- - ' . :L'AWi S :138a7 POLICY RESEARCH WORKING PAPER 138=7 Fiscal Decentralization- - - ---=Countries, such as economies in transition, tat want to and the Size of Government reduce the size of the public sector should decentralize . . . - ~~~~~~~~~~bot taxing ancl spending: An Extension with Evidence b txn d spending decisions. , from Cross-Country Data Jaber Ehdaie The World Bank Policy Reserch Department Public Econonuics Division December 1994 IPOLICY RESEARCH WORKING PAPER 1387 Summary findings Prior analyses of the relationship between fiscal government's taxing and spending powers tends to decentralization and tine size of government treat fiscal reduce the size of the public sector. decentralization as the decentralization of either taxing * Revenue-sharing arrangements in which decisions or spending powers. But decisions about taxation and about taxation are made by the national government spending are inseparable. tend to eliminate the constraining effect of the Ehdaie corrects this deficicncy and analyzes the effect decentralized spending power. of simultaneous decentralization of taxing and spending What do these findings suggest? powers - "fiscal decentralization" - on the overall size Countries, such as economies in transition, that want of the public sector using cross-country data. to reduce the size of the public sector should decentralize The econometric results of his study show that: both taxing and spending decisions. The simultaneous decentralization of the national This paper-aproductof thePublic Economics Division, Policy Research Department-is parof alargereffortin thedepartment to study the economic consequences of fiscal decentralization. Copies of the paper arc available free from the World Bank, 1818 HStreetNW, Washington, DC20433. Please contactCarlinaJones, room N10-063, extension 37699 (18 pages). December 1994. Thc Policy Research Workig Paper Series disemin the finding of wrk in progess to enage the excage of ideas aboru Wdopnma issn objectieofthe se istogetthefiiostquky, en ifthe pretaionsmare kss tban fdlypoisbe hc papers arry the names of the authors and should be used and cied accordingly. The findings, int,erpret and concluions are the asahors' own and shoud not be aibuted to t Word Bak ts Excaive Board of Dirctors, or any of its mmberncount Produced by the Policy Research Dissemination Center Fiscal Decentralization and the Size of Government An Extension with Evidence from Cross-Country Data Jaber Ebdaje Public Economics Division Policy Research Department World Bank, Washington, DC 1 ladodeeplyindebtedtoShantaDevarajuanformanyuisuggesonsandC3menslwouldalsoliketothank AnwarShah, SanjayPradhan, Edgardo Campos,PedroAJba, ShaikhHossain andVmayaSwaroopforhelpfil comments. This paper was presented at a PRDPE semiar. Comments by the seminar parmcipants are also acknowledged. Finally, Iam gratelu to Mm Zhufor compiling datawith excellent exprse and Carlinajones for typing this manuscript I. Introduction Many studies have attempted a test of the Leviathn hypothesis [Brennan and Buchanan (1980)], that fiscal decentraliztion serves as a constraint on the behavior of revenue-maximizing governments and thereby, restrains the overall size of public sector. Among others, Oates (1985), Marlow (1988), Grossman (1989; 1992), Jouifaian and Marlow (1991), and Kneebone (1992) examined the hypothesis at the national government level. A problem with all of these studies is that they treat fiscal d ali as the decentaliza of either taxing or spending powers, neglecting the inseparability of taxing and spending decisions.' TMe present paper argues that fiscal decentralization in the Leviata model of government is a composit, constitutional action contning the two inseparable elements of txig and spending decisions. It then uses inrational cros-country data to investigate whether the simultaneous decentrlzation of the national governmet's txing and spending powers tends to act as a constraining influence on the overall size of public sector. If so, wouldn't revenue sharing with taxing decisions concentrated in the hands of the reverme-maximizing national government circumvent the constraining influence of decentralization of the spending power? The answers are important for policymakers in all transition and many market-oriented economies as governments often decentralize their spending powers while pursuing the objective of a smaller public sector [for example, Bird and Wallich (1993)]. 2This inseparability issue was also ignored by other studies attempted a test of the hypothesis at the sate or country govermment level, for example, Oates (1985), Nelson (1987), Raimondo (1989), Zax (1989), and Joulfaian and Marlow (1990). 3- The organization of the paper is as follows. The next section deals with the inseparability of taxing and spending decisions in the Leviathan model of fiscal decentralization, followed by a brief discussion of the problems with the results of previous studies. The third section discusses the empirical model. The estimation method and empirical results are presented in the fourth section. The fifth section is devoted to concluding remarks. II. The Leviathan Model of Fiscal Decentralization Drawing by analogy on the conventional theory of monopoly in the private sector, Brennan and Buchanan (1980) modelled government as a monolithic entity, "Leviathan", that systematically seeks to maximize the total revenues that it extracts from the economy through the excessive tax-pricing of public goods and services it supplies. The government's ability to maximize revenue and hence expenditure, they argue, is limited only by constitutional constraints placed upon its actions- One such constraint would be the decentralization of the national (central) government's taxing and spending powers, with subnational units of government tacing and spending "independently" [Brennan and Buchanan (1980), 185]. Decntralization of taxing and spending powers provides taxpayers with options among "separate taxing-spending jurisdictions". Through the potential exercise of these options, taxpayers control the behavior of revenue-maximing governments along the lines of the Tiebout (1956) model. In a Tiebout-style world, any attempt by one jurisdiction to raise the tax price of local public goods and services it supplies will result in migration of its citizen-taxpayers to an alternative jurisdiction in the pursuit of fiscal gains. Interjurisdictional competition for mobile citizen-taxpayers and other economic resources limits govermnents' excessive tax pricing powers, -4 - encourages a more cost-efficient production-supply of local public goods and services, and thereby, restrains the overall size of the public sector. In short, the Leviathan model contends that, other things equal, "Total government intrusions into the economy should be smaller, ceteris paribus, the greater the extent to which taxes and expenditures are decentralized ......" [Brennan and Buchanan (1980), p. 1851. To further emphasize the ins.parability of tax and expenditure decentralization in their hypothesis, Brennan and Buchanan (1980) argued: "Possibility for collusion among separate governmental units .... must be included in 'other things equal'.", p. 185. They predicted that, within the constimtionally dec-ntralized fiscal structure, subnational governments would try to circumvent competitive pressures through colluding among themselves or with the natonal governmenL One obvious collusion would be agreements between subnational governments and the national government. Subnational governments would cede taxing powers to the national government. National govermment would establish a revenue-maximizing, uniform tax system across all jurisdictions. The tax revenues would be then shared among governments, with subnational governments receiving their shares in the form of intergovernmental transfers (grants) according to Grossman (1989). Revenue sharing, Brennan and Buchanan (1980) argue, subverts the primary purpose of the fiscal decentralization, which is to create competition between subnational governments. It removes one major element of the competitive government process, i.e., tax competition, by establishing a uniform tax system across jurisdictions and encourages the expansion of the public sector through the concentration of taxing powers in the hands of the revenue-maximizing -5- national government, circumventing the constraining influence of the expenditure decentralization. Each subnational unit of government must have responsibility for raising its own revenue and should be precluded from entering into revenue sharing agreements with the national or other subnational units of government [Brennan and Buchanan (1980), p. 183]. The inseparability of revenue-raising and spending responsibilities at the subnational level of government clearly requires the simultaneous assigmnent of the national government's taxing and spending powers to subnational governments. In short, the Leviathan model inplies that, other things equal, the simultaneous decentralization of the national govermnent's taxing and spending powers, Nfiscal decentra- lization", should act as a constraining influence on the overall size of the public sector.3 And 'other things equal' should include the simultaneous transfer of the national governments revemne (revenue sharing) and assignment of its spending power to subnational units of government, "fiscal collusion". The effect of fiscal collusion on the public sector size is ambiguous, depending on whether the stimulating effect of revemne sharing would exceed or fall short of the constraining influence of decentralization of the spending power. However, as mentioned earlier, -previous attempts to conduct an econometric test of the fiscal decentralization hypothesis at the national government level ignored the simultaneity of taxing and spending decisions, for example, Dates (1985), Marlow (1988), Grossman (1989; 1992), Joulfaian and Marlow (1991), and Kneebone (1992). They measured fiscal [de]centalization by a [de]centralization ratio of taxing or spending powers, i.e., the share of 3Alternative theories of goverment behavior supportng this hypothesis include Richard Musgrave's (1959) model of how the distribution funmtion of government would be carried out by subnational governments, and the more traditional public choice models (for example, Walter Hettich and Stanley Winer, 1984). -6 - [subjnational goverament[s] own-tax revenue in total subnational-national government revenues or the (sub]national govermnen[s] sbare of total subnational-national government expenditures, and collusion by the ratio of grants to total subnational governments own-tax revenues.4 In the context of the Leviathan model, the coefficient of the [de]centralization ratio of taxing or spending powers, alone, has no economic interpretation although it has been used by these studies as a statistical criterion to test the fiscal decentralization hypothesis. Contrary to this hypothesis, the collusion variable would not remain unchanged with the extent of fiscal decentralization. As an inseparable component of fiscal decentralization, the extent of decentralization of taxing powers would automatically cause an increase in suboational governments own-tax revenues and thus a decrease in the collusion variable. The present paper extends the existing literature, treating fiscal decenrtalization as the simultaneous decentralization of the government's taxing and spending powers- M. The Emipirical Model The Leviathan model predicts that, other things equal, the overall size of the public sector should inversely vary with the extent of simultaneous decenltrlization of the national government's taxing and spending decisions (fiscal decentralization). Furthermore, "other things equal" should include the simultaneous transfer of the national government's revenue and of its spending powers to subnational governments (fiscal collusion). Accordingly, this paper defines 'Regarding the collusion variable, Oates (1985) used the ratio of grants over subnational govermnents revenues for intergovernmental grants but did not discuss possibility for collusion among governments. Marlow (1988) ignored collusion. Grossman (1989) dealt with collusion but used Oates' (1985) grant variable as a measure of collusion. Other studies adopted Grossman's collusion variable. -7 - the measure of fiscal decentralization, 'FIS DEC', as the ratio of total subnational governments own-source revenues used to finance their expenditures to total subnational-national government expenditures. FIS_DEC varies only with the extent of simultaneous [delcentralization of the national government's taxing and spending powers. Its variation excludes changes in the decentralization degree of expenditures fianced through sources other than the subnational govermnents own-source revenues. As a measure of fiscal collusion, "FPS COLQ, this paper uses the ratio of the national govermnent's revenues transferred to subnational governments over total subnational-national govermnent expenditures. FIS COL only varies with the variation of simultaneous transfer of the national government's revenue and assignment of its spending responsibility to subnational governments. It remains fixed with the extent of fiscal [delcentralization. FIS DEC and IS COL are independent, policy variables. The present paper adopts the measure of public-sector size, PUB_SIZ, employed by Marlow (1988), Grossman (1989; 1992), Joulfaian and Marlow (1991), and Kneebone (1992). PIJB_SIZ is defined as the total general government expenditures share of gross domestic products. To test the fiscal decentralization hypothesis, the level (or growth) of PUB SIZ are assumed to be related to the level (or growth) of FIS_DEC, FIS COL and a set of other control variables, "Z", as follows: PUB SIZ =a0+a1FISDEC + aFISCOL + C3Z+U (1) or -8 - PUB SIZ' = a + a1FIS DEC + a2FiS COL* + a3Zx + U7 (2) where PUB SIZ = ratio of total national-subnational government expenditures to GDP; FIS DEC = ratio of total subnational governments own-source revenues over total national-subnational govermnent expenditures;- FIS COL = ratio of the national government's revenues transferred to subnational governments over total nationl-subnational govemment expenditures; Z = asector of other control variables; U = disturbance terms; and superscript asterisks in equation (2) refer to the growth rate of variables. Equations (1) and (2) are similar to the estimatn equations employed by previous studies with exception of the measurements of the FIS DEC, FIS COL, FIS DEC. and FIS COL. Tfhe Leviadtan fiscal decentralization hypothesis implies al < 0. The sign of a2 may be positive or negative, depeding on whether the stimulating effect of transfer of the national government's revenue to subnational govemments (revenue sharing) on the size of the public sector would exceed or fall short of the constraining influence of the decentralized expenditures financed through revenue-transfers. The positive sign of a2 regardless of its significance level would indicate that revenue sharing with taxing decisions concentrated in the hands of revenue- maximizing national government exterminates the constraining influence of decentalinztion of the spending power, providing further support for the Leviathan model. -9- To control for the influence of other variables, the present paper follows Oates' (1985) international cross-country analysis and uses gross domestic product per capita at the constant 1987 US dollar, "GDP-PER", for Wagner's Law and total population or the share of urban population in total population, "URB POP", as a scale variable. According to Wagner's law, demand for public goods and services is more incomelastic than demand for private goods,and services, implying a positive relationship between PUB_SIZ, the demand for public goods- services relative to total demand for public-private goods-services, and GDP PER. GDP PER also controls for the positive effect of econoniic development on the size of the public sector. The more developed (industrialized) a country, the higher would be its GDP per capita and thus public-sector size. As regards the effect of scale variable, Oates (1985) has argued that the more urbanized or populated a state, the smaller should be its public sector, reflecting some economies in providing services to more densely populated areas. The counter-hypothesis argues that growth of urban or total population would increase demand for public services, encouraging the expansion of the public sector, for example, Kneebone (1992).5 17. Estimation Methods and Emipirical Results The information on public finances in the ID Government Fiancial Statistics Yearbook (1992) were used to assemble measares of PUB_SIZ, FIS DEC and FIS COL for a sample of 30 countries in 1987, the most recent year which provided the largest size of sample (30), and 1977, the earliest year for which data were available for the most of countries in the sample (26). 5It might be argued that inclusion of some exogenous variables in the model may improve the explanatory power of the equations. But the purpose of this paper is to provide an extension of the earlier studies. - 10- Growth rates of these measures over the 1977-87 period were used to generate cross-country data for PUB SIZ, FISC DEC and FIS COLO. The longer the period, the more would be the variation in these growth rates across countries. Data for other variables were extracted from World Tables (1992). The list of countries can be found in the appendix. Equations (l) and (2) were, respectively, estimated using die level of variables in .1977 and 1987 and their 1977-87 growth rates by means of least-squares method (Table 1).6 Consistent with the Leviathan model, coefficient of fiscal-decentralization variable is negative in all of the equations. But the results are suspect due to the presence of a serious degree of multicollinearity according to Klien's rule (1962). In all of the equations, multiple correlation coefficient of the dependent variable (Y) with all independent variables (X), "R23,111 ,", falls short of umltiple correlation coefficient of at least one independent variable with all other independent variables, -"R2xi.og,xs".- Such a serious degree of multicolinearity could affect the sign or size of the parameters which are crucial for the purpose of this -study. A step-wise regression procedure was employed to search for the source of multicolinearity. The scale variable was found to be the major cause of multicollinearity in all of the equations, capturing the effects of other explanatory variables without making a significant contribution to the variation of the public-sector size. This variable, which was not important for 6Foliowing Oates (1985), equation (1) was also estimated using a logistic transformation of the dependent variable to allow the dependent variable to range over the whole set of real number. consistent with Oates' findings, the results were remarkably similar to those without the transformation; therefore, those presented here do not make use of the logistic transformation. In addition to GDP per capita, the present paper also used dummy variable, ons for industrial and zero for developing countries, to capture the effect of development on the overall size of public sector; its coefficient was found to be of insignificant in explaining variation of the public-sector size. 'For more details about the effect of multicollinearity on the signs and sizes of the parameters, see: Maddala (1977), P. 185. - 11 - Table 1. Estimation Results Equation (1) Equation (2) hId. var.2 .1977 1987 -Id. var.2 1977-87 FIS DEC -0.2592 -0.2714 FIS DEC* -0.0805 (-1.15) (-1.23) (-1.13) [0.412] [0.333] [0.130] FIS COL 0.0102 0.1672 FIS_COL' 0.0004 (0.050) (0.57) (0.366) [0.123] (0.029 - [0.048] GDP_PER 0.0141 0.0109 GDP PER! -0.6151 (2.12) (2.39) (-1.79) [0.6441 [0.514] [0.261] URB POP 0.0019 0.0019 URB PPOV 0.2219 (1.04) (1.21) (0.667) [0.5241] [0.359] [0.112] CONSTANT 0.1673 0.2199 CONSTANV 0.2405" (2.44) (2.91) (2.63) F 5-.97. 4.36 F 1.50 R2 0.57 0.45 0.25 Adj.R' 0.48 0.39 Adj. R2 0.08 1 Dependent variable is the total subnational-national government expenditures share of GDP in equation (1) and its 1977-87 growth rate in equation (2). Within the parentheses are t-statistics. Within the brackets are the multiple correlaton coefficient of each independent vanable with all other independent variables included in the model. 1. . 2 FI5_DEC = total subnational governments own-source revenues over total subnational- national government expenditures; PIS_COL = the national government's revenues tr-ansferred to subnational governments over total subational-national government expenditures; GDP PER = GDP per capita at constant 1987 $US (1000); URBEPOP t the share of urban population in total population (%). Superscript s s refer to the 197747 growth rates. - 12 - the purpose of this paper, was excluded from all of the equations and the equations were re- estimated (Table 2). Test of hypotheses about the parameters, which are also crucial for the purpose of this study, would be suspect with th- presence of heteroscedasticity, and endogeneity of fiscal decentralization and collusion variables. Heteroscedasticity is often encountered in cross-section data. A test for homoscedasticity was conducted using Glejser's (1969) procedure. The test results, which can be found in the appendix, reject the alternative hypothesis in all of the equations, that heteroscedasticity exists. Finally, the overall size of public sector might endogenously influence fiscal decentralization and collusion (independent variables), creating correlation between these independent variables and error terms. This problem may emerge from certain government activities which require or imply a national government role, e.g., military spending. Growth of these government activities and consequently the overall size .of public sector may cause declining fiscal decentralization and coilusion. To address this matter, equation (1) was estmated using instrument variable tchnique. The estdmtion results, which.are presented m the appendix, were remarkably similar to the LS ones, indicating that fiscal decentralization and fiscal collusion variables are independent of the public-sector size and consequently error terms. Table 2 reports the final estimation results. The explanatory power of equation (1), in both 1977 and 1987, is higher than the explanatory power of equation (2), with the former explaining almost 70 percent of total variations in the overall size of public sector. Consistent with Wagner's Law,: the level of per capita GDP (GDP PER) has a positive and statstically significant association with the level of the public sector size (PUB_SIZ) in both 1977 and 1987. - 13 - Table 2. The Estimation Results: After Exluding the Scale Variable1 Equation (1) Equation (2) Ind. var.2 1977 1987 Ind. var.2 1977-87 In. va.17719 FIS DEC -0.3378 -0.2788a FIS DEC* -0.0957 (-1.52) (-1.38) (-1.44) [0.371] [0.332] [0.130] FIS COL 0.0571 0.2049- FIS COL 0.0003 (0.26) (0.74) (0.28) [0.105] [0-023] [0.038] GDP PER 0.0208 0.0136b GDP PER' -0.6781a (4.38) (3.80) (-1.80) [0.309] [0.3361 [0.001] CONSTANT 0.2423b 0.3172" CONSTANr 0.2649b (5.24) (7.57) (3.21) F 6.79 5.68 F 1.91 0.54 0.44 R 0.23 Adj.R2 0.46 0.37 Adj. R2 0.11 1 Dependent variable is total subnational-national government expenditures share of GDP in equation (1) and its 1977-87 growth rate in equation (2). Within the parentheses are t-statistics. Within the brackets are the multiple correlation coefficient of each independent vanable with all other independent variables included in the model. 2 FIS_DEC = total subnational governments own-source revenues over total subnational- national government expenditures; FIS_COL = the national government's revenues transferred to subnational governments over total subnational-national govermuent expenditures; GDP_PER = GDP per capita at constant 1987 $US (1000). Superscript s refer to the 1977-87 growth rates. a Significant at the 90. percent probability level, one-tail test for the coefficients of RS DEC and FIS-DEC. b Significant at 95 percent probability level. -14- As hypothesized, fiscal decentralization (FIS DEC and FIS DEC') exerts a negative influence on the size of the public sector in all of the equations, being significantly different from zero at the above 90 percent probability level. Finally, coefficient of fiscal-collusion variable (FIS_COL and FIS COL) in all of the equations is positive but not significantly different from zero. This result, which is consistent with the Leviathan model, indicates that the stimulating effect of transfer of the national government's revenue to subnational goverments (revenue sharing) significantly neutralizes the constraining influence of the decentralized expenditures financed through revenue-transfers. V. Conclusion. Earlier attempts to examine the relation between fiscal decentralization and goverment size treat fiscal decentralization as the decentralization of either taxing or spending powers, neglecting the inseparability of taxing and spending decisions which makes interpretation of the results in the context of the model difficult. The present paper extends the existing literature, arguing that fiscal decentralization is a composite, constitutional action containing the two inseparable elements of taxing and spending decisions. It ten employs an econometric model to investigate the effect of fiscal decentalization on the overall size of public sector using intenational cross-country data. The empirical results show that (i) the simultaneous decentralization of the national govermnent's taxing and spending powers exerts a negative and significant influence on the overall size of public sector; and (ii) revenue sharing with taxing decisions concentrated in the hands of national government eliminates the negative influence of decentralization of the spending power. These findings suggest that the counties pursuing the objective of a smaller public. sector but just decentalizing their spending powers should decentralize their taxing decisions as well. -15 - APPENDIX The Test Results for Homoscedasticity-H.: au=O. t-statistics for uos' Equations: e =o0+Evjxj e J =aOr+qfxi e* =crO+Ecax' e' ao+ayx - xis - 1977 1987 1977 1987 1977-87 1977-87 Formn A: FIS DEC -1.0 0.11 -1.3 0.91 0.57 0.67 FIS COL -1.03 -1.11 -1.29 -1.28 -0.51 -0.63 GDP PER 0.39 1.31 0.54 1.73 -0.78 -0-78 Form B: FIS_DEC 1.09 0.65 -1.26 0.87 -0.59 -0.63 FIS COL 0.41 -0.79 0.41 -0.67 0.35 0.34 - GDP PER -1.17 -0.18 -0.40 0.77 -0.78 -0.78 Form C: ffIS.DEC -1.26 -0.27 -1.34 -0.63 0.98 -0.58 /rF1iCOU -0.89 -0.72 -1.13 -0.85 -0.52 -0.39 /GDPLPER -0.79 -1.18 -0.27 -1.73 -1.02 -1.25 x, is the 1977-87 growth rate of x . e = absolute values of die OLS estimates of errors in equations (1) for 1977 and 1987; e* !--absolute values of the OLS estimates of errors in equation (2). -16- Estimation Results Using Instrument Variable Technique' Id. variables 1977 1987 FIS DEC -0.3648 -0.2956a (-1.58) 0 (-1.34) FIS COL 0.0724 0.2252 (0.42) (0.76) GDP PER 0.0214 0.139b (4.43) (3.71) CONSTANT 0.2405b 0.3175b (5.19) (7.43) F 6.81 5.31 0.52 0.45 Adj.R2 0.41 0.39 * Lagged FIS_DEC and FIS_COL were used as instrument variables. 2 Significant at 90 percent probability level, one-tail test for FISC_DEC. b Significant at 95 percent probability level. List of countries: Argentina Australia Belgium Brazil Canada Chile Finland France Germany - Hungary. India Ireland Israel Italy Japan Kenya Luxembourg Malaysia Mexico Netherlands Norway Paraguay, Poland Romnania South Africa. Spain Sweden Thailand UK US - 17 - REFERENCES Bird, R. and C. Wallitch (1993). "Fiscal Decentralization and Intergovermnental Relations in Transition Economies: Toward a Systemic Framework of Analysis", Policy Research Working Paper Series No. 1122, The World Bank. Brernan, G. and J. M. Buchanan (1980). The Power to Tax: Analytical Foundations of a Fiscal Constitution, Cambridge University Press, Cambridge. Brennan, Geoffrey and James M. Buchanan. "Towards a tax constitution for Leviathan," Joumal of Public Econormics 77, 255-73. Buchanan, J.M. (1977). "Why does government grow?" in T.E. Borcherding (ed.), Budgets and Bureaucrats: The Sources of Public Sector Growth, Duke University Press. Glejser, H. (1969) "A new test for heteroscedasticity", Journal of the American Statistical Association, March. Grossman, P- (1989a). "Federalism and the size of government", Southern Economic Journal, 55, 580-93. Grossman, P. (1989b). "Fiscal decentralization and government size: an extension", Public Choice, 62, 63-9. Grossman, P. (1992). "Fiscal Decentralization and public sector size in Australia", The Economic Record, Vol. 68, No. 202, 240-6. Hettich, W. and S. Winer (1984) "A positive model of taxation", Journal of Public Economics, 24, 67-87. International Monetary Fund (1992). Government Financial Staistics, Washington, D.C. Joulfaian, D. and M. Marlow (1990). "Government size and decentralization: evidence from disaggregated data", Southern Economzic Jounal, 56, 1094-102. Joulfaian, D. and M- Marlow (1991). ."Centralization and government competition", Applied Economics, 23, 1603-12. Joulfaian, D. and R. Mookerjee (1991). "Dynamics of government revenues and expenditures mi industrial economics", Applied Economics, 23, 1839-44. Klein, L.R. (1962). An Introductionfor Econometics, Englewood Cliffs, N.J.:. Prentice-Hall, Inc. Maddala, G.S (1977). Econometrics New Yorlc McGraw-Hill Book Company. Kneebone, R. (1992). "Centralization and the size of govemment in Canada", Applied Econonmcs. 24, *1293-1300 - 18 - Marlow, M. (1988). "Fiscal decentralization and government size", Public Choice, 56, 259-69. Nelson, M.A. (1986). "An empirical analysis of state and local tax structure in the context of the Leviathan model of govermnent", Public Choice, 49, 283-294. Nelson, M.A. (1987). "Searching for Leviathan: Comment and extension", American Economic Review, 77, 198-204. Oates, W. (1985). "Searching for Leviathan: An Empirical Study", American Economic Review, 75, 748-57 Oates, W. (1989). "Searching for Leviathan: A reply and some further reflections", American Economic Review, 79, 587-83. Raimnondo, H. (1989). "Leviathan and Federalism in the United States", Public Finance Quarterly, 17, 204-15. Ram, R. (1986). "Government size and economic growth: A new framework and some evidence from cross-section and time-series data", American Economic Review 76, 191-203. Shughart, W., and Tollison, R. (1986) "On the growth of goverrnent and the political economy of legislation", Research in Law and Economics 9, 11-127. Tiebout, C.M. (1956). "A Pure Theory of Public Expenditures", Journal of Political Economics, 64, 416-24. World Bank (1992). The World Tables, Washington D.C. Zax, J. (1989). "Is there a Leviathan in your neighborhood?", American Economic Reiew, 79, 560-7. Policy Research Working Paper Series Contact Title Author Date for paper WPS1371 Tne Evolution of Trade Treaties and Sarath Rajapatirana October 1994 J. Troncoso Trade Creation: Lessons for Latin Amerca 37826 WPS1372 Administrative Charges in Pensions Salvador Vald6s-Prleto October 1994 E. Khine in Chile, Malaysia, Zambia, and the 37471 United States WPS1373 Firm Behavior and the Labor Market Simon Commander October 1994 B. Keller in the Hungarian Transition Janos Kollo 35195 Cecilia Ugaz WPS1374 Infrastructure Finance: Issues, Anand Chandavarkar November 1994 M. Gellar Institutions, and Policies 31393 WPS1375 Policy Lessons from a Simple Shantayanan Devarajan November 1994 C. Jones Open-Economy Model Delfin S. Go 37754 Jeffrey D. Lewis Sherman Robinson Pekka Sinko WPS1376 How Trade, Aid, and Remittances Maurice Schiff November 1994 J- Ngaine Affect International Migration 37947 WPS1377 Macroeconomic Adjustment to Vittorio Corbo November1994 RF Vo Capital Inflows: Latin American Style Leonardo Hemandez 33722 versus East Asian Style WPS1378 Mexico after the Debt Crisis: Daniel Oks November1994 S. Harbi Is Growth Sustainable? Sweder van Wijnbergen 37143 WPS1379 Financing Infrastructure in Developing Barry Eichengreen November1994 WDR Countries. Lessons from the Railway 31393 Age WPS1380 Transfers and the Transiton from Kathie Krunmm November 1994 K Krumm Socialism: Key Tradeoffs Branko Milanovic 34263 Michael Walton WPS1381 Welfare Economics, Political Ravi Kanbur November 1994 M. Youssef Economy arid Policy Reform in Ghana 34614 WPS1382 Saving, Investment and Growth in Klaus Schmidt-Hebbel November 1994 E. Khine Developing Countries: An Overview Luis Serven 37471 Andres Solimano WPS1383 Rural Demand for Drought Insurance Madhur Gautam November 1994 C. Spooner Peter Hazell 30464 Harold Alderman Policy Research Working Paper Series Contact lltle Author Date for paper WPS1384 Fiscal Decentralization and David Sewell November 1994 S. Langton Intergovernmental Finances in the Christine I. Wallich 38392 Republic of Albania WPS1385 Fiscal Federalism Dimensions of Tax Robin Boadway November 1994 C. Jones Reform in Developing Countries Sandra Roberts 37754 Anwar Shah WPS1386 EU Bananararna Il Brent Borrell December 1994 G. llogon 33732 WPS1387 Fiscal Decentralization and the Size Jaber Ehdaie December 1994 C. Jones of Government An Extension with 37699 Evidence from Cross-Country Data WPS1388 Does Voice Matter? For Pubic Samuel Paul December 1994 B. Moore Accountability, Yes 35261