Chapter

III Determinants of Shares in Gross Domestic Product of Functional Expenditure Categories

Author(s):
Peter Heller, and Alan Tait
Published Date:
April 1982
Share
  • ShareShare
Show Summary Details

Introduction

This section discusses the specification of the equations to predict the shares in GDP of each category of functional expenditure. The econometric results appear in Table 3. Table 1 shows the value of the IEC index. Table 4 ranks the countries by the value of their IEC index; a low ranking indicates a relatively low IEC index—namely, a low expenditure share relative to what would have been predicted for the country.

Table 3.Determinants of Functional Expenditure Categories as Share of Gross Domestic Product

(All expenditure categories as a percentage share of GDP)

Mining,Electri-
Health,Manu-city,
GeneralSocialSocialHousingAgri-factur-NaturalTransport
PublicSecuritySecurity,andculture,nig, andGas, andtation and
Ser-andandCommunityForestry,Con-WaterCommuni-
vicesDefenseEducationHealthWelfareWelfareAmenitiesFisheriesstructionSupplycations
Constant−0.59−7.00**3.18*1.31**−4.76**−3.89**0.280.470.83*0.664.80**
(−0.30)(−2.27)(1.85)(3.48)(−3.93)(−2.93)(1.45)(1.17)(1.94)(1.61)(4.49)
Income per capita
(PCI) (in thousands
of dollars)0.010.25*0.41*0.09−0.030.01
(0.13)(−0.01)(1.78)(2.60)(1.53)(−0.58)(−1.40)
Income per capita,
countries with
PCI < $1,750
(in thousands of
U.S. dollars)−0.042.11**0.82**
(−0.32)(2.92)(2.67)
Income per capita,
countries with
PCI ≥ $1,750
(in thousands of
U.S. dollars)0.20**−0.01**
(2.65)(2.68)
Percentage of popu-
lation, aged 14 and
under0.07*0.16**0.03
(1.82)(2.71)(0.81)
Percentage of popu-
lation, over age 650.12**0.84**1.02**
(2.28)(7.26)(8.01)
Infant mortality
rate0.070.06
(1.71)(1.37)
Share of labor
force in
agriculture0.02**
(3.80)
Share of labor
force in
industry0.13**0.14**
(3.12)(2.87)
Share of popu-
lation in
urban areas0.03**0.05**0.03**
(−2.28)(2.25)(−2.28)
In countries
with PCI < $1,750
(−0.48)
In countries
with PCI ≥ $1,7500.02**
(2.11)
Enrollment rate,
primary schools0.01*
(1.56)
Enrollment rate,
secondary schools,
countries with
PCI < $1,750−0.02
(−1.39)
Enrollment rate,
secondary schools,
countries with
PCI ≥ S1,7500.02**
(2.09)
Pupil-teacher ratio,
primary schools−0.04*
(−1.88)
Access to clean
water supplies
In countries with
PCI < $1,7500.02**
(−0.20)(3.20)
In countries with
PCI ≥ $1,7500.02**−0.01**
(2.97)(2.78)
Population per
hospital bed
In countries with
PCI < $1,750−0.26*
(−1.84)
In countries with
PCI ≥ $1,750−2.50*
(−1.76)
Population growth rate
in urban areas0.19**
(2.33)
In countries with
PCI < $1,750
(0.07)
In countries with
PCI ≥ $1,7500.48**
(5.22)
Share of total public
expenditure in GDP0.12**
(5.62)
Share of total public
expenditure (net of
defense)0.10**
(2.48)
Share of manufacturing
sector in GDP−0.04*−0.07**
(−1.87)(−2.24)
In countries with
PCI < $1,750−0.05**
(−2.75)
In countries with
PCI ≥ $1,750*
(1.80)
Share of other manu-
factured goods and
fuel exports in
total exports0.02**−0.01*
(3.04)(1.71)
Share of agriculture
in GDP−0.08**
(−3.76)
R20.320.150.280.620.800.840.210.170.160.440.23
(N)(91)(84)(90)(91)(91)(92)(86)(90)(72)(66)(69)

Significant at a 10 per cent level.

Significant at a 5 per cent level.

t-statistics are in parentheses.

Significant at a 10 per cent level.

Significant at a 5 per cent level.

t-statistics are in parentheses.
Table 4.Ranking of Countries by International Expenditure Comparison Index, 1977: Functional Expenditure
CountryYear of DataGeneral Public ServicesDefenseEducationHealthSocial Security and WelfareHealth, Social Security, WelfareHousing and Community AmenitiesAgriculture, Forestry, and FisheriesMining, Manufacturing, and ConstructionElectricity, Natural Gas, and WaterTransportation and Communications
Argentina19779155312612225694340
Australia1483771553335253024
Austria197773261948434465284058*54
Bahamas197657350621832720
Bahrain19776020
Bangladesh197794105081414511369
Barbados19776327840293173764655
Belgium197724367238545241135458*60
Bolivia19774345252768278212626
Botswana197765248578553827145
Burma197739751620211376722827
Burundi1977316667725457
Cameroon1976844626247665234622251
Canada1783455675157626566
Chad1977498258109613641629
Chile197777405257758263318821
Costa Rica1977158449716621169758
Cyprus1977746515253120774019274
Denmark197664421849464535664558*6
Dominican Rep.1977301884460508152603213
Ecuador197720533223273715
Egypt19772733905270798568223610
El Salvador1977389304344424633441442
Ethiopia19771644276674691543272557
Fiji1977797607580764858423428
Finland19773743686336404286592052
France1977364156645559522153193
Gambia, The19778745882356328444
Germany, Fed. Rep.1255739594547512363
Ghana197735734535394938436
Greece19774480203041334062361237
Grenada197726351861938304868
Guatemala197717291734222412
Honduras1976692838854978531129
Iceland197768227122245788
Iran1977672286120147542444722
Ireland1977632169
Israel197738169356567312239237
Italy197756195476485844546964
Jamaica1977336707411176880232830
Japan13334231864821831
Jordan19756078888764755481695667
Kenya19773448496582*681947355246
Korea19772676375154122513245
Kuwait19775049143214118016337
Lesotho19748179395049847929
Liberia1977851666771746585512644
Luxembourg19775212232666329539
Madagascar19735514485182*871060251135
Malawi19774131352156271159101723
Malaysia197754617772285153221015
Mali197659677546777312955188
Malta19775313413759606670615759
Mauritius1977721638369805590173111
Mexico197711529157371755819
Morocco197788608647798169
Nepal1977105249182*43144453
Netherlands197776398270677759
New Zealand19773215576952621873
Nicaragua197622221711828679353013
Niger1977752384538291427493048
Norway19772159654240348385511650
Oman1974901057
Pakistan1977870141327114374049
Panama197770618978883341483832
Papua New Guinea1977582176817747461503961
Paraguay197728272363251791556
Peru19774532593319456350
Philippines19764054111453192051585865
Portugal197771772141393049244651
Rwanda1977236213128292619343368
Senegal197562354226272336181211
Sierra Leone19786730475612213426144216
Singapore197786644013356126112
Somalia197780718782829086836418
Spain197713209224236283633514
Sri Lanka1977141146368289265626
Sudan1977450562510377738
Suriname19769062688560384733
Swaziland197766257458848727458
Sweden19774647816858615046655820
Syrian Arab Rep.197717424757377067695445
Tanzania19774758537382844769664934
Thailand1977769361982834339245847
Tunisia197719108979617038784143
Turkey197729514381033717675562
United Arab Emirates19778368233
United Kingdom197751568054302978505258*41
United States11279343138263962
Upper Volta19734273312916151633117
Uruguay1978893812166255815324125
Venezuela19771817642837286789564839
Yemen Arab Rep.1977618271814529
Yugoslavia19775828047647
Zambia19778283844725687433553
Number of countries in column9184909191928690726669

Asterisk denotes that this particular ranking should be treated with care as actual expenditures were extremely small and predicted expenditures negative—see text for explanation.

1973–75.

Asterisk denotes that this particular ranking should be treated with care as actual expenditures were extremely small and predicted expenditures negative—see text for explanation.

1973–75.

General Public Services

This functional category covers financial administration, external affairs (including international aid), planning, statistics, and other aspects of general administration. It also covers justice, police, public order, and safety. A believer in “Wagner’s” law9 might expect that the share in GDP of such expenditure would rise with per capita income; in fact, no statistically significant relationship was found. The only variable significant at the 1 per cent level was the share in GDP of total public -expenditure; thus, the larger the share of government, in general, the larger would be the general expenses of running government. At the same time, there is some evidence of economies of scale in the operation of the government and in the provision of judicial services; as the public sector grows, these costs do decline slightly as a proportion of total expenditure. The elasticity of public administration expenditure was found to be approximately 0.88.

The next most significant variable is the percentage of population in urban areas; apparently, as countries become more urbanized, the share in total output of general public administration expenditure decreases, suggesting some economies of scale. On the other hand, the larger the proportionate size of the young population (14 years old and younger) the higher this share is likely to be. This emphasis on the young population might be thought to be simply a proxy for nondevelopment (developing countries tend to have much larger proportions of their population in the younger age groups). However, the insignificance of per capita income as estimated in the relationship suggests this is not so and that the size of the younger population must be taken as a significant variable in its own right.

Across countries there tends to be a low standard deviation in the value of the IEC index for this expenditure category. Countries such as Argentina, The Gambia, Morocco, Suriname, and Uruguay seem to spend more than might be expected on general administration as a share of GDP (for example, IEC indices are significantly higher than 100), while the United Kingdom and Australia appear to spend as much as might be predicted. Mexico, the United States and Yugoslavia have IEC indices much lower than 100. (Does this suggest that there is less “fat” in the operation of the U.S. Government than is currently argued?) There is some tendency for Latin American, Asian, and industrial countries to spend less than would be expected on general public services and for African countries to spend more than would be expected.

Defense

This category includes all defense expenditures except those for military pensions, which are included under social security and welfare (see p. 15). Again, it might have been expected that the share of defense expenditure in GDP would be closely associated with per capita income but the relationship does not prove significant. The same variables as those influencing general administrative expenses proved to be significant for defense. The most striking difference is that, whereas urbanization had a negative impact on the share in GDP of general administrative expenditure, for defense there was a positive relationship. Defense expenditure, according to these relationships, could be expected to be higher in a more urbanized country, with a larger proportion of children of 14 years and younger and a larger public sector (net of defense spending).

While numerous influences not tested in this study (and, indeed, impossible to test) must influence defense spending, and while the low correlation coefficient suggests a large amount of “unexplained” defense expenditure, the significant variables mentioned above are interesting.10 It seems reasonable to consider that urbanized societies must spend more on defense and are willing to do so. Likewise, it is reasonable to expect that many authorities who are prepared to run a large public sector are also likely to accept the idea that a substantial part of the national budget should be spent on defense.

The country ranking confirms the expected evaluations. Sweden, the Philippines, Kenya, Tanzania, Germany, and the United Kingdom were spending roughly what could be expected in 1977 on defense and Mauritius, Barbados, and Mexico were spending much less than might be expected. The high figures for the United States, Pakistan, Portugal, Iran, Korea, Israel, the Yemen Arab Republic, and Chad reflect these individual countries’ preocupation with defense in the 1970s.

Education

It is to be expected that government expenditure on education as a percentage of GDP would be most strongly correlated with the proportion of the population in the age group affected by schooling. The largest groups attending school fall in the age group 14 years and under, followed by secondary school and university populations. Unfortunately, the population breakdown available for the large sample of countries enabled only the under 15 age group to be included; over 15 years, the population span included the active workers through age 65, negating any explanatory power of the secondary school and university groups.

An alternative measure of the effective demand for education would be the enrollment rates of the primary and secondary school age populations, respectively. The higher the enrollment rate is, ceteris paribus, the higher the expenditure share on education should be. Technological factors also influence the level of spending on education. The higher the pupil-teacher ratio is at the primary or secondary school level the lower would be the expected share of education spending in total output. Finally, it might be expected that the costs of educating a widely scattered agricultural population might be higher than the costs of educating an urban, highly concentrated school population, although this factor clearly depends on the costs of urban school development, the quality of rural education programs versus urban education programs, and the costs of urban universities. Expenditure on education might also be expected to increase with per capita income, but this influence could weaken in countries with high per capita income, where the private sector might take over some of the government’s responsibility for expenditure on education.

Per capita income proved to be a highly significant determinant of the share in GDP of public expenditure on education, mainly at incomes below US$1,750. In countries with a low per capita income, it is evident that a greater need exists for expenditure on education, but a breaking point is reached when per capita income rises to US$1,750. Further increases in per capita income tend not to lead to as great an increase in government expenditure on education as for incomes below US$1,750, probably because expenditure on education by the private sector increases to take over part of the burden or because “basic” education needs are satisfied and other priorities (economic and social) take precedence.

Another variable that was highly significant was the enrollment rate in secondary schools for those countries where per capita income was over US$1,750. A positive correlation between the primary school enrollment rate and the share of educational expenditure is also evident for these countries. This tends to bear out the observation that expenditure on education by government is believed to be important for basic primary education for low-income countries but that this attitude changes when per capita income is over US$1,750 and more importance is attached to secondary school enrollment.

Although government expenditure on education is positively correlated with the proportion of the population in the primary school age bracket, the relationship is not statistically significant. In effect, a large share in GDP of expenditure on education will not depend simply on a large number of potential students. It is particularly interesting that government expenditure on education is negatively correlated with the pupil-teacher ratio and with the percentage of the population in urban centers (significant at the 5 per cent level). However, the significant negative correlation with urbanization suggests a stronger explanatory power than might be expected for the hypothesis that it will cost the government more to educate a rural population than an urban one.

It is interesting to note that the spread between predicted and observed results for education expenditure is the smallest of all the functional categories, suggesting a greater unanimity and consensus among countries in relation to government expenditure on education. The Governments of the United Kingdom and the Netherlands appear to spend about 50 per cent more than expected, while that of the United States spends about 15 per cent less than expected. These results reflect the major differences between the three countries in their degree of state involvement in the education sector, notably universities; the United States relies far more on the private sector at this level of education (Table 1). Many Middle Eastern and North African countries seem to spend more than might be expected on education but these countries can be contrasted with their neighbors, Sudan and the Yemen Arab Republic. On balance, two thirds of the African countries spend more than would be predicted; two thirds of the Latin American countries and all of the European developing countries (Turkey, Cyprus, Greece, Malta, and Portugal) spend less than expected (i.e., have an IEC index of less than 100).

Health

This category includes government expenditure on general administration, regulation, and research for health; on hospitals, medical and dental centers, and clinics; on population control, immunization, and inoculation; and on blood donor services. It also covers the reimbursement for services of individual doctors, dentists, and paramedics under insurance schemes for individual health services outside hospitals and clinics. It excludes expenditures that would fall under social security and welfare (see p. 15).

The share in GDP of government expenditure on health might be expected to be positively correlated with factors suggesting a high basic demand for medical care, such as high infant mortality rates,11 a large population under 15 and over 65, a low life expectancy rate, a high birth and population growth rate, and poor access to clean water supplies. The higher the quality of care (for which high ratios of hospital beds, nurses, and doctors per unit of population are used as proxies) the higher would be medical expenditure. While it would also be desirable to capture the effect of any unusual country-specific disease (such as schistosomiasis, onchocerciasis, or trypanosomiasis in some of the African countries), it was impossible to collect sufficient data in this study for a significant sample of countries to test any such relationship. Finally, medical care could be expected to increase as per capita income increased.

In fact, few of these variables were significant and over 60 per cent of government expenditure on health was explained by the proportion of the population aged 65 and over (significant at the 2 1/2 per cent level) and by the ratio of population to hospital beds. Access to clean water supplies was very significant where per capita income exceeded US$1,750. In principle, one might expect poor access to clean water supplies to be associated with ill health and, thus, with a greater demand for medical care. The reverse relationship in the results suggests that access to clean water supplies may be a proxy for the overall level of economic development. Indeed, there is some correlation between the index of clean water and per capita income (R = 0.75 in the sample as a whole, and R = 0.86 for countries with per capita income in excess of US$1,750). This possible multicollinearity may also explain why per capita income proves to be an insignificant explanatory variable.

It is interesting to note that, while the proportion of population over 65 is a strongly significant factor, the variables relating to the portion of the population under 15 years, the infant mortality rate, and the birth rate were all statistically insignificant (results not shown). Again, the obvious presence of a potential demand for a sector’s services does not necessarily indicate that the services will be forthcoming. There is a statistically significant but weak positive quantitative relationship between the per capita ratio of hospital beds and the share of health expenditure at low incomes; the quantitative relationship becomes far stronger at per capita income over US$1,750, probably reflecting a greater preoccupation with the quality of medical care at higher incomes.

As with education, there is also a fairly tight bunching of IEC index values in the health sector, with a low standard deviation for the index. In terms of country rankings, most of the industrial European countries appear to spend more on health than might be expected, given their population structures, their water supplies, and their provision of hospital beds. However, it is noteworthy that government expenditure on health in the United States and Japan is some 25–30 per cent less than might be expected and that the developing countries in Europe have IEC indices less than 100. The U.S. and Japanese results again arise from the prominence of the private sector in the provision of medical care in these countries. As was true for education, two thirds of the Latin American countries spend less on health than would be expected; however, this may reflect only a problem in statistical classification. The indices of some countries with strikingly low IEC indices for health—such as Luxembourg, Argentina, Costa Rica, Paraguay, the Syrian Arab Republic, Sudan, and Nicaragua—may be misleading in their implications if account is taken of the share of their expenditure on social security (see below).

Social Security and Welfare

This category includes expenditure on social security; sickness, old age, and disability payments; and payments under contributory and noncontributory schemes; and underfunded and unfunded pension and disability plans for government employees (civil or military). It also includes unemployment, family, maternity, and child allowances, as well as any other public assistance. Welfare services include care of the elderly, disabled, mentally defective, and children.

The variables selected explain about 80 per cent of the share in GDP of government expenditure on social security and welfare. Government spending on this function would be expected to be strongly associated with the number of elderly people in the total population; indeed, this variable is significant at the 1 per cent level. The other variable that is strongly correlated with social security and welfare expenditure is the proportion of the labor force in industry; as the labor force in manufacturing expands, so does government responsibility for unemployment pay and for sickness and injury benefits. It might also be supposed that, as per capita income rises, private sector insurance might assume more responsibility for social security and welfare. This hypothesis is borne out in the results. Similarly, it could be expected that the proportion of population under 15, life expectancy, and the various medical variables might be significant; in fact, none of these was found to be particularly significant and only the infant mortality rate is included as an explanatory variable.

The German Government appears to spend on health about what would be expected, but, interestingly, the United Kingdom, often considered to be a “welfare state,” spends some 34 per cent less than would be expected. While the U.K. Government is involved in the provision of many social or welfare services, it spends less on these than many other countries, in terms of the level of benefits per recipient and in the quality of services provided. Nicaragua and Tanzania appear to spend substantially more than would be expected, given the structure of their population and their per capita income. Most OECD member countries cluster around 90–120 per cent of expected government expenditure on social security and welfare.

Health, Social Security, and Welfare Combined

These categories were combined to test whether the explanatory power of the variables improved because of overlap and possibly poor distinction between the categories of “health” and “social security” expenditure. As noted above, some countries, particularly in Latin America, have difficulty in accurately distinguishing items of health and social security expenditure; this may have led to the extremely high IEC indices for social security and the extremely low indices for health, which can be seen in Table 5. The index for the combined functional categories may be more representative of their expenditure patterns.

Table 5.IEC Indices for Health and Social Security in Selected Countries Where Medical and Social Security Systems Partly Overlap
CountryHealthSocial SecurityCombined Health

and Social Security
Paraguay19.8139.975.3
Syrian Arab Republic29.7122.186.6
Uruguay54.5139.4113.5
Costa Rica42.7210.0135.1
Nicaragua46.3400.0212.0

The proportion of the population over 65 and per capita income were both explanatory variables significant at the 1 per cent level. The percentage of the labor force in industry was also significant at the 5 per cent level. As the population over 65 increased, as the percentage of the labor force in manufacturing expanded, as income per capita rose, and as the infant mortality rate increased, expenditure on health and social security could be expected to be higher.

Again, expenditures on health, social security, and welfare by the Governments of Ireland, Japan, the United States, the United Kingdom, Norway, and Australia appear to be lower than would be expected on the basis of their population and per capita income, whereas corresponding expenditure by Germany appears to be approximately what would be expected. The expenditures of the Governments of France, Sweden, Italy, New Zealand, Mexico, Israel, Egypt, and the Netherlands all appear to be higher than expected.

Housing

Government expenditure in this area covers the provision of housing and of housing payments tied to the income level of the recipient; it also includes rent subsidies, some home purchase subsidies (exclusive of tax expenditures), and any administrative costs.

As expected, the most significant explanatory variables were those relating to urbanization and per capita income (significant at the 5 per cent and 1 per cent levels, respectively). However, the importance of these variables depends on the amount of per capita income. No matter how urbanized the country is, the share in GDP of government expenditure on housing increases as per capita income rises to US$1,750. Once this figure is reached, ceteris paribus, an increase in per capita income alone does not trigger further public sector housing involvement. (In some cases, this may reflect the increasing involvement of the private sector’s construction industry.) Once per capita income rises above US$1,750, the degree of government involvement then becomes sensitive to the extent of urbanization. Increasing urbanization triggers further increases in the share in GDP of government housing expenditure.

This seems to indicate that in countries with a low per capita income, the government cannot enter into the budgetary expense of open-ended subsidies for housing even in large urban areas. The authorities are much more likely to attempt to control this element by controls on rents and licenses to build. However, as per capita income rises and more urbanization occurs, the pressure for public housing increases and government expenditure on publicly subsidized housing becomes strongly identified with urbanization.

The standard deviation of the IEC index is extremely high for this functional expenditure category. Uruguay’s spending on housing is 92 per cent less than expected, whereas Somalia spends far more than expected (2 per cent of GDP rather than the predicted 0.25 per cent). It may be noted that the United States spends 36 per cent less than expected, France and Germany have IEC indices closer to unity, and the United Kingdom is far above (two and a half times as much) what might be predicted. In general, African, Latin American, and industrial countries appear to spend less than might be expected.

Agriculture

This covers the provision of agricultural services and financial support programs for farm prices and incomes through market intervention subsidies and price supports, and forestry and inland and ocean fishing programs, as well as research in all the sectors just mentioned.

Government expenditure on agriculture might be expected to be a function of the importance of the sector in the economy, as proxied by its share of the labor force, and might also be expected to be dependent on the type of land associated with different amounts of rainfall. Unfortunately, insufficient information for a number of countries makes it impossible to include the quality and extent of arable land as an explanatory variable. However, expenditure on agriculture might also be expected to have some functional relationship to agricultural exports or, indeed, a negative relationship to nonfood agricultural exports as a percentage of total exports. Tests were made in the study to include such variables, but it was found that the only significant variables were the percentage of the labor force employed in agriculture (significant at the 1 per cent level) and per capita income, both with a positive impact on the share of government. This is not surprising. These forces, however, work in opposite directions for some countries. For example, many European countries have a high per capita income that suggests increased spending by government on agriculture, but this is offset by the rapidly shrinking labor force in agriculture, which is a more powerful factor in reducing the impetus for governments to spend on agriculture rather than on other competing claims.

The ranking of countries by their government expenditure on agriculture confirms this outline. Some of the countries with IEC indices close to 100—the United Kingdom, which spends 95 per cent of what might be expected on agriculture, and Italy, which spends exactly what is expected—have higher per capita incomes and smaller contracting agricultural labor forces, where these two offsetting circumstances produce almost precisely the expected expenditures.12 However, a country like Mauritius, which has a large agricultural labor force and a low per capita income, spends over three times more than might be expected on agriculture, and indeed much the same is true of countries like Finland, Iceland, Japan, and Norway, all of which spend more than twice as much as might be expected—probably to assist the fishing activities of these countries. It is interesting to note that advanced countries that depend on agriculture for a major contribution to their balance of payments (for example, Denmark and New Zealand) are well above the mean. It is equally striking, on the contrary, that governments such as those of the United States and Argentina spend so much less than expected.

In general, governments in African countries seem to spend as much or more than might be expected on agriculture, Asian countries somewhat less, and Latin American countries significantly less.

Economic Services: Mining, Manufacturing, and Construction

This functional category includes expenditure for the promotion, regulation, research, subsidization, and other assistance to the mining, natural resources, manufacturing, and nonhousing construction sectors. It also includes investment grants to these sectors.

Government contributions to mining and manufacturing are strongly correlated (significant at the 1 per cent level) with the share of exports of other manufactured goods and fuel in total exports but negatively correlated with the percentage share in GDP of manufacturing. Again, it is interesting to note that per capita income does not prove to be a significant determinant of the share of such expenditure. Basically, as one would expect, the more industrially developed the country is the less likely it is to subsidize industry (under the limitations of the General Agreement on Tariffs and Trade and limitations on export credit guarantees). At the same time those countries committed to exporting manufactured products are likely to spend government revenue on attempting to help both mining and manufacturing.

Out of a sample of 72 countries in all, only 24 actually spend more government money than might be predicted on subsidizing mining and manufacturing. What is, perhaps, most interesting is the number of highly industrialized countries that apparently spend more than might be expected: Norway, 23 per cent; the United Kingdom, 31 per cent; Belgium, 38 per cent; France, 35 per cent; Sweden, 140 per cent; and Italy, over 300 per cent more. On the whole, Asian, Latin American, and African countries spend less than might be expected on subsidizing and supporting industry.

Economic Services: Electricity, Natural Gas, Steam, and Water

This category encompasses expenditure for the promotion, regulation, research, subsidization, and provision of investment grants for production, transmission, and distribution of electricity, natural gas, or steam. It does not include the mining of natural gas, which is classified under mining. This category also includes expenditure on the regulation, purification, and distribution of clean water for general use (not for irrigation).

The most straightforward hypothesis is that government expenditure in this category will rise with per capita income, the urbanization of society, the growth of manufacturing, and increased access to clean water supplies. Interestingly, per capita income was negatively and very weakly associated with expenditure on utilities, but significant variables at the 1 per cent level were urban population growth, changes in the percentage of GDP related to manufacturing, and access to clean water supplies.

Urban population growth was positively associated with this government expenditure, but only at per capita incomes over US$1,750. On the other hand, the share in GDP of government expenditure on energy and water declines as the role of the manufacturing sector increases for countries with a per capita income below US$1,750; for countries with incomes above this amount, the share in GDP of manufacturing is no longer statistically important. This initial negative relationship appears contrary to what would be expected, because increased public expenditure on electricity, steam, and gas might be expected as manufacturing increases. One possible explanation is that as the manufacturing base of the country expands, the energy supply industry becomes more profitable and the required transfers from government to these utilities on both current and capital account become less. Presumably, industry generates sufficient income to compensate utilities commercially and to enable them to operate with smaller governmental subsidies or with none at all. Similarly, in agriculturally oriented countries, the government is usually more actively involved in providing water for rural households.

The index of access to clean water supplies has a strong positive explanatory power for countries with per capita income below US$1,750; for countries with a higher per capita income, there is a negative relationship between access to clean water supplies and the share in GDP of government expenditure on water and energy. This probably reflects two influences: first, that at very low per capita income increased government expenditure leads to a rapid increase in access to clean water, but for countries with more than US$1,750 per capita income, increases in government expenditure on this overall category might improve electricity, steam, and gas more than water supply; second, at higher per capita income, charges for water, electricity, and gas reduce the necessary government subsidy for provision of these services.

The governments of countries like Egypt and Pakistan appear to be spending just about as much as would be expected on these services, given their own particular combinations of urbanization, manufacturing base, and population access to water supply. However, it is striking how governments such as those of Korea, Singapore, and Bangladesh appear to spend minimal amounts on the provision of these services,13 whereas, as might be expected, some developing countries in the process of industrialization appear to spend a great deal more on energy and water provision (for example, Mexico, the Philippines, Thailand, and Turkey). Sweden certainly seems to be in an anomalous position but its high expenditure level probably reflects the large capital investment associated with its nuclear energy program.

Economic Services: Roads, Other Transport, and Communications

This category includes expenditure on roads, railways, other transportation, and communications. Government expenditure on transport and communications could be expected to increase as per capita income rises and as urbanization increases; it could also be expected to rise as exports increase (to transport both industrial and agricultural goods to railways and harbors). In fact, one of the strongest associations of expenditure on these services (significant at the 5 per cent level) is with the growth in urban areas. Government expenditure on transport and communications was weakly associated with the share in total exports of other manufactured goods and fuel and negatively with the shares in GDP of manufacturing and agriculture. While such expenditure can be expected to rise with per capita GDP, the relationship is statistically insignificant.

In terms of country ranking, Ghana and Tanzania seem to spend close to what might be expected, but Turkey, Canada, and Italy spend approximately twice as much.

    Other Resources Citing This Publication