Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
, - Will slower population growthincreaseper capita levels of schooling and health? HUMAN CAPITAL AND ECONOMIC DEVELOPMENT Governments in virtually all developing countries are attempting to increase levels of schooling and to improve levels of health. These goals are also very important to households and other major social units. In addition to their intrinsic value as elements of human welfare, improved health and education contribute to improved economic performance. A better educated work force is more skilled, more adaptable, and more entrepreneurial. The value of education is not limited to any particular sector of the economy: better educated farmers appear more responsive to new technical possibilities, and better educated women seem more effective at allocating resources within the home, including those that enhance child survival (Schultz, 1979; Mensch et al., 1985). The importance of an educated work force for economic performance has been demonstrated both in studies that attempt to assign economic growth to various components and in studies that attempt to calculate directly the rate of return to completion of additional years of schooling. Denison (1962) attributed 42 percent of He increase in per capita GNP in He United States between 1929 and 1957 to higher average levels of education. A substantial fraction of European economic grown can also be assigned to this source. Educational upgrading contributed substantially to Japan's move from a developing to a developed economy. Denison and Chung (19763 estimate that in He decade 1961-1971 educational advance raised the grown rate of per capita GNP by 0.35 percent per year in Japan. In most of the studies attributing economic grow to various factors, an important residual 53
~4 POPUI~lON GROWTH AND ECONOMIC DEVELOPMENT remains that represents unexplained gains in output for a given level of the factors. It is likely Mat improved education accounts for a significant part of these improvements. When rates of return to investment in schooling are computed in developing countries, they invariably are very high. Psacharopoulos (1981) reviews many such computations and finds that the mean social rate of return in 22 developing countries is 27 percent for primary schooling, 16 percent for secondary schooling, and 13 percent for higher education. There are reasons to believe that these figures are biased upward by, among other things, failure to control differences in the ability and class background of persons receiving different levels of schooling. But even when these and over biases are taken into account, We returns to schooling are typically very high (Psacharopoulos, 1981~. The role of health improvements in economic growth has not received as much detailed attention as education. (For a useful review, see Barlow, 1979.) The issue is more complex than that of education because health improvements may also increase rates of population growth, introducing many additional considerations. In a simulation exercise based on the Coale- Hoover model, Barlow (1967) found that the antimalarial campaign in Sri Lanka raised the grown rate of per capita income in the short term, largely because of enhanced labor productivity, but lowered it in the long run, largely because of reduced levels of physical capital per worker (see the discussion under Question 4~. Ram and Schultz (1979) suggest that the antimalarial campaign in India during the l950s was the major reason that income grown rates were higher in that decade than during the widely publicized 1960s, when new crop varieties were introduced. For Sierra Leone, Strauss (1985) finds evidence that family nutritional intake is strongly related to family farm ouput. Deolalikar (1984) provides direct evidence on the economic payoff to improved nutritional status in south India, finding that each percentage point increase in weight-for-height of agricultural workers raised their daily wages by about 1 percent. Whatever the directly measured economic effect, it is certain that health gains would become a larger positive contributor to economic grown if measures of economic grown included an imputation for the value of gains in life expectancy (Usher, 1973~. Because of the contribution of advances in health and education to social and economic development, it is important to identify the effect of population growth on these variables. It should be recognized at the outset that the postwar era of rapid population grown has also been one of rapid gains in educational attainment and life expectancy in developing countries. In the case of life expectancy, improvements have directly contributed to more rapid population growth, a necessary relationship that makes it difficult to use time-series d~ to infer the importance of We reverse causal paths. What
SCHOOLING AND HEALTH 55 can be said is that the economic response to rapid population growth has not been so negative as to reverse the gains in life expectancy that initiated such growth. The Malthusian specter of population growth precipitating a rise in mortality that restores some kind of economic-demographic equilibrium gains little support from events occurring between 1950 and 1980. Indeed, the incidence of famine-related mortality seems to have declined throughout the twentieth century (World Bank, 1984; Simon, 1981), although famine conditions in China between 1959 and 1961 that have only recently come to light may alter this conclusion (Coale, 1984; Ashton et al., 1984~. While time-series data are instructive in setting some broad parameters within which these processes work, cross-sectional analyses permit a more precise view of the consequences of population growth by taking advantage of richer data that exhibit more independent variation among variables of interest. Such analyses have been carried out both at the family or household level and at the national level. These levels of analysis are not mutually exclusive- nations consist of families and households, and national (or regional) policies affect the decisions made at the household level~ut the distinction is widely observed in the research literature. FAMILY SIZE AND CHILDREN'S ~:ALTH AND EDUCATION Family-level relationships between family size and mean education and health of children have been widely reviewed (e.g., Wray, 1971; Terhune, 1974; Birdsall, 1977; Ernst and Angst, 1983; Rodgers, 1984; King, 1985~. Beyond a certain family size, additional children are usually associated with lower average educational attainment and reduced levels of child health, as measured by nutritional status, morbidity, and mortality. In addition to the reduced education levels, studies in Hong Kong and India show that school grades among those enrolled tend to be lower for children coming from very large families (Ernst and Angst, 1983:51~. However, negative effects of family size on child-quality variables are not always found. For example, Mueller (19843 presents evidence from Botswana and Sierra Leone that children from larger families achieve higher average levels of schooling, controlling over pertinent variables. In general, there are suggestions that the typical negative relationship between family siSe and health and education of children is larger in poorer families (Birdsall, 19803. A good deal of recent evidence on the relationship between reproductive patterns and child mortality has become available through the World Fertility Survey. As reviewed by Trussell and Pebley (1984), the evidence suggests that elimination of fourth- and higher-order birds in developing countries could reduce infant and child mortality by about 8 percent. Universal adoption of an "ideal', spacing pattern in which all births subsequent to He first
56 POP UNION GROWTH AND ECONOMIC DEVEL~PMENI are separated by at least 2 years could reduce infant mortality by about 10 percent and child mortality by about 21 percent. (Spacing patterns bear no necessary relationship to population growth, of course, and it is these patterns to which child mortality appears most sensitive.) While the Trussell- Pebley study controlled for certain important socioeconomic variables, it could not control for unmeasured "tastes" for child quality, which would be expected to be stronger among families win few children and long birth intervals. For this reason, the results may represent an upper bound on the child mortality gains to be expected from reduced fertility and longer spacing. The commonly observed negative relationship between the number of children and the "output" variables of levels of health and education probably reflects a negative relationship between the number of children and various "input" variables. Examples of studies finding negative associations between family size and per capita health and food expenditure can be found in Wray (1971) and Rodgers (1984~. It is possible Hat economies of scale in larger households offset some or all of the disadvantages of lower per capita expenditures, particularly in the area of food provision, although Here are few studies confirming such an offset. Reviewing evidence for Kenya, Colombia, and 11 Latin American cities, Tan and Haines (1983) conclude that while total household expenditures on education tend to increase with the number of children, they usually do not increase fast enough to avert a decline in schooling expenditures per child (see Birdsall, 1980, for a detailed discussion of this issue for Colombia). Complicating this relation is the fact that financial support for children's schooling is often derived from older siblings. A majority of fathers in many Asian countries expect such financial help from older sons (Bullet=, 1979), and Caldwell et al. (1982) refer to a common situation in tropical Africa wherein "sibling chains of assistance" are established so that each educated child makes it more likely that the next will be educated. In some places, parents may only have to pay for the advanced schooling of He first child, while later ones are financed by older siblings. It is important to recognize that no policy implications necessarily follow from a demonstration of a negative cross-sectional relationship between family size and child "quality." Households vely often make childbearing decisions win an awareness that having an additional child will entail a sacrifice of some other household objective: leisure, consumption of goods and services, schooling for children already bom, health of parents, investment in other household enterprises, and so on. The financial effects of additional children are widely cited concerns in childbearing decisions by families in developing couches (Bulatao, 1979) as well as in developed countries. Obviously, many couples skill consider their welfare to be increased by an additional
SCHOOLING AND HEALTH 57 birth despite the costs entailed. Most governments rarely intercede in such decisions, since households themselves bear the bmnt of the consequences. And, as illustrated by resolutions passed in United Nations' world population conferences (United Nations, 1984), governments universally proclaim the principle of family sovereigns in reproductive decisions. However, some nations have, in fact, intervened in childbearing processes in ways that violate principles of family sovereignty. It is also important to note Hat if a nation achieves lower fertility rates, the impact on the education and heals of children will be determined in part by the class distribution of the fertility reduction. If, as widely observed in Latin America, the small upper income groups have the largest proportionate reductions, then the mean levels of child health and education would be expected to decline, other things being equal, even if there are rising expenditures per child by those groups in which fertility declines. To the extent that a national policy achieves lower fertility levels through family planning programs that reduce the incidence of unwanted children, the question of impact becomes much more highly focused. The class distribution of unwanted births is generally highly skewed toward He lower income groups (Brackett, 1978; Westoff, 1978; Birdsall, 1980~. For these groups, it can be anticipated that family planning programs will enhance the mean level of child health and education through compositional effects. One of He few investigations of the effects of an unwanted birth on child quality within families in developing countries is that of Rosenzweig and Wolpin (1980~. They find that in India the bird of twins in a household significantly reduces school enrollment levels of children in the household. It also reduces household expenditure on consumer durables. Obviously, not all the twins are unwanted birds, but by the nature of the event a higher- than-average proportion are unwanted. Effects on educational enrollments were not confined to the twins themselves but extended to other children in the household. Related studies in India and Thailand using proxies for "wontedness" suggest that child mortality is higher in families in which more unwanted birds were occurring (Rodgers, 1984~. POPULATION GROWTH AND PUBLIC HEALTH AND EDUCATION ~ addition to their effects at the family level, levels of fertility can also affect the allocation of government resources to education and health. Educational attainment is produced by some combination of public and private expenditures. At the secondary school level, private contributions often assume greater importance because a child's forgone earnings increase,
58 POPUl'7ON GROWTH AND ECONOMIC DEVELOPMENT and governments are less committed to supplying that level of education to large fractions of the population. Effects of population growth on government expenditures can therefore differ from effects on enrollment. Not only do private expenditures play an important and highly variable role, but governments can alter the amount of resources spent on each enrolled child. Most work on the subject of fertility effects on educational systems is focused on the proposition that more children means that more places must be provided in school systems to maintain the population enrollment ratios. The "costs', of additional children can be readily calculated under reasonable assumptions about per pupil costs Cones, 1971, 1975; Cochrane, 1983~. Jones (1975) shows, for example, that 30-50 percent of the additional governmental education expenditures "required" over the next decade in a typical developing country is attributable to the projected grown of the school-aged population. These arguments are often among the most convincing to government planning officials about the advantages of reduced fertility. But the fact that enrollment ratios or per pupil expenditures may decline as populations grow does not mean that they necessarily will. Governments can respond to larger school-aged populations in many ways: by raising taxes, by shifting expenditure from over areas, by restructuring educational systems, and so on. Like families, governments are actors in the drama of economic-demographic relations, and their behavior cannot be readily predicted a priori. The most comprehensive examination of the effects of population growth on educational systems in developing countries is that of Schulm (1985~. In a cross-section of countries, Schultz finds that the relative size of school-aged cohorts is negatively associated with government expenditures per school- aged child in both primary and secondary schools. A 10 percent increase in the ratio of the school-aged population to the total population is associated win 11 and 17 percent reductions in expenditures per school-aged child at the primary and secondary levels, respectively. In other words, an increase in the school-aged population induces no increase in total school expenditure and may reduce it. Consistent with this result, there is no association between the relative size of the school-aged population and the share of government expenditures that are directed toward schools. However, Schultz finds no negative effect of the size of the school-aged population on enrollment rates. In fact, the relationship is positive at the primary level. These results clearly imply Nat more rapid growth produces lower expenditures per enrolled child, and Schultz shows Nat this effect takes the form of more enrolled children per teacher and lower teacher salaries. Schultz's results on expenditures are roughly consistent with an earlier cross-sectional study by Simon and Pilarski (1979) Nat finds a slight negative effect of fertility on expenditure per child. In contrast to Schultz,
SCHOOLING AND HEALTH 59' however, Simon and Pilarski find no effect on primary enrollments and a sizable negative effect of fertility on secondary enrollments. Although there is much uncertainty in these relationships, it appears that the major effects of population growth on education are on expenditures per pupil-especially teachers per pupil and teacher salaries-and not on enrollment rates. Thus, it is useful to ascertain the importance of expenditures per pupil and teachers per pupil on the quality of education. Intuition strongly suggests that both are positively associated with quality, but the evidence has not always confirmed this intuition. Simmons and Alexander (1978) reviewed multivariate microlevel studies of student performance on standardized tests in developing countries. Two of these studies found a negative or insignificant effect on student performance of per pupil expenditure for school facilities or teachers, and none showed a significant positive effect. Four studies showed a negative or insignificant effect of teacher/pupil ratios, but three showed a positive and significant effect (Simmons and Alexander, 1978:350~. These results are similar to those typically found in the United States (Hanushek, 1981; Murnane, 1981~. Additional evidence on this point comes from a study of science test scores among 13- to 14-year-olds in 29 mostly developing countries. Heyneman and Lo~ley (1983) find that 15-25 percent of the variance in test performance within developing countries can be accounted for by variables representing school and teacher quality. This fraction is larger than is typical for developed countries. The list of quality-related variables is very large and varies from country to county, and it is not possible to generalize to issues of expenditure per child or number of teachers per child. The fact that richer countries tend to have higher test scores in this study also points to die possible importance of school expenditures per child, although influences in the home are also likely to be related to this outcome. Behrman and Birdsall (1983) examine the effect of school quality (as measured by the mean years of schooling among teachers in one's area during childhood) on adult earnings in Brazil. They find a higher rate of return to investments in school quality Han to investments in school quantity. However, they were not able to control for features of the early home environment, which may be related to both quality and quantity of education, and so the net bias is indeterminate. All these studies show that the quality of schooling can be improved in developing countries (e.g., through providing more textbooks per pupil) and that the savings induced by slower growth of He school-aged population can be used to improve quality. The fact that the aggregate measures of expenditures per pupil and teachers per pupil show only a weak association with test scores in developing countries suggests that one should not be too optimistic that lower population growth will result in improved school
60 POP ULAJ7ON GROW ED ECONOMIC DEVE=PME' quality. But three of seven studies did show a significant positive effect of teacher/student ratios on student test performance, and Schultz's (lg8S) results suggest that improved teacher/student ratios are the principal route through which lower grown affects education. Therefore, it is reasonable to expect that an exogenous reduction In the growth rate of the school- aged population will in general result in some improvement in quality of education. Another element in the discussion of the aggregate relation between population growth and the average educational attainment of workers is introduced by Leibenstein (1971:188), who argues: "To the extent that entrants into the work force are of higher quality (i.e., higher education and acquired skills, etc.) than those that leave through retirement and death, the average quality of the labor force improves more rapidly if the rate of population growth is higher (other things equal) than lower." It should be noted that this effect operates only dunng a transitional period from one demographic equilibrium to another. A once-and-for-all reduction in fertility will lead to a reduced rate of labor force upgrading until a new equilibrium is established. But if a population has constant annual upgrading of amount K in its average endowment of human capital from one cohort to the next (e.g., those aged 15 in 1985 are better endowed by the factor K than those aged 15 in 1984), then regardless of its overall population growth rate, its per capita stock of human capital will be growing at rate K. In equilibrium, there is no relationship between the growth rate of the population and the growth rate of per capita human capital. Even during the transitional period to a lower fertility equilibrium, the welfare implications of Leibenstein's argument are unclear, since all cohorts of individuals are assumed to receive the same schooling in fast-growing populations as in slow-growing populations. It is only the aggregate measures that show a (temporary) deterioration. There are almost no studies of the effect of demographic variables on government health expenditures. A cursory glance at per capita national health budgets shows extremely wide variations-for example, $16.96 in Mexico and $111.88 in Venezuela (Golladay and Liese, 1980:13-which probably have more to do with variation in the nature of items that are included in the budgets than with real variations in expenditure. Such figures are not a promising vehicle for empirical investigation without furler refinement. However, Montgomery (1985) reports an insignificant, positive effect of population grown rates on government health expenditures as a share of all government expenditures in a cross-section of 49 developing countries. Whatever the effect of population grown on government heals expenditures, those expenditures are not closely related to mortality or health. Nearly all careful studies conclude dial government health expenditures per se have
SCHOOLING AND HEALTH 61 very little to do with national health conditions (Corsa and Oakley, 1971; Mosley, 1983~. Those health expenditures are usually dominated by urban- based curative services that are often modeled after health systems in developed countnes. Not only do these services often fail to reach most of the population, but they also do not appear to improve population-based measures of mortality in the urban populations that they do reach (Mensch et al., 1985~. This is not to say that government programs cannot improve health and have not done so, but only that present patterns of expenditure are not such that population growth poses a major threat to success, except, perhaps, in those countries such as Sri Lanka and Cuba, where government health programs appear better constituted and more successful. One way that population growth may actually assist governments in achieving health objectives is by increasing rural density. It is often alleged that, especially in Africa, the dispersed, low-density rural population is difficult to reach with government health services Should Health Organization, 1975:17~. These effects have not been adequately quantified. CONCLUSIONS We have depicted a multilayered relation between population growth and children's levels of education and health. Several tentative conclusions seem justified. First, larger families generally have lower levels of schooling and health per child, which probably primarily reflects lower expenditures on health, education, and nutrition per child in larger families. However, these relationships are not universal, and they do not necessarily reflect a causal impact of fertility on health and education. Rather, they may to some extent reflect deliberate parental trade-offs between family size and per child expenditure. Second, family planning programs that decrease the incidence of unwanted births are likely to raise average levels of education and health among children, both because they increase investment within the family in child health and education and because they reduce the fraction of births occurring in lower income families. Third, countries with more rapid population growth do not appear to have lower levels of school enrollment, coterie panbus; they do seem to have lower school expenditures per child and fewer teachers per student. Several (but not all) studies have found that having fewer teachers per student reduces student test performance. Therefore, some improvements in test performance may occur as a result of slower population growth. Finally, it is unlikely that rapid population growth is a major impediment to the success of government health programs as Hey are presently structured.