Munich Personal RePEc Archive
Fiscal decentralization and government
size across Europe
Makreshanska, Suzana and Petrevski, Goran
Saints Cyril and Methodius University in Skopje, Faculty of
Economics, Saints Cyril and Methodius University in Skopje,
Faculty of Economics
2 December 2016
Online at https://mpra.ub.uni-muenchen.de/82472/
MPRA Paper No. 82472, posted 08 Nov 2017 00:21 UTC
Suzana Makreshanska and Goran Petrevski
FISCAL DECENTRALIZATION AND GOVERNMENT SIZE ACROSS EUROPE
Abstract
The paper provides for an empirical study of the association between fiscal decentralization and
government size on a panel of 28 European developed and former transition countries during
1991-2011, controlling for the effects of various demographic, institutional, and macroeconomic
variables as well as for the effects of the Global Financial Crisis. The main findings from the
empirical investigation are as follows: We provide evidence for non-negligible effects of
expenditure decentralization on government size, especially in the former transition economies.
However, when we employ the revenue decentralization as an explanatory variable we cannot
provide support to the Leviathan hypothesis. We include two measures of the vertical fiscal
imbalance and provide empirical support to the common-pool hypothesis only for the former
transition countries. As for the effects of the control variables, our research results suggest that
higher public debt leads to larger government, while trade openness is associated with smaller
government size. Also, we find that the effects of population density and dependent population
on government size differs between the developed and the former transition countries, while
higher degree of urbanization reduces government size only in the developed countries subsample. Finally, we confirm that the Global Financial Crisis has strong effects on the level of
government expenditure across Europe.
Keywords: Fiscal decentralization, Government size, Leviathan hypothesis, Common-pool
hypothesis, Panel data models, Fixed-effects estimator
JEL Classification numbers: C33, H50, H71, H77.
1. Introduction
In the fiscal federalism literature there is a long standing debate on the expected effects of fiscal
decentralization on the size of the government as well as its efficiency. According to the
advocates of decentralization it should deliver both smaller and more efficient government
because the local governments usually have a comparative advantage in the allocation of
resources as compared to the central government (Oates, 1972). In other words, by bringing
government “closer to the people”, fiscal decentralization provides a better match between local
preferences and local policies. Further on, decentralization could decrease government size by
promoting the tax competition between the different levels of government. In a decentralized
system, governments’ revenue-maximizing behaviour is undermined by their need to compete
with one another for mobile sources of revenues (Brennan and Buchanan, 1980). On the other
hand, the opponents argue that the expected positive effects of decentralization on the size and
efficiency of government need not materialize since often the central government and the local
authorities engage in oligopolistic arrangements for financing from common pool of funds, thus
attempting to maximize the public expenditure (Grossman, 1989; Grossman and West, 1994).
In this paper we provide empirical evidence on the effects of fiscal decentralization on
government size for an unbalanced panel of 28 European countries during 1990-2011, estimated
by the fixed-effects estimator. Our main research task is to test empirically the two standard
hypotheses concerning the effects of fiscal decentralization: the Leviathan hypothesis and the
common-pool hypothesis. To this end, we regress the government expenditure (as % in GDP) on
several fiscal decentralization variables (local expenditure, local revenue, and vertical fiscal
imbalance), controlling for the effects of various demographic, institutional, and macroeconomic
variables, such as: the level of economic development, trade openness, population density,
urbanization, dependent population, the level of public debt, and inflation. In addition, we take
into account the effects of the Global Financial Crisis on government expenditure. In order to
check for the robustness of the results, we run three separate regressions, covering the full
sample as well as two sub-samples: one that includes the developed countries and another one
consisting of the former transition economies from Central and Eastern Europe (CEE).
The main findings from our investigation are as follows: We provide evidence for non-negligible
effects of expenditure decentralization on government size, especially in the former transition
countries. However, when we employ the revenue decentralization as an explanatory variable we
cannot provide support to the Leviathan hypothesis. In addition, we include two measures of the
vertical fiscal imbalance and provide empirical support to the common-pool hypothesis only for
the CEE countries where the reliance on tax-sharing, central government grants and other forms
of intergovernmental fiscal transfers indeed leads to larger government. As for the effects of the
control variables, our research results suggest that higher public debt leads to larger government,
while trade openness is associated with smaller government size. Also, we find that the effects
of population density and dependent population on government size differs between the
developed and the CEE countries, while higher degree of urbanization reduces government size
only in the developed countries sub-sample. Finally, we confirm that the Global Financial Crisis
has had strong effects on the level of government expenditure across Europe.
As for the organization of the paper, the next section presents the main theoretical concepts
concerning the expected effects of decentralization on government size. Section 3 provides an
overview of the empirical research in this field followed by Section covering methodological and
data issues as well as providing discussion of the regression results. As usual, the last section
concludes.
2. Fiscal decentralization and government size in theory
Since the 1950s a number of theoretical models have attempted to shed a light on the role of the
government and the effects of fiscal decentralization on government size (For instance, see
Buchanan and Wagner, 1977; Rodden, 2003). The classical approach to public finance views the
government as a “benevolent despot”, which are responsible for implementing the socially
optimal public policies (Musgrave, 1959). According to this concept of the so-called „responsive
governments“, in performing its functions the government is guided solely by the citizens’
preferences about the public goods. Therefore, in the decision-making process the government
always attempts to maximize the social welfare function. In these regards, the government size
reflects the citizens’ demand for public goods, implying that both the decentralization of the
government and tax competition between the different tiers of government could lower the
quality of public goods and, ultimately, to reduce social welfare.
Yet, there is another, presumably more realistic theoretical model arguing that the government’s
decision-making process is not necessarily based on the citizens’ preferences, but the politicians
and bureaucrats follow their own interests and aim at maximizing their power and revenues. As
can be seen, the concept of the so-called “excessive governments” introduces the principal-agent
and information asymmetry problems in the analysis of the supply of and demand for public
goods. Within this framework, on the one hand, the government pursues its own interest (to be
re-elected) so it’s maximizes the public expenditure above the demand for public goods. On the
other hand, facing the asymmetric information problem, the citizens always prefer public
expenditure to taxation. As a result, the size of the public sector shows a tendency to increase
over time (Buchanan and Wagner, 1977). The concept of “excessive governments” implies that
the decentralization is expected to lead to both smaller government and higher efficiency.
Indeed, that is the main prediction of the so-called Leviathan hypothesis, which argues that,
ceteris paribus, the role of government in the economy will be lower the higher the degree of
decentralization of government (Brennan and Buchanan, 1980). Specifically, under the
assumption of mobility of capital and labor, fiscal decentralization constrains the government
monopoly of taxation by introducing tax competition between the various tiers of government,
which ultimately reduces government size.
Another theoretical model that suggests a negative association between decentralization and the
size of government has been proposed by Oates (1972). Within this framework, under the
assumptions that the citizens have heterogeneous preferences for the locally provided public
goods and the marginal costs for providing public goods differ between the local governments,
decentralization increases the efficiency in the allocation of resources. The rationale behind this
proposition stems from the information advantage of local governments, which are capable of
identifying the true preferences of the citizens and, accordingly, to match the supply of public
goods to the demand. Consequently, the increased efficiency of public expenditure is the
mechanism by which decentralization leads to smaller government. Yet, one might argue that,
under the assumption of relatively high price elasticity of the demand for public goods, the lower
marginal costs associated with the decentralization could lead to an increase in the demand for
public goods and, accordingly, larger government.
Some authors (for instance, Grossman, 1989; Grossman and West, 1994) emphasize that the
effects of decentralization on the size of the government depends on the way the decentralization
is implemented, implying that decentralization need not result in smaller and more efficient
government. This occurs when the local and central government, instead of competing for the
scarce fiscal resources, engage in various (oligopolistic) arrangements for sharing the common
fiscal resources, such as the revenue sharing programs. In this case, the local governments accept
to substitute their fiscal autonomy for sharing the revenues with the central government.
Therefore, the tax competition among the local governments is replaced by the “race” for larger
expenditure from the “common pool” of public revenues. This is the essence of the so-called
“collusion” or “common pool” hypothesis, which claims that, when local expenditure is financed
by fiscal transfers and tax sharing, decentralization results in larger government (Grossman,
1989).
Concerning the above mentioned issue, a number of papers deal with the relationship between
intergovernmental transfers and the size of government (Ehdaie, 1994; Stein, 1998; Rodden,
2003). For instance, Rodden (2003) argues that relying on intergovernmental transfers and taxsharing as sources of local government finance lowers the degree of tax competition between the
different tiers of government, which undermines the strongest theoretical argument in favor of
decentralization. According to this view, intergovernmental transfers increase both the supply of
and demand for public goods through several channels. Specifically, when local public goods are
financed by transfers from and tax-sharing with the central government the costs are borne not
only by the local citizens (who benefits from the local public goods) but they are spread on the
whole population. In that case, the demand for local public goods will be larger because part of
the costs is transferred to the non-residents instead of being internalized. In the opposite case,
when the local public goods are financed solely by local revenue the demand for them will be
lower because all the costs are internalized (Rodden, 2003).
In addition, intergovernmental transfers lead to larger government by creating the so-called “soft
budget constraints” at the local government level. In these regards, Stein (1998) argues that,
when faced with various fiscal shocks (higher interest rates on the debt, lower local tax revenue
etc.), local governments that rely on intergovernmental transfers as a source of finance feel lesser
pressure for adjustment and could afford a higher level of local expenditure for a longer time
period (by increasing their debt). The reason for this type of behavior is that both the politicians
and the voters expect that the central government, which has already finances the current
expenditure of the local governments, would finance their debt, too, especially when the
allocation of the intergovernmental transfers is based on discretion instead of legally-binding
rules (Stein, 1998). The above arguments could be illustrated by the Italian experience with
financing public health services, which are one of the most important responsibilities of the
regional governments (Bordingnon and Turati, 2009; Josselin et al., 2012). By the early 1990s,
the public health expenditures in Italy were financed predominantly by the fiscal transfers from
the central government. As a result, these expenditures saw a significant expansion throughout
all the regions. However, following the introduction of the specific regional tax for financing the
public health services (IRAP) in 1996 both the intergovernmental transfers for this purpose and
the regional public health expenditures have declined considerably.
As can be seen, decentralization could produce both positive and adverse effects on the size of
government depending on the mode of financing the local expenditures. In other words, in order
to affect the size of government, the decentralization of expenditures needs to be accompanied
by decentralization of revenues. In these regards, there are strong theoretical arguments
suggesting that the decentralization that is based on fiscal transfers from and tax-sharing with the
central government, instead of fiscal autonomy of the local governments, does not lead to tax
competition and smaller government.
3. An overview of the empirical literature
Given the existence of competing theoretical models of fiscal federalism a large number of
empirical studies have attempted to disentangle the relationship between fiscal decentralization
and the size of government. The empirical research in this field has been pioneered by Oates
(1972, 1985). Since the 1980s, most of the empirical research within this strand of the literature
has focused on testing the Leviathan hypothesis proposed by Brennan and Buchanan (1980). The
long-lasting research interest in this filed probably reflects the controversial views of Brennan
and Buchanan (1980) about the role of government with its tendency to maximize its power,
thus, drawing the analogy with the biblical monster of gigantic size. In addition, Brennan and
Buchanan (1980) themselves have invited the researchers to test their hypothesis, thus triggering
an array of empirical studies. In which follows we provide a brief overview of the empirical
research on the relationship between fiscal decentralization and the size of government.
In his cross-section study on 57 countries, Oates (1972) found that decentralization and
government size may not be related. Similarly, Oates (1985), and Wallis and Oates (1988) also
did not find a significant relationship between decentralization and government size both in the
sample of 48 US states and in the sample of 43 developed and developing countries. Later on,
Oates’s empirical studies were replicated by many authors in different national or international
studies, covering different samples and time periods, using different measures for
decentralization and government size and different estimation techniques. For instance, focusing
on the US experience, Nelson (1986) provided some evidence that the less fragmented countries
(those with smaller number of sub-state government units) have larger state government sectors
while Forbes and Zampelli (1989) were not able to provide support to the Leviathan hypothesis.
Although there is no firm consensus about the single best measure of government size,
expenditure-based measures may be considered superior versus revenue-based measures, since
government expenditures could be financed not only by the regular sources (taxes and non-tax
revenues) but also by other sources, such as debt creation, money creation and inflation, etc.
Therefore, unlike the previous empirical studies, Marlow (1988) and Joulfaian and Marlow
(1991) used the expenditure-based measure of decentralization and government size and
provided evidence that the higher level of decentralization leads to smaller general government.
The earlier studies that empirically tested the relationship between fiscal decentralization and
government size relied on local government revenue and expenditure measures regardless
whether local governments have discretion over their revenues or expenditures. On the other
hand, as suggested by Rodden (2003), in most countries fiscal decentralization seems to have
occurred almost exclusively through increased intergovernmental grants and shared revenues
rather than through the devolution of expenditure and tax authority. Therefore, in the empirical
studies that followed, many authors started to make a distinction between expenditure and
revenue based decentralization and the size of vertical fiscal imbalance, measured by
intergovernmental grants. Grossman (1989) underlined the role of intergovernmental grants in
increasing government size, through concentrating taxing power at the central level of
government and weakening the tax competition and fiscal discipline of local governments for
financing their expenditures. Working with the same sample as Marlow (1988), Grossman
(1989) empirically confirmed that the higher federal grants to state and local government the
larger the size of the government. Similarly, Shadbegian (1999), too, supported the Leviathan
hypothesis.
Apart from the US experience, the relationship between fiscal decentralization and government
size has been tested in other intra-national studies, such as Canada and Switzerland. For instance,
Grossman and West (1994) found inverse correlation between fiscal decentralization and the
general government in Canada. Yet, on a disaggregated level, they provide evidence that the
shares of the provincial and local governments actually increase with decentralization and that
the intergovernmental grants tend to increase the size of each level of government. As for the
Swiss experience, Feld et al. (2010) found a negative relationship between fiscal decentralization
and the size of cantons.
Further on, the relationship between fiscal decentralization and government size has been
investigated in a number of multi-country studies. For instance, Ehdaie (1994) showed that
simultaneous decentralization of the central government’s taxing and spending power has
negative effect on the government size. On the other hand, revenue sharing with taxing decisions
concentrated at the central governments eliminates the negative influence of decentralization on
the government size. Hence, the countries pursuing the objective of a smaller governments,
should decentralize not only the national government spending power, but the taxing power as
well. Jin and Zou (2002) found that: both expenditure and revenue decentralization leads to
smaller government at aggregate level; and vertical fiscal imbalance increases the government
size at all levels of government. On the other hand, working with both the revenue and
expenditure-based measures, Heil (1991) rejected the Leviathan hypothesis.
Stein (1998) introduced another dimension in this subject matter by arguing that decentralization
leads to higher government spending if it is accompanied by higher local borrowing autonomy
and soft budget constraints in local financing. He tested this proposition empirically on Latin
America cross-country data, averaged for the 1990-1995 period, and found a positive
relationship between expenditure decentralization and government size. According to him, in
order to offset the positive effect of decentralization on government size, the governments should
reduce the vertical fiscal imbalance i.e. local government spending should be financed by own
local sources of revenues instead of central government grants, shared revenues and local
government borrowing.
Traditionally, the IMF’s Government Finance Statistics (GFS) has served as a data source in
most of the empirical studies on fiscal decentralization because until 1999 the GFS was the only
available international dataset on local government finance. But, although GFS provides
consistent and operational data of fiscal decentralization measures across countries and over
time, it fails to provide a full picture of fiscal decentralization (Ebel and Yilmaz, 2002). In fact,
GFS tends to overestimate the true nature of local government revenue autonomy by ignoring the
difference between local taxes upon which local governments have full tax discretion, and other
local revenues upon which central government retain control over tax rates and tax bases. On the
other hand, as pointed out by Stegarescu (2005), a decentralized system where local governments
have real autonomy to determine the allocation of their expenditures or to raise their own
revenue is more decentralized than a system in which local government expenditures and
revenues are determined by the central government, regardless of the size of local government
expenditures or revenues. OECD (1999) made an effort to improve the revenue decentralization
measure by classifying taxes by the degree of local government autonomy. Using the OECD
improved measures, Ebel and Yilmaz (2002) replicated the study of Oates (1985) and found that
local tax autonomy has a negative and significant impact on government size. Rodden (2003),
too, showed that fiscal decentralization may have a different impact on government size,
depending on whether local expenditures are funded by “own” local revenue sources or
intergovernmental grants and shared revenues from the central government. Stegarescu (2005)
went a step further, expanding the OECD data set of local government revenue autonomy to
cover 23 OECD countries over the period 1965-2001. Using Stegarescu’s improved data on tax
revenue decentralization for the OECD sample, Fiva (2006) re-examined the relationship
between fiscal decentralization and government size and concluded that tax revenue
decentralization is associated with smaller government size. On the other hand, the expenditure
decentralization for given tax revenue decentralization, is associated with a larger public sector.
Among the recent studies, Prohl and Schneider (2009) conclude that there is substantially lower
growth of public expenditure and tax burden in the countries where revenues and expenditure
responsibilities are decentralized to a large extent. They also re-examined the relationship
between decentralization and government size by introducing additional proxy variable, which
incorporates the fiscal and administrative autonomy of local governments, and showed that
higher fiscal and administrative autonomy of local governments is associated with a slower
growth of the government. Working with a large sample of 74 countries over the period 19852000, Martinez-Vazquez and Yao (2009) show that government size, measured by the number of
employees in public sector, increases with the level of fiscal decentralization. Cassette and Paty
(2010) analyze the effect of decentralization on central and local government sizes by separating
the long-run effects of decentralization from its short-run dynamics. In the long run, tax
autonomy reduces central expenditure but increases to a greater extent local expenditure, leading
to higher general public expenditure. They also show that vertical imbalances tend to increase
the size of both the local and central government. Baskaran (2011) explores the impact of fiscal
decentralization on the size of the public sector in 18 OECD countries over the 1980-2000,
depending on the government ideology. The main message of his study is that decentralization
leads to a smaller government size under a right- than under left-wing parties.
As can be seen, the empirical literature has not reached a consensus on the between
decentralization and government size. Nevertheless, we are able to extract several general
conclusions:
1) National versus multi-country studies. It seems that the national studies provide stronger
support to the Leviathan hypothesis than the cross-country studies. Indeed, most of the national
studies reviewed in this Section provide evidence for the Leviathan hypothesis (Nelson, 1986;
Marlow, 1988; Grossman, 1989; Joulfaian and Marlow, 1991; Shadbegian, 1999; Grossman and
West, 1994; Feld et al., 2010). In contrast, a small portion of the international studies is able to
this hypothesis (Ehdaie, 1994; Jin and Zou, 2002; Rodden, 2003; Fiva, 2006).
2) Different measures of the public sector. Initially, the government size had been expressed in
terms of the public revenues and these studies typically failed to provide empirical support to the
Leviathan hypothesis (Oates, 1972 and 1985; Nelson, 1986; Oates и Wallis, 1988; Forbes and
Zampelli, 1989). The recent, expenditure-based studies provide a stronger support to the
hypothesis (Marlow, 1988; Grossman, 1989; Joulfaian and Marlow, 1991; Ehdaie, 1994;
Shadbegian, 1999).
3) Alternative measures of fiscal decentralization. Recently, an increasing fraction of the studies
emphasize the composition of local revenues. As a result, the studies that measure
decentralization in terms of the local governments’ “own” revenues are more able to provide
support to the Leviathan hypothesis (Jin and Zou, 2002; Rodden, 2003; Fiva, 2006).
4. Fiscal decentralization and government size: Industrialized European countries versus
Central and East European countries
The above review of theoretical models and empirical studies suggests that the relationship
between fiscal decentralization and government size is ambiguous. In principle, the simple
transfer of public expenditures from the central government to the sub-national governments,
which occurs if sub-national governments perform the same functions previously done by the
central government, need not affect the size of the public sector. Yet, if sub-national
governments are indeed more allocative efficient, in that case decentralization should lead to
smaller government as suggested by Oates (1972). Under the third scenario, in case of a weak
fiscal discipline of sub-national governments due to the soft budget constraints and the common
pool problems, decentralization inevitably results in larger government.
Therefore, the answer to our principal research question should be sought in the empirical
investigation. Before turning to the discussion of the methodological issues as well as the
estimation results we present some general facts on the relationship between fiscal
decentralization and government size for two sub-samples.
Figure 1 plots the relationship between decentralization and the size of the public sector in the
industrialized European countries, consisting of the “old” EU members plus some other
developed countries, such as Switzerland, Iceland and Norway. As shown, the degree of fiscal
decentralization is measured by the share of sub-national expenditures in the total public
expenditures. The vertical axis depicts the size of the public sector as measured by the share of
the total public expenditures (the expenditures of the general government) in the Gross Domestic
Product (GDP). One can observe a weak positive correlation between decentralization and the
size of government, i.e. the more decentralized countries have larger public sector. Some authors
interpret the observed positive relationship as evidence that decentralization results in
overlapping functions of the central and sub-national governments, which eventually leads to a
larger public sector (Joumard and Kongsrud, 2003). An alternative explanation is that the voters
in the developed European countries simply prefer higher taxes accompanied by higher quantity
and quality of public goods. In these regards, decentralization leads to larger government by
matching the supply of public goods to the demand (Eyraud and Badia, 2013). This explanation
is probably more relevant for the Scandinavian countries, which are the most decentralized
countries and, at the same time, they are known to have the largest public sector among the
European countries (53.3% versus 48.1% in the rest of the sub-sample).
Figure 1: Fiscal decentralization and public expenditure in the developed European countries,
2013
70.0
65.0
Greece
Public expenditure as % of GDP, 2013
60.0
Finland
France
Denmark
Belgium
55.0
Sweden
Italy
Portugal
50.0
Luxemburg
45.0
Austria
UK
Netherlands
Iceland
Germany
Norway
Ireland
Spain
y = 0.0953x + 46.303
R² = 0.0522
40.0
Switzerland
35.0
30.0
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
Sub-national expenditure as % of total public expenditure, 2013
Source: Own estimates based on the EUROSTAT database.
Figure 2 plots the relationship between decentralization and the size of the public sector in the
“new” EU members, consisting mostly of the CEE countries plus Malta and Cyprus. The extent
of fiscal decentralization as well as the size of the public sector is measured in the same manner
as in Figure 1. As can be seen, the direction of the relationship between decentralization and
government size in the CEE countries slopes in the opposite direction as compared to the
developed European countries. For these countries the regression coefficient is much bigger (0.31) than in Figure 1, it has the negative sign predicted by Oates (1972) and, at the same time,
the association between decentralization and the government size is much stronger as revealed
by the coefficient of determination (0.16). This finding may be related to the massive political,
economic and institutional reforms implemented in these countries during the transition from
centrally-planed to market economies, which included the decentralization process aimed at
increasing the allocative efficiency of the government. In these regards, some argue that former
transition economies may gain larger benefits from the decentralization compared to the
industrialized countries due to the high degree of political and fiscal centralization as well as the
great inefficiency of the public sector (Shah, 2004).
Figure 2: Fiscal decentralization and public expenditure in the “new” EU members, 2013
70.0
65.0
Slovenia
Public expenditure as % of GDP, 2013
60.0
y = -0.3079x + 48.541
R² = 0.1644
55.0
Hungary
50.0
Croatia
45.0
Cyprus
Czech Republic
Poland
Malta
Slovakia
40.0
Bulgaria
35.0
Lithuania
Estonia
Romania
Latvia
30.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
Sub-national expenditure as % of total public expenditure, 2013
Source: Own estimates based on the EUROSTAT database.
The above plots of the decentralization-government size relationship are given for illustrative
purpose only and need not be interpreted as implying causality from decentralization to the
government size. For instance, one may argue that the observed positive correlation between
decentralization and government size may reflect the reverse causality, i.e. the countries with
larger government are probably better candidates for decentralizing expenditures than the
countries in which the public sector is small. In addition to decentralization, it is reasonable to
assume that government size depends on various other determinants, which have been omitted in
the above figures, such as: the demographic characteristics of the population, the level of
economic development etc. Therefore, we now turn to a more elaborate treatment of the
association between decentralization and government size.
5. Methodology and discussion
5.1. Data description
Our empirical investigation of the relationship between fiscal decentralization and government
size is based on annual data for a panel of 28 European countries during the period of 1990-2011.
Due to the data availability problems we work with an unbalanced panel, i.e. the time dimension
is not equal for all the cross-sections. Specifically, the panel consists of the following countries
with the respective time periods given in the parentheses: Austria (1991-2011), Belgium (19912011), Bulgaria (1995-2011), Croatia (2002-2011), Czech Republic (1995-2011), Denmark
(1991-2011), Estonia (1995-2011), Finland (1991-2011), France (1991-2011), Germany (19912011), Hungary (1995-2011), Iceland (1991-2011), Ireland (1995-2011), Italy (1995-2011),
Luxemburg (1991-2011), Latvia (1994-2011), Lithuania (1995-2011), Netherlands (1991-2011),
Norway (1991-2011), Poland (1995-2011), Portugal (1995-2011), Romania (1995-2011),
Slovakia (1996-2011), Slovenia (1995-2011), Spain (1995-2011), Sweden (1991-2011),
Switzerland (1991-2011), and United Kingdom (1991-2011). In order to discover whether there
are any differences in the fiscal decentralization-government size link between the industrialized
European countries and the “new” EU member states (the former communist economies from
CEE) we divide the whole sample into two sub-samples.1
Our principal research goal is to provide an empirical test of the Leviathan hypothesis, i.e. to find
out whether fiscal decentralization is associated with small government size across the European
countries. As for the dependent variable, the government size (size), we measure it by the share
of total public expenditure (general government) in GDP. We employ two measures of the extent
of fiscal decentralization: expenditure decentralization (locexp), i.e. the share of sub-national
government expenditure in total government expenditure (general government), and revenue
decentralization (locrev), i.e. the share of sub-national government revenue in total government
revenue. In addition, we like to test the relevance of the common pool hypothesis, i.e. to
investigate whether intergovernmental grants lead to a larger government. Here, we measure the
fiscal decentralization by the size of the vertical fiscal imbalance as proxied by two indicators:
first, a variable (grant) measuring the importance of central government grants as a source of
financing local government, i.e. the share of intergovernmental transfers in total sub-national
government revenues; second, a variable (autonom) representing the extent of the localgovernment tax autonomy, i.e. the ratio of sub-national tax revenues to total sub-national
expenditure. The data for the public expenditure/GDP ratio were extracted from the EUROSTAT
1
The CEE sub-sample consists of the following 11 countries: Bulgaria, Croatia, The Czech Republic, Estonia,
Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia. The other 17 countries form the developed
countries sub-sample.
database, while all the fiscal decentralization data were taken from the March 2014 issue of the
World Bank’s Fiscal Decentralization Indicators Database, which draws on the International
Monetary Fund’s Government Finance Statistics. Further on, we use several variables in the
empirical model, which serve to control for various demographic, geographic, institutional,
political, and economic factors that exert influence on the government size, such as: the level of
development, population density, urbanization, dependent population, trade openness, public
debt, and inflation. Additionally, we have constructed a dummy variable for the effects of the
Global Financial Crisis on government expenditure. In what follows we explain the rationale for
the inclusion of these variables in the regression:
According to the so-called Wagner law, government size is associated with the degree of
economic development (Peacock and Wiseman, 1961). The Wagner law implies that the income
elasticity of the demand for public goods is greater than unity, which means that the higher level
of income leads to a larger public sector. Therefore, in order to test for the Wagner law we
include the GDP per capita (gdp) as a proxy for the degree of economic development with a
positive regression coefficient. There is an obvious relationship between government size and
population density (the number of inhabitants per one square kilometer) as the latter determines
the marginal costs for providing the public goods. Consequently, we expect a negative
association between this variable (dens) and government size. Similarly, the dependency ratio
(the share of the population aged less than 14 and over 65 in the total population) influences the
demand for various fiscal transfers, such as: pensions, health care, social welfare etc. Therefore,
we include this variable (depend) in the empirical model with an expected positive sign of the
regression coefficient. The degree of urbanization (urban), measured as the share of urban
population in total population affects both the demand for public goods and the costs of their
supply. On the one hand, both the quantity and the quality of public goods increase with
urbanization while, on the other hand, urbanization reduces the costs of supplying public goods
through the effects on population density. Hence, the a priori sign of this variable is ambiguous.
Further on, Rodrick (1998) suggests that higher trade openness leads to a greater uncertainty,
thus increasing the demand for public goods because of the need for the government to offset the
effects of various external shocks. Consequently, small open economies characterized by high
trade/GDP ratios are expected to have larger government. On the other hand, international
competition and globalization puts a pressure for increased efficiency of the public sector, too,
thus, leading to small government size. As a result, the overall effect of the trade openness
variable (open) on government size is ambiguous. Also, we include two macroeconomic
variables in the empirical model, the public debt (debt) as well as the inflation rate (infl). As for
the former, the expected regression coefficient is positive due as the larger public debt is
associated with higher expenditures related with debt servicing. As for the latter, the relationship
between inflation and government size is ambiguous due to the following two opposite effects:
on the one hand, high inflation may erode tax revenues (in the case of a delay in revenue
collections or the absence of indexation mechanism) thus restraining government size; on the
other hand, in the presence of progressive taxation higher inflation leads to a drift in the tax base,
thus increasing tax revenues. Finally, in order to capture the effects of the Global Financial Crisis
on government expenditure (for instance, due to the massive costs associated with the
rehabilitation of the banking sector) we construct a slope dummy variable (crisis), which takes
the value of one during 2009-2011 and the value of zero otherwise. The expected regression
coefficient of the dummy is positive.
5.2 Specification of the empirical model
We analyse the relationship between government size and fiscal decentralization by means of a
fixed-effects panel data model, which seems to be more appropriate when working with macro
panels, especially when the cross-sections are not sampled randomly and when the research
focuses on the behaviour of the specific sample without drawing inferences about the whole
population. In addition, the fixed-effects estimator is consistent even when individual effects are
correlated with the regressors (Baltagi, 2008). In these regards, the assumption that the regressors
are not correlated with the disturbance term, which is critical for employing the random effects
model, seems to be a priori unrealistic (Wooldridge, 2002) as many of the regressors included in
the model may be correlated with the unobserved country-specific effects. For instance,
urbanization is associated with the country’s geography and history; the level of economic
development depends on the various country-specific cultural and institutional factors; the
dependent population is affected by the demographic trends in a country; inflation may reflect
the society’s aversion etc. Formally, we base our choice of the fixed-effects vis-á-vis the
random-effects model on the Hausman-test (Hausman, 1978), which in each case rejects the null-
hypothesis that the regressors and the disturbances are not correlated.2 In addition, our preference
for the fixed-effects model is supported by the results of the F-test for the joint significance of
the fixed effects, which are shown at the bottom of Table 1.
The empirical model has the following general specification:
yit = αi + γzit + xit β' + uit
(1)
where:
- y is the dependent variable (size);
- z represents the various alternative measures of fiscal decentralization (locexp, locrev,
autonom, and grant);
- x is a k-dimensional vector of explanatory control variables (urban, depend, dens,
open, gdp, debt, infl, and crisis);
- α , γ and β are the constant, the parameter before the fiscal decentralization variable
and the k-dimensional vector of parameters of the control variables, respectively;
- u are the residuals;
- i and t are the country and time subscripts, respectively.
5.3 Discussion of the regression results
Table 1 shows the estimates of the empirical model with the local expenditure as a measure of
fiscal decentralization. As can be seen, the decentralization variable turns out to be highly
statistically significant in the whole sample (at 1% level of significance) as well as the two subsamples consisting of the developed countries and CEE countries (the p-values are 1.7% and
1.4%, respectively). Also, the regression coefficient has a negative sign, suggesting that fiscal
decentralization indeed is associated with smaller government as suggested by the Leviathan
hypothesis. Its magnitude ranges from -0.0825 for the developed countries to -0.2113 for the
CEE countries, thus, implying non-negligible effects on the government size. This is especially
true for the former transition economies where the decentralization seems to offer the greatest
benefits. This result may be explained with the initially high level of centralization in the former
communist countries in which the public sector was large and highly inefficient. Under these
conditions, decentralizing government activities (accompanied by widespread reforms towards
2
The results of the Hausman-test are available from the authors upon request.
democratization of the society and introducing market economy) leads to an increase in the
efficiency in the provision of public goods and smaller government (Shah, 2004). In this regard,
the higher regression coefficient of the decentralization variable in the CEE sub-sample clearly
reflects the fact that roughly half of the time period refers to the transition phase when most of
the political, institutional and economic reforms had been implemented. As for the developed
countries, although we find a negative relationship between government size and fiscal
decentralization, the magnitude of the regression coefficient is much lower, suggesting that the
benefits of decentralization are modest in the countries with more efficient public sector.
Now we turn briefly to the estimates of regression coefficients of the various control variables
included in the empirical model. Two of them (infl and gdp) are found to be statistically
insignificant in both the whole sample as well as the sub-samples. In fact, the gdp is significant
at 10% significance level in the developed countries sample, but the magnitude of the regression
coefficient is extremely small in all the three regressions, suggesting that the level of economic
development has negligible effects on government size in the European countries. In other
words, we cannot provide empirical evidence in favour of the Wagner law. This result might
reflect the non-linear effects of the economic development on government size, i.e. it might be
possible that above some threshold level of economic development the further increase in
income per capita does not have important effects on the demand for public goods. As expected,
we find a positive and statistically significant association between public debt and government
size in all the three samples, implying that the level of public debt has important effects on
government expenditure. The regression coefficient is larger in the developed countries subsample where a ten percentage point increase in the public debt leads to a 1.5 percentage point
higher government expenditure/GDP ratio. The effects of the public debt are non-negligible in
the CEE sub-sample, too, where a ten percentage point increase in the public debt leads to almost
one percentage point higher government expenditure/GDP ratio. The larger regression
coefficient in the developed countries sub-sample probably reflects the considerably higher
indebtedness in comparison to the CEE countries.
Further on, we find statistically significant and economically important effects of trade openness
on government size. Here, the regression coefficient is negative, suggesting that the small
countries, which are exposed to the forces of globalization and international competition, cannot
obey the pressure to increase the efficiency of the public sector. Alternatively, this result could
be explained by the negative association between country size and trade openness associated
with the cultural and ethnic diversity (Alesina and Wacziarg, 1998). Population density is highly
significant in the three samples, but the sign of the regression coefficient differs across the
samples: it is positive in the whole sample and the developed countries sub-sample while it is
negative in the CEE countries sub-sample. The developed countries are able to provide the same
quantity and quality of public goods to all areas notwithstanding whether they are populated or
not. As a result, in these countries, population density does not reduce the government size
through the effects of marginal costs for the supply of public goods. On the other hand, we find
statistically significant and negative association between urbanization and government size in the
full sample and the developed countries sub-sample. As for the variable depend, it is significant
in the two sub-samples, though it has the expected positive sign only in the CEE countries
whereas it is negative in the developed. These divergent effects of depend probably reflect the
differences in both the composition of budget expenditure and the institutional characteristics
across the two groups of countries. For instance, as a result of the underdeveloped financial
markets as well as the lower level of income, private pension savings are lower in the CEE
region, so that pensions consume a large portion of government expenditure in comparison with
the developed countries. Finally, Table 1 reveals the strong effects of the Global Financial Crisis
across Europe as many of the countries included in the sample have seen large increase in the
government expenditures related to the massive costs associated with the rehabilitation of their
vulnerable banking sectors.
Table 1: Government size and decentralization of public expenditure
Variables
Whole sample
Developed
countries
CEE countries
constant
87.9461***
(7.8352)
-0.1541***
(0.0314)
0.1213***
(0.0122)
0.0000
(0.0000)
0.0654
(0.0499)
114.8096***
(8.9133)
-0.0825**
(0.0343)
0.1512***
(0.0144)
0.0000*
(0.0000)
0.0124
(0.0685)
47.6657**
(19.0416)
-0.2113**
(0.0849)
0.0995***
(0.0222)
0.0000
(0.0000)
0.0418
(0.0590)
locexp
debt
gdp
infl
-0.0949***
(0.0147)
0.1441***
(0.0399)
-0.9385***
(0.1273)
-0.6458***
(0.2147)
2.8327***
(0.5504)
83.40
(0.0000)
R2
-0.0755***
(0.0112)
0.0975***
(0.0342)
-0.6523***
(0.1113)
-0.1501
(0.1842)
3.0176***
(0.4203)
66.36
(0.0000)
0.4641
0.5519
-0.0608***
(0.0156)
-0.4956***
(0.1890)
0.0042
(0.2273)
1.2552***
(0.3773)
2.0448***
(0.5621)
30.54
(0.0000)
0.5371
Cross-sections
28
17
11
Observations
478
322
156
open
dens
urban
depend
crisis
F-test
Note:
1. ***/**/* denotes significance at 1%, 5% and 10% level of significance, respectively.
2. F-test for the significance of the fixed effects (p-value in the parentheses).
Table 2 shows the estimates of the empirical model with the local revenue as a measure of fiscal
decentralization.3 Here, the results differ widely across the three samples: the revenue
decentralization variable is statistically significant only in the developed countries sub-sample
where the regression coefficient is positive. Though it is positive in the full sample, too, its
magnitude is extremely small (virtually zero) and it is not significant. On the other hand, this
variable carries a negative sign in the CEE countries sub-samples, but it is not statistically
significant. Therefore, employing this measure of fiscal decentralization fails to conform to the
previous set of results, i.e. it does not provide support to the Leviathan hypothesis. We suspect
that these findings probably reflect the fact that the local revenue data do not adequately reveal
the true extent of fiscal decentralization, because significant portion of the local government
revenue comes from various forms of tax-sharing and/or fiscal transfers. Moreover, local
governments usually do not have full discretion over the “pure” local tax revenues, such as
property taxes, whose tax rates are normally determined by law. Therefore, even when we use
3
Alternatively, we have employed another measure of fiscal decentralization - local tax revenue, yielding virtually
the same estimates of the regression parameters. This set of results is available from the authors.
the local tax revenue as a measure of fiscal decentralization we obtain virtually the same
regression estimates.
Table 2: Government size and decentralization of public revenue
Developed
countries
Variables
Whole sample
constant
105.7489***
(8.9133)
0.0802**
(0.0377)
0.1761***
(0.0137)
0.0000**
(0.0000)
0.0095
(0.0687)
-0.1058***
(0.0143)
0.1381***
(0.0400)
-0.8903***
(0.1297)
-0.6139***
(0.2153)
2.2167***
(0.5610)
83.74
(0.0000)
R2
81.6261***
(8.4863)
0.0000
(0.0349)
0.1435***
(0.0121)
0.0000
(0.0000)
0.0883*
(0.0477)
-0.0841***
(0.0114)
0.0873**
(0.0350)
-0.6328***
(0.1157)
-0.0845
(0.1896)
2.7594***
(0.4414)
62.39
(0.0000)
0.4358
0.5500
50.4440***
(19.0416)
-0.0188
(0.0788)
0.1081***
(0.0225)
0.0000
(0.0000)
0.0695
(0.0521)
-0.0623***
(0.0158)
-0.6870***
(0.1816)
-0.0841
(0.2273)
1.6967***
(0.3738)
2.1279***
(0.5826)
31.01
(0.0000)
0.5210
Cross-sections
28
17
11
Observations
479
322
157
locrev
debt
gdp
infl
open
dens
urban
depend
crisis
F-test
CEE countries
Note:
1. ***/**/* denotes significance at 1%, 5% and 10% level of significance, respectively.
2. F-test for the significance of the fixed effects (p-value in the parentheses).
As for the regression coefficients of the control variables included in the empirical model,
generally they retain the signs and the statistical significance as before. Again, gdp is significant
at 10% significance level in the developed countries sample, but the magnitude of the regression
coefficient is extremely small in all the three regressions, suggesting negligible effects on
government size. The regression coefficient of infl is positive in the three samples, but it is
statistically significant at 10% only in the full sample. As for the regression coefficients of debt,
open, depend, and crisis, the estimates are virtually the same as in the previous set of results. The
same is true for the magnitude and statistical significance of the variable dens. The estimates of
urban remain unchanged in the full sample as well as the developed countries sub-sample, but
here we obtain a negative coefficient in the CEE sample, too (though it is not significant).
Table 3: Government size and local tax autonomy
Developed
countries
Variables
Whole sample
constant
115.0029***
(9.2232)
-0.0316
(0.0274)
0.1677***
(0.0130)
0.0000*
(0.0000)
0.0190
(0.0689)
-0.1079***
(0.0148)
0.1434***
(0.0401)
-0.9597***
(0.1291)
-0.6334***
(0.2153)
2.4552***
(0.5468)
66.85
(0.0000)
R2
82.7754***
(8.0358)
-0.0108
(0.0140)
0.1442***
(0.0117)
0.0000
(0.0000)
0.0787
(0.0512)
-0.0862***
(0.0116)
0.0898**
(0.0351)
-0.6532***
(0.1166)
-0.0671
(0.1890)
2.6797***
(0.4358)
66.36
(0.0000)
0.5765
0.5452
60.7212***
(20.3692)
-0.0182
(0.0140)
0.1097***
(0.0223)
0.0000
(0.0000)
0.0564
(0.0597)
-0.0655***
(0.0159)
-0.7672***
(0.1735)
-0.2382
(0.2500)
1.9150***
(0.3476)
1.8852***
(0.5865)
31.72
(0.0000)
0.5220
Cross-sections
28
17
11
Observations
479
323
156
autonom
debt
gdp
infl
open
dens
urban
depend
crisis
F-test
CEE countries
Note:
1. ***/**/* denotes significance at 1%, 5% and 10% level of significance, respectively.
2. F-test for the significance of the fixed effects (p-value in the parentheses).
As already mentioned, besides the Leviathan hypothesis, we aim to test the relevance of the
common-pool hypothesis, i.e. to investigate whether intergovernmental grants lead to a larger
government. To this end we employ the vertical fiscal imbalance as a measure of fiscal
decentralization. Table 3 and Table 4 present the estimates from the regression with the localgovernment tax autonomy (autonom) and intergovernmental grants (grants) as explanatory
variables. Both variables serve as proxies for the fiscal imbalance as the former shows the
proportion of local expenditure financed by local tax revenue while the latter emphasize the
reliance on central government grants as a source of finance. The greater degree of tax autonomy
means that local governments rely predominantly on their own sources of tax revenue, which
reduces the room for common-pool financing arrangements. Therefore, one expects a negative
association between autonom and size.
Table 3 reveals that the regression coefficient of the variable autonom has the expected sign in
the three samples, but it is not statistically significant. As can be seen from Table 4, similar
results are obtained for the second measure of the vertical fiscal imbalance, the variable grant,
which has the “correct” sign in the three regressions, but it is significant only in the CEE country
sample (with a p-value of 6.4%). Hence, we are able to provide empirical support to the
common-pool hypothesis only for this group of countries where the reliance on tax-sharing,
central government grants and other forms of intergovernmental fiscal transfers indeed leads to
larger government. In these regards, one may argue that the absence of strong institutions and
rule-based criteria in the allocation of fiscal transfers give rise to the common-pool phenomenon
in these countries. Finally, concerning the control variables included in the empirical model,
Table 3 and Table 4 reveal that both their sign and the magnitude have remained virtually
unchanged as in the previous specifications, so that we will not discuss them.
Table 4: Government size and central grants
Variables
Whole sample
Developed
countries
CEE countries
constant
80.7377***
(7.9882)
0.0138
(0.0163)
110.4820***
(9.1277)
0.0277
(0.0324)
55.8761***
(19.0679)
0.0310*
(0.0166)
grant
0.1680***
(0.0130)
0.0000*
(0.0000)
0.0128
(0.0690)
-0.1059***
(0.0146)
0.1438***
(0.0402)
-0.9303***
(0.1287)
-0.6416***
(0.2158)
2.4952***
(0.5450)
75.88
(0.0000)
R2
0.1447***
(0.0117)
0.0000
(0.0000)
0.0889*
(0.0476)
-0.0859***
(0.0115)
0.0940***
(0.0358)
-0.6454***
(0.1149)
-0.0701
(0.1886)
2.6888***
(0.4343)
59.44
(0.0000)
0.4368
0.5443
0.1108***
(0.0221)
0.0000
(0.0000)
0.0698
(0.0513)
-0.0659***
(0.0158)
-0.7305***
(0.1660)
-0.2540
(0.2403)
1.9382***
(0.33379)
1.8986***
(0.5707)
32.24
(0.0000)
0.5327
Cross-sections
28
17
11
Observations
480
323
157
debt
gdp
infl
open
dens
urban
depend
crisis
F-test
Note:
1. ***/**/* denotes significance at 1%, 5% and 10% level of significance, respectively.
2. F-test for the significance of the fixed effects (p-value in the parentheses).
6. Conclusions
This paper provides for an empirical study of the association between fiscal decentralization and
government size on a panel of 28 European countries during 1990-2011, estimated by the fixedeffects model. Our main research task is to test empirically the two standard hypotheses
concerning the effects of fiscal decentralization: the Leviathan hypothesis and the common-pool
hypothesis. To this end, we regress the government expenditure (as % in GDP) on several fiscal
decentralization variables (local expenditure, local revenue, and vertical fiscal imbalance),
controlling for the effects of various demographic, institutional, and macroeconomic variables,
such as: the level of economic development, trade openness, population density, urbanization,
dependent population, the level of public debt, and inflation. In addition, we take into account
the effects of the Global Financial Crisis on government expenditure. In order to check for the
robustness of the results, we run three separate regressions, covering the full sample as well as
two sub-samples: one that includes the developed countries and another one consisting of the
former transition economies.
The main findings from our investigation are as follows: The expenditure decentralization
variable is highly statistically significant in the whole sample as well as the two sub-samples.
The regression coefficient has a negative sign, suggesting that fiscal decentralization indeed is
associated with smaller government as suggested by the Leviathan hypothesis. We provide
evidence for non-negligible effects of expenditure decentralization on government size,
especially in the CEE countries. However, when we employ the revenue decentralization as an
explanatory variable we cannot provide support to the Leviathan hypothesis.
We suspect that these findings probably reflect the fact that the local revenue data do not
adequately represent the true extent of fiscal decentralization, because significant portion of the
local government revenue comes from various forms of tax-sharing and/or fiscal transfers.
Therefore, we include two measures of the vertical fiscal imbalance focusing on the composition
of local government finance, i.e. tax autonomy versus intergovernmental fiscal transfers. Here,
we provide empirical support to the common-pool hypothesis only for the CEE countries where
the reliance on tax-sharing, central government grants and other forms of intergovernmental
fiscal transfers indeed leads to larger government.
As for the effects of the control variables, our research results suggest that higher public debt
leads to larger government, while trade openness is associated with smaller government size.
Also, we find that the effects of population density and dependent population on government size
differs between the developed and the CEE countries, while higher degree of urbanization
reduces government size only in the developed countries sub-sample. Finally, we confirm that
the Global Financial Crisis has had strong effects on the level of government expenditure across
Europe.
References
Alesina, Alberto, and Wacziarg, Romain (1998), “Openness, country size and government”,
Journal of Public Economics, 69(3), pp. 305-321.
Baltagi, Badi H. (2008), Econometric Analysis of Panel Data (fourth edition). Chichester, UK:
John Wiley & Sons.
Baskaran, T., (2011), “Fiscal decentralization, ideology and the size of government”, European
Journal of Political Economy, 27(3), pp. 485-506.
Bordignon, M., and Turati, G. (2009), “Bailing out expectations and health expenditure”,
Journal of Health Economics, 28(2), pp. 305-321.
Brennan, Geoffrey, and Buchanan, James M. (1980), The power to tax: Analytical foundations of
a fiscal constitution. New York: Cambridge University Press.
Buchanan, J. M., and Wagner, R. E. (1977), Democracy in deficit. New York: Academic Press.
Cassette, A., and Paty, S. (2010), “Fiscal decentralization and the size of government: a
European country empirical analysis”, Public Choice, 143(1-2), pp. 173-189.
Ebel, R. and Yilmaz, S. (2002), “On the Measurement and Impact of Fiscal Decentralisation”,
World Bank Policy Research Working Paper Series, No. 2809.
Ehdaie, J. (1994), “Fiscal decentralization and the size of government: an extension with
evidence from cross-country data”, World Bank Policy Research Working Paper Series, No.
1387.
Eyraud, L., and Lusinyan, L. (2011), “Decentralizing Spending More than Revenue: Does It Hurt
Fiscal Performance?”, IMF Working Papers, No. 226.
Feld, Lars P., Gebhard Kirchgässner, and Christoph A. Schaltegger (2010). “Decentralized
taxation
and
the
size
of
government:
Evidence
from
Swiss
state
and
local
governments”, Southern Economic Journal, 77(1), pp. 27-48.
Fiva, J. (2006), “New Evidence on the Effects of Fiscal Decentralization on the Size and
Composition of Government Spending”, FinanzArchiv/Public Finance Analysis, 62(2), pp. 250280.
Forbes, Kevin F., and Ernest M. Zampelli (1989), “Is Leviathan a mythical beast?”, American
Economic Review 79(3), pp. 568-577.
Grossman, J. P. (1989), “Fiscal Decentralization and government size: an extension”, Public
Choice, 62(1), pp. 63-69.
Grossman, J. P., and West, G. E. (1994), “Federalism and the growth of government: revisited”,
Public Choice, 79(1-2), pp. 19-32.
Hausman, J., A. (1978), “Specification Tests in Econometrics”, Econometrica, 46(6), pp. 12511271.
Heil, J. B. (1991), “The search for Leviathan revisited”, Public Finance Review, 19(3), pp. 334346.
Jin, J. and Zou, H. (2002), “How Does Fiscal Decentralisation Affect Aggregate, National and
Subnational Government Size?”, Journal of Urban Economics, 52(2), pp. 270-293.
Josselin, Ј. М.; Padovano, Ј., and Rocaboy, Y. (2012), “Fiscal rules vs. political culture as
determinants of soft budget spending behavior, Evidence from Italian and French regions”, IREF
Working Paper No. 1/2012.
Joulfaian, D., and Marlow, M. L. (1991), “Centralization and government competition”, Applied
Economics, 23(10), pp. 1603-1612.
Joumard, I., and Kongsrud, P. M. (2003), “Fiscal relations across government levels”, OECD
Economic Studies, No.1, pp. 155-229.
Marlow, M. L. (1988), “Fiscal Decentralization and Government Size”, Public Choice, 56(3), pp.
259-269.
Martinez-Vazquez,
J.,
and
Yao,
M.
(2009),
“Fiscal
Decentralization
and
Public
SectorEmployment: A Cross-Country Analysis”, Public Finance Review, 37(5), pp. 539-571.
Musgrave, R. A., (1959), Theory of public finance. New York: McGraw Hill.
Nelson, M. A. (1986), “An Empirical Analysis of State and Local Tax Structure in the Context of
the Leviathan Model of Government”, Public Choice, 49(3), pp. 283-294.
Oates, W. E. (1972), Fiscal Federalism. New York: Harcourt Brace Jovanovich, Inc.
Oates, W.E. (1985), “Searching for Leviathan: An empirical study”, American Economic
Review, 75(4), pp.748-757.
Peacock, Alan T., and Wiseman, Jack (1961), The growth of public expenditure in the United
Kingdom, Princeton: Princeton University Press.
Prohl, S. and Schneider, F. (2009), “Does Decentralization Reduce Government Size? A
Quantitative Study of the Decentralization Hypothesis”, Public Finance Review, 37(6), pp. 639664.
Rodden, J. (2003), “Reviving Leviathan: Fiscal Federalism and the Growth of Government”,
International Organization, 57(4), pp. 698-729.
Rodrick D. (1998), “Why Do More Open Economies Have Bigger Governments?”, Journal of
Political Economy, 106(5), pp. 997-1032
Shadbegian, R. J. (1999), “Fiscal Federalism, Collusion, and Government Size: Evidence from
the States”, Public Finance Review, 27(3), pp. 262-281.
Shah, A. (2004), “Fiscal decentralization in developing and transition economies: progress,
problems, and the promise”, World Bank Policy Research Paper, No. 3282.
Stegarescu, D. (2005), “Public Sector Decentralisation: Measurement Concepts and Recent
International Trends”, Fiscal Studies, 26(3), pp. 301-333.
Stein, Ernesto H. (1998), “Fiscal Decentralization and Government Size in Latin America”,
Working Paper, Inter-American Development Bank, Office of the Chief Economist, No. 368.
Wallis, J. J., and Oates, E. W. (1988), “Decentralization in the public sector: an empirical study
of state and local government”, in Harvey S. Rosen, ed., Fiscal Federalism: Quantative Studies.
Chicago: University of Chicago Press, pp. 5-32.
Wooldridge, J. (2002). Econometric Analysis of Cross Section and Panel Data. Cambridge, MA:
MIT Press.