Working Paper (Delovni zvezek)
SIZE OF BUDGETARY AND NON-BUDGETARY GOVERNMENT:
EXPLAINING CROSS-COUNTRY DIFFERENCES
VELIKOST FISKALNEGA IN NEFISKALNEGA DELA VLADNEGA
SEKTORJA: RAZLAGA RAZLIK MED DRŽAVAMI
Dr. Primož Pevcin
Univerza v Ljubljani
Fakulteta za upravo
1000 Ljubljana, Slovenija
The purpose of the analysis presented in the paper is to identify various economic, social, political, demographic and cultural factors that could shape the differences in the size of government across countries and to verify the effect of those factors with the use of econometric analysis. The analysis focuses on ”budgetary” government, usually measured
with some government spending ratio, as well as on ”non-budgetary” government, measured
with the index of the extent of regulation in the economy. The results of the analysis revealed that economic factors are more important in explaining the variation in the size of government consumption spending and in the size of non-budgetary government, whereas political, social and cultural factors are more important in explaining the variation in the size of transfer spending. Besides, the results also indicate, that the size of budgetary government and the size of non-budgetary government move in the opposite direction.
Key words: budgetary and non-budgetary government, factors affecting differences in size, cross-country empirical analysis, transmission mechanism of government activities;
Namen analize predstavljene v prispevku je s pomočjo empirično preveriti določene teoretično utemeljene ekonomske, politične, družbene, demografske in kulturne dejavnike, ki naj bi povzročali razlike v velikosti vladnega sektorja med državami, pri čemer se analiza usmerja tako na velikost ”fiskalnega” dela vladnega sektorja, merjenega z deležem določene
kategorije vladnih izdatkov v bruto domačem proizvodu, kot ”nefiskalnega” oziroma ”tihega”
dela vladnega sektorja, merjenega z makro obsegom regulacije zasebnega sektorja. Splošna
ugotovitev analize je, da so ekonomski dejavniki pomembnejši pri analizi obsega regulacije in vladnih izdatkov za nakupe dobrin in storitev, medtem ko so ostali dejavniki pomembnejši pri analizi transfernih izdatkov. Poleg tega rezultati analize tudi kažejo, da se očitno velikosti
fiskalnega dela in nefiskalnega dela vladnega sektorja gibljeta v obratni smeri.
Ključne besede: fiskalni in nefiskalni vidik vladnega sektorja, dejavniki razlik v velikosti, empirična analiza razlik med državami, transmisijski mehanizem vladnih aktivnosti;
One of the major current issues in the social sciences is the expansion of government in recent times. Namely, absolute and relative size of government sectors grew rapidly in the majority
1 It is worth noting that the literature on the of industrialised countries over the past century.
determinants affecting the growth in size of government is quite comprehensive and numerous explanations have been developed (e.g., see Lane, 1995; Tanzi and Schuknecht, 2000; Bailey, 2002). However, growth in government spending and regulation was not fully symmetrical across countries, indicating that large cross-country differences in the size of
2government exist nowadays, even among developed countries. Consequently, the main
question that ought to be answered in the paper is, why such differences exist in the size of government and which factors shape those differences. In other words, the answer on the question concerning the reasons behind the fact why one country is economically more laissez-faire than the other should be provided, primarily to gain some insight into modern ”economic theory of government”. Therefore, the aim of the analysis presented in the
paper is to theoretically identify various economic, political, social, demographical and cultural factors influencing the variations in the size of government across countries and to empirically verify the effect of those factors on the size of government with the use of multiple regression analysis.
2. Measuring the size of government
A variety of ways exist in the literature by which the size of government is measured. These measures are usually spending based, typically with total general government expenditures or
3general government consumption expenditures. Still, the main relative measure of the size of
government is the government expenditures ratio (total government outlays as a percentage of gross domestic product) and should characterise the extent of government utilisation of domestic resources. Total government expenditures include spending on general government
4consumption, transfers and subsidies, net interest on outstanding debt, and capital goods.
The main advantage of using total government spending ratio as a measure of the size of government is that it expresses the size (and scope) of government in one single index number.
1 For the industrialised countries, the share of government spending in GDP grew, on average, from about 12 percent before World War I to about 45 percent in the middle of 1990s (Tanzi, 1997). 2 For instance, based on OECD (2001) and EBRD (2001) data, in South Korea total general government expenditures were approximately 20 percent of GDP in 2000. That share was approximately 30 percent in Ireland and something below 40 percent of GDP in the United Kingdom. Slovenia, as an example of the developed transition economy, had government spending above 40 percent of GDP. However, government appropriated even more than 50 percent of GDP in countries like Denmark and Sweden. 3 The latter is used more frequently as it is usually available for a wider range of countries. However, this measure only encompasses certain aspect of government activities and not all of them. 4 Some cross-country studies use central government expenditures as a share of GDP to measure the size of government. However, as stated by Gwartney, Lawson and Holcombe (1998), those figures can often be highly misleading, because the central government figures understate the size of government for countries where substantial expenditures are undertaken at lower levels of government.
However, a reservation exists in practice about the focus on government spending ratio as a measure of the size of government. Namely, government expenditure ratio only includes those government activities, which result in financial flows in government accounts. As Posner (1971) stresses, fiscal instruments are only one of two instruments for executing public (government) policy, the other instrument being regulation. Moreover, Leonard (1986) argues that government budgets will understate the true size of the government sector by not recording the so-called “quiet side” of government activity. Consequently, it can be derived that the influence of government in an economy goes beyond its spending activities and tax collections. For example, regulation of economic activities or governmental ownership of enterprises are two examples of government intervention that can have little impact on the size of government budget, but they can have profound economic effects. Accordingly, each of these “non-budget” items is equivalent to a tax and expenditure programme, meaning that government may be “large” with a relatively small expenditure. Namely, there is a real
possibility for similar government spending ratios to be associated with vastly different regulatory regimes. Therefore, it is of absolute importance to measure the size of both budgetary and non-budgetary government, since not just the government expenditure ratio but also the extent to which the government intervention in the form of various regulatory
5 demands affects the lives of citizens differentiates big from small government.
3. Determinants of the size of government: theoretical foundations
3.1.Size of budgetary government
As already mentioned, the literature on the determinants affecting the growth in size of government spending is quite comprehensive. On the contrary, the literature on the differences in size of government across countries and the factors behind is not so comprehensive. The more recent literature basically focuses more on the political economy determinants, which include the following:
; Size of country, trade openness, and the degree of integration in the world economy.
For example, Alesina and Warcziarg (1998) argue that the size of government correlates
negatively with country size and positively with trade openness, contrary to what most
economists would expect. They have shown that smaller countries have a larger share of
government consumption in GDP, and are also more open to trade. Moreover, they argue
that these empirical observations are consistent with recent theoretical models explaining
country formation and break up. Namely, larger countries can ”afford” to be closed, while
small countries face stronger incentives to remain open. Conversely, as trade liberalises,
5 In this context, Armey (1995) stresses that like the concept of the invisible hand of the market is well understood, also the full anatomy of the political economy should be understood – especially the invisible foot of
the government. Namely, government has the power to stimulate the economy but at the same time it can crush economic performance and incentives by excessive regulation and spending activities. Especially regulatory activities represent the kind of invisible foot that can easily crush everything and everybody it touches (steps on).
regional and cultural minorities can ”afford” to split because political borders do not
6 identify the size of market.
Similarly, Rodrik (1998) finds a strong positive association between openness and
government size. According to him, this is some sort of a paradox since it is often
assumed that closed economies have more interventionist governments, as trade
restrictions themselves are a form of government intervention. Rodrik explains this
paradox by arguing that government expenditures are used to provide social insurance
against the risk of terms of trade shocks that open economies face, meaning that
government spending obviously plays a risk-reducing role in economies exposed to a
significant amount of external risk. He shows that higher levels of government
consumption would be associated with reduced volatility in income flows. These findings
also suggest that the international economic liberalism obviously gives government a
7central role and Ruggie (1982) calls this ”the compromise of embedded liberalism”.
Moreover, Eichner and Wagener (2002) argue that the optimal size of welfare state is
larger the higher are the risks that it cannot insure (so-called background risks). This view
highly challenges classical liberal expectations that the combination of globalisation of
the economies with the onset of new technologies will transform modern governments
back into a night-watchman state. Nevertheless, according to Wolf (2001), constraints on
the government are becoming somewhat tighter and Leviathan (i.e., government as ever
growing monster) may have reached its limits, since governments will find it more
difficult to pursue an inflationary policy and they will be forced in future to relate the
taxes they raise to the benefits they provide.
; Level of economic development, preferences and the heterogeneity of taxpayers and
According to Wagner’s and Wilensky’s hypothesis, more developed countries should
have larger government. Besides, it can be argued that larger urbanisation rate of a
country should positively affect government spending, since urbanisation is likely to
facilitate increasing taxation and to create more demands for government spending.
Similarly, the share of dependent population positively affects demand for social services
and consequently government spending (Holsey and Borcherding, 1997). In addition,
6 Economic literature on country formation and break up is another interesting field within public sector economics. In more recent text, Alesina, Spolaore and Wacziarg (2003) show that the benefits of size decline relative to the cost of heterogeneity, meaning that the optimal size of a country declines with trade openness. This means that smaller countries can enjoy the benefits of cultural homogeneity without suffering the costs associated with small markets. 7 Since both international trade and the scope of government activity have expanded significantly in the post-war period in most countries of the world, Rodrik (1998) suggests that scaling governments down may actually harm the prospects of maintaining free trade on a global scale. In my opinion, this is highly controversial idea both from statistical and economic points of view. Namely, it is not logical to see public sector reform efforts as a serious threat to globalisation. Moreover, Rodrik does not provide the proof that causality goes from government size to the level of openness of the country – he takes the relationship between these two variables as identical in both ways.
governments should spend more in societies with relatively unequal income distribution because the median voter is poorer than the mean voter. The benefits to a median voter of redistributive spending outweigh the costs borne by such a voter of increased taxation to finance spending (Persson and Tabellini, 1999).
; Structure of government and political institutions.
Three fundamental features of political institutions are: (a) structure of government, contrasting centralised and decentralised governments; (b) the electoral rule, contrasting majoritarian (pluralitarian) and proportional electoral systems; and (c) the regime type, contrasting presidential and parliamentary regime. Basically, if taxpayers are mobile, the devolution of tax bases to sub-national governments (i.e. political decentralisation) encourages competition for tax bases and may help to reduce the size of government. Reliance on grants and transfers from higher levels of government to finance sub-national governments is associated with larger governments and fiscal imbalances at the sub-national level (de Mello, 1999).
Persson (2001) argues that presidential regimes should have smaller governments and countries with majoritarian elections should have smaller welfare-state programmes and less corruption. In addition, Alesina, Glaeser and Sacerdote (2001) argue that the two party system and the lack of proportionality in the United States electoral system created obstacles that blocked the formation of a strong and lasting Socialist party, which more generally limited the political power of the poor. The upheaval in continental Europe over the last century has meant that there were no durable institutions, which could protect property against popular demand for redistribution. This appears to be one of the explanations of difference in levels of redistribution and government expenditure between the United States and European countries.
; Social fractionalisation and ethno-linguistic-religious heterogeneity of the society. This factor should have two counteracting influences. On the one hand, Annett (2000) has revealed that higher social fractionalisation (as measured along ethno-linguistic and religious dimensions) leads exogenously to greater political instability, and governments use government consumption as a buffer against political instability. In addition, a high dependency ratio has been found to increase demand for government spending on education, health care, social security including pensions, and defence, and on programmes to satisfy regional and ethnic interests. This means that the effect of fractionalisation on government consumption is positive, but indirect, via the political instability channel. In a somewhat different fashion, Alesina, Glaeser and Sacerdote (2001) argue that the differences in level of redistribution between the United States and the European countries appear to be the result of racial heterogeneity, since racial
aversion in the United States obviously makes redistribution to the poor, who are
8 disproportionately black, unappealing to many voters.
3.2.Size of non-budgetary government
The term non-budgetary government indicates that government activities are not ”visible”, since they are not directly perceivable in the amount of government spending, yet they can be substantial and represent an enormous burden to individuals. The non-budgetary government usually appears in the form of various regulatory demands set by government in order to oversee individual behaviour. Formally, OECD definition states that regulation refers to the diverse set of instruments by which governments set requirements on enterprises and citizens (OECD, 1997). There is no generally accepted economic theory about the determinants of macro extent of government regulation. One of rare attempts to identify determinants of regulation was undertaken by Pryor (2002), although his approach was in its essence more expeditious and inductive. He identifies several plausible causal factors of determinants of
; Size of economy.
According to Pryor, this factor should have two counteracting influences. Because
production in large countries is more difficult to regulate than in small countries,
governments may be reluctant in implementing regulation so that laissez-faire and size
would be positively correlated. On the other hand, in a large country, the size of
enterprises is generally larger and regulation may be more necessary than in a smaller
country. Thus, size and laissez-faire would be negatively correlated, in some way
reflecting a desire by the citizenry in large countries for greater protection against
possible excesses than in smaller countries with smaller enterprises where social, in
contrast to legal, pressure can be exercised more effectively.
; The level of economic development.
This factor also should have counterbalancing influences. Namely, Pryor stresses that it is
commonly believed that governmental regulation increases as per capita income increases
because the government has more resources to carry out regulatory activities. However, it
is also possible at the same time that as the level of development rises, the economy
increases in complexity and governmental regulation becomes more counterproductive
and possibly unnecessary.
8 Accordingly, authors stress that the largest difference in the composition of government spending between the United States and Europe is in the area of transfers to households (including social security), and subsidies, which are the main tools for income redistribution. 9 Although Pryor empirically examined all those determinants on the sample of 18 OECD countries, statistical confirmation received only two factors, namely the size of the economy and income inequality. His results suggest that the degree of economic laissez-faire is inversely related to economic size and directly related to income inequality. He argues that this direct relationship is particularly important in political economy because it suggests that power relations between income groups are crucial in determining the extent of the regulatory regime. Still, the biggest disadvantage of his analysis is that it included only 18 units of observation.
; The relative importance of foreign trade in the economy.
Pryor’s hypothesis is that more open the economy, the less able or willing is the
government to regulate domestic economic activity. Similarly, Garen and Trask (2001)
argue that the scope of government is much larger in less open economies, because less
open economies tend to have less government expenditure, but have more government in
; Income inequality.
It was derived by Pryor that in situations with larger income inequality, the high-income
population has relatively more political power and, as a result, fewer regulations can be
imposed on their economic activities.
; Governmental ownership of enterprises.
Although the majority of economists see government ownership of enterprises and the
extent of regulation of the economy as substitutes, particularly in the relation to the
common fable that privatised government enterprises should be extensively regulated in
order to prevent them exercising their market power, Pryor argues that governmental
regulation and ownership are complements, since both of them spring from the same urge
to control private economic activity.
4. Determinants of government size: cross-country econometric analysis
4.1. Data and methodology
The aim the analysis is to identify factors influencing the variations in the size of government across countries and the relationship between budgetary and non-budgetary government. Therefore, cross-section modelling is used in order to focus upon international comparison in variations in the size of government. Cross-section econometric modelling is based on a sample of 32 developed (and democratic) countries for which cross-sectional data on all
10 The theory about size and relevant explanatory and dependent variables could be obtained.
scope of government implies that certain common factors exist which might affect the differences in the size of budgetary and non-budgetary government across countries, although the direction of influence may differ. The only factors that affect only the size of budgetary government are obviously demographic factors, which are short of the rationale that they possibly affect the extent of regulation in the economy.
The dependent variables in the analysis are the share of general government consumption spending in GDP (CON), the share of general government transfers and subsidies in GDP
10 The sample consists of 27 OECD member countries (out of 30 OECD countries Iceland, Japan and Turkey are excluded), four transitional non-OECD economies (Estonia, Latvia, Lithuania and Slovenia) and Chile as newly industrialised country and most developed country in Latin America.
(TRF) and the share of total general government expenditures in GDP (EXP) as measures of
11 Besides, the extent of total regulation of the economy the size of budgetary government.
(REG) is used as measure of the size of non-budgetary government. In addition, following explanatory variables are used in the regression analysis:
; Real gross domestic product per capita (GDPC):
Wagner’s law states that government spending both absolutely and relatively expands as
economies develop. Therefore it is to be expected that the level of GDP per capita would
positively affect the size of budgetary government, but the effect on the size of non-
budgetary government should not be exposed with reasonable certainty due to the
countervailing affect of the economic development on the extent of regulation.
; Country size (POP) and trade openness (OPN):
As already mentioned, Alesina and Warcziarg (1998) argued that government spending
correlates negatively with country size, whereas Rodrik (1998) found out positive
correlation between government size and trade openness. The effect of the country size on
the extent of regulation could not be predicted because of two countervailing affects.
However, it should be expected that trade openness would negatively affect the extent of
the regulation in the economy.
; The share of population above 65 (OLD) and below 19 (YOU):
The economic theory recognises the importance and side effects of population aging. This
involves increased demand for government spending on health care, social security, etc.
Moreover, a high dependency ratio in the form of a large share of young population
should for example increase demand for government spending on education, so it is to be
expected that both variables should positively affect government size, but there is no
rationale to predict any affect on the size of non-budgetary government.
; Urbanisation rate (URB):
The share of population living in urban areas should positively affect the size of
government spending, since urbanisation is likely to facilitate increasing taxation and to
create more demands for government spending. Similarly to the above two variables, no
effect on the extent of regulation is expected.
; Government ownership of enterprises (OWN) – the “non-budgetary” public sector:
Although public corporations are not included in general government sector, it is to be
expected that they should affect the level of government spending implicitly. Namely,
they operate in order to provide certain (public) goods and services for which probably
the most important costumer (but not the only one) is government. Therefore, it should be
11 The purpose of using three different measures of the size of budgetary government is in capturing possible different directions of influence of various explanatory variables on two main components of government spending, being consumption and transfer spending.
expected that government ownership of enterprises (i.e. the number of public corporations) would exert positive impacts on government consumption through purchases of goods and services produced by state-owned enterprises. Similarly, the positive effect is predicted in the relation to the extent of regulation in the economy, since larger number of public corporations would obviously require larger number of regulatory bodies, which are primarily financed with government consumption spending.
; Three political dichotomous dummy variables for country’s regime type (PRES), electoral
rule (PLUR) and political decentralisation (FED).
It is expected that presidential political regime should negatively affect the size of budgetary government and positively affect the size of non-budgetary government through income and openness ”channel”. According to theoretical predictions it is expected that the presence of plurality in electoral rules should positively affect the size of government consumption spending, negatively affect government transfer spending and positively affect through afore mentioned income and openness ”channel”. In addition, political decentralisation should positively affect both the size of budgetary and non-budgetary government. The reason should be very clear, since any additional
12 legislator would imply additional government intervention.
; Share of largest ethno-religious group in total population (FRAGM).
This variable indicates the level of ethno-religious homogeneity of society. It should have two countervailing effects. On the one hand, it should negatively affect government spending through political instability channel. Namely, in more heterogeneous societies government spending to different groups within society should serve as a means of increasing the political stability of a country, meaning that spending through this ”channel” should be minimised in more homogeneous countries. On the other hand, it should positively affect government spending, since in more homogeneous societies voters are likely to approve increased spending to certain social groups, because it is larger probability that they do not belong to certain ethno-religious minority. This effect is based on presumption that it is in human nature to prefer the people of the same kind. It is clear that the inclusion of this variable is appropriate only in the sample of democratic countries, where various social groups have possibility to express their demands in political process.
; Dichotomous dummy variable for countries in transition (TRA).
The purpose of this variable is to identify possible cultural or institutional differences that would imply different average size of government in this ”region”. Namely, the purpose for separate variable for transitional countries is to identify possible effects of the change in economic system, which largely reduced the role of government. It is to be expected
12 Regarding income and openness channel, Pevcin (2004) argues that the greater probability exists of more open
and more developed countries having parliamentary political regime and proportional electoral rules.