Wo r k i n g Pa p e r S e r i e s No 924 / August 2008 Government spending volatility and the size of nations by Davide Furceri and Marcos Poplawski Ribeiro WO R K I N G PA P E R S E R I E S N O 9 24 / A U G U S T 20 0 8 GOVERNMENT SPENDING VOLATILITY AND THE SIZE OF NATIONS 1 by Davide Furceri 2 and Marcos Poplawski Ribeiro 3 In 2008 all ECB publications feature a motif taken from the 10 banknote. This paper can be downloaded without charge from http://www.ecb.europa.eu or from the Social Science Research Network electronic library at http://ssrn.com/abstract_id=1188506. 1 We are grateful to an anonymous referee, Martine Carré-Tallon, Jürgen von Hagen, Ad van Riet, and participants at the ECB/DG-E seminar for helpful comments and discussions. We are also thankful to Silvia Albrizio for excellent research assistance. The opinions expressed herein are those of the authors and do not necessarily reflect those of the ECB or the Eurosystem. 2 European Central Bank and University of Palermo. European Central Bank, Directorate General Economics, Kaiserstraße 29, D-60311 Frankfurt am Main, Germany; e-mail: Davide.Furceri@ecb.europa.eu furceri@economia.unipa.it 3 CEPII and University of Amsterdam. CEPII - Centre d’etudes prospectives et d’informations internationales, 9, rue Gerges Pitard - 75740, Paris, France; e-mail: marcos.ribeiro@cepii.fr © European Central Bank, 2008 Address Kaiserstrasse 29 60311 Frankfurt am Main, Germany Postal address Postfach 16 03 19 60066 Frankfurt am Main, Germany Telephone +49 69 1344 0 Website http://www.ecb.europa.eu Fax +49 69 1344 6000 All rights reserved. Any reproduction, publication and reprint in the form of a different publication, whether printed or produced electronically, in whole or in part, is permitted only with the explicit written authorisation of the ECB or the author(s). The views expressed in this paper do not necessarily reflect those of the European Central Bank. The statement of purpose for the ECB Working Paper Series is available from the ECB website, http://www.ecb.europa. eu/pub/scientific/wps/date/html/index. en.html ISSN 1561-0810 (print) ISSN 1725-2806 (online) CONTENTS Abstract 4 Non-technical summary 5 1 Introduction 6 2 Theoretical model 8 3 Empirical strategy 13 4 Results 4.1 Government spending volatility by functional categories 16 5 Conclusions 23 References 26 Appendices 29 Tables and figures 30 European Central Bank Working Paper Series 39 22 ECB Working Paper Series No 924 August 2008 3 Abstract This paper provides empirical evidence showing that smaller countries tend to have more volatile government spending for a sample of 160 countries from 1960 to 2000. We argue that the larger size of a country decreases the volatility of government spending because it acts as an insurance against idiosyncratic shocks, and it leads to increasing returns to scale due to the higher ability of the government to spread its cost of financing over a larger pool of taxpayers. The results are robust to different time and country samples, different econometric techniques and to several sets of control variables. The analysis also evinces that country size is negatively related to the discretionary part of government spending and to the volatilities of most of the government spending items. Keywords: Fiscal Policy, Government Size, Fiscal Volatility, Country Size. JEL: E62, H10. 4 ECB Working Paper Series No 924 August 2008 Non-Technical Summary This paper provides empirical evidence showing that smaller countries tend to have more volatile government spending. From a theoretical point of view, we show that a negative relationship between government spending volatility and country size can be explained by two arguments: i) to the extent that government spending is used for fine tuning purposes, the size of a country acts as an insurance against idiosyncratic shocks, leading to a less volatile government spending; ii) increasing returns to scale of government spending originating from higher ability to spread the cost of financing it over a larger pool of taxpayer, allow the government to provide the public good in a less volatile way. These claims are confirmed by our empirical analysis, which is robust to different time and country samples, different econometric techniques and to several sets of control variables. Our results highlight the need for small countries to undertake fiscal adjustments in order to reduce macro-fiscal vulnerabilities and improve their economic growth prospects. In addition, to the extent that large fiscal areas reduce government spending volatility, our findings reinforce the role of fiscal coordination (or common fiscal policy) in monetary unions. ECB Working Paper Series No 924 August 2008 5 1. Introduction In recent years there has been a growing economic literature concentrating on the effects of scale and country size on various economic outcomes. From a theoretical point of view, the sign of these scale and size effects is ambiguous (Alesina and Spolaore, 2003). Empirically, even though Rose (2006) concludes that countries performance in terms of several indicators is not related with the size of the nation, Alesina and Wacziarg (1998) robustly show that smaller countries have higher levels of public consumption as a share of GDP1. This latter finding originates from economies of scale in the production of public goods and redistributive policies resulting from the higher ability of governments in big countries to spread the cost of financing public goods over a larger pool of taxpayers. Notwithstanding this level effect, to the best of our knowledge, the impact of the size of nations upon the volatility of government spending has not yet been discussed in the literature. Government spending volatility may be decreasing in the size of nations given that smaller economies are found to be more volatile and exposed to economic shocks (Furceri and Karras, 2007 and 2008). More specifically, we claim that a negative relationship between government spending volatility and country size can be explained by two arguments: 1) To the extent that government spending is used for fine tuning purposes, smaller economies, characterized by more volatile output and more exposure to idiosyncratic shocks, may use government spending more aggressively. 2) To the extent that public goods are of a non-rival nature, increasing returns to scale of varying government spending may originate from the higher 1 6 See, in addition, Bolton and Roland (1997), Alesina and Spolaore (2003), and Alesina et al. (2004). ECB Working Paper Series No 924 August 2008 ability to spread the cost of financing it over a larger pool of taxpayers. This promotes less volatile government expenditure in particular if public goods are, as desirable as, or more desired than private consumption. Nevertheless, other effects of country size may work in the opposite direction. For instance, more individual heterogeneity may prompt bigger political divergences in terms of preferences for type and size of public goods (O’Higgins and Ruggles, 1981), resulting in larger government spending volatility due to the switching of different political groups in power. Therefore, the objective of this paper is to analyze the empirical relationship between government spending volatility and country size. Such analysis is further motivated by the finding that higher volatility of public spending impacts negatively on economic growth and welfare (see, among others, Fatás and Mihov, 2003 and 2005; Furceri, 2007; Afonso and Furceri, 2008; and Loayza et al., 2007). Fatás and Mihov (2003), for example, estimate that every percentage point increase in volatility of discretionary fiscal policy lowers economic growth by more than 0.8 percentage points. In turn, Herrera (2007) estimates that the welfare loss of public spending volatility corresponds to 8 percent of consumption in developing countries2. Most of these effects of volatility occur via its negative impact on capital formation and investment as the theories of irreversible investment emphasize (see, in addition, Ramey and Ramey, 1995; Aghion and Banerjee, 2005; and Imbs, 2007). 2 For other analysis on the effects of public spending volatility on the welfare and capital formation of developing countries see Afonso et al. (2006) and Harberger (2005). ECB Working Paper Series No 924 August 2008 7 Our empirical analysis uses a panel data set that includes 160 countries with observations from 1960 to 2000 and yields as main results that: 1) smaller countries have more volatile government spending; 2) discretionary government spending volatility (corrected for output volatility) is decreasing with the size of nations; 3) consumption spending in Defense, Economic Affairs, Housing, Health, Recreation and Education is more volatile in smaller countries, but it is not in Public Services, Public Order, and Social Protection. These results are extremely robust to different time and country samples, different econometric techniques as well as to several sets of control variables. The rest of the paper is organized as follows. The next section presents a theoretical model that discusses the arguments linking country size and volatility of government spending. The third section describes the paper’s empirical methodology used to test for the relationship between country size and government spending volatility. The fourth section reports the results. Finally, Section 5 concludes the paper. 2. Theoretical Model This section presents a simple closed economy model based on Alesina and Wacziarg (1998), which illustrates why smaller countries could have more volatile government spending. We modify and extend that model in three ways. First, we use a different utility specification, even though our specification provides similar qualitative results as in Alesina and Wacziarg (1998). Second, we allow individual heterogeneity in consumption, by assuming a different income endowment for each consumer. That assumption introduces idiosyncratic income shocks in our model and it is useful to analyze how a bigger country can mitigate the effects of those shocks. Third, we use a two-period version of the model to compute the volatility of government spending. 8 ECB Working Paper Series No 924 August 2008 Consider a country composed of N individuals. The Social Planner maximizes the expected sum of utilities of all individuals: N 2 E t ¦¦ U i ,t i 1 t 1 Et ¦¦ E t 1 >u ci ,t  v Gt @ , N 2 (1) i 1 t 1 where Et is the expectation operator conditional on information at time t, E is the social discount factor, ci,t is the private consumption of individual i in period t, and Gt is the level of non-rival public goods in period t. The functions u and v are further assumed to be increasing in c and G, strictly concave and twice continuously differentiable. In each period households are endowed with an income level yi,t, on which they have to pay taxes. The resulting net income is assumed to be consumed at the same period, so that the individual household flow budget constraint reads: ci ,t d 1  W yi ,t , (2) where W denotes the constant and exogenous (income) tax rate. In this society each individual is further assumed to live in a distinct region that faces an idiosyncratic income shock Hi,t. Thus, in each period the stochastic income endowment is given by: yi ,t y  H i ,t , (3) where y is the average income level assumed for simplicity to be constant over time. Moreover, for every period, the income shock Hi,t is independently and identically distributed among the individuals (regions) with expected value equal to zero and standard deviation equal to VH. Hence, by the Law of Large Numbers, the country’s income shock (sum of idiosyncratic shocks) converges to its expected value the larger is the number of individuals in the country. ECB Working Paper Series No 924 August 2008 9 The government, in turn, raises tax revenues Tt and purchases goods Gt every period. For simplicity, we also assume that the government does not borrow, which makes the government’s flow budget constraint equal to3: Gt Tt œ Gt N W ¦ y i ,t . i 1 Without any further constraints and using (2) and (3), the period-by-period resource constraint in this economy reads: N ¦ y i ,t N N i 1 i 1 N ¦ c i , t  G t œ N y  ¦ H i ,t i 1 ¦c i ,t  Gt . (4) i 1 The Social Planner maximizes then (1) subject to (4) with respect to ci,t and Gt, which by assuming perfect foresight leads to: N ¦ u' c Nv' Gt . i ,t (5) i 1 This condition shows that the average marginal utility of consumption is equal to the marginal utility of government spending when welfare is maximized in this economy. Further, to assess the overall effect of changes in the population size N on government spending volatility, we resort to the following quadratic utility function: u ( x) vx Z  [  1 x 2 / 2  [x, Z ! 0 , [ ! 1 , and x  [ / [  1 , (6) where the restriction x  [ / [  1 ensures that the marginal utilities of private consumption and public spending are always positive, and the parameter Z regulates the desirability of 3 Notice that, in fact, governments use public debt management to cushion the effect of income shocks on its revenues and to keep government expenditures more stable (see Herrera, 2007). However, not all countries can rely on such instrument. Moreover, this simplifying non-borrowing assumption is useful here to test how the country size impacts on that volatility. 10 ECB Working Paper Series No 924 August 2008 public spending relative to private consumption. The higher is Z the more desirable is government expenditure compared to private consumption4. Then, using (6), we obtain from (5) that Gt 1 ZN N ¦c i ,t i 1  N Z 1 [ , ZN [  1 which by using (4) becomes5: Gt N · N Z 1 [ § . ¨ N y  ¦ H i ,t ¸  1  ZN © i 1 ¹ 1  ZN [  1 1 (7) Further, from (3) and (7) the effect of country size on the government expenditure over aggregate income (GDP) is: w Gt / Yt wN  Z 1  ZN 2  Z 1 ª [ º N Z 1 [ y . » 2 « 2 1  ZN ¬ [  1 Yt ¼ 1  ZN [  1 Yt (8) This expression is negative whenever Z t 1 and the sum of idiosyncratic shocks N ¦H i ,t is not too high6. If government spending is as desirable as, or more desirable than i 1 private consumption, then an increase in country size leads to a fall in the government spending-income ratio. As Alesina and Wacziarg (1998) discuss, an increase in country size raises the optimal level of public spending provision, which can be interpreted as an income effect; but it also reduces per capita cost of public goods for a given level of provision, allowing more private consumption (substitution effect). This latter effect comes from the higher 4 For more details and another application of equation (6) see, among others, Poplawski Ribeiro and Beetsma (2008). 5 Notice that [ can always be chosen such that equation (7) provides a larger Gt when government expenditure compared to private consumption becomes more desirable (higher Z). 6 More precisely this expression is always negative if Z t 1 and N ¦H i ,t  yZN 2 . i 1 ECB Working Paper Series No 924 August 2008 11 ability of the government to spread the cost of financing public goods over a larger pool of taxpayers (higher N) leading to increasing returns to scale. Therefore, expression (8) shows that if government expenditure is as desirable as private consumption, the substitution effect dominates and the ratio Gt /Yt falls when N increases. In addition, we can easily obtain the variance of government spending in this simple two-period model. For that, we first compute the average value of that variable: G Ny [ N Z 1 1   1  ZN [  1 1  ZN 2 1  ZN N ¦H i ,1  H i,2 , i 1 which makes the variance of government spending equal to: var G 2 1 4 1  ZN 2 ªN º «¦ H i ,1  H i , 2 » . ¬i 1 ¼ (9) Hence, the effect of an increase in N on the variance of government spending becomes: w var G wN Z  2 1  ZN 2 3 ªN º «¦ H i ,1  H i , 2 »  0. ¬i 1 ¼ (10) Equation (10) shows that the larger the country size, the lower the variance of government spending. That is due to two main effects. First, by the Law of Large Numbers, the income shocks Hi,1 and Hi,2 converge to their expected values the bigger the country size (higher N), thus moving that variance towards zero. Intuitively, larger countries are less exposed to specific idiosyncratic shocks, and therefore, government revenues and expenditures become less volatile (see also Rodrik, 1998). Moreover, it is possible to argue that, the larger the country the less exposure to “shock surprises” (Hi,1 - Hi,2) and the lower the output volatility VH see Furceri and Karras, 2007 and 2008). Second, an increase in country size eases the provision of a less volatile government expenditure, which is preferred the more desired is the public good compared to private 12 ECB Working Paper Series No 924 August 2008 consumption. That is again due to the increasing returns to scale of that non-rival good, and the consequent reduction in the per capita cost of public goods for a given level of provision when N goes up. In fact, as previously argued, if government spending is as desirable as, or more desirable then private consumption ( Z t 1 ), then an increase in the country size leads to a fall in government spending-income ratio. Similarly it is possible to see from equation (9) that an increase in the desirability of public consumption over private consumption ( Z n ) will lead to a decrease in spending volatility. In sum, our model illustrates reasons for less volatile government expenditure in larger countries, namely lower exposure to idiosyncratic risks and economies of scale in public goods provision. Nevertheless, the magnitude and the sign of the effect of country size on the volatility of government spending remains an empirical question, on which the next sections delve into. 3. Empirical Strategy Data for government expenditure is retrieved from the Penn World Table 6.2. The dataset consists of 160 countries, which had available data for each of the years from 1960 to 2000. We use the log of its population as our measure of a country’s size, and the standard deviation of annual growth of government consumption spending7 as our measure for government spending volatility8. 7 We use the annual growth rate of total government expenditure as dependent variable rather than annual growth rate of government total expenditure (or total revenue), since the latter is not available for such an extensive set of countries for a long time span. Moreover, government consumption accounts usually for approximately 4/5 of total expenditure. 8 The choice of the standard deviation of the growth rate of real government spending as measure of spending volatility could be criticized since, usually, countries with higher growth rates of government spending have higher standard deviations. An alternative measure to control for this “scale” effect could be to consider the coefficient of variation as a measure of volatility. However, there is an obvious problem when we compute the coefficient of variation: for some countries (with highly volatile government spending) the average growth ECB Working Paper Series No 924 August 2008 13 We set up our estimated models in a number of different ways. In particular, we use (i) OLS both in a bivariate model and in models controlling for a country-specific volatility effect; (ii) Fixed Effects estimation; (iii) Random Effect estimation; and (iv) Instrumental variables (IV) estimation both in a bivariate model and in models with control variables. Similarly to Rose’s (2006) and Furceri and Karras (2007, 2008) strategy, we use four different sets of control variables, most of them obtained from Rose’s website (www.haas.berkeley.edu/~arose)9. The first set of controls includes: (a) the urbanization rate, (b) population density, (c) the log of absolute latitude (kilometers from the equator), (d) a binary dummy variable for a landlocked country, (e) an island-nation dummy, (f) a high income country dummy, (g) regional dummies for developing countries from 1) Latin America, 2) Sub-Saharan Africa, 3) East Asia, 4) South Asia, 5) Europe-Central Asia, 6) and Middle East-North Africa, and (h) language dummies for countries that speak 1) English, 2) French, 3) German, 4) Dutch, 5) Portuguese, 6) Spanish, 7) Arabic, and 8) Chinese. Many of these variables are related to the quality of governments. In fact, as pointed out by La Porta et al. (1998), it is likely that latitude from the equator, income and regional dummies are related to the quality of government and institutions. Moreover, by including language dummies rate over some time spans turns out to be negative, implying thus a very low measure of volatility in contrast with the evidence. Therefore, we check the robustness of our results with two other measures of government spending. The first is the standard deviation of the cyclical component of real government spending (Furceri, 2007; Afonso and Furceri, 2008). Its use avoids the “scale” problem since the time average of the cyclical component by construction is zero for each country. The second measure is the ratio between the standard deviation of real government spending and the average level of government spending. Its use avoids businesscycle effects resulting from the employment of annual data. All results of this paper are qualitatively unchanged if we use these measures of volatility. 9 See Data Appendix for a more detailed description of the variables and their source. 14 ECB Working Paper Series No 924 August 2008 we are able to capture (at least in part) different level of language fractionalization among countries10. The second set of control variables augments the first set including also dummies to control for the effect of new, decolonized, and COMECON countries (see Alesina and Wacziarg, 1998): (a) a dummy for countries created post-World War 2, (b) a dummy for countries created after 1800 but before 1945, (c) a dependency dummy, (d) an OPEC dummy, and (e) a COMECON dummy. The third set of controls includes four other macroeconomic variables that are associated with government volatility: (a) GDP per capita11, (b) Openness12, (c) CPI Inflation, and (d) Government size13. In fact, as pointed out by Fatás and Mihov (2003) it is likely that poor countries have shorter and more volatile business cycles due to less developed financial markets, for example, and at the same time they may resort more often to discretionary policy (see also Rand and Tarp, 2002). Similarly, economies with a higher degree of openness, and thus more exposed to external shocks, may use more frequently discretionary countercyclical fiscal policies (Rodrik, 1998). Moreover, countries with bigger government are usually characterized by bigger automatic stabilizers and thus are less tempted to use discretionary measure of fiscal policy for fine tuning purposes (Fatás and Mihov, 2001). The main advantage of this set of controls is that they are variables usually associated with government volatility, which are available for all the period under study. 10 In the following of the analysis we will use other variables as proxy of ethnic fractionalization. The use of language dummies to this purpose, at this stage, is justified for the greater data availability. 11 Although the inclusion of GDP per capita could lead to multicollinearity since both population and GDP per capita may account for scale effects, in our sample these two variables result to be scarcely correlated (0.07). 12 We use as proxy for openness the GDP’s share of total exports and imports. 13 Government size is here measured as the ratio of government spending to GDP. ECB Working Paper Series No 924 August 2008 15 Moreover, other variables for which we have data just for the last decade could also be important determinants for government volatility. For this purpose, we consider a fourth set of controls for which we have data only relatively to the last time period 1991-2000. The variables included are those of the third set of controls plus: (a) an index of the level of Democracy, (b) an index for the level of Corruption, (c) an index for Political Stability, (d) an index for Government Effectiveness, (e) an index for Country Risk, and (f) an index for language fractionalization. To summarize, we estimate the effect (ȕ) of country size on government spending volatility using the following regression model: ln(ıi,t-t+W) = E ln(Popit) + D + {JtTt} + ȈjGjXijt + Hit (11) where ı measures government spending volatility for country i at time t, Pop denotes population, {Tt} denotes a set of time- specific fixed effects, and {Xj} denotes a set of control variables. H is a well-behaved residual, and D, {J}, {G}, are the coefficients of our other control variables. 4. Results Figures 1 provides the scatter plot of government spending volatility (measured by the standard deviation of the annual growth rate of government expenditure) against country size (measured by the natural logarithm of population) for the entire period 19602000. The figure exhibits negative and statistically significant relation between these two variables. In particular, the estimate of this simple bivariate relation for the full sample gives us: ıi = 0.207 - 0.011 ln(Popi) (7.77) (-3.40) 16 ECB Working Paper Series No 924 August 2008 with R2 = 0.06, and t statistics shown in parenthesis. The relationship is clearly negative and statistically significant, even though the relatively low value of the R-squared coefficient suggests that other factors could have a significant impact on volatility of government spending 14. Moreover, the coefficient of country size does not seem to be affected by outliers such as those countries with volatility higher than 0.3. To confirm this, running again the regression, this time excluding outliers, the relationship is still negative and actually strengthened15: ıi= 0.169 - 0.008 ln(Popi) (9.90) (-3.92) with R2 = 0.08, and t statistics shown in parenthesis. We now proceed with more formal statistical evidence. Table 1 reports the estimated slope coefficient (ȕ) of country size, along with the associated t-statistics in parentheses for several specifications of equation (11). In particular, the four columns of Table 1 correspond to: (i) bivariate OLS; (ii) OLS including the first set of controls; (iii) OLS including the second set of controls; and (iv) OLS including the third set of controls. Focusing on the full-period (pooled) 1961-2000, it can be readily seen that the relation between country size and government expenditure volatility is negative and 14 Since our dependent variable is based on estimates (sample standard deviation) the regression residuals can be thought of as having two components. The first component is sampling error (the difference between the true value of the dependent variable and its estimated value). The second component is the random shock that would have been obtained even if the dependent variable was directly observed rather than estimated. This would lead to an increase of the standard deviation of the estimates, which will lower the t-statistics. This means that any correction to the presence of this un-measurable error term will increase the significance of our estimates. A second concern is the possibility of heteroskedasticity. However, in most of our estimations heteroskedasticity does not seem to be a problem. When it does, we correct for that by using White standard errors. 15 Estimating a non linear relation, the relation is still significant and negative: ıi= 0.169 +0.022 ln(Popi)- 0.022 ln(Popi)2 (9.90) (1.34) (-1.94) ECB Working Paper Series No 924 August 2008 17 statistically significant: the larger the size of the country, the less volatile its government expenditure. It is noteworthy that the coefficient on size remains negative and significant in every specification. In particular, two considerations are important. First, the magnitude of the coefficient is broadly constant over the different set of controls. Second, the coefficient remains significant even after controlling for an exhaustive set of regional, geographical, and macroeconomic variables16. In fact, we believe it is significant that country size is shown to reduce government spending volatility even when we control for openness, since trade openness is the only variable found to be robustly and significantly related with country size (Rose, 2006)17. The interpretation of the coefficient relative to country size is the following. By our estimations, an increase of one percent in population will determine a decrease of 0.2 percent in government expenditure volatility (on average). In other words, just because Germany is ten times the size of Belgium, means that Germany has 50 percent less volatile government expenditure than Belgium. We have also examined the robustness of the relation between country size and fiscal volatility with respect to different time periods. In particular, we considered six different time samples: 1961-1970, 1971-1980, 1981-1990 and 1991-2000. Table 2 presents, across the above mentioned time periods, the coefficient on country size obtained 16 In our estimations, Island, Arabic language, OPEC, and Government Size are other variables that we find to be highly significant. For Island countries that could be attributed to the fact that they are more open to foreign trade, even though expenditure volatility is very high for some of these countries (Le Borgne and Medas, 2007). In turn, Arabic and OPEC economies are rich in oil revenues and contingent upon that commodity. Hence, the volatility in oil price might explain the higher volatility of government spending on those countries. 17 As robustness check, we also include private consumption volatility and public debt in the third set of controls of Table 1. The first variable turns out highly significant and positive (not shown here), but country size still remains highly significant when controlling for it. The link between public expenditure and household consumption results from the transfers made by the governments or the taxes paid by households (Herrera, 2007). Public debt, in turn, is insignificant in our estimations. Further, its inclusion reduces substantially the number of countries in the sample, which hams the significance of all other variables, including that of country size. 18 ECB Working Paper Series No 924 August 2008 using the same specification as in Table 1. Our results suggest that while the effect of country size on government expenditure volatility remains negative and statistically significant, the magnitude increases over time, especially in the last decade. From a statistical point of view, this could be attributed to a lower number of degrees of freedom for this sample period (for the first sample period), and to the fact that government expenditure has been poorly measured during the first years. From an economic point of view, a possible interpretation, as suggested by Alesina and Wacziarg (1998), is that many new decolonized had to “build up” their public sector during the first time samples, and as their level and volatility of government expenditure converged to a sort of steady state level, the effect of the fundamental determinants of government volatility started to play a larger role. Another robustness check that we provide involves the use of different estimation techniques. Tables 3 and 4 report the estimated slope coefficient of country size for the first, second, and third set of controls with: (i) Fixed Effects and Time Random Effects; and (ii) IV estimation, respectively18. Analyzing these tables we can immediately see that the effect of country size on government volatility is still robust to all methods of estimations. In particular, while the magnitude of the coefficient is broadly unchanged over the different 18 We use the logarithm of the country’s total area as instrumental variable for the log of its population, as did Rose (2006), Furceri and Karras (2007, 2008) and as argued by Drazen (2000). The F-statistic of the simple regression of log of population on log of total area is 2070.43, which suggests that the possible bias of the IV is substantially lower than the one of the OLS (Staiger and Stock, 1997). There is also very little concern of reverse causality. In fact, it is very unlikely that people choose where to live based on consideration of government spending volatility. In contrast, there could a more serious issue of endogeneity for other controls variables (as inflation). We address this issue (and also the one for our variable of interest) considering the starting value of the control between time t and time t+W, while we use a measure of volatility of time(t, t+W  ECB Working Paper Series No 924 August 2008 19 techniques of estimation and set of controls, its significance level increases with respect to OLS and IV when we control for time effects both Fixed and Random19. The analysis presented so far has shown that the effect of country size on government spending volatility is very robust to different econometric techniques and sets of controls. However, other variables for which we have data only for the last decade, such as Democracy, Corruption, Political Stability, Government Effectiveness, Country risk and language fractionalization, can account for higher fiscal volatility. To check for robustness, we consider these variables in the OLS and IV estimation. The results are reported in Table 5. Again the results are robust. In particular, while the coefficient on population is still statistically significant its magnitude is increased. It is possible to argue that most of the variation in many determinants of fiscal volatility (such as political constraints, income, inflation and etc.) occurs between the rich and the poor countries. Thus, both from a theoretical perspective and (especially) from a policy point of view is important to assess whether the relationship between country size and government spending volatility is still negative within each group (Rich and Poor20). While, we have already shown that our analysis still holds when we include as control variable the level of GDP and income dummies, it would be important also to run two different regressions for each group of countries. Table 6 conveys the results. They show 19 According to the Hausman test, the Fixed Effects specification is preferred to the Random Effects. However, we cannot reject the hypothesis of absence of time effects at 5% significant level. Similarly, the inclusion of country effect does not improve the fitness of our model either the significance of our estimates. This is mainly due to the fact that country effects are to some extent captured by language and regional dummies. However, by including only country effects in the regression with the third set of controls the magnitude of the coefficient of country size increases (to -0.77) and its significance level remains high (tstatistic=-4.50). 20 We use the World Bank classification to differentiate among Rich and Poor countries. In particular, we includes in Poor countries those countries classifies as “Low Income”, “Lower Middle Income”, and “Upper Middle Income”; and we include in the Rich countries those classifies as “High Income-non OECD” and “High Income-OECD”. 20 ECB Working Paper Series No 924 August 2008 that while the coefficient on population has the same sign across the two different groups, the magnitude and significance level is bigger for Poor countries. Finally, our empirical analysis regarding volatility of aggregate government consumption concludes using a proxy for discretionary spending volatility, instead of general government spending volatility, as our dependent variables. It is important to stress the fact that there is no consensus in the literature on the appropriate measure of discretionary (cyclically adjusted) fiscal policy21. The difficulty mainly comes from the simultaneity in the determination of output and government spending volatility. To this purpose we use a measure of discretionary fiscal policy that is not affected by output volatility. In more detail, following Fatás and Mihov (2003, 2006) our measure is obtained by estimating, for each country i, the following equation: 'Gi ,t D i ,t  E i 'Yi ,t  J i 'Gi ,t 1  G iWi ,t  H i ,t , (12) where G is the logarithm of real government spending, Y is the logarithm of real GDP, and W includes a time trend, inflation and inflation squared. The estimated standard deviation of the residuals (i.e. V i ,t W var H i ,t t W ) is assumed as a quantitative estimate of discretionary fiscal policy volatility. In order to estimate equation (11) we include the contemporaneous value of output growth and we use past values as instrumental variable to avoid the possibility of endogeneity bias. We instrument current output growth with lagged GDP growth, the index of oil prices, lagged inflation, and the lagged value of government spending growth. Table 7 presents the coefficient on country size obtained using the same specification used in Table 1. Our results point out that the effect of country size on 21 See Alesina and Perotti (1996), Blanchard (1993) and Fatás and Mihov (2003, 2006) for a detailed discussion on alternative measures of discretionary fiscal policy. ECB Working Paper Series No 924 August 2008 21 discretionary government spending volatility is still negatively and statistically significant. Thus, not only smaller countries have more volatile general government spending but they also, independently of automatic stabilizers, tend to use government spending more actively. This could suggest that the relation between spending volatility and country size is negative not only to the extent that government spending is used for fine tuning purposes, but also to the extent that public goods are of a non rival nature. 4.1 Government Spending Volatility by Functional Categories Our analysis, so far, has pointed out a clear negative relation between government spending volatility and country size. However, to better understand this relation it is useful to analyze the different components of government consumptions. For this purpose, we consider the following categories: i) General public services; ii) Defense; iii) Public order and safety; iv) Economic affairs; v) Housing and community amenities; vi) Health; vii) Recreation, culture and religion; viii) Education; and ix) Social protection22. As we discussed in our theoretical section, a larger country size may reduce government spending volatility because of the higher returns to scale of the non-rival good. To this extent, we should expect spending volatility related to non-rival public goods (such as general administration) to be more associated with country size than spending volatility related to rival public goods (such as education, health, and order and safety). However, our theoretical model also evinces that larger countries are more able to mitigate idiosyncratic shocks and stabilize its government spending. Therefore, we should expect, to a certain extent, all items of government spending to be negatively associated with country size. 22 22 Data for government consumption classified by function are retrieved by the UN and OECD data sets. ECB Working Paper Series No 924 August 2008 Table 8 shows the results of the regression between government consumption volatility classified by economic function and country size for the period 1971-2000 and using the third set of control variables23. Each of the columns of the table corresponds to a different economic function of government consumption. Analyzing the results, we can observe that the relation between government consumption and country size is negative for each of the different categories. Thus, these results seem to confirm the idea that smaller countries tend to have more volatile government spending also because they are more exposed to idiosyncratic shocks. Moreover, from all spending items analyzed, economic affairs and public order are the one whose the coefficient of country size is more significant, which might be due to the high level of non-rivalry of these goods. Summarizing, this analysis has confirmed the findings of our theoretical model that due to both, the higher scale economies in the provision of non-rival public goods and to a lower exposure to idiosyncratic shocks, larger economies are more able to stabilize their government spending. 5. Conclusions This paper provides empirical evidence showing that smaller countries tend to have more volatile government spending. From a theoretical point of view, we show that a negative relationship between government spending volatility and country size can be explained by two arguments: i) to the extent that government spending is used for fine tuning purposes, the size of a country acts as an insurance against idiosyncratic shocks, 23 The results are qualitatively robust also to the inclusion of the additional variables present in the fourth control set. ECB Working Paper Series No 924 August 2008 23 leading to a less volatile government spending; ii) increasing returns to scale of government spending originating from higher ability to spread the cost of financing it over a larger pool of taxpayer, allow the government to provide the public good in a less volatile way. These claims are confirmed by our empirical analysis, which is robust to different time and country samples, different econometric techniques and to several sets of control variables. In particular, disaggregating government consumption by function, it emerges that government consumption spending in all functions is more volatile in smaller countries. In addition, the empirical analysis evinces that the discretionary (not reacting to the state of economy for fine tuning purpose) government spending volatility is also decreasing with the size of nations. Our paper highlights the need for small countries to undertake fiscal adjustments in order to reduce macro-fiscal vulnerabilities and improve their economic growth prospects (see also Le Borgne and Medas, 2007; and Medina Cas and Ota, 2008). In addition, to the extent that large fiscal areas reduce government spending volatility, our findings reinforce the role of fiscal coordination and the move towards common fiscal policy in monetary unions, even though other factors may undermine and overcome such fiscal manoeuvre (see, among others, Beetsma and Bovenberg, 1998; Beetsma et al., 2001; and von Hagen et al. 2002). The current analysis also offers various possibilities for further research. On the theoretical side, a more structural model would be helpful to better understand the mechanisms underlying the economic and political effects of country size on the government spending volatility. For instance, modeling the political side of the economy could be useful to investigate the impacts of country size and political heterogeneity on our variable of interest. On the empirical side, an analysis of the effects of country size on the 24 ECB Working Paper Series No 924 August 2008 volatility of taxes revenues, transfers, and debt management could ratify our findings that that variable indeed acts as an insurance against idiosyncratic shocks, and show how strong this effect is indeed. ECB Working Paper Series No 924 August 2008 25 References Afonso, A., Furceri, D., 2008. Government Size, Composition, Volatility and Economic Growth. ECB Working Paper 849. Afonso, A., Skuchnecht, L., Tanzi, V., 2006. Public sector efficiency: evidence for new EU member states and emerging markets. ECB Working Paper 581. Aghion, P., Banerjee, A., 2005. Volatility and Growth, Clarendon Lectures in Economics, Oxford: Oxford University Press. Alesina, A., Perotti, R. 1996. Fiscal Expansions and Adjustment in OECD Economies. Economic Policy XXI, pp. 205–240. Alesina, A., Wacziarg, R., 1998. Openness, country size and government. Journal of Public Economics 69(3), pp. 305-321. Alesina, A., Spolaore, E., 2003. The Size of Nations. MIT Press, Cambridge, MA. Alesina, A., Spolaore, E., Wacziarg, R., 2004. Trade, growth and the size of countries. In P. Aghion and S. Durlauf, Editors, Handbook of Economic Growth, North-Holland, Amsterdam. Beetsma, R., Bovenberg, A.L., 1998. Monetary Union without Fiscal Coordination May Discipline Policymakers. Journal of International Economics, Vol. 45, pp.239-258 Beetsma, R., Debrun, X., Klaassen, F., 2001. Is Fiscal Policy Coordination in EMU Desirable? Swedish Economic Policy Review, Vol. 8, No. 1, pp. 57-98. Blanchard, O., 1993. Suggestions for a New Set of Fiscal Indicators. In H. A. A. Verbon and F. A. A. M. van Winden, editors, The New Political Economy of Government Debt, Amsterdam, The Netherlands: Elsevier Science Publishers. Bolton, P., Roland, G., 1997. The breakup of nations A political economy analysis. Quarterly Journal of Economics 112, pp. 1057–1090. Drazen, A., 2000. Political Economy in Macroeconomics. Princeton, NJ: Princeton University Press. Fatás, A., Mihov, I., 2003. The Case for Restricting Fiscal Policy Discretion. Quarterly Journal of Economics 118, pp.1419-1447. Fatás, A., Mihov, I., 2005. Policy volatility, institutions and economic growth. CEPR Discussion Paper 5388. Fatás, A. Mihov, I., 2006. The Macroeconomics Effects of Fiscal Rules in the US States. Journal of Public Economics 90, pp. 101-117. 26 ECB Working Paper Series No 924 August 2008 Furceri, D., 2007. Is Government Expenditure Volatility Harmful for Growth? A CrossCountry Analysis. Fiscal Studies 28 (1), pp. 103-120. Furceri, D., Karras, G., 2007. Country Size and Business Cycle Volatility: Scale Really Matters. Journal of Japanese and International Economies 21(4), pp. 424-434. Furceri, D., Karras, G., 2008. Business cycle volatility and country size: evidence for a sample of OECD countries. Economics Bulletin, 5(3), pp. 1-7. Harberger, A., 2005. On the process of growth and economic policy in developing countries. PPC Issue Paper No. 13. USAID Development Experience Clearinghouse. December. Herrera, S., 2007. Public Expenditure and Growth. World Bank Policy Research Working Paper 4372, October. Imbs, J., 2007. Growth and Volatility. Journal of Monetary Economics, 54 (7), pp. 18481862. La Porta, R., Lopez de Silanes, F., Shleifer, A., Vishny, R., 1998. The Quality of Government. Journal of Law, Economics and Organization, 15 (1), pp. 222-79. Le Borgne, E., Medas, P., 2007. Sovereign Wealth Funds in the Pacific Island Countries: Macro-Fiscal Linkages, IMF Working Paper 07/297 (Washington: International Monetary Fund). Loayza, N., Ranciere, R., Serven, L., Ventura, J., 2007. Macroeconomic volatility and welfare in developing countries: an introduction. World Bank Economic Review 21, pp. 343-357, April. Medina Cas, S., Ota, R., 2008. Big Government, High Debt, and Fiscal Adjustment in Small States, IMF Working Paper 08/39 (Washington: International Monetary Fund). O'Higgins, M., Ruggles, P., 1981. The Distribution' of Public Expenditure Among Households in the United States. Review of Income and Wealth, Vol. 27. Poplawski Ribeiro, M., Beetsma, R., 2008. The political economy of structural reforms under deficit restrictions. Journal of Macroeconomics, Vol. 30, pp. 179-198. Ramey, G., Ramey, V.A., 1995. Cross-country evidence on the link between volatility and growth, American Economic Review, Vol. 85, pp. 1138–51. Rand, J., Tarp, F., 2002. Business Cycles in Developing Countries: Are They Different? World Development, Vol. 30, Issue 12, pp. 2071-88. ECB Working Paper Series No 924 August 2008 27 Rodrik, D., 1998. Why Do More Open Economies have Bigger Governments? Journal of Political Economy, 106, pp. 997-1032. Rose, A.K., 2006. Size Really Doesn’t Matter: In Search of a National Scale Effect Journal of the Japanese and International Economies, 20 (4), pp. 482-507. Staiger, D., Stock, J.H., 1997. Instrumental variables regressions with weak instruments. Econometrica 65, pp. 557–586. von Hagen, J., Hughes Hallett, A., Strauch, R., 2002. Budgetary consolidation in Europe: quality, economic conditions and persistence. Journal of Japanese and International Economies 16(4), pp. 512-535. 28 ECB Working Paper Series No 924 August 2008 Data Appendix Table A. Summary Statistic and Source for the Main Variables Description Source # Obs. Mean Government Spending Volatility PWT6.2 451 0.015 Log of Population PWT6.2 832 14.852 Urbanization Rate Rose 819 48.842 Density Rose 710 253.421 Latitude Rose 832 9.577 GDP per capita Rose 612 5220.501 Openness Rose 582 76.572 CPI Inflation Rose 504 55.799 Democracy Rose 531 3.902 Corruption Rose 184 -0.004 Political Stability Rose 165 -0.004 Government Effectiveness Rose 184 -0.006 Country Risk Rose 139 67.937 Language Fractionalization Rose 191 0.394 Notes: PWT6.2 refers to the Penn World Table v. 6.2. Rose refers to A.K. Rose’s website. St. Dev. 0.017 2.303 24.839 1300.324 15.208 7780.298 45.310 499.7929 4.190 1.001 1.001 1.000 11.743 .0280 Table B. Correlation between Government Spending Volatility Categories GS PU DE OS EA HO HE RE ED SP GS 1 PU 0.215 1 DE 0.164 0.044 1 OS 0.173 0.591 0.092 1 EA 0.440 0.320 0.249 0.561 1 HO 0.088 0.207 0.078 0.255 0.341 1 HE 0.026 0.397 0.162 0.753 0.423 0.21 1 RE -0.045 0.044 0.192 0.177 0.266 0.30 0.394 1 ED 0.088 0.234 0.073 0.610 0.565 0.16 0.696 0.128 1 SP 0.076 0.141 0.082 0.375 0.322 0.32 0.531 0.715 0.416 1 GS= Government Spending; PU= General public services; DE= Defense; OS= Public order and safety; EA=Economic affairs; HO=Housing and community amenities; HE=Health; RE=Recreation, culture and religion; ED=Education; SP=Social protection. ECB Working Paper Series No 924 August 2008 29 Figures and Tables FIGURE1. CORRELATION OF GOVERNMENT VOLATILITY AND POPULATION Standard deviation of goverment growth 0.7 ALB 0.6 KWT 0.5 TKM ARE 0.4 GNQ QAT 0.3 0.1 0.0 OMN COG SAU GEO LBR KGZ RUS BIH RWA PNG BHR GAB IRN GNB CVT TTO BWA YEM SCG GMB ZAR NGA ZMB SYC BRN SWZNAM TGO AFG NIC KIR DZA UKR CPV HRV MRTERI KAZ DMAVUT DJI BTN SLE TCD MDA SOMBGRVEN UZB MUS LVA MNG MOZTZA KNA EST JOR SYR ECU BLZ LUX CYP LSO ROM NER CHL LBN BDI MWI ZWE CIV SDN DOM CAF IDN PAN LCA STP BHS MAC ARM NLI CUB GRD MYS UGA WSM BOL BEN CMR SGP COM PER HKG HUN LKA BLT FJI KHM PRK BRA MKD ETH FSM ISR POL LAO KOR PRY SEN GIN GHA ISL SVN MDV NOR TWN JAM SLV CZE TUR THA MAR BFA FIN URY PHL MEX SVK BRB HND KEN CRI GRC PAK ANT ZAFEGY PRT BMU NZL ARG JPN NPL MDG PRI GTM LTU CAN IRLDNK BEL COL SWE AUT CHE BGD AUS ITA GER NLD USA FRA VNM ESP GBR BLR ATG 0.2 PLW TJKAZE IRQ SUR 0 30 IND 14 Ln Pop ECB Working Paper Series No 924 August 2008 CHN Table 1-Government Spending Volatility and Country Size Bivariate -0.098 (-6.09)*** Control1 -0.153 (-7.47)*** Control2 -0.160 (-6.53)*** Control3 -0.208 (-5.97)*** Urban - -0.002 (-0.70) -0.002 (-1.08) -0.003 (-0.99) Density - -0.001 (-1.57) -0.001 (-0.29) -0.001 (-0.75) Landlocked - -0.131 (-1.30) -0.071 (-0.72) -0.078 (-0.70) Island - -0.303 (-2.90)*** -0.238 (-2.09)*** -0.223 (-1.85)* English - -0.079 (-1.01) -0.033 (-0.41) 0.026 (0.31) French - -0.127 (-1.34) -0.015 (-0.16) -0.047 (-0.47) Spanish - -0.224 (-1.96)** -0.110 (-0.84) -0.144 (-1.02) Portuguese - -0.456 (-2.62)** -0.210 (-1.02) -0.249 (-0.94) Arabic - 0.382 (3.43)*** 0.195 (1.70)* 0.335 (2.38) German - -0.338 (-1.59) -0.236 (-1.18) -0.307 (-1.27) Dutch - -0.276 (-1.31) -0.062 (-0.28) 0.101 (0.43) Swedish - -0.742 (-1.82)* -0.547 (-1.43) -0.375 (-1.09) Chinese - 0.656 (2.33)** 0.780 (2.07)** 0.544 (0.97) Latitude from Equator - -0.003 (-1.21) -0.004 (-1.50)* -0.006 (-2.03)** Income - -0.132 (-3.28)*** -0.124 (-2.84)*** -0.114 (-2.19)** Lnpop ECB Working Paper Series No 924 August 2008 31 Opec - - 0.982 (6.63)*** 0.746 (5.67)*** Comecon - - 0.212 (0.97) -0.072 (-0.20) Independence - - 0.000 (0.30) -0.000 (-1.00) Post war - - 0.085 (0.64) 0.063 (0.41) Inflation - - - 0.029 (1.72)* Openness - - - -0.003 (-0.03) GDP per capita - - - -0.001 (-1.02) Government Size N R2 Adjusted-R2 545 0.064 0.062 438 0.162 0.130 Notes: t-statistics in parenthesis; *,**,*** respectively significant at 10%,5% and 1%. 32 ECB Working Paper Series No 924 August 2008 376 0.372 0.337 -0.013 (-3.38)*** 275 0.445 0.392 Table 2. Government Spending Volatility and Country Size (OLS)-Robustness over time Lnpop N R2 Adjusted-R2 Lnpop N R2 Adjusted-R2 Lnpop N R2 Adjusted-R2 Lnpop N R2 Adjusted-R2 Bivariate -0.096 (-2.26)** 94 0.052 0.042 Bivariate -0.059 (-1.79)* 140 0.022 0.016 Bivariate -0.119 (-4.38)*** 146 0.118 0.111 Bivariate -0.108 (-3.42)*** 160 0.069 0.063 1961-1970 Control1 -0.109 (-2.25)** 94 0.315 0.183 1971-1980 Control1 -0.099 (-2.69)*** 137 0.334 0.246 1981-1990 Control1 0.165 (-4.94)*** 144 0.321 0.235 1991-2000 Control1 -0.188 (-4.88)*** 149 0.333 0.252 Control2 -0.081 (-1.67)* Control3 -0.054 (-0.63) 91 0.385 0.215 66 0.472 0.227 Control2 -0.002 (-2.04)** Control3 -0.182 (-2.11)** 123 0.354 0.227 74 0.423 0.189 Control2 -0.149 (-3.71)*** Control3 -0.137 (-2.43)** 126 0.431 0.322 93 0.638 0.516 Control2 -0.216 (-4.54)*** Control3 -0.221 (-3.51)*** 124 0.415 0.301 109 0.471 0.320 Notes: t-statistics in parenthesis; *,**,*** respectively significant at 10%,5% and 1%. ECB Working Paper Series No 924 August 2008 33 Table 3. Government Spending Volatility and Country Size (Fixed & Random Effects) Lnpop 1961-2000 (FE) Bivariate Control1 Control2 -0.096 -0.149 -0.157 (-5.94)*** (-7.22)*** (-6.47)*** N R2-within R2-between R2-overall Lnpop 545 0.062 0.858 0.064 438 0.277 0.562 0.274 1961-2000 (RE) Bivariate Control1 -0.098 -0.153 (-6.09)*** (-7.47)*** N R2-within R2-between R2-overall 545 0.062 0.858 0.064 438 0.276 0.428 0.275 Hausman Test (FE vs RE) p-value 0.24 0.99 Notes: t-statistics in parenthesis; *,**,*** respectively significant at 10%,5% and 1%. Control3 -0.190 (-5.42)*** 376 0.377 0.619 0.371 275 0.456 0.998 0.440 Control2 -0.160 (-6.53)*** Control3 -0.208 (-5.97)*** 376 0.375 0.494 0.372 275 0.452 0.867 0.445 1.00 1.00 Table 4. Government Spending Volatility and Country Size (IV) Lnpop N R2 R2-adjusted Bivariate -0.054 (-2.56)*** 545 0.051 0.049 1961-2000 Control1 -0.139 (-4.76)*** 438 0.274 0.246 Notes: t-statistics in parenthesis; *,**,*** respectively significant at 10%,5% and 1%. 34 ECB Working Paper Series No 924 August 2008 Control2 -0.161 (-4.50)*** Control3 -0.183 (-3.20)*** 376 0.372 0.337 276 0.304 0.242 Table 5. Government Spending Volatility and Country Size Lnpop 1991-2000 OLS & Control4 -0.200 (-2.59)*** N R2 R2-adjusted IV & Control4 -0.138 (-1.39) 100 0.503 0.298 100 0.499 0.291 Notes: t-statistics in parenthesis; *,**,*** respectively significant at 10%,5% and 1%. Table 6. Government Spending Volatility and Country Size (Rich and Poor countries) Lnpop N R2 R2-adjusted Lnpop N R2 R2-adjusted 1961-2000 (Rich) Bivariate Control1 -0.159 -0.092 (-6.70)*** (-2.96)*** 228 0.166 0.162 190 0.492 0.445 1961-2000 (Poor) Bivariate Control1 -0.075 -0.154 (-3.53)*** (-4.25)*** 317 0.038 0.035 248 0.126 0.070 Control2 -0.024 (-0.65) Control3 -0.069 (-1.61)* 166 0.599 0.544 133 0.632 0.553 Control2 -0.202 (-4.60)*** Control3 -0.307 (-5.24)*** 210 0.181 0.099 146 0.350 0.231 Notes: t-statistics in parenthesis; *,**,*** respectively significant at 10%,5% and 1%. ECB Working Paper Series No 924 August 2008 35 Table 7-Discretionary Government Spending Volatility and Country Size Bivariate -0.075 (-2.32)*** Control1 -0.067 (-3.50)*** Control2 -0.029 (-1.43) Control3 -0.076 (-3.14)*** Urban - 0.005 (2.60)** 0.005 (2.85)** 0.005 (2.50)** Density - 0.003 (1.77)* 0.005 (3.18)*** 0.006 (3.90)*** Landlocked - 0.116 (1.42) 0.169 (2.39)** 0.135 (1.96)** Island - 0.002 (0.02) 0.104 (1.31) -0.002 (-0.02) English - -0.030 (-0.46) -0.053 (-0.93) -0.046 (-0.89) French - -0.082 (-1.17) -0.034 (-0.56) -0.038 (-0.66) Spanish - -0.002 (-0.02) 0.072 (0.94) -0.038 (-0.49) Portuguese - 0.107 (0.74) 0.098 (0.80) -0.109 (-0.83) Arabic - 0.052 (0.54) 0.005 (0.06) -0.005 (-0.07) German - -0.520 (-3.29)*** -0.524 (-3.93)*** -0.427 (-2.73)*** Dutch - -0.570 (-2.76)*** -0.693 (-3.91)*** -0.654 (-3.77)*** Swedish - -0.545 (-2.26)** -0.473 (-2.34)** -0.399 (-2.20)** Chinese - -1.624 (-1.74)* -2.573 (-3.20)*** -3.505 (-3.49)*** Latitude from Equator - 0.000 (0.09) 0.000 (0.43) 0.001 (0.58) Income - -0.260 (-8.82)*** -0.220 (-8.20)*** -0.146 (-4.39)*** Lnpop 36 ECB Working Paper Series No 924 August 2008 Opec - - 0.148 (1.35) 0.214 (2.10)** Independence - - 0.003 (5.45)*** 0.002 (3.86)*** Post war - - -0.041 (-0.39) -0.103 (-1.05) Inflation - - - 0.015 (2.43)** Openness - - - -0.013 (-1.39) GDP per capita - - - -0.002 (-2.85)*** 91 0.057 0.046 90 0.790 0.743 83 0.871 0.832 Government Size N R2 Adjusted-R2 -0.002 (-0.69) 80 0.905 0.866 Notes: t-statistics in parenthesis; *,**,*** respectively significant at 10%,5% and 1%. ECB Working Paper Series No 924 August 2008 37 Table 8. Government Spending Volatility by Functional Classification and Country Size Lnpop N 2 R 2 PU -0.241 (-2.43)** DE OS EA HO HE RE -0.180 -0.474 -0.352 -0.192 -0.284 -0.266 (-1.69)* (-2.43)** (-3.81)*** (-2.11)** (-3.46)*** (-2.60)** ED SP -0.315 -0.252 (-3.42)*** (-2.72)*** 102 83 60 94 95 95 76 100 88 0.342 0.554 0.555 0.533 0.460 0.524 0.632 0.233 0.342 R -adjusted 0.159 0.391 0.290 0.388 0.295 0.378 0.479 0.027 0.132 Notes: t-statistics in parenthesis; PU= General public services; DE= Defense; OS= Public order and safety; EA=Economic affairs; HO=Housing and community amenities; HE=Health; RE=Recreation, culture and religion; ED=Education; SP=Social protection. *,**,*** respectively significant at 10%,5% and 1%. 38 ECB Working Paper Series No 924 August 2008 European Central Bank Working Paper Series For a complete list of Working Papers published by the ECB, please visit the ECB’s website (http://www.ecb.europa.eu). 893 “Sticky wages: evidence from quarterly microeconomic data” by T. Heckel, H. Le Bihan and M. Montornès, May 2008. 894 “The role of country-specific trade and survey data in forecasting euro area manufacturing production: perspective from large panel factor models” by M. Darracq Pariès and L. Maurin, May 2008. 895 “On the empirical evidence of the intertemporal current account model for the euro area countries” by M. Ca’Zorzi and M. Rubaszek, May 2008. 896 “The Maastricht convergence criteria and optimal monetary policy for the EMU accession countries” by A. Lipińska, May 2008. 897 “DSGE-modelling when agents are imperfectly informed” by P. De Grauwe, May 2008. 898 “Central bank communication and monetary policy: a survey of theory and evidence” by A. S. Blinder, M. Ehrmann, M. Fratzscher, J. De Haan and D.-J. Jansen, May 2008. 899 “Robust monetary rules under unstructured and structured model uncertainty” by P. Levine and J. Pearlman, May 2008. 900 “Forecasting inflation and tracking monetary policy in the euro area: does national information help?” by R. Cristadoro, F. Venditti and G. Saporito, May 2008. 901 “The usefulness of infra-annual government cash budgetary data for fiscal forecasting in the euro area” by L. Onorante, D. J. Pedregal, J. J. Pérez and S. Signorini, May 2008. 902 “Fiscal consolidation in the euro area: long-run benefits and short-run costs” by G. Coenen, M. Mohr and R. Straub, May 2008. 903 “A robust criterion for determining the number of static factors in approximate factor models” by L. Alessi, M. Barigozzi and M. Capasso, May 2008. 904 “Does money matter in the IS curve? The case of the UK” by B. E. Jones and L. Stracca, June 2008. 905 “A persistence-weighted measure of core inflation in the euro area” by L. Bilke and L. Stracca, June 2008. 906 “The impact of the euro on equity markets: a country and sector decomposition” by L. Cappiello, A. Kadareja and S. Manganelli, June 2008. 907 “Globalisation and the euro area: simulation based analysis using the New Area Wide Model” by P. Jacquinot and R. Straub, June 2008. 908 “3-step analysis of public finances sustainability: the case of the European Union” by A. Afonso and C. Rault, June 2008. 909 “Repo markets, counterparty risk and the 2007/2008 liquidity crisis” by C. Ewerhart and J. Tapking, June 2008. 910 “How has CDO market pricing changed during the turmoil? Evidence from CDS index tranches” by M. Scheicher, June 2008. 911 “Global liquidity glut or global savings glut? A structural VAR approach” by T. Bracke and M. Fidora, June 2008. ECB Working Paper Series No 924 August 2008 39 912 “Labour cost and employment across euro area countries and sectors” by B. Pierluigi and M. Roma, June 2008. 913 “Country and industry equity risk premia in the euro area: an intertemporal approach” by L. Cappiello, M. Lo Duca and A. Maddaloni, June 2008. 914 “Evolution and sources of manufacturing productivity growth: evidence from a panel of European countries” by S. Giannangeli and R. Gόmez-Salvador, June 2008. 915 “Medium run redux: technical change, factor shares and frictions in the euro area” by P. McAdam and A. Willman, June 2008. 916 “Optimal reserve composition in the presence of sudden stops: the euro and the dollar as safe haven currencies” by R. Beck and E. Rahbari, July 2008. 917 “Modelling and forecasting the yield curve under model uncertainty” by P. Donati and F. Donati, July 2008. 918 “Imports and profitability in the euro area manufacturing sector: the role of emerging market economies” by T. A. Peltonen, M. Skala, A. Santos Rivera and G. Pula, July 2008. 919 “Fiscal policy in real time” by J. Cimadomo, July 2008. 920 “An investigation on the effect of real exchange rate movements on OECD bilateral exports” by A. Berthou, July 2008. 921 “Foreign direct investment and environmental taxes” by R. A. De Santis and F. Stähler, July 2008. 922 “A review of nonfundamentalness and identification in structural VAR models” by L. Alessi, M. Barigozzi and M. Capasso, July 2008. 923 “Resuscitating the wage channel in models with unemployment fluctuations” by K. Christoffel and K. Kuester, August 2008. 924 “Government spending volatility and the size of nations” by D. Furceri and M. Poplawski Ribeiro, August 2008. 40 ECB Working Paper Series No 924 August 2008