.540 1.000 Grant dependence -0.311 0.211 1.000 Oil 0.123 -0.058 -0.278 1.000 Gas -0.059 -0.082 -0.180 0.459 1.000 Real GRP 0.130 -0.123 -0.447 0.937 0.646 1.000 Poverty -0.406 -0.039 0.736 -0.373 -0.266 -0.524 1.000 Urbanization 0.465 0.167 -0.652 0.261 0.140 0.366 -0.710 1.000 Privatization 0.543 0.208 -0.498 0.110 -0.022 0.193 -0.570 0.515 1.000


34

effect of other related variables.29 Essentially such a group of factors has a confounding

effect that we cannot apportion among individual variables. This is not however a

concern for this study aiming at identifying a group of factors explaining the extent of

intra-regional equalization rather than pinpointing impacts of individual factors.

Table 6. OLS regression of equalization effort using the I0 measure

Specification (1) (2) (3) (4)

Inequality in own-source-revenue, 2001

-7.005** (3.495)

-7.026** (3.474)

-7.921** (3.209)

-7.027** (3.393)

Local share of regional-local expenditures, 2001

13.760** (6.739)

12.386* (6.635)

18.232** (7.610)

10.901* (6.279)

Local share of regional-local tax revenue, 2001

-16.770** (6.768)

-16.106** (6.719)

-14.441** (6.585)

-10.500 (6.668)

Dependence on federal grants, 2001

6.303* (3.560)

6.064* (3.644)

3.869 (3.858)

Oil extraction, tones per capita, 2001

-0.023* (0.013)

-0.045*** (0.010)

-0.024* (0.012)

Gas extraction, mil. m3 per capita, 2001

0.006*** (0.002)

0.003*** (0.001)

0.006*** (0.001)

GRP, mill. RUR per capita, 2001

0.001 (0.113)

Urbanization -0.073 (0.045)

Population with higher education, 1994 share

-61.210** (26.331)

-61.087** (26.225)

-63.432** (26.094)

-67.616*** (23.815)

1996 votes for the Communist candidate

0.128** (0.057)

0.126** (0.058)

0.122** (0.055)

Privatization sales since 1993

-16.881** (8.382)

Constant 9.170* (4.858)

9.757* (5.018)

15.532*** (4.812)

15.484*** (4.261)

Sample size 71 71 71 71

R2 0.25 0.23 0.23 0.24

Notes: Dependent variable is equalization effort measured as the ratio of inequality levels before and after equalization correspondingly. Robust standard errors are provided in parentheses.

29 We have also considered other explanatory variables including tiers of local government, land area, population density,

regional population, average district population, ethnic fragmentation. However, they were not found statistically significant in any of our econometric specifications.


35

Table 7. OLS regression of equalization effort using the I2 measure

Specification (1) (2) (3) (4)

Inequality in own-source-revenue, 2001

-0.313*** (0.090)

-0.311*** (0.087)

-0.290*** (0.063)

-0.318*** (0.073)

Local share of regional-local expenditures, 2001

14.703** (5.761)

13.628** (5.688)

18.326** (6.720)

13.431** (6.174)

Local share of regional-local tax revenue, 2001

-13.935*** (5.128)

-13.464** (5.122)

-11.991** (4.923)

-8.967 (5.470)

Dependence on federal grants, 2001

5.296* (2.699)

5.296* (2.792)

3.345 (2.790)

Oil extraction, tones per capita, 2001

-0.022* (0.012)

-0.038*** (0.010)

-0.022* (0.011)

Gas extraction, mil. m3 per capita, 2001

0.005*** (0.001)

0.003*** (0.001)

0.005*** (0.001)

GRP, mill. RUR per capita, 2001

0.002 (0.012)

Urbanization -0.078* (0.040)

Population with higher education, 1994 share

-47.262** (23.145)

-46.713** (23.210)

-46.415* (23.308)

-50.523** (20.291))

1996 votes for the Communist candidate

0.099** (0.046)

0.100** (0.048)

0.093** (0.046)

Privatization sales since 1993

-13.575** (6.687)

Constant 5.414 (4.456)

5.560 (4.658)

9.841 (4.740)

11.064*** (3.931)

Sample size 71 71 71 71

R2 0.24 0.23 0.23 0.24

Notes: Dependent variable is equalization effort measured as the ratio of inequality levels before and after equalization correspondingly. Robust standard errors are provided in parentheses.


36

We employ these variables in an attempt to explain the regional governments’

equalization effort in 2001 for a sample of 71 regions for which data are available.30 The

equalization policy for each region is modeled with OLS regressions, seeking to identify

the determinants of equalization efforts undertaken by different regions. Tables 6 and 7

report OLS estimates for the impact of the potential determinants on I0 and I2 measures of

inequality respectively. The sign and statistical significance of the estimates are roughly

the same for both measures of inequality (i.e. I0 and I2). The only exception is the

negative effect of the urbanization rate, which is more significant for equalization in the

upper tail (I2-inequality).

The empirical results suggest that regions with higher initial disparities in local

government’s own and assigned revenue tend to undertake less equalization.31 Also

regions where local governments play a larger role in the provision of public services

exert more equalization effort. At the same time, regions allowing localities on average to

retain more tax revenue at the point of collection, achieve less equalization. Although

dependence on federal transfers is associated with more equalization, the effect might be

confounded with that of poverty and urbanization. Regions producing more oil undertake

more equalization while regions producing more natural gas undertake less equalization.

The opposite effects of oil and gas on equalization probably cause the insignificant effect

of gross regional product on equalization.

As expected, reform-minded regions undertake less equalization as revealed by

the coefficients on the pro-communist vote and extent of privatization. Similarly, regions

30 We had to drop Astrakhan Oblast from our initial sample of 72 regions due to the lack of data on educational attainment. 31 This is parallel to inter-personal redistribution, where at least one study finds that US states with heavier concentration

of poverty decrease their budget choices in favor of distribution on either revenue or expenditure side ( Bahl et al 2002).


37

with more educated electorate undertake less equalization. Regions with stronger

dominance of urban population exhibit less equalization in the upper tail.

In summary, it would appear that regional differences in the extent of equalization

actually pursued can be partially explained by a combination of structural, economic and

political factors. With adjusted R-squares in tables 6 and 7 under 0.25 it is obvious that

there are many other factors behind the different regional practices that have not been

captured in our regression estimates.

5. Conclusion

Our empirical analysis shows wide variations in per capita revenue of Russia’s

local governments from their own sources. Most of this variation is driven by disparities

within regions rather than between them. The intergovernmental reforms of 2001 further

widened these disparities and our results attribute this to the federal policies rather than

interference from regional authorities. The federal decisions resulted in heavier

dependence of local budgets on revenue from the corporate income tax, which has

historically been a predominantly subnational tax in Russia against any normative

wisdom. The allocation of shared tax revenues to local budgets and even more so the

distribution of regional-local grants considerably decreases total variation in fiscal

outcomes. However most of equalization occurs within regions while disparities between

regions decrease only slightly. Thus, the centralization of tax revenue in the federal

budget did not lead to more effective equalization by means of federal grants.

The results in this paper generally reject the conjecture that national equalization

efforts— in particular the 2001 reform of tax revenue sharing— might be blocked at the


38

regional level. On average, regional authorities face larger disparities in own-source

revenues of constituent localities than respective disparities across regions. However,

after the allocation of regional-local grants, within-region inequality in nominal fiscal

resources becomes smaller than the corresponding between-region inequality. In fact, the

vast majority of regions achieve a larger reduction of within-region disparities than is the

case for between-region disparities. Thus, if anything, the national equalization target is

overshot by regional authorities. However, it is possible that regional authorities upset

national government goals other than equalization, such as creating incentives for local

economic development.32 If there is a trade-off between equity and growth, then regional

equalization policies might weaken incentives for local development sought by the

national government. This trade-off is indirectly revealed in our results as less

equalization achieved in regions allowing localities to retain a larger share of tax revenue

at the point of collection. Tax revenue retention has been identified as an important fiscal

incentive for local governments to promote development of the private sector in

transitional economies (see Desai et al., 2005, and Jin et al., 2005).

We find that a dramatic decrease in local fiscal disparities occurred in some

regions while almost no equalization achieved in others. Regions undertaking a larger

equalization effort are those where local governments are responsible for a larger share of

public services. Political factors seem to be an important determinant of the equalization

effort; in particular, pro-communist regions appear to undertake higher equalization

efforts. Finally, less equalization is undertaken in regions with higher initial disparities in

local government’s own and assigned revenue.

32 See, for example, Qiao et al. (2003) for analysis of this possibility in the case of China.


39

Our results generally support those obtained by Bahl and Wallace (2003), the only

other study on this topic that we know, who measured the extent of equalization by the

amount that the income elasticity of own-source revenue exceeds the income elasticity of

fiscal outcomes. They find that central governments in China, Russia, and USA achieve

weaker equalization than 1-2 provinces selected for studying in each of the three

countries.33 Because our data covers most of Russia’s regions, we show that this result on

average holds for all regions and not just for the one selected in Bahl and Wallace’s

study.34

Although the actual degree of equalization undertaken by regional governments

vis-à-vis the central government may differ in other countries from those we obtain for

Russia, the policy implication of our findings is that overall decentralization design into a

hierarchical system does not have to underperform in terms of local government

equalization in comparison to other intergovernmental arrangement, such as bi-furcated

systems.

33 For China, further evidence is provided by Tsui (2005) showing that, while allocation of intergovernmental transfers

increases disparity in local fiscal outcomes in China, this counter-equalizing effect is larger for between-provincial disparity than for within-province disparity.

34 In addition, our results only partially support another hypothesis by Bahl and Wallace that more equalization in wealthier provinces is driven by larger disparities within wealthier regions rather than by the level of income. While we show that the level of income has no significant impact on equalization effort by regional authorities, we also find that larger disparities in own-source revenue lead to less equalization undertaken.


40

Appendix

Decomposing the aggregate inequality by factor components and subgroups

In mathematical terms, the total variation computed for the entire sample

comprises the sum of the squared deviations of a locality level variable yi from the grand

mean y of this variable:

( )2∑∑ −i j

ijij yyn .

This total sum can be rearranged into two components:

( ) ( )22 ∑∑∑ −+−i j

iijiji

ii yynyyn

Here, ni and nij denote population of region i and locality j in that region

respectively, iy stands for the region i’s mean value of variable yij. Further, if we divide

this expression by ( Σni –1), we get the total variance, which is again a sum of two

components. The first component, called “between-group variance,” characterizes the

average dispersion of regional means from the grand mean. The second, component,

called “within-group variance,” characterizes the average dispersion of data points from

the corresponding regional means. Note that the “between-group variance” weights

regions in proportion to their population, unlike the variance computed over the sample

of regional means equally weighing all regions. To make this measure mean independent,

we can divide the variance by the square mean 2y and thus obtain the square coefficient

of variation I2= I2b+ I2

w, which is a sum of its between-region and within region

components.

In a similar manner we can decompose mean log deviation I0 =I0b+ I0

w:


41

( ) ( ) ( )∑∑∑∑∑ −+−=−i j

ijiiji

iii j

ijij yynyynyyn lnlnlnlnlnln .

In addition, all general entropy measures, including I0 and I2, are also amenable to

decomposition by various components of our variable of interest yij. Thus, if yij=Σykij,

than we can express ∑=k

ckc IsI ,

where ( )( )y

yysk

k var,cov

= . Moreover, the relative contribution of each component of

y can be further broken down into a between-region subcomponent and a within-region

subcomponent: sk =skb+ sk

w.


42

References

Alexeev, Michael and Galina Kurlyandskaya. (2003). ‘Fiscal Federalism and Incentives in a Russian Region,’ Journal of Comparative Economics 31: 20–33.

Bahl, Roy and Sally Wallace. 2003. “Fiscal Decentralization, the Provincial-Local Dimension.” Public Finance in Developing and Transitional Countries: A Conference in Honor of Richard Bird, James Alm and Jorge Martinez (ed.), Edward Elgar.

Bahl, Roy and Sally Wallace. forthcoming. "Fiscal Decentralization and Fiscal Equalization within Regions: The Case of Russia,” Journal of Public Budgeting, Accounting and Financial Management.

Bahl, Roy, Jorge Martinez-Vazquez and Sally Wallace (2002) State and Local Government Choices in Fiscal Redistribution, National Tax Journal, 55, 4, 723.

Bahl, Roy, and the Intergovernmental Fiscal Relations Team. 1999. “Intergovernmental Fiscal Relations in Leningrad Region.” International Studies Working Paper #99-2, Andrew Young School of Policy Studies, Georgia State University, Atlanta.

Desai, Raj M., Lev Freinkman and Itzhak Goldberg (2005) Fiscal federalism in rentier regions: Evidence from Russia, Journal of Comparative Economics, 33, 4, 814.

Gauri, Varun (2004) Social Rights and Economics: Claims to Health Care and Education in Developing Countries, World Development, 32, 3, 465.

Greene, William H. 2000. Econometric Analysis. Fourth Edition. Prentice Hall: New Jersey. Hofman, Bert and Susana Cordeira Guerra. (2004). “Ensuring Inter-regional Equity and Poverty

Reduction,” International Studies Working Paper 04-11, Andrew Young School of Policy Studies, Georgia State University, Atlanta.

Jin, Hehui, Yingyi Qian and Barry R. Weingast (2005) Regional decentralization and fiscal incentives: Federalism, Chinese style, Journal of Public Economics, 89, 9-10, 1719.

Le Grand, Julian (1975) Fiscal Equity And Central Government Grants To Local Authorities., Economic Journal, 85, 339: 531.

Martinez-Vazquez, Jorge, Andrey Timofeev and Jameson L. F. Boex. (2006)Reforming Regional-Local Finance in Russia. WBI Learning Resources Series, Washington, D.C.: The World Bank.

Qiao, Baoyun, Jorge Martinez-Vazquez, and Yongsheng Xu.Qiao. 2002 "Growth and Equity Tradeoff in Decentralization Policy: China's Experience." with International Studies Program Working Paper 02-16. Atlanta, Georgia: Andrew Young School of Policy Studies.

Plekhanov, Alexander and Lev Freinkman (2005) What determines the extent of fiscal decentralization? The Russian paradox. The World Bank Policy Research Working Paper Series: 3710. Washington, D.C.: The World Bank.

Prud'homme, R. (1995): "The Dangers of Decentralization," World Bank Research Observer, 10, 201-20. Shorrocks, Anthony F. (1982) Inequality Decomposition by Factor Components, Econometrica, 50, 6, 193-

211. Shorrocks, Anthony F. (1984) Inequality Decomposition by Population Subgroups, Econometrica, 52, 6,

1369-1385. Timofeev, Andrey. 2003 “Determinants of Decentralization within Russian Regions,” LGI Discussion

Paper No. 23. Budapest: LGI. Treisman, Daniel. 1996. “The Politics of Intergovernmental Transfers in Post-Soviet Russia.” British

Journal of Political Science 26(2): 299–335. Tsui, Kai-yuen (1998) Factor Decomposition of Chinese Rural Income Inequality: New Methodology,

Empirical Findings, and Policy Implications, Journal of Comparative Economics, 26, 3, 502. Tsui, Kai-yuen (2005) Local Tax System, Intergovernmental Transfers and China's Local Fiscal

Disparities, Journal of Comparative Economics, 33, 1, 173. Stewart, Kitty. 1997. "Are Intergovernmental Transfers in Russia Equalizing?" EUI Working Papers, ECO

No. 97/22. European University Institute, Florence, Italy. Zhuravskaya, E. V. (2000). ‘Incentives to Provide Local Public Goods: Fiscal Federalism, Russian Style,’

Journal of Public Economics 76, 337–68.

Documents

Regional – Local Dimension of Russia’s Fiscal … · Regional-Local Dimension of Russia’s Fiscal Equalization 5 may also be driven by political goals unrelated to economic objectives