41105 July 2007 ˇ Number 108 A regular series of notes highlighting recent lessons emerging from the operational and analytical program of the World Bank`s Latin America and Caribbean Region. Public Sectors in the Americas: How big are they? Mauricio Carrizosa This note provides selected measurements of public sec- vices. Table 1 ranks the 50 countries and territories in tor size for the countries and territories of the Americas the Americas by the ratio of general government con- and discusses the resulting differences across countries sumption to GDP (Column 2). Territories are included and territories and across the different measurements in order to provide a wider variety of experience, as of size. The key conclusions are that public sector sizes financial dependence on higher-level governments is differ considerably across countries and territories, that predicted to raise public sector sizes in the dependent there are differences in the public sector size ranking governments. of countries and territories depending on which mea- surement is looked at, and that there is a wide variety General government consumption ranges from 6 per- of reasons that may account for or explain why a given cent of GDP in Haiti and Guatemala to 110% of GDP country has a large or a small public sector. in Falklands-Malvinas. Even if the latter's extreme value is discarded , the highest (French-Guiana) is still The size of the public sector in a given country is a ten times higher than Haiti. Not surprisingly, if territories matter of considerable concern from the economic per- are excluded, the highest level of general government formance perspective. Within the public sector, a gov- consumption is Cuba's, at 38% of GDP. General govern- ernment takes up resources by taxing economies in one ment consumption in more than half of all cases (28 of way or another and provides public goods and services. the 50) is within the 10%-20% of GDP range. Govern- Taxation often undermines economic performance by ment consumption in territories tends to be relatively changing incentives in the wrong direction (say by en- higher, exceeding 35 percent of GDP in seven of the couraging too much leisure). Expenditure may improve fourteen territories. performance by, for example, improving property rights. It may also undermine performance by diverting resourc- Differences in government consumption are high- es from better uses. lighted when expressed in per capita terms. The ratio of government consumption to GDP is a relative measure In addition to government units, the public sector in- that abstracts from the absolute level of the country's cludes public corporations. The impact of the public sec- per-capita GDP. The corresponding per capita measure is tor on economic performance is correspondingly broader obtained by (a) expressing general government consump- than that of government. To the efficiency effects of taxes tion in per-capita terms after (b) converting general gov- and expenditures we must add the efficiency effects of ernment consumption to international (purchasing power public financial and non-financial enterprises. parity, PPP) dollars. The resulting measure (Column 3 of Table 1) vividly shows the enormous differences in Several measures of the public sector or of its compo- the per capita cost of public service attention, from $110 nents are available. These comprise measures of rev- in Haiti to $13.795 in Greenland. The country rankings enues, expenditures, employment, and institutions. The change considerably, with richer countries and territories most widely used measure of government size is general having larger per-capita government consumption levels. government consumption, defined as the sum of general government expenditures on wages, goods, and services. The sum of general government consumption and pub- It is a measure of the cost of providing government ser- lic investment, expressed as a percentage of GDP (Col- umn 4 of Table 1), provides a measure of public sector pension benefits. As a percent of GDP, interest and size from the perspective of final aggregate demand. transfers tend to be higher in small countries (as high as Because it excludes public expenditures on transfers, 24% of GDP in one island). Including the overall burden public sector final demand is an indicator of public sector of debt and pensions alters a few of the rankings consid- size that is tailored to assess the potential macroeconom- erably as compared with the consumption measure (e.g., ic effects of public expenditure. In comparison with gov- Jamaica, one of the highly indebted countries, ranks 24 ernment consumption, it is likely to be more unstable be- instead of 38 and Colombia ranks 31 instead of 40). cause it includes public sector investment. The shares of public sector final demand in GDP suggest that in most The measure of total public sector expenditure will not countries final public expenditure is an important policy fully capture the size of the public sector from a produc- tool. In half of the countries or territories in the Ameri- tion perspective because general government consump- cas, the share of final public expenditure in GDP exceeds tion includes wages only to general government employ- that in the USA (22%), where public expenditure is con- ees. A broader measure is the ratio of total public sec- sidered a central tool of macroeconomic management. tor employment to the labor force (Column 6 of Table 1). It is not easy to obtain homogenous data for this Rankings based on public sector final demand differ measure, although the increasing availability of house- somewhat from those based on government consump- hold and labor market surveys has made it less difficult. tion. Montserrat and Guyana are noteworthy cases of Under this measure, Haiti's remains the smallest public very large investment programs, exceeding 20% of GDP. sector (1.6%) and Cuba's is the largest (84%). In 30 of In the first, public investment is driven by UK-funded the 50 countries or territories in the Americas public efforts at reconstruction from the devastation caused by employment is within 5%-15% of the total labor force. the eruption of the Soufriere Hills volcano that began in 1995. In Guyana, public investment is driven by govern- In contrast to previous indicators, the employment mea- ment efforts to restructure the economy, particularly the sure is based on physical units (employees), not on mon- sugar sector. At the other end of the spectrum, public in- etary units. When the public sector is measured from vestment in Argentina, Aruba, the Netherlands Antilles, the production perspective, rankings change, some and Peru was less than 2% of GDP in the year of refer- times considerably. For example, Argentina, Chile, and ence. Public investment of course may change sharply Mexico rank similarly in terms of government consump- from year to year, as it is one of the few expenditure tion; yet Chile's public employment is a much smaller items that governments can change easily in response to share of the labor force than Argentina's or Mexico's. changing needs and resources. Colombia's government consumption is larger than Costa Rica's; yet Colombia's public employment share Adding net lending, interest on the public debt, and is much lower than Costa Rica's. These differences in general government transfers to the private sector ranking may reflect differences in the number and size (e.g., pension benefits) to consumption and investment of public enterprises, in the "labor intensity" of govern- provides a measure of total public sector expenditure ment, in the relative average wage of public sector em- (Column 5 of Table 1). Total public expenditure is ployees or in errors of observation. roughly what governments must finance by getting grants or levying taxes, including future taxes (borrowing) and What explains differences in public sector size across the implicit 100% taxation of state-owned enterprises countries? A myriad reasons account for these differ- (SOEs). Total public sector expenditure exceeds 30% of ences. These include differences in relative public sector GDP in more than half of the cases (30), including Brazil wages, debt levels, the size of the bureaucracy, the size and Colombia. The highest levels of public expenditure of the elderly population receiving pension benefits, and (as a share of GDP) occur primarily in territories that so on and so forth. For example, the size of the public receive grants from higher-level governments (e.g., Gua- debt accounts for more of Jamaica's public sector size deloupe from the French government). than for Puerto Rico's similarly sized total public sector expenditures. There are very large gaps in many countries between total public expenditure and total final public sector Fundamentally, however, public sector size results demand. The difference reflects the burden of debt and from a mix of political and economic factors. Tradi- 2 ˇ July 2007 ˇ Number 108 tional "demand" factors that have been advanced to ex- constrained by large degrees of informality, which partly plain the growth of the public sector include GDP growth offsets their natural-resource-based tax buoyancy. Sixth, ("Wagner's Law, the hypothesis that Government goods weak institutional or absorptive capacity may well limit and services are income-elastic), population (to reflect public expenditure in countries like Haiti. Of course, in possible scale economies in public sector provision) and any given country, several factors are at work simultane- changes in the relative price (cost) of its goods and ser- ously or at one time or another. vices. Other, less traditional variables are also invoked, including the population's age structure (e.g., older popu- The net result of the factors outlined above may well be lations require more social security), informality (which public sector sizes that are not optimal (i.e., that do not raises the cost of collecting taxes), openness (which maximize welfare). Indeed, cross-country econometric raises the demand for income security or may lower evidence often suggests that increases in the share of tax/tariff collection), labor intensity ("Baumol's disease", general government consumption reduce growth. Under whereby relative labor intensity in Government raises the Barro (1990) model, this implies that government government size by constraining government productivity consumption levels are larger than those that would growth), income inequality (which triggers more redistri- maximize growth. Under certain assumptions, the size at bution through the public sector), and transfer revenues which growth is maximized can be regarded as "optimal" (the "flypaper" effect that transfers, which are politically (i.e., as maximizing household welfare). cheaper than taxes, encourage spending). Ideology (with governments that lean towards the left being relatively The wide range of government sizes shown by the data more inclined to public provision of goods and services) may be consistent with the view that the optimal size of and veto power may also play a role. government is not unique. Presumably the optimal size of a government with poor governance or effectiveness Short of a full panel econometric exercise, we can only in the delivery of public sector goods and services would highlight some empirical regularities or illustrations be smaller than that of a government with good institu- linked to various hypotheses. A look at Table 1 suggests tional capacity. Nevertheless, while the data collected in some. First, plotting per-capita government consump- this note shows a wide range of sizes, most tend toward tion against per-capita income shows that as we move a central range: whether optimal or not, countries tend to richer countries, government consumption rises more to choose government consumption levels between 10% than proportionately, with an elasticity of 1.2. This argues and 20% of GDP and public employment between 5% in favor of Wagner's Law that the "demand" for govern- and 15% of the labor force. ment services is income elastic. Second, most of the large public sectors (e.g., those with government consumption References greater than or equal to 20% of GDP) are in small coun- Barro, Robert J. (1990). "Government Spending in a tries or territories. This suggests that economies of scale Simple Model of Endogenous Growth," Journal of Politi- in delivering public services play a role. Third, financially cal Economy, 98, 5 (October), part II, S103-S125. dependent territories (more often than not also small), such as the French departements also tend to have larger IMF (2001). Government Finance Statistics Manual, public sectors. This points to the possible role of grants http://www.imf.org/external/pubs/ft/gfs/manual/pdf/all. in encouraging large public sectors, the so-called flypaper pdf, p.10. effect. In contrast, most UK territories (Montserrat ex- cluded) receive little help and have accordingly smaller public sectors. Fouth, in Cuba and indeed in some other countries in past historical periods (e.g., Guyana until the 1980s, Nicaragua during the 1980s), weak veto power About the Author may help explain their large government sizes at the time. Mauricio Carrizosa is Sector Manager of the Economic However, weak veto power in Chile after 1974 actually Policy group for Latin America and the Caribbean at the helped explain the opposite - a decline in public sec- World Bank. His e-mail address is Mcarrizosa@worldbank. tor size -, pointing to the interaction of veto power with org. Comments and suggestions provided by Tito Cordella, Nick Manning, Errol George Graham, Ulrich Lachler and ideology or the policy stance. Fifth, public sector size Elizabeth Ruppert are gratefully acknowledged. in countries such as Bolivia, Peru and Mexico may be July 2007 ˇ Number 108 ˇ Table 1: Public Sector Size in the Americas (2005) Percent of GDP except for Columns (3) and (6) Share of Government Government Consumption Government Consumption Total Public Share of Public Country/Territory Consumption in GDP PPP$ Per Capita plus Investment Expenditure Sector in Labor (1) (2) (3) (4) (5) Force (6) 1 Falklands-Malvinas 110.1% 21566 126.2% 150.0% 90.2% 2 French Guiana (France) 59.6% 4843 68.1% 81.0% 17.0% 3 Greenland (Netherlands) 51.1% 13795 57.6% 73.1% 43.3% 4 Montserrat (UK) 49.9% 3372 72.0% 90.3% 83.6% 5 Martinique (France) 38.1% 5855 42.7% 60.0% 11.8% 6 Guadeloupe (France) 38.0% 2990 42.8% 62.3% 12.0% 7 Cuba 38.0% 1256 47.6% 59.6% 84.2% 8 Virgin Islands (US) 35.2% 4833 40.7% 47.2% 24.1% 9 Suriname 31.0% 1436 35.5% 44.7% 37.4% 10 Aruba (Netherlands) 26.6% 5982 28.2% 36.2% 12.9% 11 Canada 23.7% 7792 27.7% 36.7% 16.9% 12 St. Kitts and Nevis 23.3% 3368 29.9% 43.3% 41.6% 13 St Lucia 23.0% 1466 31.4% 38.0% 12.1% 14 Antigua and Barbuda 22.0% 2714 27.8% 44.3% 32.0% 15 Guyana 21.6% 975 35.5% 57.7% 9.1% 16 Barbados 21.0% 3771 25.1% 42.0% 18.1% 17 Brazil 20.1% 1940 23.1% 39.0% 10.4% 18 St. Vincent and the Grenadines 20.0% 1370 33.2% 39.7% 11.6% 19 Dominica 19.0% 1142 27.8% 40.0% 21.2% 20 USA Netherlands Antilles 19.0% 7952 21.6% 36.4% 13.9% 21 (Netherlands) 18.9% 2899 19.6% 31.3% 8.3% 22 Turks and Caicos Islands (UK) 17.1% 1732 23.6% 31.5% 11.0% 23 Grenada 17.0% 1398 32.8% 38.9% 9.4% 24 Virgin Islands (UK) 16.8% 4091 20.7% 24.5% 38.0% 25 Trinidad and Tobago 15.0% 2064 22.2% 34.2% 23.6% 26 Bahamas 15.0% 2645 18.5% 23.1% 18.8% 27 Cayman Islands (UK) 15.0% 6463 17.5% 22.4% 7.1% 28 Anguilla (UK) 14.7% 1267 25.5% 35.0% 25.0% 29 Honduras 14.0% 390 20.1% 29.8% 5.6% 30 Panamá 14.0% 1104 16.5% 25.2% 11.3% 31 Belice 13.9% 1057 18.0% 28.0% 11.8% 32 Venezuela 13.0% 849 22.0% 31.4% 9.4% 33 Ecuador 12.5% 534 18.2% 23.8% 6.5% 34 Puerto Rico (USA) 12.0% 2225 17.8% 33.2% 23.0% 35 Chile 12.0% 1516 15.4% 27.4% 6.2% 36 Mexico 12.0% 1225 14.8% 23.3% 11.1% 37 Argentina 11.9% 1713 13.9% 22.0% 11.2% 38 Jamaica 11.0% 484 16.0% 35.5% 10.5% 39 Uruguay 11.0% 1118 13.3% 33.0% 14.5% 40 Colombia 11.0% 855 16.8% 31.0% 4.9% 41 Nicaragua 11.0% 405 17.8% 27.0% 6.9% 42 Bermuda (UK) 11.0% 6568 14.4% 21.3% 10.6% 43 El Salvador 11.0% 583 13.3% 19.4% 7.3% 44 Paraguay 10.9% 541 14.8% 22.3% 8.3% 45 Bolivia 10.8% 308 16.5% 33.9% 6.3% 46 Perú 10.0% 623 11.7% 16.8% 6.0% 47 Dominican Republic 9.8% 741 14.5% 19.2% 9.6% 48 Costa Rica 9.0% 899 12.4% 20.5% 12.9% 49 Haiti 6.0% 110 10.5% 13.7% 1.6% 50 Guatemala 6.0% 268 8.2% 12.8% 4.7% Sources: Official Government websites, World Bank Public Expenditure Reviews and Purchasing Power Parity (PPP) GDP statistics, IMF Stand-by reports, CIA World Factbook, Household Surveys, Miscellaneous country and press reports, and own estimates. "en breve" is produced by the Knowledge and Learning Team of the Operations Services Department of the Latin America and the Caribbean Region of the World Bank - http://www.worldbank.org/lac ˇ July 2007 ˇ Number 108