Journal Issue

Government employment and pay: some international comparisons: Data from 85 countries examined

International Monetary Fund. External Relations Dept.
Published Date:
September 1983
  • ShareShare
Show Summary Details

Peter Heller and Alan Tait

One aspect of the growth of the government in recent decades has been the emergence of the public sector labor force as a key element in the economies of both developed and developing countries. This is manifested in several ways. In countries where government employment occupies a significant place in the modern labor force, the government’s demand for skills and its wage policies are likely to have an important influence on wages throughout the economy and on the distribution of income. Since wages are a significant share of government expenditure, wage and employment decisions are often central factors in the adjustment of budgetary policies. Government employment is frequently seen as a policy instrument for achieving macroeconomic objectives, both in the short and medium term. Government wages and salaries also have an immediate impact on the economy through their claim on the government budget and thus indirectly on prices, investments, and the balance of payments.

Yet it is remarkable how little information is readily accessible on public sector employment and pay in most countries. The work on which this article is based relied almost entirely on the data drawn from 64 developing countries and 21 countries belonging to the Organization for Economic Cooperation and Development (OECD). The range and depth of these national statistics proved extremely varied. International comparisons are further complicated by lack of agreement on how to solve such conundrums as defining the unit of government or the provision of public services or measuring public sector employment. Comparisons of wages and salaries, too, are complicated by special allowances for food, transport, and housing, and further biased by the payment of bonuses and special pension arrangements.

This article is based on a longer study on the same topic, published by the fund as Occasional Paper No. 24 and is obtainable from the Publications Unit. See page 9 for ordering details.

This article does not solve definitively these knotty problems but it pulls together, as far as we know, the best available statistics and presents some of the more interesting results on patterns of public sector employment and wages. It is hoped that by establishing the importance of this topic, it will stimulate further debate and analysis.

How large is government?

The share of government employment in total nonagricultural employment is an important indicator of the likely impact of public sector wage policies on modern sector wage rates, the distribution of income, and the structure of output. It appears that the threshhold for potential leverage is a public sector employment of about 20-25 per cent of nonagricultural sector employment; over that proportion, public wage awards seem to affect national wage rate determination, under it, the leverage appears less important.

The clear message from the study’s statistics is that government wage policy is more likely to have a significant impact on overall remuneration of nonagricultural employees in developing countries, where total public sector employment averages 44 per cent of nonagricultural employment (Table 1). In some countries—Benin, Ghana, and Zambia, for instance—the ratio is over 70 per cent. The ratio in the OECD countries implies less leverage with public sector employment at about 24 per cent.

Table 1Relative size of government employment
Developing countries
OECD countriesTotal sampleAfricaAsiaLatin America
As per cent of nonagricultural employment
Central government8.723.430.813.920.7
State and local government11.
General government19.
Nonfinancial public sector enterprises4.113.918.715.75.5
Total public sector employment24.243.954.436.027.4
As per cent of total population
Total public sector employment9.
Source: The full data set, as well as standard deviations for each of these estimators, is provided in the longer study cited on the first page of the article.

The proportion of government employees to population is a rough measure of the magnitude of public services provided, though not of their quality or efficiency. The OECD countries have, on average, seven government employees per hundred inhabitants at the central, state, and local governmental levels (excluding employees of nonfinancial public enterprises) in contrast to only three in the developing world.

Typically, the developing countries are much more centralized than developed countries; on average, their central government constitutes about 85 per cent of total government employment, compared with only 42 per cent in the OECD countries. The difference primarily reflects higher employment at the state and local government levels in the OECD countries. Nonfinancial public enterprises loom much larger in developing countries—where their share of total public sector employment averages approximately 29 per cent, compared with 17 per cent in the OECD countries. One might conclude that the greater fragmentation of authority over wage decisions between the central and local authorities in the developed countries would limit the impact of government wage policies on the private sector relative to developing countries, where the central government is so much more dominant. The extent of central authority over nonfinancial public enterprise wage policies is likely to depend on the legal structure and the relative autonomy of enterprises.

These results are paralleled in the distribution of wages and salaries by level of government. About 70 per cent of total general government wages in federal economies is paid at the state and local governmental levels. But the share of the local government wage bill is also strikingly high in many countries normally considered as centrally dominated. For instance, in Japan, 69 per cent of the government wage bill is paid to local government employees and almost 70 per cent in Denmark. In some developing countries, the wages paid by nonfinancial state enterprises absorb almost 50 per cent of total public sector wages; Brazil and Zambia are examples. This emphasizes the importance of wage settlements in the public enterprise sector as well as at the local authority level.

Some useful facts emerge from the figures relating public sector wages to national income and gross domestic product. It seems that even where civil servants represent a high proportion of nonagricultural employment, they may not have a particularly large claim on total GDP, particularly in the more agriculturally oriented economies. In India, for example, central government wages account for only about 4 per cent of GDP, though central government employees constitute 20 per cent of non-agricultural employment. Central government wages tend to have a relatively similar share of GDP in the developing countries and the OECD group, 8 per cent relative to 5 per cent, but the share of state and local governmental wages, which averages over 6 per cent among the OECD countries, is below 1 per cent in the less developed countries. An alternative perspective is to look at government wages as a per cent of total wages in the economy (Table 2).

Table 2Relative level of government wages(In per cent of total wages in the economy)
OECD countriesDeveloping countries
Total sampleAfricaAsiaLatin America
Central government8.719.822.617.214.7
State and local government11.
General government20.720.024.817.0
Nonfinancial public sector enterprises4.
Total public sector22.932.0
Source: The full data set, as well as standard deviations for each of these estimators, is provided in the longer study cited on the first page of the article. In this table, the means for the more general units of government do not equal the sum of the means of the component units, reflecting differences in the size of each sample of observations.

Determinants of government size

In the more detailed study referred to on the first page, statistical tests were made to ascertain whether the magnitude of services provided directly through government employment increased as a country developed. The results show clearly that on a per capita basis the number of government employees at the central, state, and local levels rises with per capita income, particularly for countries with per capita income above US$800. This is despite the known tendency of developed countries to rely on transfers and subsidies and on the use of consultants in the provision of such services. However, in terms of our measure of potential “leverage,” the share of general government employees in total nonagricultural employment declines up to per capita incomes of $1,400, but then rises after that point, reflecting the relatively greater importance of state and local government employees at higher per capita income levels. The strength of the relationships is remarkable, given the cross-sectional nature of the data. The larger the population, the lower the proportion of the population employed by central government and the less the impact on nonagricultural sector employment. Surprisingly, state and local government employment per capita is not affected by the size of a country’s population.

How high are government wages?

The most obvious, and, more important, the most readily calculable, measure of the relative pay of civil servants is the ratio of their average wage to GDP per capita (although the average includes military wages and is probably understated relative to civilian earnings). The variation of this ratio is remarkable (Table 3). Whereas in the OECD countries the government average wage is almost twice per capita income, in the developing countries it is more than four times as large, and much larger in some countries. In Burundi, a civil servant earns 15 times the average per capita income. Undoubtedly, such factors as the degree of dualism in an economy, the educational content of government sector jobs, the relative importance of expatriates in the government workforce, and the demographic structure of the population contribute to the difference between developed and developing countries in this ratio. The results of a test aiming to explain these variations for 44 countries in the sample suggest that this ratio rises with per capita income for countries with average incomes less than $600. Beyond that, there is no statistically significant relationship between this ratio and per capita income.

Table 3Average government wages: some comparisons
Developing countries
OECD countriesTotal sampleAfricaAsiaLatin America
Ratio of average central

government wage to GDP per capita
Ratio of average central

government wage to average wage in

manufacturing sector
Ratio of average central

government wage to implied average

wage outside the central government1
Source: The full data set, as well as standard deviations for each of these estimators, is provided in the longer study cited on the first page of the article.

Another measure of the level of government wages is the ratio of the average wage in central government to the average wage outside—for example, in state and local governments, nonfinancial public enterprises, and the private nonagricultural wage sector. This comparison relies on the relative weight of central government wages and employment in total wages and nonagricultural wage employment in the economy, respectively. In most countries, developed as well as developing, central government employees seem to be better paid on average than employees outside the central government. However, the difference between the average wage of employees at all levels of general government (that is, central and state and local government) and the private sector is markedly smaller. An interesting anomaly is Zambia, where central government wages appear to be only one third those in the private sector.

The structure of government wages

The structure of wages within government has broad ramifications for many important policy issues. The spread between the bottom- and top-paid civil servants and among occupational categories determines the incentives for productivity and advancement within the government and can affect relative wages and employment outside.

The limited data on average salaries at different levels of the public sector show that, on average, central government employees are almost uniformly better paid than state or local government employees, while nonfinancial public enterprises tend to pay best of all. (Some notable exceptions to this last rule include Benin, Canada, India, Italy, and the Republic of Korea.) In part, this difference in pay may reflect differences in the skill mix requirements across levels of government.

Another measure of wage spread was calculated by comparing the starting salary for specific occupations with that for a clerical officer. Although it would be unreasonable to expect every country to have the same differences between positions, the extent to which they vary is nevertheless striking. In many cases, the reasons are not readily apparent. Why, for instance, does a primary school teacher in Cyprus earn 48 per cent of a clerical officer’s salary, while one in New Zealand makes 414 per cent? Why does a doctor in Sweden make 154 per cent of the salary of a clerical officer, and one in Kenya 708 per cent? There are also wide and inexplicable variations in relative salaries in the same profession. For example, in some countries, the difference in the wages of primary and secondary school teachers is very small, say, from zero to 12 per cent, yet, in others, it is closer to 50 or 60 per cent—in Kenya, a secondary school teacher appears to be paid almost three times as much as a primary school teacher.

Yet some variations may be more predictable; countries with major dependence upon agriculture, such as Kenya, tend to reward their agriculture officers more generously than others, such as the Bahamas, Canada, Cyprus, and El Salvador. The relative scarcity of a particular skill category in a country may also explain some of the cross-country variations in salaries across positions.

To measure the “equity” of government salaries, “Lorenz” curves on the civil service salary structures of countries were calculated to measure the cumulative distribution of the number of employees below different salary levels and the cumulative level of total salaries paid to such employees. The degree of inequality in government will influence the relative impact of such measures to control government wage expenditures as a general or selective freeze on vacancies. The greater the inequality, the greater the necessity that the freeze on vacancies cover employees at the upper end of the salary range if its fiscal impact is to be significant.

The analysis shows large variations in the degree of inequality in the overall salary structure of different countries. On balance, developed countries tend to have a more egalitarian wage structure, with only 15 per cent of wages earned by the top 10 per cent of employees, in contrast to 23 per cent in the developing countries. Conversely, the bottom 70 per cent of employees earn 57 per cent of wages in the OECD countries, compared to 52 per cent among developing countries in our sample. Most of the more developed countries group slightly more of their civil servants in the middle of the salary range, while the developing countries tend to skew their employment more toward the lower end of the range.

Employment and wages by function

To evaluate whether certain sectors are excessively large or small, or excessively well or badly paid, estimates were made of central government employment in various sectors. In order to allow for differences in federal structure, an adjusted measure of central government employment was estimated, which includes state and local government employees in the police, health, and education sectors. The data showed, surprisingly, that the mean number of administrators per 100 inhabitants was remarkably similar for developed and developing countries. African countries tend to have the highest administrative burden, and Asian countries the lowest, while within the OECD, the most overadministered country (by this measure) is easily the United Kingdom, with 0.78 administrators for every one hundred people in the population. Countries that have been more influenced by the British Commonwealth system of government appear to employ more government civil servants in administration than other countries, but this could be the outcome of a domestic definition of “administration.” Those employed in administration also appear to be paid rather more than the average for the public service, particularly in developing countries.

The proportion of adjusted central government employees involved in education is dramatically higher in the OECD countries than in the developing world—the mean for the OECD is 39 per cent in contrast to only 29 per cent in developing countries. Belgium commits 58 per cent of its central government staff to teaching, for example. The number of employees per capita in the education and health sectors in the OECD countries is three to four times that in the developing countries (though there is considerable variance in the health statistics, reflecting institutional differences in the role of the government in this sector).

For both defense and police, mean levels of employment per capita are higher in the developing countries than in the OECD. In the developing countries about 21 per cent of the adjusted central government work force is likely to be committed to defense and 12 per cent to police, whereas the comparable figures for the OECD are about 16 per cent and 7 per cent. But without Cyprus, Korea, and Singapore in the developing country group, the developed countries emerge as employing twice as many defense personnel per capita. For the countries where defense data are available, soldiers and civilian defense employees do not appear to be systematically paid markedly lower salaries than those in the other functional sectors, despite compulsory military service in some countries. On average, police and defense forces receive comparable pay in the sample of developing countries. In the OECD, police are better paid, while in some developing countries—Argentina and Sri Lanka—they appear to be paid much less than the defense forces, and in some—Korea and Swaziland—they are paid far more.

Perhaps the most important implication of this study is that our level of ignorance on a vitally important area of public sector policy remains astonishingly high. Governments should be able to learn from each other, but to do so, reliable data must be collected, collated, and disseminated. Surely statistics on public sector employment and pay should be made widely available to the public that finances the employment. It is an area where a natural curiosity should be satisfied and where, by doing so, benefits might accrue to governments in many parts of the world and help crucial areas of macroeconomic planning.

Other Resources Citing This Publication