IV Determinants of Shares in Gross Domestic Product of Economic Expenditure Categories


This section analyzes the determinants of the shares of alternative economic categories of public expenditure as a share of GDP. The principal approach in specifying the equations is to assume that a specific technological bias exists in respect of the provision of different functional expenditure categories and that the relative importance of these functional categories in a given country will determine the relative importance of the different economic categories of expenditure used to realize these objectives.


This section analyzes the determinants of the shares of alternative economic categories of public expenditure as a share of GDP. The principal approach in specifying the equations is to assume that a specific technological bias exists in respect of the provision of different functional expenditure categories and that the relative importance of these functional categories in a given country will determine the relative importance of the different economic categories of expenditure used to realize these objectives.

Table 6 provides the basic econometric results used for calculating the IEC indices; to maximize the size of the sample of countries, these equations use the aggregate functional category of expenditure on economic services. To obtain a clearer picture of the relative impact of spending on the different economic services subsectors—transport, electricity and water, agriculture, and mining and manufacturing—equations have been estimated by using these more disaggregated variables (Table 7).

Table 6.

Determinants of Economic Categories of Expenditure as Share of Gross Domestic Product

article image

Significant at a 10 per cent level.

Significant at a 5 per cent level.

t-statistics are in parentheses.
Table 7.

Determinants of Economic Categories of Expenditure as Share of Gross Domestic Product Using Disaggregated Categories of Expenditure on Economic Services

article image

Significant at the 10 per cent level.

Significant at the 5 per cent level.

t-statistics are in parentheses.

Table 2 provides the IEC index, and Table 8 ranks countries by the value of the IEC index (as Table 4 does for the functional expenditure shares).

Table 8.

Ranking of Countries by International Expenditure Comparison Index, 1977: Economic Expenditure

article image
article image

Goods and Services

Wages and Salaries

This expenditure category covers all payments in cash before the deduction of withholding taxes, social security payments, or pension fund contributions. It does not include income in kind such as the value of food, clothing, or lodging provided free of charge or below market prices; such income in kind is included under “Goods and Services Other Than Wages” (see p. 21).

Wages and salaries are a substantial part of all government payments, as is evident from the significant constant term in the estimation. As might be expected, government expenditure on education is a significant explanatory variable. Most countries find that teachers’ salaries are an important, and often controversial, component of government expenditure.

Less obvious, but clearly important, is government expenditure on economic services. When the latter is disaggregated by sector, expenditure on mining and manufacturing as a percentage of GDP prove to be significant at the 5 per cent level (Table 7). It seems probable that the staffing costs of administering and monitoring the numerous public sector schemes associated with mining and manufacturing can impose significant costs on government in terms of wages and salaries. The other economic subsectors—electricity, agriculture, water, and transport—do not appear to be important determinants of wage and salary expenditure. Equally interesting, expenditures on health and on public administration as a share of GDP did not prove to be very significant influences on the shares of wages and salaries.

It is interesting to note that per capita income is negatively related and significant. That is, as per capita income rises, it can be expected that government wages and salaries as a proportion of GDP will fall. Presumably, in developing countries government employees form a significant part of the income earners and hence of GDP, but as the country develops, the relative importance of direct government provision of services, and thus the government’s role as an employer, falls and other types of expenditure (e.g., transfers) become more important. Appendix Table 13 shows that generally the ratio of government wage expenditure in total expenditure is relatively higher in poorer countries.14

In testing the impact of a country’s economic structure on public wage and salary expenditure, the proportion of the labor force in agriculture proves to be negatively correlated with government expenditure on wages and salaries (significant at the 5 per cent level) which is to be expected. In effect, across countries at a given level of development, those with a large labor-intensive agricultural sector are likely to provide fewer direct government services.

Some large European countries spend less than expected on wages and salaries (for example, Italy, the United Kingdom, and Austria). Overall, defined in terms of an IEC index below 95, Asian, African, and Latin American countries allocate less to wages and salaries than might be expected, Middle Eastern countries tend to allocate more. The dispersion of IEC indices is fairly narrow for this category of expenditure, suggesting that countries tend to be more likely to spend what would be expected on wages and salaries than on other categories.

Goods and Services Other Than Wages

This category covers all goods and services bought on the market or received through loans or grants (materials, office supplies, rent, fuel, electricity, travel, telephones, equipment with a life less than a year, and goods and services distributed to employees free of charge); not included are fixed capital assets, stocks, land, and intangible assets.

For most countries, it seems that a strong determinant of increased expenditure on other goods and services will be increases in defense expenditures (significant at the 1 per cent level). Large shares in GDP of expenditure on health, economic services, and public administration also seem to lead to a large share of purchases of other goods and services. Interestingly enough, a large share of expenditure on education is significantly and negatively correlated with such purchases. In effect, it may be that some sectors require a fixed complement of nonlabor inputs for the provision of services, whereas other sectors may be able to substitute labor or, more realistically, squeeze nonwage expenditures, for a given amount of services provided. Not surprisingly, expenditure on social security is not a significant factor in determining such expenditure.

Overall it seems that there is no systematic pattern across regions in terms of a bias toward such expenditure. There is a slightly higher dispersion in IEC index values for this category of expenditure, but the standard deviation of the IEC index seems significantly below that of other current or capital expenditure categories.

Total Goods and Services

This expenditure category is the aggregate of government spending on wages and salaries and on other purchases of goods and services. Probably the most interesting point about this equation is that, despite the high level of explanatory power (R2 = 0.62), per capita income has an insignificant value. The most important influences are government expenditures on defense, education, agriculture, manufacturing and mining, and health—broadly as presented above for each of the separate categories. Social security expenditure is not an important factor in determining such expenditure. In a cross-section of expenditure on economic services, expenditure on mining, manufacturing, and agriculture tends to lead to significant spending on goods and services. With respect to countries with IEC index values five points above or below 100, there is some tendency for African and industrial countries to spend more than would be expected and for Asian and Latin American countries to spend less. Another interesting aspect of these results is that the dispersion in IEC index values is lower for the aggregate category of expenditure than for its disaggregated subcomponents.

Interest Payments

This category covers all interest payments to domestic and foreign holders of government debt. Again, defense spending emerges as one of the most significant explanatory variables. As unexpected defense expenditures associated with emergency circumstances cannot (or are not) financed through current taxation, it would be expected that the associated debt financing would greatly increase interest payments. Surprisingly, countries that have a large share in GDP of public expenditure on health also tend to have larger expenditures on interest.

It could be hypothesized that the larger the proportion of government expenditure on such economic services as electricity and on current transfers to mining and manufacturing the lower interest payments would have to be. The government, instead of nationalizing such concerns (which would involve large capital sums raised through the bond market), provides current subsidies and investment grants, so that a larger share of such expenditures would generally be associated with lower interest payments. The coefficient for expenditure on government economic services is negative and significant, supporting the suggestions made above. This result appears even more clearly when, for a more limited set of countries, expenditure on economic services is disaggregated. The coefficient (–0.38) for expenditure on electricity, gas, and water is even more strongly negative (Table 7).

In country rankings, France seems to spend much less on its interest payments than might be expected, given its per capita income, its defense spending, and its support of electricity, mining, and manufacturing. On the other hand, the United Kingdom spends more than might be expected. Regionally, Asian countries clearly appear to spend more on interest than would be expected, while Middle Eastern countries spend less. The dispersion in IEC indices for this category of expenditure is very high, suggesting possibly that time plays such an important role in determining eventual annual interest payments unique to each country that cross-section analysis is not appropriate for this functional category.

Subsidies and Other Current Transfers

Subsidies include all transfers on current account to private industries and grants to public enterprises for offsetting operating losses stemming from government action. Other current transfers include transfers to other levels of government for current purposes, grants to private nonprofit institutions, and, most important, cash transfers to households (including payments for social security, unemployment benefits, family allowances, civil service pensions, and scholarships).

Such expenditure should be associated with the expansion of social services and welfare (for example, social security, welfare, and education), as society is more capable of subsidizing the provision of such services or providing transfers made to improve income distribution. It might also be expected that the expansion of both a modern agricultural sector and a manufacturing sector would lead to subsidies (although insofar as these sectors proved efficient and profitable, subsidies would become unnecessary).

The result is that almost 88 per cent of government expenditure on subsidies and transfers as a share of GDP is explained by the proportion in GDP of government expenditure spent on social services, education, defense, and per capita income. Health expenditure is not significant. Per capita income, although positively correlated, is only significant at the 6 per cent level. As mentioned above, expenditures on social security and education might be expected to be important, but expenditure on defense as a powerful explanatory variable is somewhat surprising. Perhaps the industries that are needed to ensure domestic defense require industrial subsidies (for example, for steel and shipbuilding) or large defense expenditures might require complementary inputs at subsidized prices (transport, energy).

In the country rankings, it is surprising that Mexico seems to offer almost 50 per cent less in subsidies than might be expected, although this is possibly explained by aid to industries in other ways, such as tax concessions, and much the same may be true of the United States. At the same time, it would not be expected that the United Kingdom spends some 24 per cent more than predicted on subsidies, Korea 73 per cent more, the Philippines and Egypt more than twice as much, and Sudan and Pakistan over four times as much. Equally interesting, African countries tend to spend far more than expected on subsidies and transfers, but it is not known which sector is benefited most—the urban or the rural. More than half of the Latin American countries spend less than expected. Although the predictive power of the equation is high, the dispersion of IEC index values is also high.

Total Current Expenditure

This expenditure category represents the aggregate of the categories discussed above, that is, expenditure on wages and other goods and services, interest, subsidies, and transfers. Countries that allocate a large share of GDP to public expenditure on defense and social services (health, education, and social security) rely on current expenditure as the main instrument for realizing these objectives.

The dispersion of IEC index values is the lowest of all the economic variables. Almost all the Latin American countries and two thirds of the Asian countries spend far less on current expenditure than would have been predicted; conversely, most of the African countries spend more than expected.

Capital Assets

This category covers the acquisition of new and existing durable goods (with a normal life in excess of one year) but excludes all military goods. Two key areas of functional expenditure give rise to the purchase of capital assets: expenditure on economic services and on general public administration. The higher the share in GDP of expenditure on general public services (police, general administration, the judiciary, legislature) the higher is government investment (significant at the 1 per cent level). This is similarly true for expenditure on economic services, with the key subsectors being public expenditure on utilities and transport (Table 7). Other key functional categories, such as health, education, or social security, prove to be unimportant as factors explaining the share in GDP of public capital investment. The amount of per capita income is not an important factor in determining whether public sector investment is an important share of GDP.

A problem with the capital investment variable is that a figure for any one year can be misleading. It is in the nature of government acquisitions of capital assets that they are made sporadically; governments change and with them the prevailing views on the role of government ownership of capital. Thus, the low actual ranking of the United Kingdom, compared with what might be expected given that nation’s size of public administration and its expenditure on utilities and roads, is misleading and does not reflect the substantial capital investment already made by the Government of the United Kingdom in the 1950s and 1960s. Similarly, the high figure for Sweden might be predicted, but that for Spain may simply represent a catching up after years of a deliberately contained government investment in capital projects. Finally, across regions, a slight majority of African countries tends to spend less than expected on capital investment.

Capital Transfers

Capital transfers are unrequited payments to help the recipients (other branches of government, public enterprises, or the private sector) to buy capital assets or to compensate for loss, damage, or some extraordinary problem. To some extent, they could be viewed as a reciprocal of the acquisition by the State of capital assets. Such transfers could be expected to be associated positively with the growth of a modern agricultural sector and the mining and manufacturing sectors. To the extent that the central government is involved directly in capital acquisition, such transfers would need to be less.

The key functional expenditure determining the share in GDP spent on public capital transfers is expenditure on economic services. Within this, the principal economic subsectors are government spending on roads, agriculture, and mining and manufacturing. Expenditure on capital transfers is positively related to the amount the government spends on agriculture as a proportion of GDP (significant at the 5 per cent level) but negatively correlated with the proportion of the labor force in agriculture, suggesting that the smaller the labor force and the more modern the agricultural sector the more likely it is that government expenditure may take the form of capital transfers. Such transfers enable the smaller agricultural labor force to use more modern equipment and to improve its capital stock for processing and storage and allow it to be protected from the effects of natural disasters. Much the same appears true for the mining and manufacturing sectors.

The negative coefficients for defense expenditure and government expenditure on roads tend to suggest that such spending preempts government allocation for capital transfers. Per capita income does prove to be modestly significant for this variable; the negative coefficient suggests that, again, with higher per capita incomes the need for government expenditures on capital transfers is reduced.

In the country rankings, it is surprising that a country such as the Netherlands spends almost twice what might be expected, given the relative importance of its agricultural, mining, and manufacturing sectors. However, this may be explained by the sporadic nature of capital transactions, although capital transfers—often made under entitlement programs—would be expected to be less responsive to major fluctuations than purchases of capital assets directly by government. Across regions, countries in Africa and Latin America tend to spend less than would be expected on capital transfers.

The dispersion in IEC index values is the highest for this type of expenditure; only 4 of the 62 countries in the sample have IEC values between 95 and 105 (Table 3).

Total Capital

This category is the aggregate of government expenditure on the acquisition of capital assets and capital transfers. Expenditures on the more capital-oriented functional categories—economic services, housing, and community amenities—prove to be the most important determinants of the share in GDP of public capital expenditure. When the economic services category is disaggregated, it is found that expenditure on electricity, gas, and water, on mining and manufacturing, on roads, and on agriculture gives rise to a significant amount of capital expenditure (Table 7).

Across regions, the key imbalance appears in Africa and in the industrial countries, where a significant majority of countries seem to spend less on capital expenditure than expected. Perhaps because of the importance of the members of the Organization of Petroleum Exporting Countries, the Middle Eastern countries spend more on capital expenditure than expected.