Welfare Theory Versus Public Choice Theory

Classical welfare theory gives us a normative view of what government should do. The main economic role of government is to correct market failure by funding public goods, by subsidizing (or taxing) goods that generate (positive or negative) externalities, and by compensating for capital market or insurance market failure, in addition to simply setting the framework within which private enterprise will function. ¹

With respect to distribution, the "maximum" point of social welfare is acknowledged by most economists to depend on equity as well as efficiency. Opinions vary widely on whether a "social welfare function" exists, what an "equitable" distribution would be, and whether it is possible to aggregate diverse preferences to arrive at a consensus on this matter.² However, despite this ambivalence about the redistributive role of government, most economists agree that if there is to be any redistribution, it should be from rich to poor and not vice versa—the ''Robin Hood'' function of government (Birdsall, 1992). We use the term "equity" as a shortened form of Robin Hood redistribution in this paper.

A second, more recent, and less benevolent view of government activities stems from public choice theory, which gives us a positive model of what the government will do, under the presumption that the chief agents act to maximize individual utility rather than social welfare. According to this theory, politicians do not seek to maximize efficiency but rather to maximize their own chances of staying in power; bureaucrats seek to maximize their budgets; and individuals use governments to augment their real income via the creation of protected market positions and the direct provision of services and transfers (Mueller, 1979; Borcherding, 1985).

Politicians and political parties have some discretionary power because of barriers to entry and because they are in a position to shape as well as respond to people's tastes. At the same time, threats from actual or potential competitors limit the scope of their monopoly power. Thus, natural selection operates in political life as well as in economic or biological life. In terms of the entire spectrum of issues, among which different groups of voters have different trade-offs, the politicians who survive to make policy are those who assess these trade-offs correctly and give influential groups what they want on issues that are most salient to them. Where democracy does not exist, a similar process often occurs, but with greater discretionary monopoly powers for government officials who control the political market.

Public policies designed to maximize private interests will not necessarily be inefficient. Indeed, politically influential groups would have a potentially larger pie to capture if the Pareto frontier were reached; compensatory mechanisms could then make everyone better off. Taxes could be imposed on some (less influential groups) and transferred to others.

However, the allocation of resources resulting from public choice politics is often inefficient, for the following reasons:

Public Choice in Developing Countries

How could we expect these divergent forces to sort themselves out in the developing country context? On the one hand, the gulf between rich and poor and the relative number of poor people are much greater there, so under "one man, one vote" we would expect to find the poor gaining from politically induced redistributions. Indeed, in a few countries (e.g., Malaysia) an economically disadvantaged group has used its political advantage effectively to increase its share of the national income. However, differences in education (hence, in organizational and communication skills) are also much greater in developing countries, and democratic institutions are often primitive, limiting the power of the poor. We would expect the latter tendency to dominate in most cases.

This is not to say that there will be no redistribution to the poorer classes. In fact, even when the rich are in control, we would expect to find some such redistribution of income. For example, people voluntarily donate to beggars and use the government as an efficient mechanism for donating to disadvantaged groups, in part because the extremes of poverty and socioeconomic immobility raise fears of crime or revolution that will ultimately hurt the rich. In developed countries, historically, the provision of certain merit goods to the poor (e.g., basic health and education services, social insurance) has been viewed as an effective way to combat these problems.

In addition, in developing countries, where there are many more poor people than rich, the desire to constrain the popularity of opposition parties encourages some distribution to lower-income groups on grounds of expediency. Out-of-power groups must be appeased by giving them "just enough" to prevent opposition parties from gaining strong support (a "contestable market" view of political equilibrium), but just enough may not be very much. For example, it may imply that the poor are given low-cost services or very limited access to high-cost services from which the rich are the main beneficiaries. Governmental expenditures on high-quantity, low-quality primary-level school systems and on selective high-cost universities are common illustrations of these two phenomena. In short, in many situations, perverse distributional rather than efficiency or equity criteria determine the allocation of government funds, and these criteria imply large benefits to powerful upper-income groups, combined with small redistributions to the poor (Behrman and Birdsall, 1988). We believe that these pessimistic predictions of public choice theory are consistent with the observed actions of developing countries in health today.

Correlation Between Mortality and Poverty

The most obvious success of the state in reducing mortality in developing countries has come via programs based on new technologies, programs to immunize people and to control malaria and other endemic diseases. In some cases (e.g., malaria control), these programs provided "public goods" that were automatically available to all. In other cases (e.g., immunizations), they provided "quasi-public goods" that were ostensibly available to all but, in fact, had a large private component that was rationed to the people (typically middle and upper class) who were most likely to perceive and capture these benefits. This group was the easiest clientele to reach and serve; hence, such expenditures bore a high rate of return, as well as strong political support.

Further mortality declines will be much harder to achieve; they will depend much more than in the past on behavioral changes, particularly among people who have been most resistant to such change, (i.e., the poor and uneducated), and on public expenditures specifically targeted toward these groups.⁷ Moreover, these are the politically disenfranchised groups in many societies. Thus, such policies will be costlier, will bear a lower rate of return, and, for political economy reasons, are unlikely to be adopted.

What is the evidence that further mortality declines will be harder to achieve—because they will have to reach the poor? First, within developing countries, the differentials in mortality associated with income and other measures of socioeconomic status have persisted, even when overall mortality has fallen substantially. In Brazil, deaths from infectious and parasitic diseases have declined drastically. These accounted for 45 percent of all deaths in 1930 but only 11 percent in 1980 (Briscoe, 1990). Other causes of death, however, are highly concentrated among the poor. Infant mortality rates among the rural poor in the northeast of Brazil are two to three times higher than rates in the wealthier, more urbanized south. For eight developing countries in which households were surveyed in the mid-1980s, infant mortality among mothers with no education was two to three times higher than among mothers with secondary education and three to six times higher than among mothers with more than secondary education (Mosley et al., 1990). The poverty rate works better than average income as a predictor of infant mortality and life expectancy in a sample of 86 developing countries (Anand and Ravallion, 1993). These continuing differences suggest that progress in reducing mortality among the very poor and uneducated requires new, and probably more expensive, initiatives (including the reduction in poverty) than those that succeeded in reducing aggregate mortality in the past.

Second, there is a difference in the nature of diseases that kill people in developing countries today and those that predominated in developed countries when they had similar overall mortality levels. For example, the proportion of deaths due to diarrhea is much higher in the developing world than it was in the West when overall mortality levels were similar, in large part because the prevalence of other diseases has been reduced by technological interventions (Preston et al., 1972; World Health Organization, 1989; J.L. Bobadilla, personal correspondence, 1991). Diarrhea is a disease of the poor; it is found along with, and contributes to, malnutrition, and it is caused by lack of access to clean water, simple health services, and basic education. To date, diarrhea has been relatively impervious to the programs and technologies that have reduced other causes of mortality. It has been hoped that the "new technology" of oral rehydration therapy (ORT) can reduce mortality due to diarrhea, but use of ORT itself requires change in the behavior of mothers and other caretakers. Moreover, it is not clear that repeated handling of diarrhea through ORT actually reduces mortality in environments in which infants are likely to die from other diseases of poverty (Mosley and Chen, 1984).

Third, there is evidence across countries of continuing huge disparities in disease risks, including risk of chronic disease, that are also clearly associated with overall differences in income, education, and the effectiveness of the public health infrastructure. For example, the annual risk of infection from tuberculosis is 50 to 200 times greater in developing than in developed countries (Mosley et al., 1990).

Reductions in mortality will continue in developing countries at a moderate rate, if further increases in educational opportunities, especially for women, and in family income are assumed. But the dramatic declines in mortality of the past are unlikely—unless public expenditures are concentrated much more heavily on the poor, to ensure delivery of good-quality personal health services. Of critical importance are family planning, nutritional supplementation for children, prenatal and obstetrical care, clean water, and the correct use of effective drugs (against respiratory infections, tuberculosis, and malaria). However, for such services to be effective, not only must they be readily available, they must be sought, understood, and used voluntarily by their clientele. Such behavioral change will occur only if the time and money costs of health services to the poor decline and/or if the poor receive information that changes their underlying tastes and choices.

The time and money costs will decline only if government targets its health spending to the poor, making basic health services easily accessible in the rural areas and urban neighborhoods where the poor are concentrated. For the reasons of political economy given above, such targeting appears unlikely. Similarly, it will be difficult to inform the poor and change their health habits (e.g., concerning nutrition, sanitation, and drug regimes to handle tuberculosis) unless their educational levels increase, which also requires a heavy and unlikely targeting of public (educational) expenditures toward the poor. Indeed, if the poor had access to substantially improved and heavily subsidized health, education, water, and sanitation facilities, there would be drastic improvements in their mortality rates and therefore the overall mortality rates of their countries, but a drastic redistribution of real income and utility would be implied. In this sense, continued high rates of health improvement and income redistribution go hand in hand, and both will be opposed by powerful groups intent on maintaining the distributional status quo.

Population Aging and Disease Patterns

The aging of the population in developing countries, a result of fertility declines over the last two decades and falling death rates, means that the prevalence of such chronic "adult" diseases as cancer, hypertension, and heart disease (in comparison to parasitic diseases and childhood infectious and diarrheal diseases) is increasing. The number of people over age 65 in developing countries will more than double between 1985 and 2015. By the year 2025, this elderly group will exceed 10 percent of the population in many developing countries, a proportion close to that in the United States today (Mosley et al., 1990; World Bank, 1991).

Brazil provides a telling example of the effects of past improvements in health technology and future changes in demographic structure on the pattern of disease. In 1950, heart disease, stroke, cancer, and accidents accounted for 20 percent of deaths there; in 1980, they accounted for approximately 50 percent. The increase in their relative contribution over the period was due largely to a decline in the proportion of deaths caused by infectious and parasitic diseases. However, the changing age structure means that from now on their prevalence is likely to rise as the proportion of the elderly in the population increases. Age structure changes alone imply a 60 percent increase in deaths due to these diseases from 1980 to 2020 (Briscoe, 1990).

Increases in age-specific death rates from some of these chronic diseases will further change epidemiological patterns. Age-specific death rates are likely to rise because of increasing exposure to such risks as smoking, poor diet, and urban pollution. For example, increases in tobacco consumption in the developing world over the last 40 years are likely to cause increases in the incidence of lung cancer comparable to the large increases experienced in Great Britain in the 1960s and 1970s, and in the United States more recently (in both places the peak has already been passed; Jamison and Mosley, 1991).

These changing demographic and epidemiological patterns will put new financial pressure on health systems in developing countries. By 1980 in Brazil, the allocation of public health resources for curative care had increased to 85 percent of all spending (from 36 percent in 1965), and the treatment of patients with heart disease accounted for an estimated 25 percent of all inpatient costs. Per capita health expenditures on persons over 60 were 3.5 times greater than the average for the population.⁸

Unfortunately, continuing increases in spending on hospital services for the care of chronic disease, although very expensive, will have little effect on the high death rates in developing countries among the poor.⁹ The estimated cost (per discounted healthy life years gained) for such services as maternal and child health care, immunization, vitamin A supplementation, antismoking campaigns, and short-course therapy for tuberculosis is $50 to $150; the estimated cost per life saved through curative treatment (e.g., at the extreme for coronary bypass surgery and hospital management of lung and stomach cancers) is more than $1,000 (Jamison and Mosley, 1991; Akin et al., 1987). The latter costs are likely to rise still further because of continuing pressure to use high-cost technologies, such as open-heart surgery, organ transplants, and kidney dialysis.

Thus, the marginal cost of reducing mortality would appear to be much lower if societies concentrated on diseases of the young (and the poor), and the marginal benefit, in terms of years of life expectancy added, would be much higher; hence such an allocation would seem to be efficient by these criteria. However, political pressures from influential groups have led most societies to spend disproportionately on diseases of adults, particularly curative hospital care, as described in detail later. The problem is exacerbated because the young come disproportionately from lower-income groups and the old from middle-and upper-income groups; the problem is currently being experienced in developed countries as well. However, its consequences are (and will be) even more marked in developing countries, where income and health disparities are greater. The argument of the previous section suggests that unless funding mechanisms change, the battle will be won by the wealthier, older groups whose rapidly accelerating health care costs will use up public resources that could more productively, and equitably, be spent on the younger and poorer members of society.

Endogeneity of Medical Research

This argument is strengthened if we ask: Why are expensive high-technology procedures aimed disproportionately at curing diseases of the old? We believe that the allocation of research and developmental resources in this direction is not accidental. Instead, we believe that the allocation process within the medical research arena is itself endogenous, governed by the same political economy forces described above with regard to service delivery. (For a much fuller discussion of the endogeneity of medical research, particularly its response to alternative insurance systems, see Weisbrod, 1991.)

A large share of medical research is carried on by pharmaceutical and other private companies in industrial countries, with the object of patenting and selling the life-prolonging drugs and equipment that they develop. The generous health insurance coverage received by the old, in part because of their political power, combined with the rapidly aging population profile in these countries, assures these companies of a ready market for their new products. On the other hand, the proportion of children in the population is declining, and they come disproportionately from low-income families who are less likely to have health insurance or to use it if they have it. Along similar lines, insurance often covers curative treatment rather than preventive. Therefore, companies in developed countries have less incentive to direct their research efforts toward children and preventive treatment than toward old people and cures. As a result, the developing countries, which utilize the new products and methods that emerge from this process, are faced with expensive medical technologies aimed at prolonging the life of the old rather than preventing illness among the young, and also with political pressures from the influential older groups in society for public spending to make these technologies accessible to them.

Other Reasons for Pessimism

Other reasons for pessimism include increased fiscal pressures on the government, leading it to divert expenditures from health care; political pressures from workers that often cause these cuts to be made in inefficient ways; and cost escalations in health care that diminish the productivity of the remaining expenditures, ceteris paribus.

The expenditure of central governments in developing countries rose dramatically from 5 percent of the gross national product (GNP) in 1945 to almost 20 percent in the early 1970s. It has continued to rise since then, though more slowly, and in 1985 was about 22 percent of GNP (World Bank, 1988b). Excluding expenditures on social security (which are higher in industrial countries) the share of the public sector in GNP in developing countries now exceeds that in industrial countries. In general, public expenditures have risen faster than revenues, so that annual deficits in developing countries have increased from less than 3 percent to more than 4 percent of GNP since the early 1970s (World Bank, 1988b).

The growing deficits do not bode well for spending on health or on education, which has potentially positive effects on health in addition to its other benefits. Between the early 1970s and 1985 in developing countries, the share of central government budgets directed toward health fell from 7 to 4 percent; the share for education fell from 14 to 10 percent (World Bank, 1988b). In some countries, particularly in Africa, the falling shares translated into real overall declines and even larger declines on a per capita basis. Though state and local governments may have taken up some of the slack, the dominance of the central government in most of the developing world means that expenditures of other governmental levels are not likely to have compensated.

Fiscal pressures in the 1980s also contributed to a reduction in the efficiency of government spending on health. Although in the long run the public sector may react to reduced resources for health by altering its mix of services, the short-run reaction has been to maintain the product mix and protect spending on personnel while cutting expenditures on nonpersonnel operating costs, including drugs, fuel for vehicles, and maintenance of physical plant and equipment (Akin et al., 1987). People are not easily laid off from public service jobs. Moreover, scarce foreign exchange is critical to the purchase of drugs, fuel, and equipment but not to the payment of wages. Because the nonpersonnel inputs are usually a small portion of total costs (less than 20 percent), they must be cut drastically to reduce total spending by a relatively modest amount. The price of a small financial saving is a large drop in the effectiveness of the system as a whole.

Finally, these problems have been exacerbated by the extension of health insurance without adequate cost containment measures in some developing countries. For example, in Latin America, the system of automatic reimbursement by the public sector insurer to charges of private providers has fueled the rapid growth of prices and spending on curative care, leaving fewer resources to be spent on more productive basic health services that are not covered by insurance.

All of these forces lead us to conclude that governments will simply not perform as effectively in the future as they have in the past; that this is due in part to growing tensions between the health needs of the rich and poor, old and young, in the face of increasingly constricted state resources; and that a radical restructuring of public-private roles may simultaneously improve equity, efficiency, and health outcomes.

Examples of Inefficient and Inequitable Public Health Spending

As discussed earlier, efficiency criteria would dictate government expenditures for such programs as immunization (generating externalities), improved water supply (a public good), monitoring minimum standards for pharmaceuticals and pesticides, and generating publicity about lifestyles that promote good health (such as antismoking and pronutrition campaigns) to help consumers make better-informed utility-maximizing decisions (Birdsall, 1989). The provision of basic medical services to low-income groups and to rural regions that cannot support a private competitive market in medical services is also warranted if people care about the health of others and wish to reduce the overall incidence of illness and mortality. Maternal and child health programs are particularly important examples of the latter since these affect the health of entire generations, in which there may be a large societal interest. Because they have public good characteristics, all the above programs are not likely to be provided by the private market; hence they are logical candidates for public funding.

We have argued that these basic services and informational programs would raise health standards and reduce mortality in the most cost-effective way (Birdsall, 1989; Akin et al., 1987). They would most help the poor, where mortality loss is currently highest. They would also help the young, where potential gain in life expectancy is the greatest. On both efficiency and equity grounds, these are the programs that merit public spending.

However, in many countries we observe relatively little public health money going to these cost-effective programs. Instead, a large proportion of public health budgets is spent on hospitals, usually located in urban areas, even in countries where the vast majority of the population lives in rural areas and suffers from high mortality rates caused by diseases that need not be treated in hospitals. In virtually all developing countries, the total health expenditure for curative care is between 70 and 85 percent, leaving only 15 to 30 percent for spending on preventive care and community services.

More specifically, in Bangladesh in 1986 hospitals consumed more than 80 percent of recurrent public health spending. In Brazil in 1982, 70 percent of public health funds was spent on reimbursement for physician and hospital care, including expensive high-technology procedures (kidney dialysis, coronary bypass, caesarean section). In Zimbabwe, which has tried to make its health sector more egalitarian, two-thirds of Ministry of Health expenditures are for hospital services, and 60 percent of these expenditures was absorbed by four hospitals in Harare. In Tanzania, which has made a special effort to improve rural clinics, 60 percent of the recurrent health budget was nevertheless spent on hospitals in 1983-1984 (Griffin, 1989; World Bank, 1988b).

Typically, these hospitals are located in urban centers of population and they serve the middle and upper classes; the superior public hospitals (e.g., armed forces or social security hospitals) serve the elite. Most of their patients are middle aged or older. Because hospital services are parceled out to their patients, they have a large private benefit component and could therefore be financed privately. But once government undertakes the task of financing hospitals, resources are crowded out and financing absorbs a large share of the public budget, because of the high cost of modern medical technology.

How Shifting to Private Spending Could Help

Suppose instead that many urban hospitals were turned over to private bodies, with fees to be covered by mandated health insurance (which might be administered by government but financed by premiums paid by the beneficiaries or their employers). Along similar lines, user charges could be instituted at the remaining public facilities, and the use of competitive, privately managed services might improve their effectiveness. Public funds would then be freed up to provide the externality-generating health programs listed above and also to subsidize health insurance for the poor. We contend that these changes would bring about a net improvement in health indicators, far beyond that which would be experienced under current funding mechanisms.

Examples of countries with such experiments are Zambia, where the university hospital at Lusaka is being turned into a parastatal that charges clients for services, with public funds thereby released to finance new maternal, child health, and family-planning services; Zimbabwe, where a fee has been introduced for patients who bypass lower levels of the health system and those who want a private hospital room; and The Gambia, where fees charged for drugs are turned over to village development councils for further health improvement (Akin et al., 1987). In Jamaica, costs declined when housekeeping and food services at public hospitals were contracted to private firms (Griffin, 1989). In Chile, increased reliance on private hospitals during the past decade was accompanied by a shift toward less expensive medical personnel (more nurses and midwives, fewer doctors), by structural changes to improve incentives, and by the targeting of government services toward primary health care and other services for the poor (Griffin, 1989).

Moreover, to the extent that reliance on government funds has limited hospital expansion, access to private funds (including insurance reimbursement) may increase the supply of hospital services and thereby improve overall access to health services. In the Philippines in the 1970s, after a policy change that allowed private expenditures, the greatest expansion of hospitals occurred in the poorest regions served (Griffin, 1989).