POVERTY THE WORLD BANK REDUCTION AND ECONOMIC MANAGEMENT NETWORK (PREM) Economic Premise JUN 2012 DECEMBER 010 • Numbe 99 • Number 1 74074 Public Spending for Long-Run Growth: A Practitioners’ View Norman Gemmell, Florian Misch, and Blanca Moreno-Dodson By financing public goods and services that enhance productivity and promote private investment, public spending is widely believed to be critical for long-run growth. Such effects are distinct from any short-run Keynesian response to a public spending stimulus. While a short-run response generally operates through aggregate demand, long-run growth effects alter aggregate supply conditions. While academic literature generally supports the belief that public spending promotes growth in the long run, understanding which public expenditure allocations can trigger such effects in a particu- lar country setting is challenging in practice. The objective of this note1 is to review the trade-offs faced by fiscal policy makers in developing countries who are considering using public expenditure policy as an instrument to promote long- run growth, provide guidance from the empirical literature, and review the types of data sources that are helpful in this context. The existence of long-run growth2 effects from public spend- put level as a function of factor inputs (capital and labor), as ing is hardly contested in policy debates. At the international well as the productivity through which these inputs are level, assistance to developing countries comes mostly in the combined, have been extended by incorporating various ele- form of loans and grants to create fiscal space to ultimately ments of fiscal policy. In particular, the growth models as- increase public spending, at least temporarily. In the Europe- sume that the government raises taxes to finance various an Union, the second largest budget item is the cohesion poli- types of public expenditures that raise the marginal product cy, which essentially consists of grants that are mostly given to of factor inputs in the output production process. This may, relatively less developed regions to spur growth. Finally, at the for instance, be motivated by the fact that private equip- national level, similar systems exist in many countries, where ment, such as machinery and vehicles, can be employed relatively rich regions or the central government make trans- more productively when public infrastructure is in place. fers to support long-run growth in economically lagging parts Other models embody additional transmission channels of the country. through which public spending affects aggregate private in- vestment. In neoclassical models, there are effects on growth Supporting Theoretical and Empirical only for a transitional period, as the economy moves to its Evidence new level of output—though the length of this transition re- Growth literature that has emerged over the last 20 years mains subject to debate and may last a long time. The 1990s, supports the belief that public spending affects long-run however, saw the development of endogenous growth mod- growth.3 First, standard growth models explaining total out- els in which, under certain assumptions, various fiscal policy 1 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise parameters have more persistent effects by affecting the tion, the growth effects of alternative offsetting changes are long-run growth rate of the economy. likely to differ in magnitude and time horizon. Second, empirical evidence increasingly suggests that, in The composition of public spending both developed and developing countries, fiscal policy affects Since not all types of public spending have the same level of long-run growth, which broadly confirms the predictions of importance, increasing certain public spending types at the the theoretical literature. These studies consider fiscal policy expense of others, on a pro rata basis, affects long-run growth. parameters as explanatory variables and identify various pub- Therefore, the relative productivity of different public expen- lic expenditure categories that tend to promote long-run diture categories is critical for policy makers to determine the growth. Although most of the existing studies are based on composition of public spending. To this end, policy analysis cross-country data, which have been somewhat challenged by sometimes distinguishes between capital and current govern- various papers, more recent studies have resolved a number of ment spending to predict the growth effects of public spend- econometric issues and appear to be more robust. In addition, ing. Underlying this categorization is the belief that capital a few more recent papers use national and subnational fiscal spending leads to the accumulation of public capital and policy data from one country. There is also a large body of lit- therefore to higher economic growth, whereas current or con- erature on the impact of the public infrastructure stock on sumptive spending affects, at best, welfare, while being economic performance, which has been reviewed in detail by growth neutral, or in a more pessimistic scenario, is growth Romp and de Haan (2007). inhibiting and may not even affect welfare. Third, there is strand of the growth literature, including However, this approach is being increasingly questioned Temple (1999) and Rodrik (2003), that identifies more fun- by economic research. In models of growth and public finance, damental determinants of growth.4 This literature has identi- the growth effects of public spending depend on whether the fied three deep determinants of growth: trade, institutions, particular types of public spending affect the productivity of and geography. Here, anecdotal evidence also suggests that labor and private capital, and on the magnitude of these ef- many of these factors and/or their effects are influenced by fects. In other words, whether public spending results in the public expenditure. Poor geography may at least partially be accumulation of public capital is less relevant. Rather, through overcome through, for example, public spending on better in- affecting private productivity, certain types of public spend- frastructure links, public health programs, and agricultural ing potentially raise the returns to investment and thereby the research. The quality of institutions also depends on public rate of total (private and public) capital accumulation. It is the services such as the judiciary and public administration, latter transmission channel that is essential to understand the which are both, at least partially, financed through public effects of public spending on growth. Further, alternative ap- spending. Finally, Anderson and van Wincoop (2004) show proaches emphasize that it is essential to consider capital and that trade costs are large and likely to affect trade patterns. current expenditures together, given the fact that they usually Further, trade costs are influenced by the level and quality of have a joint effect on growth and that their distinction is often publicly financed services, partially or totally, such as cus- mechanical and artificial in practice. toms, trade logistics, and transport links. In this note, the term “growth-enhancing expenditure�5 refers to public spending categories that affect private sector Fiscal Policy Challenges in Practice productivity and private investment. In the context of devel- From a policy perspective, it is essential to understand what oping countries, empirical studies usually consider public types of public spending allocations promote long-run growth spending on education and research, health, transport, com- in a particular country setting. This is challenging, however, munication, water supply, energy, and law enforcement to be, given that fiscal policy is subject to inherent trade-offs that a priori, growth enhancing. However, both the intersectoral arise in part due to government budget constraints, which are and intrasectoral composition of public spending that are difficult to evaluate. The fundamental property of the govern- conducive to growth need to be determined in the specific ment budget constraint is that, like any other identity, it re- country context. By contrast, there are those public spending quires every fiscal change to be offset by a compensating categories that are, a priori, not expected to affect private sec- change. Increases in a certain category of public spending tor productivity, and, taken by themselves, are therefore ex- must be financed by increases in tax revenue, in the deficit, or pected to be growth neutral, at best. However, it is important in the level of grants. Otherwise, other types of spending to recognize that nonproductive public spending may fulfill would have to be lowered, or the offsetting mechanism could important roles from a welfare point of view. be a combination of some or all those elements. The net ef- The level of public spending fects of any fiscal change on growth therefore depend on the Given the government budget constraint, the effects of in- way it is offset. The direct and indirect growth effects of the creasing the overall level of public spending on growth are am- offsetting element may have similar or opposite signs. In addi- biguous. First, they depend on whether there are across-the- 2 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise board changes or changes to the particular types of public taxation, or deficit, used to finance them. Positive, negative, or spending being increased. However (and second), even if the zero effects are each possible, depending on the combinations overall spending increase is driven by spending categories that of different elements of fiscal policy used. are supposedly growth promoting, the government budget Table 1 suggests that reallocating public resources from constraint dictates that the “net effects� of public spending growth-neutral categories to growth-enhancing categories can increases depend on the exact financing mechanism, that is, be expected to have a positive impact on growth. Similarly, an how such an increase is financed. This implies that even if cer- increase in growth-enhancing expenditures financed by non- tain types of public expenditures are expected to be beneficial distortionary taxation is also likely to promote long-run for growth, governments should not automatically finance growth. By contrast, it is more difficult to judge whether debt- them. financed increases in growth-enhancing public spending are Short-run versus long-run growth beneficial or not. While unsustainable public debt levels do During economic downturns, governments may increase endanger macroeconomic stability and therefore undermine public spending as a means of stabilizing aggregate demand. private investment and growth, it can be argued that growth- Ideally, stimulus packages meet both the objectives of short- enhancing spending, for instance, will pay for itself through run output stabilization and long-run growth promotion, faster long-run growth, which, in turn, generates higher pub- but, in practice, numerous trade-offs arise. For example, pub- lic revenue. However, this is only the case if the returns on lic investment items are often considered a key element of public spending, reflected in higher growth rates, exceed the short-run stimulus programs. However, existing evidence on interest on debt to be repaid, which is often relatively high in their impact is mixed, and in some cases negative. The objec- developing countries, reflecting the perceived danger of sover- tive of short-run stabilization often dictates that any public eign default. The literature therefore suggests that the levels investment projects financed be labor intensive and able to be of both the deficit and the accumulated debt matter for the quickly implemented. By contrast, large and complex invest- growth effects, as shown in table 1. ment projects (such as major roads, railways, or electricity In addition to the direction of the growth effects, esti- generation facilities), which are often required to remove mates of the magnitude of some (but not all) types of public bottlenecks for growth, are more likely to be subject to long spending changes can be extracted from this literature. How- implementation lags and may be less labor intensive, and ever, observation shows a fairly high degree of robustness is therefore less attractive as stimulus plans. Finally, stimulus still lacking. Some of the differences in the results can be at- programs aimed at short-run stabilization may be designed to tributed to the econometric specifications used and the ex- be more labor intensive than required, thereby creating inef- penditure changes being estimated, but this feature makes it ficiencies even if, in principle, they could contribute to long- hard to compare and to verify individual parameter estimates. run growth. In addition, using the reported estimates to predict the effects of changes of functionally disaggregated intrasectoral expen- Policy Insights from Fiscal Policy Growth Literature Table 1. Expected Growth Effects of Public Spending Changes Empirical growth literature presents predictions for fiscal changes that promote long-run growth; table 1 includes a Increase of / Growth-enhancing Growth-neutral summary of these predictions. The literature distinguishes financed by expenditure expenditure between growth-enhancing and growth-neutral public Growth-enhancing Negative growth spending, as described above, and differentiates between tax - expenditure effects types labeled as “distortionary,� expected to adversely affect Growth-neutral Positive growth growth through lowering the returns to private investment, - expenditure effects and those labeled as “nondistortionary,� which, in principle, should not affect private investment choices and can there- Ambiguous growth Negative growth Distortionary taxation effects effects fore be expected to be growth neutral. In the context of this literature, corporate and personal income taxes are consid- Positive growth Nondistortionary taxation Zero growth effect ered more distortionary, while consumption taxes are usu- effects ally seen as nondistortionary with respect to economic activ- Debt accumulation (deficit Positive growth Negative or zero ity and growth. < 1.5% of GDP) effects growth effects Categorizing public expenditures and tax types in this Debt accumulation (deficit Negative growth Negative growth way allows analysis to visualize potential long-run growth ef- > 1.5% of GDP) effects effects fects of fiscal policy that depends on both the level and the Source: Adam and Bevan (2005). composition of public expenditures, as well as the forms of Note: GDP = gross domestic product. 3 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise diture categories—for instance, expenditure types within the 2. In this context, “long-run� as a time period is defined as lon- education sector (primary, secondary, and so forth)—is also ger than three years. problematic for various reasons. Gemmell, Misch, and More- 3. Gemmell, Misch, and Moreno-Dodson (2012) briefly sum- no-Dodson (2012) provide detailed summary tables contain- marize this literature. ing the estimates of the growth effects. 4. Other prominent contributions include Sachs and Warner Especially in cross-country studies, unobserved heteroge- (1995); Krueger (1998); Acemoglu, Johnson, and Robinson neity may be one of the reasons why some of the estimates are (2001); and Sachs (2001). not robust. In turn, one potential source of heterogeneity is 5. In the literature, this type of public spending is often called related to unobserved differences in the actual execution/im- “productive� and “core.� plementation of public spending—that is, the extent to which 6. Examples of tracking surveys include López-Cálix, Alcazar, public spending actually translates into public services deliv- and Wachtenheim (2002) and Reinikka and Svensson (2004). ered (such as teachers performing their duties in school and 7. Gemmell, Misch, and Moreno-Dodson (2012) outline students learning more) and public infrastructure assets being these approaches in greater detail. built (such as roads being constructed and used by the public). Poor governance is the prime cause of low levels of pub- References lic spending execution/implementation, which in turn may Acemoglu, D., S. Johnson, and J. Robinson. 2001. “The Colonial be a significantly more important policy issue than both the Origins of Comparative Development: An Empirical Investiga- composition and level of public spending. Evidence from tion.� American Economic Review 91 (5): 1369–1401. public expenditure tracking surveys that trace the flow of Adam, Christopher S., and David. L. Bevan. 2005. “Fiscal Deficits and Growth in Developing Countries.� Journal of Public Eco- public resources from origin to destination suggests that low nomics 89: 571–97. accountability and public spending leakages are the primary Anderson, James E., and Eric van Wincoop. 2004. “Trade Costs.� concern of policy makers and the area in which improve- Journal of Economic Literature XLII: 691–751. ments may have greater effects on growth.6 For example, Gemmell, Norman, Florian Misch, and Blanca Moreno-Dodson. studies may find that public spending on roads has hardly 2012. “Public Spending and Long-Run Growth in Practice: any growth effects simply because the resources are ulti- Concepts, Tools, and Evidence� (chapter 2). In Is Fiscal Policy the mately deviated for other purposes and not used for road Answer? A Developing Country Perspective, ed. Blanca Moreno- construction, although they still appear in the government Dodson. Washington, DC: World Bank. budget as infrastructure spending. Krueger, A. 1998. “Why Trade Liberalisation Is Good for Growth.� Economic Journal 108 (450): 1513–22. As a consequence, it is important that any country-level López-Cálix, J. R., L. Alcazar, and E. Wachtenheim. 2002. “Peru: analysis of the growth effects of fiscal policy exploits data Public Expenditure Tracking Study.� World Bank, Washington, sources that are more innovative compared to cross-country DC. http://siteresources.worldbank.org/INTPUBSERV/Re- macrodata, which existing literature predominantly uses. Ex- sources/peru.pets.concept.sep9.2002.pdf. amples of such alternative data include subnational fiscal Moreno-Dodson, Blanca (ed.). 2012. Is Fiscal Policy the Answer? A policy data, firm-level data, and internationally comparable Developing Country Perspective. Washington, DC: World Bank. indicators of government effectiveness that are available for Reinikka, R., and J. Svensson. 2004. “Local Capture: Evidence from the majority of countries and that may be helpful to evaluate a Central Government Transfer Program in Uganda.� Quarterly Journal of Economics 119 (2): 679–705. the effects of fiscal changes.7 Rodrik, D. 2003. “Institutions, Integration, and Geography: In About the Authors Search of the Deep Determinants of Economic Growth.� In In Search of Prosperity: Analytical Narratives on Economic Growth, Norman Gemmell is Holder of the Chair in Public Finance at ed. D. Rodrik. Princeton, NJ: Princeton University Press. Victoria University in Wellington, New Zealand. Florian Misch Romp, W., and J. de Haan. 2007. “Public Capital and Economic is Deputy Head of the Corporate Taxation and Public Finance Growth: A Critical Survey.� Perspektiven der Wirtschaftspolitik 8 Department at the Centre for European Economic Research in (S1): 6–52. Mannheim, Germany, and a Consultant at the World Bank. Sachs, J. 2001. “Tropical Underdevelopment.� Working Paper 8199, National Bureau of Economic Research, Cambridge, MA. Blanca Moreno-Dodson is a Lead Economist at the World Bank. Sachs, J., and A. Warner. 1995. “Economic Reform and the Process Notes of Global Integration.� Brookings Papers on Economic Activity 26: 1–118. 1. This Economic Premise summarizes Gemmell, Misch, and Temple, J. 1999. “The New Growth Evidence.� Journal of Economic Moreno-Dodson (2012). Literature 37 (1): 112–56. The Economic Premise note series is intended to summarize good practices and key policy findings on topics related to economic policy. They are produced by the Poverty Reduction and Economic Management (PREM) Network Vice-Presidency of the World Bank. The views expressed here are those of the authors and do not necessarily reflect those of the World Bank. The notes are available at: www.worldbank.org/economicpremise. 4 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise