A. Expenditure Incidence

Measuring the incidence of benefits associated with the provision of public services is complex. At a minimum, we want to know which individuals avail themselves of publicly provided services. Answering this most basic question is demanding in terms of detailed micro-level surveys which collect data on visits to public health care facilities and attendance at public school.

The most complex issue in benefit incidence studies is determining the value of the benefits to individuals making use of a service or participating in a program. The simplest approach, which we rely on heavily in this paper, uses a binary indicator of whether or not one accesses a service. Implicit in this method is that all who use a service or participate in a program receive the same benefits. This is obviously not correct, and most likely introduces a systematic bias in the results. Specifically, it is likely that the poor attend lower quality schools and receive lower quality health care, in part because the services they have access to are not financed as well. This commends trying to go beyond the simple yes/no characterization of use, and instead place a differential value on the service received by individual, or sub-group in the population, and thus, the extent it is welfare improving. This is sometimes done based on the unit cost of the subsidy (i.e., the cost incurred by the government of an individual attending school, or visiting the clinic) (see for example, Meerman 1979; Selowsky 1979; Demery, Dayton and Mehra 1996; Castro-Leal, et al 1997; and Demery 1997). Numerous tenuous assumptions are required in this valuation exercise. Most obvious is the difficulty of measuring the cost of service delivery accurately. Ideally, we would arrive at unit values based on expenditure data from individual schools and health facilities. There is, however, a paucity of such information. We therefore generally rely on government budget data, divided by the estimated number of individuals going to schools or attending a specific health facility in a region or community. The correspondence between budgets and expenditures, however, is often weak. We are also forced to make strong assumptions about the homogeneity of large clusters of clinics or schools, usually defined by the geographical areas on which it is possible to derive the government’s cost of delivering the service, usually a province or region. This reflects the reality that unit cost data (on various categories of services) are only available for a few regions of the country.

A related problem is that using budget or even accurate government expenditure data inevitably assumes that marginal benefits equal average benefits. This problem is illustrated by the case of governments increasing the expenditure on primary health care clinics. Such increased spending could be primarily on clinics in good neighborhoods, not in the neighborhoods where the poor reside. Conversely, the poor may receive most of the benefits of marginal spending on health and education, particularly to the extent that the upper income households are approaching satiation at the margin, or as second round investments in schools and clinics are in more remote areas which were neglected initially.³ Thus, the information on unit subsidies is inherently inaccurate and limited in terms of the level of disaggregation.

Given our interest in measuring the extent to which a service improves individual or household utility (or some other notion of welfare), another problem in benefit incidence studies is the dubious notion that the value of a public service is equivalent to the cost incurred by the state in providing the service. In theory, the benefit to the recipient using a health clinic or attending a public school should be equated to the amount that the individual would pay for the service, or similarly, receive for re-selling the service in the open market. In practice, if we could estimate this, the outcome may have little resemblance to the cost incurred by the government in delivering the service. A number of reasons can explain this potential inconsistency. On the cost side, corruption, inefficiency, and misallocation of funds can lead to public expenditures far in excess of what actually goes to the intended beneficiaries. On the benefit side, standard economic theory indicates that unless demand is completely inelastic, the value of a transfer in kind is less than the value of a cash transfer with the same cost. On the other hand, there are often quantity constraints associated with the provision of public services which makes their value greater to rationed consumers. Even if we could make an accurate accounting of individuals’ reservation prices, we still face the possibility that the amount consumed is not discretionary, further complicating its valuation. In addition, there are other short-term benefits of using public facilities, such as feeling more fit as a result of treatment at the clinic, which are difficult to value in money metric utility terms.

But perhaps most important reason that valuing accessed services at the cost of the state subsidy deviates from the true value of the benefit is that government may not provide the “optimal” amount of service, failing to take into account the externalities, and related long-term benefits of spending on health and education. To the extent that the life-long benefits of more education and better health on the enhancement of labor productivity are not factored into the government’s decision on social sector spending, the level of expenditure will deviate from the optimum.⁴ This is particularly likely given the obvious difficulties of capturing the externalities and non-exclusivities associated with spending on health and education in the measurement of benefits. Likewise, there are also less direct longer-term benefits that may result from, for example, investing in technical schools and universities where the entrepreneurial talents of the graduates contribute to future employment possibilities in manufacturing and industry for workers in the lower end of the income distribution.

The reliance on costs as a proxy for benefits to the individual also generally fails to account for the fact that economic agents respond to available subsidies in ways that often render incorrect simple accounting of first round benefits and costs. There are a number of ways in which individuals respond to the provision of public services. Some are in their control, such as adjusting consumption and savings decisions. Important labor market responses in terms of labor-leisure tradeoffs are also expected. The benefits of public spending are also reduced by the changes in level of private transfers that result from public spending. Thus, going beyond such first approximations, and at least taking into account the behavioral response of individuals to changes in the price and availability of publicly provided services, is useful to get a more accurate estimate of the incidence of public expenditures.

The above problems amply illustrate the limitations of benefit incidence studies that use simple indicators of facilities, or disaggregated unit subsidies to value school attendance or health visits. To surmount these limitations, we need to estimate individual valuations of visits to health facilities and attendance at schools via demand function for these services. Armed with these models, we can calculate compensating variations, or willingness-to-pay for social services. These types of estimations of individuals valuations make it possible to not only examine more accurately who benefits from specific subsidies, but to engage in counterfactual experimentation and simulations of the impact of alternative pricing policy regimes on household welfare and the treasury. For example, the welfare effects of applying user fees to certain classes of social services can be determined on the basis of such models of health and education demand (see for example, Gertler, Locay and Sanderson 1987). They can also be used to determine to what extent user fees are a viable—that is, whether cost-recovery schemes represent a relatively non-distortionary means of raising revenues that can in turn increase the supply and quality of other services to the poor (Litvack and Bodart, 1993).⁵ And most important, perhaps, the price parameters provide policy makers with information on the changes in the welfare of the households affected, as per the earlier discussion of willingness to pay.

Despite the value of such models of health and education demand, the comparability of results is quite limited since the prices used are generally imputed from costs of travel, queuing, etc., given that either the services are free or money prices are difficult to accurately measure. Myriad other econometric problems also plague behavioral response models.⁶ So while we encourage further research efforts that go beyond simple benefit incidence analysis, we do so cautiously. Such exercises are demanding in terms of data and analytical capacity. Furthermore, for policy makers who are primarily interested in ranking the progressivity of benefits associated with various categories of public expenditure, or whether a service is progressive, available evidence indicates that little value is added in going beyond the simplest binary approach that assesses who makes use of what service (Younger, 1997).

B. Tax Policy

In this section we briefly discuss the methodological issues in determining the “incidence” of taxation. Our objective is to determine whose real purchasing power falls when the government imposes different types of taxes. In analyzing the economic incidence of taxes in Africa we confront a number of challenges. First, economists have understood that the entities that are legally required to pay a tax are not necessarily those that suffer a reduction in real purchasing power when the tax is imposed. They may successfully “shift” the tax onto other households. A clear example is a firm. Governments in developing countries collect most taxes from firms, but the firms do not suffer reductions in purchasing power. Either the households that own them do, or the firm shifts the tax to its customers or suppliers through changes in its prices. For example, it is standard to assume that if an industry is competitive, then a tax on its product will be passed on to consumers via a price increase equal to the tax rate. On the other hand, a tax on firms’ profits probably falls mostly on firms’ owners. The other common example is the ability to avoid a tax by changing one’s consumption or income pattern. For example, households that have high elasticities of demand for gasoline, say, can avoid a tax on gasoline consumption by switching to substitutes with little loss in welfare, while those with an inelastic demand cannot do the same so easily.

In trying to measure the economic incidence of taxes, we adopt a number of rather strong assumptions. For direct taxes, we assume that the factors that produce the associated incomes pay the taxes. Thus, wage workers pay the withholding tax on wage income. This assumption is equivalent to assuming that households supply the associated factors completely inelastically so that they cannot shift the tax. Selden and Wasylenko (1992) defend this elasticity assumption on the grounds that, while restrictive, it often produces results similar to those of more sophisticated models, but at a substantially lower cost in terms of the time and effort required.

For indirect taxes, we assume that households that consume the taxed items pay the associated taxes. Thus, smokers pay taxes on tobacco, households that use kerosene for lamps pay the taxes on kerosene, etc. There are, however, two exceptions to this general rule, largely because of the controversy that surrounds two types of taxes. For gasoline taxes, no one doubts that direct consumption of gasoline is highly concentrated in the upper end of the expenditure distribution, yet critics of gasoline taxes argue that the secondary impact of such a tax is regressive because an increase in gasoline prices causes increases in other prices, especially transport, on which poor people depend more than the rich. To include this effect, we assume that the gasoline tax falls on both direct consumers of gasoline and also consumers of public transportation services.⁷

Import duties are the other tax that is difficult to manage. Household surveys do not ask whether goods consumed are imported or not, so we cannot identify import consumers directly. Rather, we assume that the prices of all goods for which imports are a large share of the market go up by the amount of the tariff when it is imposed. Thus, those who “pay” the tax are consumers of the good, whether it is actually imported or produced domestically. However, not all of this payment goes to the government. A share of the benefits from the import duties goes to protected local producers of the same good who get to charge a higher price for their output. Thus, the costs to consumers that we identify are not equal to the government’s revenue.

Finally, for the most part, our analysis uses statutory tax rates rather than any estimates of taxes actually paid. The importance of the informal sector, smuggling, and corruption mean that taxes actually collected are far below what perfect compliance with the tax code would yield. We have made some allowances for this, mostly by assuming that certain informal purchases, mostly food and services, and informal incomes escape taxation altogether. But for other products and incomes, we assume that the taxes are paid as per the tax laws.

C. Dominance Testing

We are primarily interested in ranking the progressivity of benefits of categories of social expenditure, and different types of taxation. Furthermore, we want to evaluate the distribution of expenditures and taxes against two benchmarks: whether they are progressive (i.e., inequality reducing relative to our welfare benchmark), and whether they are per capita progressive, implying that those at the lower (upper) end of the income distribution receive (pay) at least an equal level of benefit (taxes) as upper (lower) income individuals. To do so, we use two tests for the progressivity of health and education expenditures, and the revenues that mainly finance that spending. The first involves the statistical comparison of concentration curves for the types of expenditures and taxes. These curves are similar to Lorenz curves in that they plot households from the poorest to the wealthiest on the horizontal axis against the cumulative proportion of benefits received, or taxes paid, for all households. The second employs cardinal measures in the form of the extended Gini coefficients which provides a middle ground between the normative generality (and consequent indeterminacy) of the welfare dominance approach and the precision (and lack of normative generality) of the Gini coefficient (Yitzhaki, 1983).

To amplify first on the testing of welfare dominance through the comparison of the concentration curves, Yitzhaki and Slemrod (1991) prove that for any social welfare function that is anonymous and favors an equitable distribution of income, changing the structure of expenditures (taxes) by slightly increasing (decreasing) one transfer (tax), x, and reducing (increasing) those on another, y, by just enough to keep total expenditures constant will improve social welfare when x’s concentration curve is everywhere above (below) y’s.⁸ The intuition is straightforward. If poorer households tend to receive (pay) more of the benefits (taxes) associated with a particular type of social sector expenditure (revenue measure), say primary education (export taxes), and less of another, say secondary education (VAT), then reducing (increasing) expenditures (taxes) on the latter to pay for more of the former will improve the distribution of welfare. Yitzhaki and Slemrod refer to this as welfare dominance because of the analogy with the concept of second order stochastic dominance in the finance literature.

In addition to comparing the concentration curves for different type of social services and categories of taxes, we also compare each concentration curve to two benchmarks: the Lorenz curve for per capita expenditures and the 45-degree line. We can say that an expenditure (tax) is progressive if it benefits (taxes) poorer households more (less) than wealthy ones, relative to their income, and regressive if it does not. At the same time, public expenditures, especially in the social sectors, are often held to a higher standard than taxes in their being considered well-targeted to the poor only if the benefits go disproportionately to the poor in absolute terms, not relative to income. We will call such transfers “per capita progressive” and note that they have a concentration curve that is above the 45-degree line (concave rather than convex). We will call social services whose concentration curve is above the Lorenz curve but below the 45-degree line simply “progressive” and those below the Lorenz curve are “regressive,” analogous to the standard tax literature.

Because the concentration curves are constructed from sample data, comparisons between them are, or should be, statistical.⁹ Beach and Davidson (1983) first derived distribution-free standard errors for comparison of independent Lorenz curves. However, while such standard errors are adequate for comparing distributions across independent populations, a problem arises in the case of testing dominance of social services and taxes that may be correlated with income, as well as each other. In a recent paper, Davidson and Duclos (1996) derive distribution-free standard errors for the difference between two concentration curves which may be dependent. We use the Davidson and Duclos’ estimator to establish a confidence interval around the estimated concentration curves and then test for significant differences between them.

In addition to accounting for the possible dependence between concentration curves, our tests differ from most of the literature in the way that we use the covariance matrix for the ordinate estimates. Typically, researchers who apply statistical tests use t-tests for the difference between the ordinates of two concentration curves at several abscissa (usually 0.1 to 0.9). Then they reject the null hypothesis of non-dominance when one of the ordinates differs statistically in the direction of dominance, as long as none of the other pairs indicates a statistically significant result in the opposite direction.¹⁰ Howes (1996a) shows that we can only be sure that the probability of type I error is no more than the critical value if we reject the null hypothesis in the case that the difference in the ordinates of the two curves is non-zero for every ordinate tested and, obviously, that the difference be of the same sign. This decision rule is clearly less likely than the more common one to reject the null in favor of dominance. In practice, we find that it leads us to accept the null quite often, leaving us with little to conclude about the relative progressivity of categories of expenditures or taxes. However, bounding the size of the test at the risk of low power is consistent with standard econometric practice, and we follow it here. Of course, as indicated above, failure to reject the null leaves us with an indeterminate result, unless we can establish that the two concentration curves cross, something shown by two significant differences in ordinates of opposite signs.

In another paper, Howes (1996b) criticizes the use of (rather wide) quantiles for dominance testing. In theory, we can establish welfare dominance only if one concentration curve is above another at every point. In practice, when we determine dominance by relying on t-statistics that test for the difference of ordinates in two concentration curves, it is almost always the case that as we approach the extremes of the distribution (0 and 1), t-statistics go to zero. Statistical testing of very small quantiles is also limited by the sample size. As a result, establishing dominance at each point on the concentration curve is not feasible, and instead, we rely on what Howes refers to as “restricted” dominance. This involves excluding the extreme tails so that we reject the null of non-dominance even if the curves cross or are not significantly different in that range (i.e., the 99^th percentile). Choosing how restrictive to be is difficult and arbitrary. Most papers use ordinates at the deciles (0.1 to 0.9), which ignores fairly large sections of the income distribution and thus weakens the economic significance of any conclusion that one transfer dominates another. On the other hand, choosing very small quantiles reduces the power of the test as standard errors become based on very few observations per quantile. Based on relatively small sample sizes in our surveys, and the even smaller number of individuals who, for example, are enrolled in post-secondary education, we will extend the range of values over which we test dominance only to the fifth percentile of the income distribution at the bottom and the ninety-fifth percentile at the top.¹¹ In sum, our decision rule is this: using 20 equally spaced ordinates from 0.05 to 0.95, we reject the null in favor of dominance if all the t-statistics are greater than the critical value and of the same sign; or, we reject the null in favor of crossing if there are at least two significant t-statistics with opposite signs. Rejecting the null of non-dominance using the above procedure implies that one distribution is preferred over the other under any social welfare function that favors progressivity. This is indeed a demanding criteria, especially in light of the low power of the test.¹²

A. Benefits for Social Spending

Within country comparisons

An example of the concentration curves for social sector benefits is presented for Côte d’Ivoire in Figure l.¹³ The visual examination of curves is suggestive of the progressivity of various types of services relative to each other as well as the 45-degree line and the distribution of expenditures. The curves for Côte d’Ivoire, like most countries, indicate that the most progressive of the social expenditures is primary education. At the other extreme, the concentration curve for post-secondary education is most convex. This implies that the benefits associated with post-secondary schools are regressive, being more concentrated among the rich than consumption in general, itself already highly concentrated. For most of the countries, the concentration curves for health services and secondary education fall between the Lorenz curve and the 45-degree line so that they are progressive, but the rich still enjoy a greater share of the benefits of spending in absolute terms. We also observe that, like in the case of Côte d’Ivoire, the concentration curves for non-hospital based health care are generally above those for hospital care. Many of the concentration curves for the social services cross each other, as well as the 45-degree line and the expenditure Lorenz curve, suggesting that at least in these cases, we cannot establish a clear dominance ordering.

Concentration Curves for Health and Educationin Côte d’Ivoire

Citation: IMF Working Papers 1999, 172; 10.5089/9781451858556.001.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Concentration Curves for Health and Educationin Côte d’Ivoire

Citation: IMF Working Papers 1999, 172; 10.5089/9781451858556.001.A001

• Download Figure
• Download figure as PowerPoint slide

Concentration Curves for Health and Educationin Côte d’Ivoire

Citation: IMF Working Papers 1999, 172; 10.5089/9781451858556.001.A001

• Download Figure
• Download figure as PowerPoint slide

We next examine the country-specific dominance test results following the methods outlined in the previous section to gain a clear story of whether social services (1) are per capita progressive (i.e., where the concentration curve is above the 45-degree line implying that the poor receive more benefits than the rich in absolute terms), (2) are progressive (i.e., where the concentration curve is above the expenditure distribution, implying that the poor benefit more in relative terms), and (3) can be ranked or ordered by their degree of progressivity (Tables 2 through 9). Based on t-tests for the difference between ordinates of two concentration curves at 20 abscissa, we find that with the exception of primary education in South Africa, no services are per capita progressive (i.e., we cannot reject the null that their concentration curves are equal to or below the 45-degree line). Conversely, there are many examples of the 45-degree line statistically dominating services—that is, where the poor receive less benefit from the service in per capita terms than individuals at the upper end of the expenditure distribution. We reject the null in favor of the dominance of the 45-degree line for: post-secondary school in Ghana, Guinea, Madagascar, South Africa, and Uganda; secondary school in Guinea, Tanzania, and Uganda; primary school in Guinea; hospital care in Ghana, Guinea, and Tanzania, and non-hospital care in Madagascar. In addition, there are a number of cases where we find statistically significant crossings with the 45-degree line:¹⁴ primary education in Côte d’Ivoire, Madagascar, Mauritania, Tanzania, and Uganda; secondary school in South Africa; hospital care in South Africa; and non-hospital care and South Africa.

Table 2.

Dominance Results for Public Education and Health Services in Côte d’Ivoire

Source: Government of Côte d’Ivoire (CILS), 1985; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 2.

Dominance Results for Public Education and Health Services in Côte d’Ivoire

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	45° line		X				D
(2)	Primary	X			D	D	D
(3)	Non-hospital						D
(4)	Secondary
(5)	Hospital
(6)	Household expenditure
(7)	Post-secondary

Source: Government of Côte d’Ivoire (CILS), 1985; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 2.

Dominance Results for Public Education and Health Services in Côte d’Ivoire

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	45° line		X				D
(2)	Primary	X			D	D	D
(3)	Non-hospital						D
(4)	Secondary
(5)	Hospital
(6)	Household expenditure
(7)	Post-secondary

Source: Government of Côte d’Ivoire (CILS), 1985; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 3.

Dominance Results for Public Education and Health Services in Ghana

Source: Government of Ghana (GLSS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 3.

Dominance Results for Public Education and Health Services in Ghana

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	Primary				D		D	D
(2)	45° line					D	D	D
(3)	Non-hospital							D
(4)	Secondary							D
(5)	Post-secondary
(6)	Hospital
(7)	Household expenditure

Source: Government of Ghana (GLSS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 3.

Dominance Results for Public Education and Health Services in Ghana

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	Primary				D		D	D
(2)	45° line					D	D	D
(3)	Non-hospital							D
(4)	Secondary							D
(5)	Post-secondary
(6)	Hospital
(7)	Household expenditure

Source: Government of Ghana (GLSS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 4.

Dominance Results for Public Education and Health Services in Guinea

Source: Government of Guinea (EIS), 1994; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 4.

Dominance Results for Public Education and Health Services in Guinea

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	45° line			D	D	D	D	D
(2)	Non-hospital				D	D	D	D
(3)	Primary					D	D	D
(4)	Secondary					X
(5)	Household expenditure				X		X
(6)	Hospital					X
(7)	Post-secondary

Source: Government of Guinea (EIS), 1994; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 4.

Dominance Results for Public Education and Health Services in Guinea

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	45° line			D	D	D	D	D
(2)	Non-hospital				D	D	D	D
(3)	Primary					D	D	D
(4)	Secondary					X
(5)	Household expenditure				X		X
(6)	Hospital					X
(7)	Post-secondary

Source: Government of Guinea (EIS), 1994; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 5.

Dominance Results for Public Education and Health Services in Madagascar

Source: Government of Madagascar (EPM), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X’ indicates that the concentration curves cross.

Table 5.

Dominance Results for Public Education and Health Services in Madagascar

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	Primary		X	D		D	D	D
(2)	45° line	X		D			D	D
(3)	Non-hospital						D	D
(4)	Hospital
(5)	Secondary						X
(6)	Household expenditure					X		D
(7)	Post-secondary

Source: Government of Madagascar (EPM), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X’ indicates that the concentration curves cross.

View Table

Table 5.

Dominance Results for Public Education and Health Services in Madagascar

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	Primary		X	D		D	D	D
(2)	45° line	X		D			D	D
(3)	Non-hospital						D	D
(4)	Hospital
(5)	Secondary						X
(6)	Household expenditure					X		D
(7)	Post-secondary

Source: Government of Madagascar (EPM), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X’ indicates that the concentration curves cross.

Table 6.

Dominance Results for Public Education and Health Services in Mauritania

Source: Government of Mauritania (EPC), 1995/6; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 6.

Dominance Results for Public Education and Health Services in Mauritania

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	45° line				X			D
(2)	Non-hospital							D
(3)	Hospital
(4)	Primary	X				D	D	D
(5)	Secondary
(6)	Post-secondary							X
(7)	Household expenditure						X

Source: Government of Mauritania (EPC), 1995/6; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 6.

Dominance Results for Public Education and Health Services in Mauritania

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	45° line				X			D
(2)	Non-hospital							D
(3)	Hospital
(4)	Primary	X				D	D	D
(5)	Secondary
(6)	Post-secondary							X
(7)	Household expenditure						X

Source: Government of Mauritania (EPC), 1995/6; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 7.

Dominance Results for Public Education and Health Services in South Africa

Source: Government of South Africa (SALS), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

^1/School attendance is for public and private schools.

Table 7.

Dominance Results for Public Education and Health Services in South Africa

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	Primary 1/		X		D		D	D
(2)	Non-hospital	X			X		D	D
(3)	Hospital				X		D	D
(4)	45° line		X	X		X	D	D
(5)	Secondary 1/				X		D	D
(6)	Post-secondary 1/
(7)	Household expenditure

Source: Government of South Africa (SALS), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

^1/School attendance is for public and private schools.

View Table

Table 7.

Dominance Results for Public Education and Health Services in South Africa

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	Primary 1/		X		D		D	D
(2)	Non-hospital	X			X		D	D
(3)	Hospital				X		D	D
(4)	45° line		X	X		X	D	D
(5)	Secondary 1/				X		D	D
(6)	Post-secondary 1/
(7)	Household expenditure

Source: Government of South Africa (SALS), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

^1/School attendance is for public and private schools.

Table 8.

Dominance Results for Public Education and Health Services in Tanzania

Source: Government of Tanzania/World Bank (HRD), 1995; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 8.

Dominance Results for Public Education and Health Services in Tanzania

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	Post-secondary				X	X		X
(2)	Primary			X		D	D	D
(3)	45° line		X			D	D	D
(4)	Non-hospital	X						D
(5)	Hospital	X
(6)	Secondary
(7)	Household expenditure	X

Source: Government of Tanzania/World Bank (HRD), 1995; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 8.

Dominance Results for Public Education and Health Services in Tanzania

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	Post-secondary				X	X		X
(2)	Primary			X		D	D	D
(3)	45° line		X			D	D	D
(4)	Non-hospital	X						D
(5)	Hospital	X
(6)	Secondary
(7)	Household expenditure	X

Source: Government of Tanzania/World Bank (HRD), 1995; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 9.

Dominance Results for Public Education and Health Services in Uganda

Source: Government Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X’ indicates that the concentration curves cross.

Table 9.

Dominance Results for Public Education and Health Services in Uganda

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	Primary		X			D	D	D
(2)	45° line	X				D	D	D
(3)	Non-hospital					D	D	D
(4)	Hospital
(5)	Post-secondary
(6)	Secondary							X
(7)	Household expenditure						X

Source: Government Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X’ indicates that the concentration curves cross.

View Table

Table 9.

Dominance Results for Public Education and Health Services in Uganda

		(1)	(2)	(3)	(4)	(5)	(6)	(7)
(1)	Primary		X			D	D	D
(2)	45° line	X				D	D	D
(3)	Non-hospital					D	D	D
(4)	Hospital
(5)	Post-secondary
(6)	Secondary							X
(7)	Household expenditure						X

Source: Government Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X’ indicates that the concentration curves cross.

Comparisons between the expenditure Lorenz curve for household expenditures and various categories of social services reveal a number of cases where the latter dominate (i.e., where the services are progressive). Foremost, we can reject the null of non-dominance between public primary schools and the expenditure Lorenz curve in all countries. The same is true for non-hospital health care. That is, the benefits of primary school and health care outside hospitals are more progressive than the distribution of expenditures in African countries. This general pattern, however, does not apply to the benefits of hospital care, secondary and post-secondary education. Specifically, public secondary schools are only progressive relative to the expenditure distribution in the cases of Ghana and South Africa. The only other case of expenditure progressivity is hospitals in South Africa.¹⁵

Pair-wise comparisons of social services also reveal some common patterns. Primary education dominates secondary education in all cases except Guinea and South Africa; although, only in the case of South Africa can we statistically prove that secondary schooling is more progressive than post-secondary schooling.¹⁶ We can only show that hospital care is less progressive than other facilities (i.e., clinics) in the case of Guinea despite that comparison with the Lorenz curves suggest that the latter are more progressive. When we compare primary education with non-hospital based health services we cannot reject the null of non-dominance, except in Madagascar, indicating no general ordering in terms of the progressivity of the two types of benefits.

Regional disaggregation

While the results above are based on national data, it is also possible to disaggregate the data regionally, and by gender. In the case of primary education, we reject the null of dominance between rural and urban areas in all countries except Côte d’Ivoire. And in the case of non-hospital health services, we do the same for the Côte d’Ivoire, Guinea, Madagascar, and Uganda. This implies that services provided in rural areas are more progressive than those in urban areas. One can infer that, on the margin, directing more services to rural areas will likely contribute to a more progressive distribution of welfare.

In contrast, a comparison of the male and female concentration curves in all countries reveals few differences, as illustrated by Mauritania and Tanzania. This applies to both education and health. In fact, a review of the dominance test results indicates only one case where we reject the null that the concentration curves for males and females are the same—for primary education in Uganda, where the equality of the benefits of men’s education exceeds that of women’s education. Thus, unlike geographical targeting, there is no evidence here that social sector spending on men is more or less equitable than that on women.¹⁷

Comparing methods for service valuation: binary indicators vs. disaggregated unit costs

In this section we compare the results of analyzing benefit incidence based on a simple dichotomous variable of whether or not an individual uses a service (i.e., goes to a clinic or attends school), with the unit subsidy valuation derived from dividing government budget data by government estimates of the number of individuals who use a service.¹⁸ Our interest in making this comparison is to explore the extent to which the two methods differ, and to understand why.

We are able to compare unit subsidies to the binary approach for health and education services in Guinea, Madagascar, and to a lesser extent, South Africa. In Guinea, our unit subsidies are disaggregated on the basis of the 5 regions of the country (see Appendix Tables 33 and 34). In terms of dominance testing, there is only one change in dominance orderings from the results that rely on the binary variable (Table 10): the 45-degree line no longer dominates primary education.

Table 10.

Dominance Results for Unit Values and Binary Methods for Guinea

Source: Government of Guinea (EIS), 1994; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 10.

Dominance Results for Unit Values and Binary Methods for Guinea

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
(1)	45° line					D	D	D	D	D	D
(2)	Non-hospital, binary				D		D	D	D	D	D
(3)	Primary, value				X	D		D	D	D	D
(4)	Non-hospital, value			X					D	D	D
(5)	Primary, binary								D	D	D
(6)	Secondary, value							D
(7)	Secondary, binary								X
(8)	Household expenditure							X		X	X
(9)	Hospital, value								X
(10)	Hospital, binary								X

Source: Government of Guinea (EIS), 1994; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 10.

Dominance Results for Unit Values and Binary Methods for Guinea

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
(1)	45° line					D	D	D	D	D	D
(2)	Non-hospital, binary				D		D	D	D	D	D
(3)	Primary, value				X	D		D	D	D	D
(4)	Non-hospital, value			X					D	D	D
(5)	Primary, binary								D	D	D
(6)	Secondary, value							D
(7)	Secondary, binary								X
(8)	Household expenditure							X		X	X
(9)	Hospital, value								X
(10)	Hospital, binary								X

Source: Government of Guinea (EIS), 1994; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Before presenting the dominance results from Madagascar, we once again refer to the unit subsidy figures in Appendix Tables 35 and 36. We are particularly skeptical about what we find to be somewhat implausible figures from Madagascar: that the unit subsidy for basic health care facilities in Antananarivo is far less than that in 4 of the other 5 regions. Conversely, the unit value of hospital visits is substantially more in Antananarivo than other regions, as we would expect, being more than 4 times greater in two instances. The reason for our skepticism is that we can think of no a priori reason non-hospital care is so much less expensive in the capital city, while hospital care is much more so. With this qualification, we find no changes either relative to the Lorenz curve, the 45-degree line, or in the ordering of the progressivity of services (Table 11).

Table 11.

Dominance Results for Unit Values and Binary Methods for Madagascar

Source: Government of Madagascar (EPM), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 11.

Dominance Results for Unit Values and Binary Methods for Madagascar

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
(1)	Primary, value				D	D		D	D	D	D
(2)	Primary, binary			X	D	D		D	D	D	D
(3)	45° line		X		D	D				D	D
(4)	Non-hospital, value										D
(5)	Non-hospital, binary										D
(6)	Hospital, binary
(7)	Secondary, value										X
(8)	Secondary, binary										X
(9)	Hospital, value
(10)	Household expenditure							X	X

Source: Government of Madagascar (EPM), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 11.

Dominance Results for Unit Values and Binary Methods for Madagascar

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
(1)	Primary, value				D	D		D	D	D	D
(2)	Primary, binary			X	D	D		D	D	D	D
(3)	45° line		X		D	D				D	D
(4)	Non-hospital, value										D
(5)	Non-hospital, binary										D
(6)	Hospital, binary
(7)	Secondary, value										X
(8)	Secondary, binary										X
(9)	Hospital, value
(10)	Household expenditure							X	X

Source: Government of Madagascar (EPM), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

In the case of South Africa, we have unit subsidy information, by region, only for health services. More specifically, we can distinguish between nine regions of the country, in terms of the unit costs of a visit to hospitals and health centers/clinics. These are shown in Appendix Table 36, where extremely large regional differences are noted. To no surprise, the unit subsidies are highest in the Northern Cape, and lowest in Eastern Transvaal. For health clinics, the difference is more than seven times, and for hospital benefits, the difference is almost fivefold. Despite these dramatic regional differences in unit subsidies, and the fact that an examination of the concentration curves indicates that over most of the range of values the binary approach makes services appear more progressive, we find no statistical differences from the binary approach in the ordering of health care, or in the comparisons relative to the 45-degree line or Lorenz curve (Table 12).

Table 12.

Dominance Results for Unit Values and Binary Methods for South Africa

Source: Government of South Africa (SALS), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

^1/School attendance is for public and private schools.

Table 12.

Dominance Results for Unit Values and Binary Methods for South Africa

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
(1)	Primary, binary 1/		X			D			D	D	D	D	D
(2)	Non-hospital, binary	X				X					D	D	D
(3)	Non-hospital, value					X			X		D	D	D
(4)	Hospital, binary					X				D	D	D	D
(5)	45° line		X	X	X		X	X	D		D	D	D
(6)	Secondary, binary 1/					X			X	D	D	D	D
(7)	Hospital, value					X			X		D	D	D
(8)	Primary, value			X			X	X			D	D	D
(9)	Secondary, value												D
(10)	Post-secondary, binary 1/
(11) Post-secondary, value
(12)	Household expenditure

Source: Government of South Africa (SALS), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

^1/School attendance is for public and private schools.

View Table

Table 12.

Dominance Results for Unit Values and Binary Methods for South Africa

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
(1)	Primary, binary 1/		X			D			D	D	D	D	D
(2)	Non-hospital, binary	X				X					D	D	D
(3)	Non-hospital, value					X			X		D	D	D
(4)	Hospital, binary					X				D	D	D	D
(5)	45° line		X	X	X		X	X	D		D	D	D
(6)	Secondary, binary 1/					X			X	D	D	D	D
(7)	Hospital, value					X			X		D	D	D
(8)	Primary, value			X			X	X			D	D	D
(9)	Secondary, value												D
(10)	Post-secondary, binary 1/
(11) Post-secondary, value
(12)	Household expenditure

Source: Government of South Africa (SALS), 1993; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

^1/School attendance is for public and private schools.

In addition to this regional disaggregation, we also have disaggregated unit subsidies, by race, for education. As show in Appendix Table 38, the spending per student is dramatically higher for whites than Africans, with that for coloreds falling in between. Also noteworthy is that the spending in the Homelands on Africans is far lower than the non-Homelands. When examining the dominance results for the binary approach versus unit subsidies that take into account these dramatic racial differences, we find that the 45-degree line now dominates primary education, while just the opposite was true based on the binary approach (Table 12). Another difference is that when the binary approach is used we find that secondary school is more progressive than university education. This is not the case when unit subsidies are employed. Thus, when there is a high correlation between income and the benefits of a service received by different segments of the population, employing unit values can have an important impact on the findings.

Cross-country comparisons

In this section we explore inter-country comparisons of the progressivity of certain categories of social sector expenditure. Prior to doing so, however, we admonish caution in drawing inferences from these results. While all surveys in this study are quite similar in terms of the questionnaire design, the surveys undoubtedly differ in terms of sampling and non-sampling errors. These types of errors are not expected to effect significantly intra-country comparisons of the progressivity of expenditure, as presented above. However, they will detract from the quality of inter-country comparisons, as this study is not immune from the limitations of all similar exercises that examine inequality across different countries.

Statistical tests of dominance of the country Lorenz curves reveal that South Africa’s expenditure inequality is significantly worse than other countries’ (Table 13). Likewise, Ghana’s inequality is less than all but Mauritania’s. Both Mauritania’s and Tanzania’s expenditure distribution is less concentrated than Guinea, Madagascar, and Côte d’Ivoire. Expenditure inequality, based on a statistical comparison of the 20 pairs of ordinates is also found to be less in Uganda than Madagascar and Côte d’Ivoire.

Table 13.

Cross-Country Dominance Results for Household Expenditures

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 13.

Cross-Country Dominance Results for Household Expenditures

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	Ghana			D	D	D	D	D	D
(2)	Mauritania				X	D	D	D	D
(3)	Tanzania					D	D	D	D
(4)	Uganda		X				D	D	D
(5)	Guinea								D
(6)	Madagascar								D
(7)	Côte d’Ivoire								D
(8)	South Africa

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 13.

Cross-Country Dominance Results for Household Expenditures

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	Ghana			D	D	D	D	D	D
(2)	Mauritania				X	D	D	D	D
(3)	Tanzania					D	D	D	D
(4)	Uganda		X				D	D	D
(5)	Guinea								D
(6)	Madagascar								D
(7)	Côte d’Ivoire								D
(8)	South Africa

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Comparisons of the progressivity of primary education reveal that the concentration curve for South Africa (which unlike the other countries includes private as well as public schools) dominates those in Guinea, Tanzania, and Uganda (Table 14). The findings about the relative progressivity in South Africa are particularly interesting in light of the extremely unequal expenditure distribution in South Africa. Also, in regard to education, the distribution of benefits associated with primary schools in all countries, except Mauritania, is more progressive than in Guinea.

Table 14.

Cross-Country Dominance Results for Use of Public Primary Schools

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 14.

Cross-Country Dominance Results for Use of Public Primary Schools

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	South Africa			X	D	D			D
(2)	Ghana								D
(3)	Madagascar	X							D
(4)	Uganda								D
(5)	Tanzania								D
(6)	Côte d’Ivoire								D
(7)	Mauritania
(8)	Guinea

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 14.

Cross-Country Dominance Results for Use of Public Primary Schools

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	South Africa			X	D	D			D
(2)	Ghana								D
(3)	Madagascar	X							D
(4)	Uganda								D
(5)	Tanzania								D
(6)	Côte d’Ivoire								D
(7)	Mauritania
(8)	Guinea

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Dominance testing further indicates that secondary schooling is more progressive in South Africa than Côte d’Ivoire, Guinea, Madagascar, and Uganda; and in Ghana than Guinea and Uganda (Table 15). Post-secondary schools in South Africa dominate Guinea and Madagascar, and those in Ghana also dominate Madagascar (Table 16).

Table 15.

Cross-Country Dominance Results for Use of Public Secondary Schools

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 15.

Cross-Country Dominance Results for Use of Public Secondary Schools

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	South Africa			D			D	D	D
(2)	Ghana							D	D
(3)	Côte d’Ivoire
(4)	Mauritania
(5)	Tanzania
(6)	Madagascar							X	X
(7)	Uganda						X
(8)	Guinea						X

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 15.

Cross-Country Dominance Results for Use of Public Secondary Schools

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	South Africa			D			D	D	D
(2)	Ghana							D	D
(3)	Côte d’Ivoire
(4)	Mauritania
(5)	Tanzania
(6)	Madagascar							X	X
(7)	Uganda						X
(8)	Guinea						X

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 16.

Cross-Country Dominance Results for Use of Public Post-Secondary Schools

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Govemmentof South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 16.

Cross-Country Dominance Results for Use of Public Post-Secondary Schools

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	Tanzania		X	X
(2)	Ghana	X							D
(3)	South Africa	X						D	D
(4)	Uganda
(5)	Mauritania
(6)	Côte d’Ivoire
(7)	Guinea
(8)	Madagascar

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Govemmentof South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 16.

Cross-Country Dominance Results for Use of Public Post-Secondary Schools

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	Tanzania		X	X
(2)	Ghana	X							D
(3)	South Africa	X						D	D
(4)	Uganda
(5)	Mauritania
(6)	Côte d’Ivoire
(7)	Guinea
(8)	Madagascar

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Govemmentof South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

The only dominance found in cross-country comparisons of non-hospital based health care is that, in Madagascar, services are less progressive than in Guinea and Uganda (Table 17). When it comes to the distribution of benefits associated with hospital care, dominance results indicate that benefits are less concentrated in South Africa than Côte d’Ivoire, Ghana, Guinea, Madagascar and Tanzania (Table 18).

Table 17.

Cross-Country Dominance Results for Use of Public Non-Hospital Health Services

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 17.

Cross-Country Dominance Results for Use of Public Non-Hospital Health Services

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	South Africa					X
(2)	Mauritania					X
(3)	Uganda								D
(4)	Côte d’Ivoire
(5)	Guinea	X	X						D
(6)	Ghana
(7)	Tanzania
(8)	Madagascar

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 17.

Cross-Country Dominance Results for Use of Public Non-Hospital Health Services

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	South Africa					X
(2)	Mauritania					X
(3)	Uganda								D
(4)	Côte d’Ivoire
(5)	Guinea	X	X						D
(6)	Ghana
(7)	Tanzania
(8)	Madagascar

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 18.

Cross-Country Dominance Results for Use of Public Hospitals

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

Table 18.

Cross-Country Dominance Results for Use of Public Hospitals

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	South Africa				D	D	D		D
(2)	Mauritania
(3)	Uganda
(4)	Côte d’Ivoire								D
(5)	Ghana								D
(6)	Tanzania								D
(7)	Madagascar
(8)	Guinea

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

View Table

Table 18.

Cross-Country Dominance Results for Use of Public Hospitals

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1)	South Africa				D	D	D		D
(2)	Mauritania
(3)	Uganda
(4)	Côte d’Ivoire								D
(5)	Ghana								D
(6)	Tanzania								D
(7)	Madagascar
(8)	Guinea

Sources: Government of Côte d’Ivoire (CILS), 1985, Government of Ghana (GLSS), 1992; Government of Guinea (EIS), 1994, Government of Mauritania (EPC), 1995/6; Government of Madagascar (EPM), 1993; Government of South Africa (SALS), 1993; Government of Tanzania/World Bank (HRD), 1995; Government of Uganda (HIS), 1992; and authors’ calculations.Notes: All measures are scaled by household size (per capita); the “D” indicates that the row dominates the column, and the “X” indicates that the concentration curves cross.

In the cases of Guinea and Madagascar, where we have unit value information on health and education, we sum up the benefits across types of social services, and across all social services. Of the four services we compare (hospitals, other health care, primary education, and secondary education), only one is significantly different: primary education subsidies in Madagascar are more progressive than those in Guinea. Further, because primary education is a large share of all subsidies, the sum of all subsidies for these four services is more progressive in Madagascar than in Guinea.