Intra-regional equalization and growth in Russia.
Martinez-Vazquez, Jorge ; Timofeev, Andrey
INTRODUCTION
Until 2009, the Russian economy had been enjoying above 5% annual
growth since it hit bottom along with oil prices in 1998. At the same
time the national poverty headcount dropped from 30% in 1999 to under
14% in 2007. Partially this is explained by the favorable commodity
prices enjoyed over this period as illustrated by a close relationship
between Russia's economic performance and oil prices depicted in
Figure 1. Indeed the two oil price dips of the 1990s coincide with the
two financial crises in Russia: the Black Tuesday of 1994 and the
Default of 1998. While it has become a commonplace to link Russia's
economic performance to oil--and some even go far back to explain the
collapse of the Soviet Union (Gaidar, 2007)--oil does not explain
everything that happens in Russia. Three-fourths of Russian regions
recorded positive growth in 1999 compared to three-fourths of regions
showing negative growth a year before. At the same time only one-third
of regions produced any oil and the oil price increased only modestly
from US$12/barrel in 1998 to US$17/barrel in 1999.
[FIGURE 1 OMITTED]
There is now wide consensus among observers of the Russian economy
that the long-term viability of Russia's economic growth hinges on
the improvement of productivity in the non-resource-based sectors of the
economy. Besides removing barriers for market competition, the most
effective government contribution to the productivity growth would be
through improving economic infrastructure, especially roads. According
to a recent World Bank study, the impact on productivity from bringing
the level of infrastructure to the OECD average would be higher than the
combined impact of closing the gap in terms of financial development,
quality of institutions, and education--in that order of magnitude
(World Bank, 2007). It has to be noted that in Russia, sub-national
governments are responsible for a considerable part of that
infrastructure.
Thus, Russia's economic recovery and long-term growth
prospects are more complex than can be suggested by a simple
macroeconomic story, such as oil price boost. What interests us in this
paper is that the dynamics of economic recovery have been very uneven
across Russian regions. Therefore, the determinants of long-term growth
need also be sought at the sub-national level. Among those potential
determinants we are particularly interested in this paper in the
potential role played by fiscal equalization among municipalities within
Russian regions.
Martinez-Vazquez and Timofeev (2008) find that the vast majority of
Russia's regions achieve a larger reduction of within-region fiscal
disparities than is the case for between-region disparities being
addressed by federal grants. In this paper we are interested in
understanding what may be the economic consequences of different degrees
of equalization being pursued in Russia's regions.
Fiscal equalization theory predicts economic inefficiency to arise
from disparities in local government revenues from source-based taxes
and local rents (Albouy, 2012). Therefore, fiscal equalization has the
potential to mitigate those disparities and prevent fiscally induced
misallocation of mobile economic resources. However, according to the
same theoretical models, disparities in residence-based taxes should be
equalized only in that part arising from differences in the skill
composition of local residents in terms of their income-earning
abilities, but not based on their actual income, as for example
determined by local income-generating opportunities. However, just like
in tax policy, governments might be willing to sacrifice some economic
efficiency and growth for more equity by equalizing actual revenues due
to inability to observe potential income. As we explain in this paper,
during the period covered by our data, most of Russia's regions
were equalizing actual revenues of constituent local governments. Not
surprisingly we find a negative relationship between intra-regional
equalization and regional economic growth.
The rest of this paper is organized as follows. First, we survey
theoretical links between equalization and economic outcomes conjectured
in the traditional fiscal federalism literature. Next we review those
past studies that link intergovernmental relations to economic outcomes
in Russia. Then we present empirical evidence on the extent of fiscal
inequality and equalization between and within Russian regions. Finally,
we examine the empirical relationship between the extent of
intra-regional equalization and economic growth in Russian regions. In
the final section we conclude.
THEORETICAL LINKS BETWEEN EQUALIZATION AND ECONOMIC EFFICIENCY
In the first-best theory of fiscal federalism there is no place for
general-purpose intergovernmental grants and revenue sharing. Local
governments are prescribed to engage 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
not be feasible for local governments to rely exclusively on benefit
taxation. (1) 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.
In practice, unconditional grants constitute a prevalent form of
intergovernmental transfers in many countries. Thus England's
Grants in Aid go as far back as 1830 (Webb, 1920). Moreover, the notion
of 'fiscal equity' has been the prime goal for grant schemes
in many countries since the beginning of the twentieth century, when the
notion of a 'national minimum' level of services for all
citizens entered the policy debate. This is attributed to the switch of
local governments from protective to social services thus moving further
away from benefit taxation (Buchanan, 1950). Federal countries with more
recent constitutions often have a constitutional requirement of
equalization payments to federating units. (2)
The traditional fiscal federalism literature does not provide
clear-cut implications of inter-jurisdictional equalization for economic
growth. Theoretically, equalization grants can be used to improve
economic efficiency by reducing horizontal inequities that otherwise
would lead to inefficient allocation of resources. Such efficiency case
for equalization was first articulated by Buchanan (1950, p. 585),
arguing that, if equals are 'pressed more in one area than in
another [through higher taxation and/or lower value of public services],
there will be provided an incentive for migration of both human and
non-human resources into the areas of least fiscal pressures'.
Furthermore, subsequent elaboration of this argument in the
literature concluded that differences in the value of public services
among different jurisdictions are determined by the differences in the
composition of residents in those jurisdictions in terms of
income-earning abilities and availability of revenue from local rents
(see Boadway, 2004 for a comprehensive review). Because the inefficient
migration of labor due to fiscal disparities is caused by differences in
jurisdiction characteristics (ie composition of labor force, local
rents), in order to induce efficient migration patterns, it suffices to
provide corrective subsidies to local governments rather than to
individual local residents. Note, however, that according to these
models, equalization grants should compensate a locality that has a
higher proportion of low-ability labor but not for the
cross-jurisdictional differences in wages of the same ability type of
labor. (3)
Similar to the case of equalization of revenue capacity,
compensating for disparities in demand for government services or the
costs of those services does not have clear-cut efficiency implications.
The issue of higher demand for government services, for example stemming
from a larger number of children per household, is analogous to
equalization of per capita fiscal capacity discussed above. Instead of
equalizing per capita fiscal capacity, here we need to consider
disparities per eligible resident (eg, per school-age child). (4) Thus
efficiency would improve from preventing fiscally induced migration to
localities where per child tax price of education is cheaper due to
access of the local government to some rent revenue, such as natural
resource royalties. While no theoretical model has been developed in the
literature to examine the efficiency implications of equalization of
service needs, Boadway (2004) suggests that efficiency-inducing grants
should take into account the composition of residents in terms of
eligibility for public services similar to the composition in terms of
income-earning abilities accounted for when equalizing revenue capacity.
However, compensating for differentials in unit costs of public
services, for example due to economies of scale, might impede efficiency
enhancing migration and thus prevent achieving the optimal scale of
producing government services (eg, the optimal school size).
To our knowledge, only one theoretical study has formally modeled
differences in the unit costs of a public service provision in addition
to differences in the levels of its congestion. In his model, Albouy
(2012) allows the unit cost of the public good provision to vary among
regions due to differences in productivity. Thus per capita costs of the
public service provision are equal to pG/N, where p is the marginal rate
of transformation of the private good (a numeraire) into a public good.
At the same time per capita benefit g from the public good output G is a
function of population size N: g = G/[N.sup.a], as in the previous
literature. Then the equalization grant term accounting for expenditure
needs looks like (1-a) x
([p.sub.1][G.sub.1]/[N.sub.1]-[p.sub.2][G.sub.2]/[N.sub.2]). Thus, it
appears that efficiency-enhancing grants should take into account
differences in the local costs of producing the public good. However,
the derived equalization grant takes into account per capita costs of
the local public goods provided at the levels that are optimal from the
social planner's perspective. As Boadway (2004) conjectures that it
would be optimal to have a lower level of public service provision in a
high-cost region, then assessed expenditure needs
[P.sub.i][G.sub.i]/[N.sub.i] can be higher or lower in a high-cost
region depending on the interplay of [p.sub.i] and [G.sub.i]. Albouy
(2012) concludes that variations in the costs of private or public goods
due to production efficiency or factors costs ought to be ignored in
equalization formulas.
Economic performance of Russian regions
Figure 2 shows the disparities in economic growth among
Russia's regions over the period 2001-2007. Over this period, the
median regional growth rate fluctuated between 3.9% and 8.2%, but with
wide disparities in the growth rate across regions. In any year the
inter-quartile range of regional growth rates was in excess of 2
percentage points and the coefficient of variation was above 0.5.
[FIGURE 2 OMITTED]
Overall, the dynamics of the 1999-2008 growth streak had been very
uneven across Russian regions. The question is what may account for
that. According to Matheson (2005), sub-national public investments in
Russia are affected by fiscal equalization. The transition literature
has also identified tax revenue retention as an important fiscal
incentive for local governments to promote development of the private
sector in their localities (see Desai et al, 2005, and Jin et al.,
2005). In Russia, less equalization is achieved in regions that allow
localities to retain a larger share of tax revenue at the point of
collection (Martinez-Vazquez and Timofeev, 2008). This is suggestive of
tradeoff between equity and growth as being hypothesized in the
literature (Qiao et al., 2008).
In the literature there is a discussion on the determinants and
outcomes of regional-local transfers in Russia. As until recently only a
few regions used explicit formulae for transfer allocation, most studies
had to analyze regional policies implicitly revealed through the
incidence of grants. For instance, Zhuravskaya (2000) finds that
regional governments offset completely changes in localities' own
revenues with changes in transfers. She also finds that claw-backing of
local revenues by regional authorities is negatively related to the
creation of new businesses in the region. However, Alexeev and
Kurlyandskaya (2003) argue that a rational regional government that is
averse to transfers would never want to completely compensate a locality
for a fall in local revenues as long as local authorities' efforts
affect local revenues. Using data for localities of one region, they
find no evidence that regional transfers tend to completely offset
changes in local revenues during the fiscal year. They do find, however,
evidence of the ratchet principle in a multi-year framework, that is an
increase in own revenue relative to the previous year is associated with
a decrease in transfers. Equity considerations might explain such
divergence of regional government policies from the efficiency
rationale.
Since 2000, more studies have been examining regional growth in
Russia. (5) In particular, Desai et al. (2005) link regional economic
recovery and structural adjustments to differences in tax revenue
retention by the regions interacted with their dependence on federal
transfers and natural resources. They find that revenue retention and
availability of grants and natural resources have a positive association
with economic recovery; however, the presence of federal transfers and
natural resources inhibit the positive effect of revenue retention.
FISCAL DISPARITY AND EQUALIZATION BETWEEN AND WITHIN RUSSIAN
REGIONS
For the measures of inequality and equalization between and within
Russian regions we draw on Martinez-Vazquez and Timofeev (2008), who
find a wide variation in local governments' per capita revenue from
their own taxes and from federal and regional tax sharing sources in
2001. In this paper, we use fiscal data from the same source, which is a
survey of sub-national budget reports for fiscal years 1999 and 2001.
The survey was conducted by the Center for Fiscal Policy (Moscow) by
requesting from individual regional governments the annual budget report
for the regional government and constituent local governments. In 1999
and 2001, the number of responding regions totaled about 70 out of a
total 89 regions of Russia, with about 1,900 out of total 2,500 local
governments. In regions with two-tier local governments, municipal data
are aggregated at the top tier.
We use two indices of inequality belonging to the class of general
entropy measures: square coefficient of variation ([I.sub.2]) and mean
log deviation ([I.sub.0]), also known as Theil's second measure.
These measures have a number of useful properties. In particular, these
indexes allow us to quantify relative contributions of between-region
and within-region disparities to the nation-wide level of inequality. In
addition, these two measures nicely complement each other as [I.sub.0]
gives more weight to disparities in the lower tail and [I.sub.2] gives
more weight to disparities in the upper tail of the distribution
(Martinez-Vazquez and Timofeev 2008).
Table 1 reports values of the two inequality measures for per
capita fiscal resources (before and after cumulatively adding the three
main categories of intergovernmental revenues to the own-source
revenues) among local governments of the 70 regions of Russia (1,953
localities in total) for 1999. The different categories of revenues are
discussed in Martinez-Vazquez et al. (2006), who find that the share of
own-source revenue in local budgets increased from 12% in 1999 to 14% in
2001. However, in 1999 the bulk of own-source revenues was accounted for
by a gross sales tax ('Housing maintenance tax') while in 2001
it was replaced by a local surtax on the federal Corporate Income Tax
(CIT). 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. The category
of 'regulated revenue' refers to tax-revenue sharing at
variable rates determined by the higher-level government as part of the
annual budget process. The final category of intergovernmental grants
compiles all discretionary transfers to local governments whether
formula-based or completely discretionary.
Table 1 reveals that the coefficient of variation in own-source and
assigned revenue stands at a rather high level of 1.55 (which is a
square root of the I2 measure, equal to 2.4 as reported in column 3).
This disparity is somewhat lessened after the allocation of regulated
tax revenues among localities, so that the coefficient of variation
decreases from 1.55 to 1.41. However, considerable equalization is
achieved only after the distribution of regional-local grants, after
which the coefficient of variation drops to 1.13.
The true extent of cross-jurisdictional equalization achieved with
the allocation of regulated tax revenue and grants is somewhat disguised
in the descriptive statistics computed over the sample pooling
localities from different regions. Decomposition of the square
coefficient of variation ([I.sub.2]) reveals that the between-region
disparity in local fiscal outcomes is actually larger mainly due to the
counter-equalizing effect of assigned revenue - than it was for the
initial between-region disparity in own-source revenue (see the top
panel in Table 1). For both [I.sub.0] and [I.sub.2] 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.
The same story holds if we adjust per capita resources of local
governments for the region-wide index of cost of living (see the bottom
panel of Table 1). (6) Generally, because of the differences in input
prices and socioeconomic 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 (eg,
heating fuel) to achieve the same outcome of public provision. While
accounting for inter-regional cost differences reduces total variation,
it nevertheless undergoes a similar transformation at each stage of
resource allocation as the inequality in nominal per capita resources.
We can consider one additional stage of resource allocation, which
is constituted by direct expenditures undertaken by regional
governments. Assuming, for the sake of argument, that expenditures from
the regional budget uniformly benefit residents of all constituent
localities, we can add a per capita amount of regional expenditures to
the per capita amount of local expenditures. In particular, such
imputation will account for the different extent of local
governments' involvement in the provision of public services in
different regions, that is, expenditure decentralization within regions.
Inclusion of regional expenditures into local fiscal outcomes
considerably decreases within-region disparities as shown in the last
column of Table 1. However, this equalizing effect hinges on our
assumption of uniform distribution of benefits from regional
expenditures. Thus, this imputation marks the range of the extent of
equalization when considered alongside the estimates produced under the
opposite assumption of no equalizing effect of direct regional spending.
At the same time, the measures of between-region disparities increase
with the inclusion of regional expenditures, regardless of the
assumptions about the intra-region incidence of the direct spending of
regional governments. 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.
When adjusted for differences in the cost of public services and
the scope of local government responsibilities, the local budget data
can reveal disparities in the level of public services among
Russia's localities. The next question we 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. (7) To answer this
question, we transform the 1999 actual tax collections into a measure of
revenue capacity by adjusting for the local level of tax effort. (8) The
measure of revenue capacity captures the ability of the jurisdiction to
raise revenue for public spending given the level of economic activity
within its boundaries and the devolved authority to translate this
economic activity into public revenues.
The constructed measure of revenue-raising capacity is estimated
for a sum of taxes that are commonly retained by local governments and
on average account for 75% of their pre-transfer revenue. These taxes
include: personal income tax, CIT, retail sales tax, property taxes,
land taxes, and a number of minor local taxes. To estimate tax yields in
specific localities, figures for local government receipts from each tax
were adjusted for the regionally set retention rates to derive the
amount of total collections for each tax in the locality. Information on
tax revenue retention rates derives from regional budget laws for the
year 1999.
Out of several potential proxies for local taxable bases, the best
explanatory power was shown by the combination of manufacturing output
and average wages, jointly explaining over 60% of variation in local
government own-source revenue in 1999. We use the obtained estimates to
impute revenue capacity of local governments in 1999. Because
municipal-level statistical data are not available for some regions, the
adjustments for tax effort can be only performed for a somewhat smaller
sample comprised of 1,602 localities in 59 regions. In this smaller
sample, the inequality in nominal and price-adjusted per capita
resources (top and middle panels of Table 2) undergoes a similar
transformation at each stage of resource allocation to the inequality
measures computed for the larger sample (Table 1).
Adjusting for the local tax effort produced slightly larger
measures of inequality but still following the same pattern at each
stage of resource allocation (see the bottom panel of Table 2). Overall,
if we account for differences in costs and tax capacity, the impact of
equalization on the between-region variation appears somewhat stronger.
However, the degree of between-region equalization is still short of
what we observe for within-region equalization as manifested in a
smaller share of the total inequality of fiscal outcomes accounted for
by the within-region inequality when compared to the respective share
before equalization.
Finally, we note a dramatic increase in fiscal inequality that
occurred between 1999 and 2001, as is evident from Table 3. This is due
to the reshuffling of tax assignments between the levels of government.
In particular to compensate municipalities for the loss of revenue from
the Housing Maintenance Tax (up to 1.5% of enterprise gross sales,
accounting for 14% of local pre-transfer revenues in 1999), in 2001
local governments were allowed to 'piggy-back' on the federal
CIT by introducing a local surtax of up to a maximum of 5%. To create
fiscal space for the local CIT surtax, the ceiling on the regional CIT
surtax was lowered and the requirement for regions to share CIT revenue
with municipalities was lifted.
As the gross sales tax had had a more evenly distributed tax base
than the CIT, concentrated in the (mainly urban) places of
incorporation, the within-region variation accounts for a larger share
of the total inequality in 2001 than in 1999.
FISCAL EQUALIZATION AND GROWTH
We now set out to examine how these cross-regional differences in
the extent of inequality and equalization translate into differences in
economic performance. Following Martinez-Vazquez and Timofeev (2008), we
measure the extent of equalization by regional governments as the ratio
of the inequality indexes (either [I.sub.2] or [I.sub.0]) before and
after equalization. Moreover, the extent of equalization is measured
separately first, in the narrow definition and second, in the broader
definition inclusive of direct regional spending. Because the scope of
own sources in local revenues changed between 1999 and 2001, resulting
in incomparable values for the inequality and equalization measures, in
our growth regressions we perform separate estimation on data for 1999
and 2001. (9) Furthermore, for 1999 we also attempt to perform
estimation for the equalization measure computed on figures adjusted for
local tax effort in addition to the estimation with the equalization
measure computed on nominal figures. However, the equalization measure
computed on figures adjusted for local tax effort turns out to be
insignificant in explaining regional growth and therefore we do not
report the results of those estimations.
Because the fiscal equity theory hinges on the mobility of
individuals across local jurisdictions, in an alternative specification
we allow for an interaction between fiscal equalization and migration.
(10) For each region, migration is measured as the fraction of regional
population who moved in, out, or within the region in a given year.
As control variables we have considered regressors commonly used in
growth equations (eg Qiao et al., 2008; Desai et al., 2005): population
growth, capital accumulation, education attainment, foreign trade,
ethno-linguistic fractionalization, urbanization, etc. In addition, we
tried other control variables identified in previous studies of regional
economic performance in Russia. The real per capita economic product is
included to control for regional income convergence. Regional tax effort
is included to isolate the equalization effect from tax policy effects.
Following Desai et al. (2005), we also include tax revenue retention by
localities, dependence on federal transfers and revenues from natural
resources, and the interaction among those variables. Finally, following
Brown et al. (2009), we include the number of civil servants per capita
to control for the interference by state actors, or alternatively
institutional support for private economic activity. To address possible
endogeneity, all regressors are lagged by 1 year relative to the
beginning of the period over which economic growth is computed as the
dependent variable.
Locality-level information on non-fiscal indicators derives from a
survey of regional statistical offices conducted by the federal
statistical office in 1999. Region-level non-fiscal data derive from the
2007 Russian Statistical Yearbook. Information on the region-wide tax
effort is taken from the 2001 calculations of federal grants.
Descriptive statistics for our explanatory variables are reported in
Appendix A.
RESULTS
Table 4 reports the estimation results for the 2000-2004 average
growth rate of the Gross Regional Product (GRP) as the dependent
variable and the 1999 inequality and equalization measures as the main
explanatory variables.
The top panel of Table 4 reports regressions using [I.sub.0]
indices while in bottom panel the regressions utilize [I.sub.2] indices.
Many of the control variables turned out to be of little statistical
significance and thus were omitted from the final regressions without
substantial impact on the fit of the model or the estimates of the
impact of remaining variables. According to our estimates, inequality in
own-source revenue in 1999 is associated with higher average growth rate
in the next 5 years. At the same time equalization through shared taxes
and grants in 1999 is associated with lower average growth rate in the
next 5 years. The effect of equalization through centralization of
resources to the regional budget is not statistically significant.
Table 5 reports the estimation results for the 2002-2006 average
growth rate of GRP as the dependent variable and the 2001 inequality and
equalization measures as the main explanatory variables, which are
computed using [I.sub.0] and [I.sub.2] indices, respectively. The 2001
inequality in own-source revenue is associated with lower average growth
rate in the next 5 years. Equalization through shared taxes and grants
seems to retard growth while equalization through the centralization of
resources to the regional budget is associated with higher average
growth rate in the next 5 years.
There are many similarities in results from the two periods of
time. Thus, in both time periods our regressions have higher explanatory
power when using the [I.sub.0] measures of inequality, which are more
sensitive to the disparities in the lower tail. Thus, fiscal incentives
faced by poorer localities seem to matter more for regional growth than
the treatment of the rich localities. In both time periods, equalization
through shared taxes and grants is associated with lower average growth
rates. In both time periods, the negative impact of equalization on
growth is amplified by higher mobility of population. In fact, according
to the 1999 regression, equalization in the upper tail ([I.sub.2]) does
not impede growth if the extent of mobility is below the average by more
than half a standard deviation.
Despite the aforementioned similarities, the results from the two
time periods differ on several accounts. First, subsequent economic
growth is positively related to the initial level of inequality in the
1999 own-source revenue dominated by the gross sales tax--while it is
negatively related to inequality in the 2001 own-source
revenue--dominated by the local surtax on corporate income. Besides
pointing out that these are very different revenue sources having
different spatial incidence, one can speculate about the fact the CIT
surtax is more of a source-based tax on local rents than the gross sales
tax is. The CIT tax is collected at the location of corporate
headquarters not where the sales (business activities) take place.
According to the fiscal equalization theory, disparities in local
government revenues from source-based taxes and local rents can cause
fiscally induced misallocation of mobile economic resources (Albouy,
2012).
Another difference between the two time periods is that most
regressors have higher statistical significance in the 2001 regression
than in the 1999 regression. In particular, the positive impact of
equalization through the centralization of resources to the regional
budget is highly statistically significant in 2001.
The impact of statistically significant control variables suggested
in other studies have the expected signs in our analysis: tax burden
(negative), dependence on federal transfers (negative), oil extraction
(negative), interaction of tax revenue retention with availability of
oil and federal grants (opposite to the sign at the uninteracted term).
DISCUSSION
We have found that regional intergovernmental policies dramatically
reduce within-region inequality in fiscal outcomes compared to the
inequality in own-source revenue. Thus, in a median region the squared
coefficient of variation drops by a factor of five after the
distribution of regional-local grants and, to a lesser extent, after the
allocation of regulated tax revenues among localities. For comparison,
in the United States state-local grants reduce between-county inequality
only by a factor of 2 (Martinez-Vazquez and Timofeev, 2010). As a result
the inequality in local expenditures in a median Russian region is only
two times higher than the inequality in a median US state while for
own-source revenue the inequality in a median Russian region is four
times higher than in a median US state.
Overall, our results suggest that intra-regional fiscal
equalization have a substantial impact on regional growth in Russia. For
example, in 1999, a one standard deviation increase in the measure of
regional equalization through shared tax revenues and grants translated
into a decrease of subsequent regional growth rate in the magnitude of
0.18 of its standard deviation (0.37 for the [I.sub.2] measure). In
2001, one standard deviation increase in regional equalization through
shared tax revenues and grants translated into 0.45 of a standard
deviation decline in regional growth rate. Finally, one standard
deviation increase in equalization through centralization of resources
to the regional budget translated into more than one standard deviation
increase in regional growth rate (0.8 for the [I.sub.2] measure).
Notwithstanding the dramatic equalization undertaken by Russian
regions, the nation-wide inequality (between--and within-state combined)
in fiscal outcomes is still rather high. For international comparison,
the squared coefficient of variation for per capita local expenditures
in Russia is three times higher than in China (Tsui, 2005) and 10 times
higher than in the United States (Martinez-Vazquez and Timofeev, 2010).
Such relatively high inequality in local fiscal outcomes is mainly due
to unmitigated between-region disparities, accounting for 55% of
expenditure inequality in Russia. This is comparable to just over half
of total inequality accounted for between-state inequality in the United
States and 61% of expenditure inequality accounted for by the
between-province disparities in China. However, in the initial
inequality in own-source revenues, between-state disparities account for
less than 40% in Russia compared to 59% in China and 32% in the United
States.
The most direct policy implication of our study is that
policymakers may face a tradeoff; choices about higher equalization
levels can translate in lower rates of economic growth as a result.
CONCLUSION
The purpose of this study has been to explore how cross-regional
differences in the extent of fiscal equalization within Russian regions
translate into differences in their economic performance. Overall, our
results suggest that intra-regional fiscal equalization in Russia has a
substantial impact on regional growth. Furthermore, in our regressions,
equalization in the lower tail of the distribution has a higher
statistical and economic significance, suggesting higher importance of
fiscal incentives for less affluent localities in a region.
APPENDIX A
Descriptive statistics
Tables A1 and A2 report descriptive statistics for the explanatory
variables for our 2000-2004 and 2002-2006 regressions, respectively.
Table A1: Summary statistics for the 2000-2004 regresssors
Mean Std. Dev.
Inequality in own-source 0.083 0.040
revenue ([I.sub.0])
Equalization ([I.sub.0]) 7.020 5.750
Equalization through 20.013 22.011
centralization ([I.sub.0])
Inequality in own-source 0.435 0.460
revenue ([I.sub.2])
Equalization ([I.sub.2]) 6.921 6.271
Equalization through 19.335 20.221
centralization ([I.sub.2])
Migration 0.037 0.034
Tax effort 1.097 0.537
Local retention of tax 0.576 0.125
revenue
Dependence on federal grants 0.209 0.183
Oil extraction 3.553 17.557
GRP per capita 20.253 7.308
Federal bureaucracy 3.055 1.080
Regional bureaucracy 1.319 0.944
Local bureaucracy 3.659 1.096
Judicial bureaucracy 0.892 0.274
Population growth -5.843 4.761
Capital growth 0.038 0.024
Min Max
Inequality in own-source 0.015 0.262
revenue ([I.sub.0])
Equalization ([I.sub.0]) 1.095 32.222
Equalization through 1.716 134.410
centralization ([I.sub.0])
Inequality in own-source 0.049 3.686
revenue ([I.sub.2])
Equalization ([I.sub.2]) 0.521 36.933
Equalization through 0.857 112.291
centralization ([I.sub.2])
Migration 0.009 0.271
Tax effort 0.543 5.256
Local retention of tax 0.211 0.851
revenue
Dependence on federal grants 0.000 0.820
Oil extraction 0.000 132.035
GRP per capita 0.000 41.249
Federal bureaucracy 1.545 8.705
Regional bureaucracy 0.203 4.523
Local bureaucracy 1.201 7.079
Judicial bureaucracy 0.240 1.976
Population growth -14.600 9.200
Capital growth 0.014 0.167
Table A2: Summary statistics for the 2002-06 regresssors
Mean Std. Dev.
Inequality in own-source revenue ([I.sub.0]) 0.156 0.189
Equalization ([I.sub.0]) 7.445 5.957
Equalization through centralization ([I.sub.0]) 29.418 48.527
Inequality in own-source revenue ([I.sub.2]) 1.418 3.949
Equalization ([I.sub.2]) 6.691 5.520
Equalization through centralization ([I.sub.2]) 25.539 37.328
Migration 0.035 0.032
Tax effort 1.005 0.341
Local retention of tax revenue 0.550 0.109
Dependence on federal grants 0.329 0.226
Oil extraction 3.794 18.253
GRP per capita 46.183 43.865
Federal bureaucracy 3.366 2.291
Regional bureaucracy 1.664 2.336
Local bureaucracy 3.920 2.435
Judicial bureaucracy 0.974 0.496
Population growth -5.582 5.422
Capital growth 0.068 0.043
Min Max
Inequality in own-source revenue ([I.sub.0]) 0.016 0.929
Equalization ([I.sub.0]) 0.157 23.709
Equalization through centralization ([I.sub.0]) 0.291 383.175
Inequality in own-source revenue ([I.sub.2]) 0.049 25.254
Equalization ([I.sub.2]) 0.471 23.826
Equalization through centralization ([I.sub.2]) 1.145 263.940
Migration 0.009 0.201
Tax effort 0.545 2.645
Local retention of tax revenue 0.232 0.831
Dependence on federal grants 0.007 0.905
Oil extraction 0.000 138.535
GRP per capita 11.049 303.661
Federal bureaucracy 1.576 19.333
Regional bureaucracy 0.207 14.889
Local bureaucracy 0.124 20.778
Judicial bureaucracy 0.244 3.690
Population growth -14.900 15.200
Capital growth 0.015 0.294
REFERENCES
Albouy, D. 2012: Evaluating the efficiency and equity of federal
fiscal equalization. Journal of Public Economics 96(9-10): 824-839.
Alexeev, M and Kurlyandskaya, G. 2003: Fiscal federalism and
incentives in a Russian region. Journal of Comparative Economics 31 (1):
20.
Berkowitz, D and DeJong, DN. 2003: Policy reform and growth in
post-soviet Russia. European Economic Review 47(2): 337-352.
Berkowitz, D and DeJong, DN. 2005: Entrepreneurship and
post-socialist growth. Oxford Bulletin of Economics & Statistics
67(1): 25-46.
Berkowitz, D and DeJong, DN. 2011: Growth in post-soviet Russia: A
tale of two transitions. Journal of Economic Behavior & Organization
79(1-2): 133-143.
Boadway, R. 2004: The theory and practice of equalization. CESifo
Economic Studies 50(1): 211.
Brown, JD, Earle, JS and Gehlbach, S. 2009: Helping hand or
grabbing hand? State bureaucracy and privatization effectiveness.
American Political Science Review 103(02): 264-283.
Buchanan, JM. 1950: Federalism and fiscal equity. The American
Economic Review 40(4): 583-99.
Desai, RM, Freinkman, L and Goldberg, I. 2005: Fiscal federalism in
rentier regions: Evidence from Russia. Journal of Comparative Economics
33(4): 814.
Energy Information Administration. 2009: World crude oil prices,
http://tonto.eia.doe.gov/dnav/pet/ pet_pri_wco_k_w.htm, accessed 16
October 2009.
Fisher, RC and Papke, LE. 2000: Local government responses to
education grants. National Tax Journal 53(1): 153.
Gaidar, Y. 2007: Collapse of an empire: Lessons for modern Russia,
translated by Bouis, AW Brookings Institution Press: Washington DC.
Golubchikov, O. 2007: Re-scaling the debate on Russian economic
growth: Regional restructuring and development asynchronies. Europe-Asia
Studies 59(2): 191-215.
Goskomstat. 2008: Russian statistical yearbook. Goskomstat: Moscow.
Jin, H, Qian, Y and Weingast, BR. 2005: Regional decentralization
and fiscal incentives: Federalism, Chinese style. Journal of Public
Economics 89(9-10): 1719.
Martinez-Vazquez, J and Timofeev, A. 2008: Regional-local dimension
of Russia's fiscal equalization. Journal of Comparative Economics
36(1): 157.
Martinez-Vazquez, J and Timofeev, A. 2010: 'The long and
winding road to local fiscal equity in the United States: A fifty year
retrospective'. International Studies Program Working Paper Series,
at AYSPS, GSU paperl027, International Studies Program, Andrew Young
School of Policy Studies, Georgia State University.
Martinez-Vazquez, J, Timofeev, A and Boex, J. 2006: Reforming
regional-local finance in Russia. WBI Learning Resources Series. World
Bank: Washington DC.
Matheson, T. 2005: Does fiscal redistribution discourage local
public investment? Evidence from transitional Russia. Economics of
Transition 13(1): 139.
Qiao, B, Martinez-Vazquez, J and Xu, Y. 2008: The tradeoff between
growth and equity in decentralization policy: China's experience.
Journal of Development Economics 86(1): 112.
Tresch, RW. 2002: Public finance: A normative theory. Elsevier
Science, Academic Press: San Diego, CA and London.
Tsui, K. 2005: Local tax system, intergovernmental transfers
China's local fiscal disparities. Journal of Comparative Economics
33(1): 173.
Webb, S. 1920: Grants in aid: A criticism and a proposal. Longmans,
Green and co: London, New York.
World Bank. 2007: Russian Economic Report #15. November 2007. The
World Bank Group.
Zhuravskaya, EV. 2000: Incentives to provide local public goods:
Fiscal federalism, Russian style. Journal of Public Economics 76(3):
337-368.
JORGE MARTINEZ-VAZQUEZ & ANDREY TIMOFEEV
International Center for Public Policy, Georgia State University,
PO Box 3992, Atlanta, 30302-3992, USA.
E-mail: atimofeev@gsu.edu
(1) There are many practical constraints to the implementation of
benefit taxation. Theoretically, with decreasing costs, benefit taxation
would not be sufficient to cover total costs with marginal costs being
below average costs (Tresch, 2002, p. 844).
(2) For example, Section 36 of Canada's Constitution Act of
1982. Other examples include Germany, Spain, and South Africa.
(3) Cross-jurisdictional differences in wages for the same type of
labor can arise due to differences in the costs of local public goods as
a result of congestion (Boadway, 2004). Given these differences in the
level of local public goods, the same type of labor will have different
wages in different regions. Because of the concavity of the utility
function, different types of labor will require different amounts of
wage adjustment to compensate for the same difference in the local-level
public good provision. In the special case of a pure public good and
constant migration costs, the productivity of labor (wages) and the
optimal level of service provision will be the same in both regions.
(4) Indeed, in the USA court-mandated schemes of state-wide
equalization of education finances are often set up in terms of taxable
property value per student (Fisher and Papke, 2000).
(5) These include Berkowitz and DeJong (2003); Berkowitz and DeJong
(2005); Golubchikov (2007), and Berkowitz and DeJong (2011) to name just
a few.
(6) Unfortunately, for the time period in our sample, there are no
municipal-level data that would allow us to make adjustments for
differences in relative prices across localities. Adjustments using the
region-level price index will not affect our measures of within-region
equalization, which is the main focus of this study. However,
regional-level adjustments help strengthen our finding of weak
between-region equalization reveled by between-region inequality
measures.
(7) We are indebted to Vladimir Popov for this point.
(8) We do not have municipal-level non-fiscal data to estimate tax
effort and revenue capacity for 2001.
(9) As we have explained, these data derive from a one-off survey
of sub-national budget reports for fiscal years 1999 and 2001 and
therefore are not available for other years. While this constraints our
analysis to cross-sectional estimations, it provides unique insights
into intergovernmental relations within Russian regions.
(10) We are indebted to an anonymous referee for pointing out the
importance of controlling for interaction with factor mobility.
Table 1: Disparities in per capita revenue of local governments within
and between regions, 1999
Own Plus Plus
source assigned regulated
Nominal fiscal resources
Square coefficient of variation 1.238 2.395 1.99
([I.sub.2])
Within regions 0.640 1.162 0.722
Between regions 0.597 1.233 1.268
Mean log deviation ([I.sub.0]) 0.154 0.163 0.155
Within regions 0.083 0.076 0.067
Between regions 0.070 0.086 0.087
Fiscal resources adjusted for the price level
Square coefficient of variation
([I.sub.2]) 0.833 1.286 1.147
Within regions 0.528 0.735 0.558
Between regions 0.305 0.551 0.588
Mean log deviation ([I.sub.0]) 0.134 0.136 0.127
Within regions 0.083 0.076 0.067
Between regions 0.050 0.059 0.060
Plus Plus regional
grants expenditures
Nominal fiscal resources
Square coefficient of variation 1.276 1.201
([I.sub.2])
Within regions 0.310 0.121
Between regions 0.966 1.080
Mean log deviation ([I.sub.0]) 0.084 0.117
Within regions 0.016 0.007
Between regions 0.067 0.110
Fiscal resources adjusted for the price level
Square coefficient of variation
([I.sub.2]) 0.655 0.599
Within regions 0.232 0.089
Between regions 0.423 0.510
Mean log deviation ([I.sub.0]) 0.059 0.061
Within regions 0.016 0.007
Between regions 0.042 0.053
Note: The sample includes 1,953 localities in 70 regions (covering
population of 108,399) for which fiscal data are available.
Source: Authors' calculation based on data from the Center for Fiscal
Policy, Moscow
Table 2: Disparities in per capita revenue of local governments within
and between regions, 1999
Own Plus Plus
source assigned regulated
Nominal fiscal resources
Square coefficient of variation 1.343 2.282 2.273
([I.sub.2])
Within regions 0.625 0.998 0.813
Between regions 0.718 1.284 1.459
Mean log deviation ([I.sub.0]) 0.156 0.163 0.162
Within regions 0.079 0.075 0.068
Between regions 0.076 0.088 0.094
Fiscal resources adjusted for the price level
Square coefficient of variation 0.845 1.286 1.317
([I.sub.2])
Within regions 0.484 0.684 0.632
Between regions 0.361 0.601 0.684
Mean log deviation ([I.sub.0]) 0.132 0.134 0.129
Within regions 0.079 0.075 0.068
Between regions 0.052 0.059 0.061
Fiscal resources adjusted for the local tax effort and price level
Square coefficient of variation 0.971 1.213 1.071
([I.sub.2])
Within regions 0.599 0.707 0.540
Between regions 0.372 0.505 0.531
Mean log deviation ([I.sub.0]) 0.170 0.170 0.159
Within regions 0.110 0.107 0.100
Between regions 0.059 0.063 0.058
Plus Plus regional
grants expenditures
Nominal fiscal resources
Square coefficient of variation 1.449 1.344
([I.sub.2])
Within regions 0.349 0.140
Between regions 1.099 1.204
Mean log deviation ([I.sub.0]) 0.091 0.100
Within regions 0.016 0.007
Between regions 0.074 0.092
Fiscal resources adjusted for the price level
Square coefficient of variation 0.746 0.647
([I.sub.2])
Within regions 0.261 0.102
Between regions 0.485 0.545
Mean log deviation ([I.sub.0]) 0.061 0.059
Within regions 0.016 0.007
Between regions 0.045 0.052
Fiscal resources adjusted for the local tax effort and price level
Square coefficient of variation 0.660 0.589
([I.sub.2])
Within regions 0.258 0.108
Between regions 0.402 0.481
Mean log deviation ([I.sub.0]) 0.070 0.063
Within regions 0.027 0.013
Between regions 0.043 0.050
Note: The sample includes 1,602 localities in 59 regions for which
statistical data are available to estimate the tax effort.
Source: Authors' calculation based on data from the Center for Fiscal
Policy, Moscow
Table 3: Disparities in per capita revenue of local governments within
and between regions, 1999 and 2001
Own Plus Plus
source assigned regulated
Nominal fiscal resources in 1999
Square coefficient of variation 1.148 2.373 1.886
([I.sub.2])
Within regions 0.575 1.075 0.624
Between regions 0.572 1.297 1.261
Mean log deviation ([I.sub.0]) 0.152 0.161 0.151
Within regions 0.082 0.076 0.066
Between regions 0.069 0.084 0.084
Nominal fiscal resources in 2001
Square coefficient of variation 3.209 4.082 3.068
([I.sub.2])
Within regions 1.936 2.129 1.474
Between regions 1.272 1.953 1.594
Mean log deviation ([I.sub.0]) 0.261 0.257 0.218
Within regions 0.160 0.143 0.120
Between regions 0.100 0.113 0.097
Plus Plus regional
grants expenditures
Nominal fiscal resources in 1999
Square coefficient of variation 1.154 0.976
([I.sub.2])
Within regions 0.311 0.120
Between regions 0.842 0.856
Mean log deviation ([I.sub.0]) 0.077 0.130
Within regions 0.016 0.007
Between regions 0.061 0.123
Nominal fiscal resources in 2001
Square coefficient of variation 1.894 1.691
([I.sub.2])
Within regions 0.858 0.477
Between regions 1.036 1.214
Mean log deviation ([I.sub.0]) 0.124 0.103
Within regions 0.057 0.031
Between regions 0.067 0.072
Note: The 2001 sample includes 1,688 localities in 62 regions
(covering population of 95.5 million) for which fiscal data are
available. The 1999 sample includes 1,693 localities in 62 regions
(covering population of 95.1 million) for which fiscal data are
available.
Source: Authors' calculation based on data from the Center for Fiscal
Policy, Moscow
Table 4: OLS regression of 2000-2004 regional growth
LHS: Average growth rate 2000-04 (1) (2)
[I.sub.0] measure
Inequality in own-source revenue 36.701 *** 43.533 ***
(8.457) (8.468)
Equalization -0.205 * -0.228
(0.110) (0.303)
Equalization through centralization 0.018 0.025
(0.034) (0.039)
Migration -- -30.988
(65.174)
Migration x Equalization -- -1.083
(9.137)
[R.sup.2] 0.63 0.67
# of regions 64 64
[T.sub.2] measure
Inequality in own-source revenue 0.745 4.590 ***
(0.997) (1.098)
Equalization -0.153 0.269 *
(0.110) (0.149)
Equalization through centralization 0.013 0.019
(0.041) (0.040)
Migration -- 55.065 **
(24.639)
Migration x Equalization -- -17.697 ***
(4.424)
[R.sup.2] 0.56 0.65
# of regions 64 64
Note: Robust standard errors are provided in parentheses: *
statistically significant at 10%; ** statistically significant at 5%;
*** statistically significant at 1%. Federal district dummies included
in all specifications. In addition to the shown variables of interest,
the regressions also include control variables: Tax effort; Local
retention of tax revenue; Dependence on federal grants; Oil
extraction; Oil extraction x Tax retention; Grants x Tax retention;
GRP per capita; Local bureaucracy; Judicial bureaucracy; Other
bureaucracy; Population growth; Capital growth.
Table 5: OLS regression of 2002-2006 regional growth
LHS: Average growth rate 2002-06 (1) (2)
[I.sub.0] measure
Inequality in own-source revenue -3.427 *** -3.362 **
(1.285) (1.305)
Equalization -0.239 *** -0.190
(0.076) (0.133)
Equalization through centralization 0.079 *** 0.077 ***
(0.026) (0.027)
Migration -- 1.788
-19.824
Migration x Equalization -- -1.422
-3.645
[R.sup.2] 0.64 0.65
# of regions 69 69
[I.sub.2] measure
Inequality in own-source revenue -0.104 ** -0.100 **
(0.042) (0.040)
Equalization -0.243 *** -0.085
(0.082) (0.152)
Equalization through centralization 0.070 ** 0.064 **
(0.030) (0.031)
Migration -- 6.848
(12.597)
Migration x Equalization -- -4.446
(3.226)
[R.sup.2] 0.62 0.64
# of regions 69 69
Note: Robust standard errors are provided in parentheses: **
statistically significant at 5%; *** statistically significant at 1%.
Federal district dummies included in all specifications.
