Susan Hill Cochranco Fet0ility and Eduction What Do We Really Know? LO58/^of demand _ N Time costs for, bi of children supplof of chilnchildren Kind of work'"' Economic Wife's s n benefits desired wife powerrt of children fLculy size Wife's and reg tion reerencef suppl d of ccation Attitude X owled Wife's for, and/ prefe'rences supply of, children Access Attitude KCnowledge TIIEORETICAL DE I ERMINANTS OF FERTILITY 61 Blake (1956), Rosen and Simmons (1971), Holsinger and Kasarda (1976), Haas (1974), and Namboodiri (1972). Variables in the model Thlic schemaltic representation of fertility determnination is presentcd in Figurc 2.2. As mentioned above, this model rcally has thrce parts: dcmanid, supply, and supply regulation. Demand factors are prcsented in the left-hand portion of the diagram. Supply factors operating through current family size and fccundity are presented in the uppcr- central and right-hand portion of the diagram. Fertility rcgulation is represented in the lower right-lhand portion of the diagram. Thlie proportion of space devoted to various clusters of factors docs not represent their relative importance. THE SUPPLY OF CHILDREN. Supply hias a stock and flow componcnt. 'lle potential numiiber of births in any time period depends on wlhat is gencrally called fecundity or natural fcrtility. This dcpcnds on bio- logical and behavioral factors." Important biological factors include the age and healtlh of the mothcr, whether lactation is occurring, and the interval since the last birth. Tlihe behavioral variables primarily regulate sexual activity. Among the important variablcs arc taboos about sexual intercoursc. 3lhese variables depend on the marital status of the individual."1 Sexual activity also depends on the residcncc of the spouses and the type of living arrangemncnts."'1 llhe stock of living childrcn in a family is determined by the births in each previ- ous period, the number of periods (gencrally dctermined by the duration of marriage), and the mortality of chlildren. Ilhc stock of childrcn and demand factors determine wvhetlhcr another child is desired. Fccundity and/or fertility regulation dctermine whether an additional birth occurs. 8 These are behavioral variables that are not designed to regulate fertility although they may have that cffect. 9 Relevant marital circumstances depend on not only whether a person is single, married, divorced or widowed, but also whethier it is a legal or consensual union, or a monogamous or polygamous union 10. In many developing countries men frequently leave their wives for long periods to obtain work. In addition, the practice of living in parental households may also affect sexual activity. 62 FERTILITY AND EDUCATION * WHIAT DO WE REALLY KNOW? THE DEMAND FOR AN ADDITIONAL CHILD. The demand for an addi- tional child can be considered as either a decision to adjust the stock of children or one which had been made after the marginal cost and benefits of an additional child had been weighed. In the stock- adjustment model, each spouse is assumed to have a desired family size determined by their individual preferences and perceptions of the cost and benefits of children. (These are actually cost-benefit functions and need not be assumed to be constants.) This desired stock of children can then bc compared with the actual stock. This approach has most of the disadvantages of the static model of fer- tility. It is preferable simply to examinc the decision to add a child in terms of marginal costs and benefits. Unfortunately, although this latter incremental approach requires fewer constraining assumptions, it is not very useful for organizing a review of the empirical literature because what little evidcnce does exlst on the demand for children has been collected using conccpts more closely related to the static concept of demand rather than on the incremental decision to add a child. Although some data has been collected on whether couples desire an additional child, no corresponding work has been done on the marginal, rather than the total, costs and benefits of children. Some data has been collected on the perceived levels of costs and benefits. Therefore the model used here is the stock-adjustment model rather than a theoretically superior incremental model. The stock- adjustment model is, however, superior to a purely static demand model because it allows for the possibility of altering the desired stock as more information becomes available. The demand for a stock of children, or a desired size of family, is in some ways more problematic than supply, and it cannot be dis- cussed in the same way as the family's supply of children. Since the husband and wife have desired family sizes which necd not coin- cide,"' the family's demand for childrcn cannot be defined. Thus the demand for children is determined by examining each spouse's desired family size, which depcnds on the individual's preferences, perceptions of the costs and benefits of childrcn, and perceptions of the ability to afford children. Neither the preferences nor the per- 11. In polygamous marriages the male and female fertility will differ, and the biological supply of chlildren will diffcr for husbands and wives. THEORETICAL DETERNIINANTS OF FERTILITY 63 ceptions of the spouses need agree, and so fairly large discrepancies may exist between the desired family sizes of individual spouses. 'T'he ability to afford children depends on family income and wealth as well as on the "price" of children. Income depends on the husband's and wife's wages and labor supply as well as nonlabor in- come and possibly the income produced by children or other famlly members. Thc "price" of children is far more complex. Money ex- penditures on children are not determined by the market nor com- pletely prcscribcd by society. While some elements of cost are determined by exogenous factors, such as the pricc of food, others are determined by parents' aspirations for their children.'2 In addi- tion to monetary cost, children entail the expenditure of time. The time cost of children depends on the amount of time devoted to each child"3 and on the value of that time. Generally in Western societies such time is spent by the mother. In developing societies siblings and extended family members or domestic servants may perform these functions as well. The value of the time devoted to children depends on the alternative uses of time available, on the value of time in those uses, and on the compatibility of those alternative uses with child care. These factors, in turn, depend on the wages of women and children and the kinds of work availablc. All of these variables thus need to be examined as well as important nonmarket uses of time which may limit the ability of individuals other than the mother to care for children, such as the school attendance of older siblings. If cach spouse has a family size target, it can be compared with cur- rcnt family size to determine the desirability of another birth. It is at this point that child mortality affects demand. Rcductions in child 12. Thls might be identified with the demand for chlild quality, in the Chicago tcrminology, but following the lead of Turchi ("Micro-economic Theories of Fertility") and the pragmatic approach of empiricists, these elements of cost will not be scparated into the required (price) and discretionary (quality). Parents' aspirations for their children perhaps at best reflect what the economists mean by quality. But such aspirations affect fertility through the perceived cost of children. 13. The amount of time devoted to a child is not fixed in any society but varies enormously depending on wage rates, child care substitutes, aspirations for children and perceptions of the most effective methods for raising children, in particular, the perceived ability to substitute between the mother's time and that of others. 64 FERTILITY AND EDUCATrION : WIIAT DO WE REALLY KNOW? mortality increase the supply of living childrcn. It is also probable that such reductions reduce the demand for additional children.'4 FERTILITY REGULATION. If another birtlh is desired and timing fac- tors are ignored, then supply factors predominate. If an additional birth is not desired by one spouse or is not desired at the current time, then there is a potential demand for fcrtility regulation.15r WlVether a potential dcmand for fertility regulation leads to actual regulation depends on several interveniing variables: (a) which spouse does not wish an additional chlild, (b) the power of that spouse in making marital decisions, (c) spouses' ability to communicate fcrtility dcsires and contraceptive information to each other, (d) the kinds of techniques of fertility regulation available or known to the couple,", and (e) access to, and cost of, fertility regulation, including psycho- logical cost of the various techniqucs. Direction of the Relations between Intervening Variables and Fertility Ilhc direction of the relation between the variables in Figure 2.2 and the addition of a chlild are the samc as those between such variables and completed fertility.17 Althouglh scholars do not agree completely on the direction of thcsc relations, there is some general agreement. Column 1 of Trable 2.1 summanzes the directions of the hypotlhesized effects of the variables of the model on fertility. Column 14 It is possible for suci reductions to increase the demand for additional children T'his latter result is only likely if the cost of children before the average age of child deaths is relatively high. See Coclralne and Cochrane (1974). 15 Strictly speaking, fertility rcgulation should include stimulation of fertility in the infertile as well as its suppression in the fertile 16 If the wife wants no more children, but all contraceptie methods are dominated by the husband, fertility is unlkely to be suppressed. The converse is of course also true, so that a wide miLx of techniques will increasc the number of families practicing contraception. 17. Fertility during a certain period depenids on age and time since last birth, which in the long run will not affect completed fertility THEORETICAL DETERMINANTS OF FERTILITY 65 2 gives citations supporting the hypothesized effect.'8 Williams (1976) has recently published a review of the empirical literature that examined many of the variables. The papers by United Nations (1973), McGrcevey and Birdsall (1974), and Mason and others (1971) also summarize existing research rather than being original empirical studies. Ambiguous relations Thle two most ambiguous of these relations are the effect of the husband's income and the husband's power in marital decision- making. Although there are theoretical reasons for believing that increases in the husband's income xNill increase the demand for children and thus completed family size, other things being equal, the cmpirical evidence does not support this. This probably results from the fact that income is correlated both with wages and with other factors such as preferences, aspirations for childrcn, and access to birth control, and few studies have controlled for these factors (Bean and others, 1977). The effect of the husband's power is ambiguous on theoretical as well as on empirical grounds. If a man wants more children than his wife, then the greater the husband's power relative to his wife, the greatcr fertility will be. However, desired family size of each spouse depends on preferences and on perceived costs and benefits. Although wvives may bear more of the cost of childbcaring, if they have stronger preferences for children, their desired family size may exceed that of their husband's. In suchi circumstances, increases in the wife's power wvould not lower fertility. Unfortunately, not enough surveys have collected data on the preferences and power of both husbands and wives to resolve this issue. Weller (1968) found the wife's power reduced fertility among working women in Puerto Rico, and Mitchell (1972) found this to be true among women in Hong Kong. However, both of these are special groups in wNhich a wife's desired family size is likely to be below her husband's. Rosen and Simmons (1971) found that the wife's power decreased desired family size but increased or was increased by actual family size. 18 Only published studies have been cited here, and the listing is far from complete. 66 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? Table 2.1. Evidence on the Effect of Intervening Variables on Completed Fertility Direction Intervening variable of relation Empincal support Supply factors Probability of marriage + Schultz (1972), Mazur (1973), Maurer (1973) Wife's age at marrtage - McGreevey-Birdsall, Encarnaci6n (1968), Kim and others (1974), Davidson (1973), Yaukey (1972), Palmore and Anffin (1969) Health + Butz (1976), Baird (1965) Separate location of spouse - Williams (1976) Joint family living ? (-) United Nations (1973), Williams (1976) Legal, monogamous marnage ? (+) Mason and others (1971) (+), Nerlove and Schultz (1969) (+), Miro and Mertens (1968) (mixed), United Nations (1973) (mixed) Taboos on sexual activity - United Nations (1973) (several studies cited) Infant and child mortality + McGreevey-Birdsall (1974), Snyder (1974), Williams (1976) Demand factors Preferences for children ? (+) Husband's wage 7 Simon (1974), Williams (1976) Relations based on theory Several of the other relations in Table 2.1 have little or no empirical basis. Some of these are ambiguous because of measurement prob- lems, others have mixed support. In a number of cases, however, the theoretical reasoning is so compelling that, despite lack of data, the signs will have the credibility of those which were established em- pirically. Among these variables are health, location of spouse, preferences toward children, and knowledge of and attitude toward contraception. The reason for lack of appropriate data differs in each case. Health and preferences are of course difficult to measure. In addition, health, preferences, and knowledge and attitude toward THEORETICAL DETERMINANTS OF FERTILITY 67 Table 2.1 (continued) Direction Intervening variable of relation Empirical support Money cost of children - Mueller (1972), Bulatao (1975), Arnold and others (1975) Wife's wage - Mason and others (1971) DaVanzo (1972), Snyder (1974), Rosenzweig and Evenson (forthcoming) Incompatibility of wife's work - Coldstein (1972), United Nations (1973), Bindary and others (1973), Williams (1976) Cost of child care substitutes - Cain and Weinenger (1973), McCabe and Rosenzweig (1976) Economic benefits of children + Mueller (1972), Harmon (1970), McCreevey-Birdsall (1974) Fertility regulation Husband's marital power ? (±) Weller (1968) (+), Mitchell (1972) (+) Husband-wife - Mitchell (1972), Michel (1967), comnmlunication Hlll and others (1959), Ramakumar and Copal (1972) Knowledge of birth control - Attitude toward birth control - Access to birth control - Masoni and others (1971), Schultz (1972) Sources: For completc refercnces, see the sources for this chapter. contraception may all be affected by fertility as well as affecting it. Tlhus, after-the-fact cross-section rescarch designs that do not meas- ure these variables before childbearing do not provide good evidence of their effect on fertility. Only longitudinal analyses would disen- tanglc these effects. Absence of spouse seems to be a neglected vari- able that could be collectcd in the process of obtaining fertility histories.'9 The dircction of the effect of all thcse variables on com- pleted fertility seems to be intuitively obvious even if empirical sup- 19. Williams cites a study by Rele which shows such short-term male migra- tion reduces fertility by one-half a child. 68 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? port is lacking. The strength of those relations, however, is much less certain. Relations with weak empirical support Several other relations have weak support, such as joint family living, type of marriage, and the cost of child care. In the first instance, there are theoretical reasons for believing there might be a positive or negative effect.20 Thus its sign is not specified. The evidence on type of marriage is mixed, but there seems to be slightly more evi- dence suggesting that legal, monogamous marriages have higher fer- tility than concensual or polygamous marriages, so a tentative positive sign is specified. There is little empirical evidence, but strong theo- retical evidence, that increases in the cost of substitutes for the wife's time in child care should reduce fertility. Therefore, a negative rela- tion is assigned here. Important relations needing more evidence For sevcral variables the evidence is fairly good, but given the im- portance of the interaction of these variables with education (to be discussed later), more direct evidence is needed. These are the vari- ables related to the time cost of children, particularly the wife's wage, and the compatibility of market work and chilld care. The four studies cited here with respect to the wife's wage are less than perfect. 'T'he study by Mason and others relies on data from the United States. DaVanzo's study (1972) uses regional aggregate data, and Snyder's study (1974) uses wages for working wives only and sets the wages of nonworking wives equal to zero. Rosenzweig and Evenson (1977) use district wage averages in rural India. Thlle scarcity of data results in part from the fact that many researchers have concentrated on labor force participation and fertility rather than on the wife's wage and fertility. Economists have rejccted this approach because the wage could simultaneously affect labor partici- pation and fertilitv. In addition, family size may affect labor par- 20 Joint family living makes child care substitutes more available and may increase fertility However, the lack of privacy is believed by some, and has been partially confirmed, to reduce sexual activity In addition there is some evidence that such families feel morc constraints on resources Thus fertility may be limited. THEORETICAL DETERMINANTS OF FERTILITY 69 ticipation as well as the converse (Shields, 1977). Another reason for rejecting the labor participation approach is that contradictory evidence exists on whether such an inverse relation with fertility pre- vails in developing countries where mzarket work may not be incoin- patible with child rearing. The evidence of incompatibility is based mostly on inferences, since it relies on evidence from broad occupa- tional groupings. Better data on actual or potential wages of womien and compatibility of market work and fertility would improve the understanding of both fertility and female labor participation. The last variable, which slhould perhaps receive more attention, is access to contraception. The studies cited here use aggregate data on the availability of publicly sponsored family planning, and the evidence is convincing. NV/here no such programs exist, however, ac- cess to contraception may be highly limited by income, which in turn is supposedly affected by education. Thus, somc of the differences in fertility across educational groups, particularly the husband's educa- tion, may be reduced by providing public family planning. This is an issue that needs to be addressed in discussion of policy. Relations with strong empirical support Five variables that appear to have effects on completed fertility are strongly confirmed by empirical evidence. These are the proba- bility of being married, wife's age at marriage, child and infant mortality, the perceived economic benefits of children, and husband- wife communication. Tfhe cost of children and the observance of taboos on sexual activity are less well-documented but still convincing. The perceived costs and benefits of children are being studied in- creasingly, and the evidence continues to confirm the expected rela- tions, It seems that it is primarily the educational cost of children that dominate parents' thinking (Mueller, 1972). The time cost of children has not been wcll integratcd in these studies, but one of the modules in the World Fertility Survey attempts to correct for this. The economic benefits of children are derived from child labor and old age security. Communication between husband and wife seems to be important for contraceptive use. Some of this evidence was collected wlhen methods related to coitus such as the diaphragm and condom predominated. In those cases cooperation was essential. However, some of the more recent studies indicate that this is also an important variable when more modern methods are used. 70 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? Probability of marriage and wife's age of marriage seem to have strong effects, but these are not strictly linear nor are they equally strong in all areas. The nonlinear effect of age of marriage results from the fact that in societies with very early marriages, relatively few women are fertile when they first marry. Thus, increasing the age of marriage from 14 to 16 may have minor impact whereas in- creasing it from 20 to 22 may have quite a substantial effect.21 Thne effect of proportion married on fertility depends on the cxtent to which extramarital sexual activity and childbearing are accepted or supported by society. Thus, in Moslem countries the proportion married would have a much stronger effect than in the United States. Child mortality appears to have a biological and a behavioral effect on the number of children ever born. Increasing the proportion of children who survive would tend to diminish the demand for addi- tional births and tberefore wvould tend to initiate behavioral changes. In addition, there is evidence that lactation suppresses ovulation. Therefore, a living infant Whlo is nursing increases the intervals be- tween births. Having discussed the support for the relation between these inter- vening variables and completed fertility, the effect of education on these variables and the consequent indirect effects of education on completed fertility through these variables is examined. There is no reason to expect the effect of education to be all in one direction. As explained in Chapter 1, education has multiple effects that affect access to information, market opportunities, nonmarket efficiency, attitudes, behavior patterns, and status. In addition, the relative im- portance of these effects differ for men and women and also differ depending on the social and economic setting. Therefore, the effect of education on fertility through the intervening variables will be complex and varied. The Hypotlhesized Effect of Education on Intervening Variables In Table 2.2 the direction of the relations between husband's and wife's education and the intervening variables are shown in columns 21. If all women under 16 were not yet fertile, increasing the age of marriage from 14 to 16 would not affect fertility. THEORETICAL DETERMINANTS OF FERTILITY 71 Table 2.2. Effect of Education on Fertility through Intervening Variables Effect of Effect of education education on fertility on the through intervening Effect of intervening variables zntervening variables variables on Intervening variable Male Female fertility Male Female Supply factors Probability of being married + - + + Wife's age at marriage 7 + - ? Health + + + + + Separate location of spouse - ? - + ? Joint family living - - ? (-) ? (+) ? Legal, monogamous marriage ? + ? (+) ? ? (+) Taboos on sexual activity - - - + + Infant and child mortality - - + Demand factors Preferences for children' - - 7 (+) ) Husband's wage + 0 ? ? 0 Money cost of children + + - - - Wife's wage 0 + - - Incompatibility of wife's work 0 + - 0 - Cost of chlld care substitutes' 0 + - 0 - Economic benefits of children - - + - - Fertility regulation 1-Husband's marital power + - ? (+) ? ? Husband-wife communication + + - - - Knowledge of birth control + + - - - Attitude toward birth control' + + - - - Access to birth control + + - - - a. Depends in part on whether the education is religious or secular. b Depends on community level of female education. 1 and 2. The effect of thc intervening variable on fertility (from Table 2.1) is given in column 3. T'he hypothesized multiple effects of education on completed fertility acting through these intervening variables are shown in columns 4 and 5. These latter effects are ob- 72 FERTILITY AND EDUCATION WHIAT DO WE REALLY KNOW? tained from combining the effects of intervening variables on fertility shown in Table 2.1 and the effects of education on these variables in colunmns 1 and 2 of Table 2.2 (tlhat is, multiplying the sign in column I or 2 by the sign in column 3). Several genera] observations can be derived from Table 2.2. First, columns I and 2 slhow that for many variablcs, the effect of educa- tion differs for men and woomen. Thne directions of the effects are the same in only eleven of twventy cases. Columns 4 and 5 show that the education of females is more likelv to reduce fertility than is the education of males. In thirteen cases the indirect effect of female education is negative, indicatinig redtuced fertility, but in only eighit cases is the impact of male education negative.22 This provides hypo- thetical support for the observation that female education reduces fer- tility morc than male education. Tllhe evidence supporting the signs of columns 1 and 2 is discussed in Chapters 3, 4, and 5, but first the reasons supporting the hxpothesizcd relations are discussed. Thcoretical support of hypothesized sLp plV relatiolns 'T'he probability of marriage, wife's age at nmarriage, and type of marriage are all important variables in determining completed fer- tility. In each case, male and fcmale education appear to affect the variable differently. The age of marriage is thel most widely discussed of the three nmarriage variables, perlhaps because it has suclh a strong negative association withl fertility. T'lhe wife's age of marriage is af- fected primarily by the wife's education.23 Education may affect females' age of marriage in several wavs. Education in general raises wage rates and increases access to better lobs, making market work more attractive. Therefore, women may desire to work for some time 22. The ratio of negative to positive impacts for male education is 8 to 5, but for females it is 13 to 4. If husband's income has positive effects on fer- tility as some believe, the male ratio would only bc 8 negative to 6 positive effects. 23. Although male education may affect male's age of marriage, the latter is much less important to fertility than female age of marriage If there were fixed differences between the age of brides and grooms, male education would increase male's age of marriage and thus would pull up wife's age of marnage, but this seems very uncertain. Thus the effect of male's education on wife's age of marriage is hypothesized to be unknown THEORETICAL DETERMIINANTS OF FERTILITY 73 before marriage, thus postponing it. This effect of education would apply throughout all levels of education, but because certificates or degrees are used as screening devices in hiring, this cffect may not be equally strong for each increment in years of schooling. Education may also narrow the range of potential marriage part- ners and may thus increase the waiting time involved in finding the right suitor. This effect may operate at all levels of cducation, but thc impact may be concentrated at levels of certification or degree lcvels. Education beyond the primary level may conflict directly wvith early marriagc since married women are generally not enrolled in schools in developing countries. Thus, regardless of the content of secondary or post-secondary education, it will have a very direct effect on the age of marriage in countries which traditionally have low marriage ages. In countries with late marriage ages, this effcct of education is unlikely to be important. It would be quite difficult to separate out these differcnt effects of education on the wife's agc of marriage, empirically, but such a separation wvould be quite valuable for designing policy. PEducation will affect not only the age of marriage, but the proba- bility of ever marrying. More educated males wvill be more likely to marry because their incomes will be higher and the choice of part- ners increased. For femalcs, education increases the altcrnatives to marnage and reduces the pool of acceptable marriage partners if women must marry men at least as well educated as themselves. 'T'hese increased altcrnatives may also increase the probability that a woioman will divorce her spouse. These altcrnatives may also decrease the ac- ceptance of ccrtain forms of marriage such as consensual or polyga- mous marriages into traditional joint households. Thus, althouglh education of women reduces the probability of being married and thus reduces fertility, it may also incrcase fertility of those women wvho do marry becausc the kinds of marriages they will accept-legal, monogamous, nuclear marriages-are more conducive to higher fer- tility; but thcse effects are less certain.24 If these effects are very 24. For males, the income associated with higher education may permit them the luxury of consensual and/or polygamous marriages If this were the case, it would tend to raise their o%vn fertility but perhaps to reduce the fertility of their first (legal) wives. The net effect on fertility is fairly uncertain, depending greatly on the kind of valucs incorporated in the education process. Thus this effect is listed as unknown in Table 2 2 74 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? strong, the effect of education on the fertility of all women may be more negative than its effect on the fertility of married or ever-married women. Since most of the individual studies reviewed in Chapter I were limited to married women, the effect of education on overall fertility would tend to be underestimated if education of women re- duces the proportion ever marrying. The other major effects of education on the biological supply of children act through the health of the parents and children and the behavioral variables affecting supply, such as living arrangements and observance of taboos on sexual activity. The biological effects are somewhat clearer. Education would seem to increase the health of parents and children through better knowledge of hygiene and nutrition and better access to modern medicine and adequate food supplies. Although the health of males might be expected to have a smaller effect on fertility than the health of females, ultimately poor health means earlier death of the husband and shorter mamage duration. The behavioral variables that may affect the supply of births through sexual activity are location of spouse, residence in extended families, and the acceptance of taboos on sexual activity. The prac- tices of living in joint households and of having the husband leave the wife for long periods to work in other areas are associated with low income and low education. Since education, particularly the education of males, tends to reduce these practices, it will tend to increase fertility. In addition, since education changes adherence to traditional patterns of behavior, it will tend to reduce compliance with traditional sexual taboos such as postpartum abstinence and thus will tend to raise fertility. Therefore, on the supply side, education has positive and negative effects on fertility. The positive effects act through health and the level of sexual activity. The negative effects result primarily from the effect of female education on the age of wife at marriage and the probability of marrying. Education's effect through infant and child mortality is more complex. Its negative effect on completed family size probably results, at least in part, from the reduced de- mand for additional births. Overall male education is generally asso- ciated with increases in the supply of children, but female education has mixed effects on supply. THEORETICAL DETERMINANTS OF FERTILITY 75 Theoretical support of hypothesized demand relations Although education has positive and negative effects on fertility on the supply side, the vast majority of its effects on the demand for children are negative as shown in Table 2.2. The effect of education on demand also depends primarily on the wife's, as opposed to the husband's, education. The least certain effects of education on the demand for children or desired family size are those which act through preferences and the husband's income. While it is quite clear that education is likely to increase the husband's incomc, it is unclear whether this will increase fertility. However, therc is a little uncer- tainty about the effect of education on the preferences for children. Gcnerally, education rcduces such preferences by changing traditional values and increasing awarencss of alternative sources of satisfaction. However, the effect of family sizc prefcrenccs, such as ideal family size, on behavior is not well established empirically. Although the income effects of education on demand are unclear, the cost cffccts are strongly negativc. Education very clearly tends to raise the perceived cost of childrcn and to reduce the economic re- turns from them. The economic returns from children accrue in the form of child labor and support by childrcn in old age. 'llese costs and benefits can be altcred either by chaniges in extcrnal circum- stances such as opportunities for child labor or by clhanges in values on attitudes. Since education results in a shift of occupations from agricultural self-employcd to urban employcc, child labor is lcss im- portant within the family, and cmploymiient of childrcn outside the family becomes less acceptable as aspirations for the higlh educational achievement of children increase. 'llicsc shifts in occupation pri- marily rcsult from changes in male occupation, but aspirations may be affected by male or female education. 'llTe security motivc for having children may bc affected by malc and female education. Tllhe more educated generally have bctter access to capital mlarkets and arc morc likely to be employees in organizations that provide some form of retirement. In addition, women with job skills necd rcly less on children for support if thcir spouse dics. The moncy cost of children also depcnds on the education of both parents. By raising the aspirations for children, education raises the 76 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? expenditures that parents make on these children-particularly edu- cational expenditures. The time cost of children is affected primarily by the education of the wife and of women in general. The higher the wife's education, the higher her potential wage, and the more likely it is that a job will be available in the modem sector. Such jobs have two important characteristics-they are more attractive, and they are less likely to be compatible with child rearing. Both of these factors raise the cost of time devoted to child care.25 In addition to the effect of the wife's education on her own opportunity cost of time, female education in general will affect the cost of substitutes for mother's time in child carc. Higher education of adult women means that many women have alternatives that are better paying and perhaps more attractive than domestic service. The enrollment of female children in school makes it difficult for mothers to rely on their older daughters to care for younger children. 'This is a possible source of the interaction between individual fertility and community education noted in earlier studies. If women in general have very little education, an educated woman can afford massive child care substitutes and thus can combine the benefits of education with high fertility. Once the general level of female education is high, this is no longer possible becausc the child care substitutes are in school themselves or have qualified for higher paid work than child care service. Thus, unless a positive effect of male wages on the demand for children can be established, it is fairly evident that the general impact of education on the demand for children is negative. This applies to the education of the wvife and to the general level of education of women. The education of males reduces the perceived economic re- turns of children and raises the cost by increasing aspirations for the children. Theoretical support of hypothesized relation of education and fertility regulation Education increases the ability to regulate fertility, and thus its effect on completed fertility is negative for male and female educa- tion. It is hypothesized that the wife's education is more important 25 The more attractive a job, the higher the satisfaction denved and the higher the benefits from work. THEORETICAL DETERNIINAN'I S OF FER rILITY 77 than the lmusband's in gaining knowlcdge and forming attitudes. Education is hypothesizcd to incrcasc the knowledge of contracep- tive methods, and it would be a morc importan-t determinant of knowledge in arcas witlhout active public family planning programs. Education is also expectcd to make people more reccptive to ncw idcas and more likely to approve the use of contraceptives. Sinlcc edu- cation is also associated with higher income, it is cxpccted to be associatcd with increascd acccss to contraccption and abortion, par- ticularly in areas without widesprcad public family planning programs. Education is also hypotlhesized to affect contraceptive bchavior indirectly by altering how spouscs make decisions. Incrcascd educa- tion is expected to increasc discussion between husband and wife. Such discussion is necessary for successful contraception use and also for a consensus on practicing contraception. In addition, thc wifc is expected to have more responsibility and influencc in family decisionmaking if she has more cducation. Thc effcct of the 11us- band's education on his wife's role in decisionimlakiing depends largely on the kind of education. Summary The most important indircct effccts of education would bc as- sumed to be those acting througlh proportion marricd or probability of marrying, the age of marriagC,20 infanit and child mortality, the pcr- ceived cost of childrcn, economic bencfit of children, timc cost of children (wages of wife, compatibility of wife's work, availability of chlild care substitutes), communiicationi bctween husband and wife, and the birth control variables. Hcalth and prcferences are more intractable. However, the effcct of cducation on the ability to regulate fer- 26. Duza and Baldwin (1974) give some impressive evidence on the impor- tance of marital behavior in fcrtility declines. This evidence indirectly supports the emphasis oni such behavior in this model of fertility determination. In Korea between 1960 and 1970, changes in marital structure explained 35 pcrcent of the change in fertility In West Malaysia it explained 67 percent, in Iaiwan it ex- plained 23 percent In the Philippines marital structure accounted for two-thirds of the reduction. Thesc figures were cited from Cho and Rutherford (1973). R. Freedman and others (1970) attribute 10 of the 40 percent dccline in I-Iong Kong's crude birth rate (1961-68) to delayed marriage 78 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? tility is only relevant if the potential supply of children exceeds the demand. There may be certain situations where this does not occur, and this may explain the lack of a relation between education and fertility in those situations. In certain situations where the demand for children exceeds the supply, education may be associated with a greater biological supply of children (as explained above) without having any affect on the demand for children or even occasionally increasing it. It is evident that a large part of the negative effect of female edu- cation on fertility is believed to result from the employment oppor- tunities for women-higlher wages, greater attractiveness of jobs available, and the incompatibility of childbeanng and market work. If for some reason these factors arc not present, then a large part of the effect of female education on fertility wvill not operate. It seems more likely that these market effects of female education will emerge in urban rather than rural areas. Modern sector, high-status jobs are rare in rural areas, and if female education increases status, cducated women may in fact be less likely to work at available jobs than un- educated women (Goldstein, 1972). Tllhis may explain part of the interaction of education and urban residence. It may be that the conditions whlicl stimulate all the negative effects of education on fertility occur only in urban areas. Unfortunately, the relcvant kind of information on female market alternatives is fairly scarce. This study attempts to gather what evidence is available on the relation between female education and job opportunities, as well as other relations, but much more empirical research on this topic is needed. The next chapters establish the extent to which education affects intcrvening variables by examiniing the empirical evidence on the relation betwecn education and the relevant variable within the model established in Figure 2.2. In some cases the evidence is quite substantial; in other cases, it is nonexistent. In addition, education is related also to some intermiiediate variables-desired family size and contraceptive use. Education cannot directly affect these variables but acts through the more elcmentary intervening variables. How- ever, examination of the relations betxveen these intermediate vari- ables and education allows the relative importance of the effect of education through various channels to be judged. 3 Education and the Biological Supply of Children As described in the previous chapter, education affects fertility partially through factors determining the supply of children or natural fertility. The effects of education on the age of marriage and the proportion ever marrying are expected to reduce the bio- logical supply of children, but its effects on health of parents and children as well as on certain behavioral factors such as lactation practices and observance of sexual taboos are expected to increase the potential supply of living children. Thus, the net effect of educa- tion on the potential supply of children through these factors is impossible to determine theoretically. However, the magnitude of these effects on various supply variables can be determined by sur- veying the rclevant empirical work. The literature on the effects of education on marital behavior and on the other biological and be- havioral supply factors is examined below. 79 8o FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? Education and Marriage T'he proportion married in any group affects the fertility of that group. At the individual level, education may affect the probability of marrying at each age, and thus the age of marriage, as well as the probability of completing one's reproductive life without ever being married. Effect on the proportion never married In Chapter 2 the probability of ever being married was hypothe- sized to be inverscly related to education for women and directly related for men. Several studies confirm this pattern for women, but the pattern is not universally valid. CROSS-NATIONAL PATTERNS. On a cross-national level, Ruth Dixon (1971) has analyzed the proportion of womiien 40 to 44 whlo have never been married. She included measurcs of female literacy as well as the gainful cmployment of women 20 to 24 along witlh other ex- planatory variablcs. Female literacy was lower in countries with higher female rates of spinsterhood in the Westem European and English overseas countrics. In Eastern Europe, the Middle East, and Asia, however, the higlher the female literacy, the higher the propor- tion of women not married in the 40 to 44 age group.' Female literacy had little effect on the proportion of males ever married in this age group.2 Thle proportion of employed females aged 20 to 24 had no relation to the proportion never marned, aged 40 to 44 in the West, but there was a high positive correlation in the East.3 Tlle very different effects of female literacv and female employment on 1. Dixon uses an East-West distinction rather than the more conventional Western-versus-tradition distinction of Hajnal. Dixon's distinction is based on the greater cmphasis on individual responsibility and its effect on the feasibility of marriage in the West 2. In general, female behavior was much better explained by mcasures of the feasibility and desirability of marriage than was male marriage behavior. 3. Female employment or female literacy were never the most important factors, but they were more important in the East than in the West. In the East, female employment was a more important variable than female literacy. THE BIOLOGICAL SUPPLY OF CHILDREN 8i the proportion of women married in the East and West is part of a general pattern observed by Dixon. In the West, the poorer the country, the higher the proportion never married. In thc East, the richer the country, the higher the proportion never married. Dixon hypothesized that this resulted from the differences in the feasibilities of marriage in poor societies which have predominantly nuclear fami- lies and in those which have predominantly extended families. If nuclear families are the norm, as in the West, the poorest countries have fewer couples who can afford to set up the nuclear households necessary for marriage. CROSS-REGIONAL PATTERNS. Since the developing countries are the greatest concern, it is interesting to determine if, for those countries, the Eastern pattern of an inverse relation between education and probability of marriage persists. The studies listed in Table 3.1 give information on Thailand, Chile, Taiwan, and 19th century Russia. The methodology of the studies differs, but generally female educa- Table 3.1. Evidence on Education and the Proportion Ever Married Location Study (type) Direction (date published) (sample size) Method of relation Clhojnacka Russia, correlation (M) inverse (1976) 19th century (F) inverse (cross-regional) DaVanzo Chile multiple (F) legal direct (1972) (cross-regional) regression (F) consensual inverse (M) legal inverse (M) consensual direct estimated net (F) direct effect (M) inverse Dixon (cross-national) correlation (F) East-inverse (1971) (F) West-direct Maurer and Thailand multiple (F) inverse others (cross-regional) regression (M) direct (1973) Speare and Taiwan multiple (F) inverse others (individual) classification (1973) (3,579) Note: (F) = education of women; (M) = education of men Sources: For complete references, see the sources for this chapter. 82 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? tion can be said to increase the proportion never marrying except for Chile. For 19th century Russia, Choinacka (1976) found that literacy was negatively correlated with proportion married for men and women in urban and rural areas, but the effect was much stronger for women than for men. For Taiwan, Speare and othcrs (1973) found all increases in female education reduced the proportion married even when age, work experience, and residence were controlled. The other two studies have more complex results. In a cross-regional analysis of Thailand, Maurer and others (1973) found that for women, education decreased the proportion married at all ages. For men, education decreased the proportion married for those under 30 and increased it for those over 30. Thus it increased the proportion eventually marrying. This conforms to the hypothesis specified in columns 1 and 2 of Table 2.1. DaVanzo (1972) has analyzed the effects of male and female edu- cation in a cross-regional study of Chile. Unfortunately, her results cannot be compared directly with those of Thailand since the analysis of marriage in Latin America requires the separate analysis of legal and consensual marriages. Female schooling increases the proportion of females legally married for women in five-year age groups between 15 and 40 and for all women 15 to 49, but the coefficicnt is not sig- nificantly different from zero for women in their 40s. Male education decreases the proportion legally married; the coefficients have some- what lower values than female education, but for the ordinary least squares equations the coefficients are significant for males under 40. For consensual unions, the results are reversed. For all ages, female education has a significantly negative correlation. For males, the coefficients are positive but are significant only for males 15 to 19. It is quite obvious from DaVanzo's work that education of males and females does have a different impact on marriage behavior. For the purposes of this study, however, an analysis of the percent not married (legally or consensually) would have been more useful. Some information on this can be gained by comparing the size of the co- efficients for the legal and consensual unions.4 Since the coefficient 4. Since people must be either not married, legally married, or consensually married, the coefficients must add to zero, since a change of 1 year of schooling can increase the proportion in one group only at the expense of the others. This assumes that the dependent variable has been appropriately constrained. THE BIOLOGICAL SUPPLY OF CHILDREN 83 for female education is a larger absolute number for legal than con- sensual unions, increases in schooling must increase those legally married more than it decreases the number consensually married, and therefore it must reduce the proportion single. For males, education increases the proportion single at all ages. Both of these results are contrary to thc hypotheses set forth and to the other evidence. The only likely explanation would seem to be that Chile does not follow the Eastern pattern, but rather that, as in the West, cducation in- crcases the probability of marriage for women. The male behavior, howevcr, is very atypical. To usc education as a way to reduce fertility through lower propor- tions married, it would be important to identify more specifically the kinds of countries in whichn marriage is made more likely by higher education and thc countrics wvliere it is made lcss likcly. Effcct on the age of marriagc While little additional evidence of the relation betvecn education and proportion nevcr married xvas found, somc infercnccs can bc made from other kinds of studies. Since therc is gencrally a closc positive relation betwcen the proportion of women nevcr married and female age of marriage,5' much can bc infcrred from reviewing data on thc age of marriage. The substantial evidence on thc cffect of education on the age of marriagc is summarizcd in Table 3.2. CROSS-NATIONAL PATIrERNS. To gain a general perspective, Dixon's cross-national study is again useful. Rather thani using agc at marriage, Dixon used the proportion unmarried in the 20 to 24 age group. rlhis, shc felt, could reflect the pattcrn of agc of marriage. Again, the effect of female literacy on proportion married at young ages differs bc- twecn the East and WVcst. In thc West, higlher literacy is positively correlated wvith the proportion of Nvoicen and men married, but this variablc is not important in multiple regrcssion. In thc East, fcmalc literacy is negatively correlatcd with thc proportion of men and wvomen marricd in this age group, and female literacy is thc most imnportant variable in multiple regression analysis for women but not for men. In the East, but not the WNest, the proportion of womcn 20 5 Dixon found a correlation of 0.78 for womeil but only 0 48 for men. For Asia where the proportion-never-miarrying is generally low, Smithi found a cor- relation of 0.64 84 FERTILITY AND EDUCATION WHAT DO WE REALLY KNOW? Table 3.2. Evidence on Education and the Age of Marriage Study Location Direction (date published) (sample size) Method of relation Cross-national Dixon 57 countries multiple (F) direct' (1971) Westem Europe regression and English multiple (F) inverse overseas regression eastern Europe, multiple (F) direct' Middle East, regression Asia Cross-regional Smith 10 Asian correlation (M) and (F) direct (1977) countries (pooled) 10 Asian correlation (M) and (F) direct countries (individually) Cross-tabular-individual data Amani Iran cross-tabular (F) direct (1971) urban survey (2,000) Caldwell Ghana cross-tabular (F) direct (1968) urban elite (M) irregular (627) Chander and West Malaysia cross-tabular (F) direct Palan WFS (1974) (1977) (5,360 house- holds) Chung Korea cross-tabular (F) direct' (1972) 1971 survey (1,933) Hawley and Thailand cross-tabular (F) irregular Prachuabmoh 1964 survey (1966) (1,207) Lapierre- Korea cross-tabular (F) urban-direct Adamcyk and (F) rural-direct Burch (1972) Mott western Nigeria cross-tabular (F) direct (1976) (395) Nayar India cross-tabular (F) urban-direct (1974) 1961 Kerala (F) rural-direct THE BIOLOGICAL SUPPLY OF CIIILDREN 85 Table 3.2 (continued) Study Locatlion Direction (date published) (sample size) Method of relation Olusanya Nigeria cross-tabular (F) direct (1971) (4,408) (M) curvilinear Palmore and West Malaysia cross-tabular (F) direct Ariffin (5,457) (1969) Stycos Peru (1968) (699) cross-tabular (F) Lima-direct (177) cross-tabular (F) Chiibote- curviliiear (202) cross-tabular (F) Viru-Huaylos- direct Stycos and Turkey cross-tabular (F) urban-direct Weller 1963 survey (F) rural-direct (1967) (2,700) World Fertility Nepal cross-tabular (F) direct Survey (5,665) (M) inverse (1977) Yaukey Lebanon cross-tabular (F) Moslemi-direct (1963) (613) (F) Christian-- no relation Yaukey and San Jose cross-tabular (F) legal-curvilinear Tlhorsen (2,132) (F) consensual- (1972) imverse Caracas cross-tabular (F) legal-direct (2,087) (F) consensual- direct Mexico City cross-tabular (F) legal-direct (2,353) (F) consensual- direct Bogoti cross-tabular (F) legal-direct (2,259) (F) consensual- direct Rio de Janeiro cross-tabular (F) legal-direct (2,512) (F) consensual- direct Regression-Individual data 9 CELADE cities (See table 5) correlation (M and F) direct (1972) ideal age of multiple (F) direct' marriage regression except San Jose and Quito (M) directb (Table continues on the following page) 86 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? Table 3.2 (continued) Study Location Direction (date published) (sample size) Method of relation Harmon Philippines multiple (F) direct' (1970) 1968 survey regression (7,237) Knowles and Kenya multiple (F) direct' Anker 1974 survey regrcssion (1975) (1,074) Kogut Brazil mulhple (F) direct' (1974) 1960 census regression (M) direct Paydarfar Iran correlation (1975) ideal age for daughter (1,062 urban) (F and M) urban- direct' (176 rural) (F and M) rural- none (146 tribal) (F and M) tribal- none Note. (F) = education of women; (M) = education of men a Statistically significanit except in San Jose and Quito. b Statistically significanit only in Quito c Statistically significant Souirces. For complete references, see the sources for this chapter. to 24 employed was important in explaining proportion married at thesc ages. Thlus even though the proportion married at 20 to 24 may not be perfcctly correlated with age of marriage, possible East-West differenccs should be cxpected, and the East pattern should corre- spond to the hypotheses about cducation and age of marriage dis- cussed in the previous section. Since only developing countries are considered, the East pattern should predominate. The studies listed in Table 3.2 diffcr greatly in geographical loca- tion and metlhodology. Latin America is best represented in the studics because of the CELADE work. Asia is represented by five coun- tries and Africa by three. 'lTlc Middle East is represented by Lebanon, Iran, and Turkey. The methodology of these studies ranges from two-category cross-tabulations to multiple regression analysis. THE BIOLOGICAL SUPPLY OF CHILDREN 87 CROSS-TABULAR PATTERNS. The data reported on Iran, Lebanon, and Western Nigeria involve simple comparisons between two educa- tional groups. Mott's study of Western Nigeria (1976) shows that a lower proportion of women 15 to 19 with some education are married than are those with no education, 33 and 70 percent, respectively. Mott and Amani (1971) show that literate women in the city of Isfahan marry on the average a year later than illiterate women. Yau- key (1963) shows that among urban Moslem women in Lebanon, the educated marry a year later than the uncducated, but there is no difference in age of marriage bv education among Christian women., More detailed cross-tabular comparisons are reported in eight studies that present intcresting information on the direction and magnitude of the relation. For the most part thlis evidence shows uniformly positive relations between the education of women and agc at marriage. There are three irregulanties in these results. For one area of Peru in Stycos' study (1968), and for San Jose in the Yaukey-Thorsen study (1972), the lowest educational group marries later than some other groups. Yaukey and Thorsen found this to be true of both legal and consensual unions in San Jose. In their study of rural Thailand, Hawlcy and Prachuabmoh (1966) found several irregularities in the relation for various age groups and for the sample as a whole. On the average, those with no education married some- what later than those with 1 to 4 years of school, but this pattern was not uniform for all age groups. The two studies wvhich include male education in cross-tabular analysis shoxv irregular results. In Nigeria, Olusanya (1971) found that males with no education marry later than all other men except those who have attended a university. Caldwell (1968) used propor- tions of males marrying in each age group and found that fewer of the males with middle school had married by 25 and by 30 than those of any other group. However, Caldwell's data are not completely representative, since he studied an elite group in urban Ghana. The data from the Nepal World Fertility Survey (1977) shows that men with some schooling marry a little earlier than males with no school- ing. Tlhus, thcre is mixed evidence of the relation between male edu- cation and age of marriage. 6. Christian women marry on the average two years later than educated Mos- lem women. It may be that the higher the average age of marriage, the lower the effect of education as hypothesized in Chapter 2. 88 FERTILITY AND EDUCATION: WIIAT DO WE REALLY KNOW? What can be said about the size of the effect of education on age of marriage? Thlc studies whichi show a uniform relation between education and age of marriage differ in the categories used, making comparisons difficult. However, two patterns emerge. Tllc least uni- formity appears in comparisons between those Withl no education and those witlh primary schooling. 'llhe two studies of Korea (Chung, 1972; Lapierrc-Adamcyk-, 1972) show about a 2-year difference in the age of marriage for these groups. The Malaysian World Fertility Survey (1977) shIows differences of slightly more than 1 year. The Nepal World Fertility Survey shIows a differcnce of 0.6 years. The Indian study (Nayar, 1974) shows differences of between 0.5 and 0.9 years. This is close to the 0 6 diffcrence found in the Nigerian study. Stycos found differences as low as 0.5 for women whlo attended only 1 year of schIool, as large as 1.8 for those wvho attended 4 years, and slightly larger differences for those wvho attended 5 or 6 years. How- ever, in Chimbote those who had attended 5 to 6 years had about the same age of marriage as those with no schooling, and those with betwecn 1 and 5 years schooling had lower ages of marriage. Thus, the effect of small amounts of schooling differs greatly from 2 years to less than a year, and in some cases cven a lower age of marriage is observed for those with a small amount of schooling. 'Tfle difference in age of marnage bctween primary completion and secondary completion scems to be substantial in all cases. In the five Latin Amcrican cities in the Yaukey and Thorsen study and in the two cities for which data exist in Stycos's studv of Peru, the difference ranges from 0.7 to 3 years (1.8 years is the unweighed average differ- ence in these seven cities). In Nigeria the difference was 4.5 years. In India, it ranged from 1.3 to 2.8 depending on area, with larger differences in rural areas. T'he Lapierre-Adamcyk study of Korea shows the difference to be about 1.7 years. Tlle Malaysian World F'ertility Survey shows differences of almost 3 years except for women over 45 where sample sizes were vcry small.7 No study shows that those attending secondary school marry younger than primary graduates. Some studies indicate the relation between post-secondary educa- tion and age of marriage, but suchI information secms to be irrelevant for population policy considerations in developing countries. By far 7 These differences are not uniform at all age groups, since the youngest age groups exclude women who have yet to marry, that is, have ages of marriages far above the mean for each group. THE BIOLOGICAL SUPPLY OF CHILDREN 89 the largest differences and the most uniform effect is that of second- ary education. This results, no doubt, from the fact that secondary education directly conflicts with marriage and, in itself, is an alterna- tive to early marriage which is independent of job markcts. However, the fact that in all cases those completing secondary school have an average age of marriage in excess of 20, and generally over 21, indicates that the impact of secondary school goes far beyond the simple act of keeping adolescent girls off the street or, rather, out of the marriage markct and in thc classroom. In two of the three studies that report on urban and rural data scparatelv, there is some evidence that the rclation of education to age of marriage differs somewhat by residence. Among those who have been married 1 to 4 ycars in the Kerala India study (but not among those married 10 to 14 years), the difference in age of marriage betwecn those with elementary and those with high school is much larger in rural than in urban areas. In Turkey (Stycos and Weller, 1967) the educated always marry later than the uneducated, but among working women the diffcrence in urban areas is more than twice the difference in rural areas. Among unemployed wvomen the differences by educational status are nearly identical in urban and rural arcas. In the Lapierre-Adamcyk and Burch studies of Korea there appear to bc very few differences in urban and rural areas in the relation between education and age of marriage. Smith (1977) also approached the issuc of thc interaction of education and urban residence in the analysis of agc of marriage in Asia without finding a clear pattern of interaction. In addition to these urban-rural differences in actual age of mar- riage by educational group, Paydarfar (1975) has reported on the ideal age of marriage that Iranian men reported for their sons and daughters. In urban areas thCse ideals for daughters wcre significantly positively related to the father's (and his wvife's) education. Among rural and tribal groups no significant relation was reported. In none of the groups was the idcal age for sons significantly related to education.8 MULTIPLE REGRESSION PATTERNS. The most soplhisticated studies methodologicallv are those few multiple regression studies of the age 8 The laclk of significance in rural and tribal areas may result from the smaller sample sizes in that area. It would have been helpful if Paydarfar had reported the value ot correlations coefficients as well as the significance 90 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? of marriage. Kogut's study of Brazil (1974), Knowles and Anker's study of Kenya (1975), and Harmon's study of the Philippines (1970) provide straightforward confirmation of the effect of the wife's edu- cation on the age of marriage. In each case the coefficient is positive and significant.9 Kogut also found that the male's education had a positive significant coefficient for equations of the male's age at mar- riage, but in each instance the wife's education had a greater effect. The magnitude of the effect of education on the age of marriage differs by region and age group. In Brazil, the coefficient was 0.3 or larger, indicating that, at the average level of education, an increase of 1 year of schooling would increase the age of marriage by about a third of a year. The only other coefficient of this magnitude was among 25- to 29-year-olds in the Philippines. In Kenya, the coefficient was only 0.08. Part of the difference in effect may reflect the different average level of education in each group. In Kenya, the average level of education was only 3.63, but in the Philippines it ranged from 3.24 for women over 44 for whom education had no significant effect to 6.44 among women 30 to 34 where the coefficient was 0.28. Average education of women is not given by region for the Brazilian data. Although the CELADE study (1972) does not provide zero and multiple correlation analysis of age at first marriage, it does contain such analysis for ideal age of marriage.10 The wife's education and that of her husband are positively correlated with the ideal age at first marriage in all nine cities. In multiple correlation, wife's educa- tion is significantly positively correlated in all cases except San Jose and Quito.1" In multiple correlation, husband's education is only significantly correlated in one case, San Jose, and here it is positively correlated. In summary, the bulk of evidence indicates that increases in female education increase the age of marriage, as hypothesized. The excep- tions to this pattern tend to result from the fact that, in certain circumstances, those with no education at all tend to marry later than 9. The only exception is for women over 44 in the Philippine's study where the coefficient is not significant. 10. This is the age that women report as being ideal for first marriage. Yaukey and Thorsen show that "whereas there was a marked association between high education and later ideal marriage, there was at least marked association in that direction for real marriage." 11. Yaukey and Thorsen found that in San Jose the least-educated married later. THE BIOLOGICAL SUPPLY OF CHILDREN 91 those with some primary education. In some instances, those with no education also had fewer chlildren than those with a few years of schooling. The effect of male education on age of marriage is much less clear because of the scarcity of data, but male age of marriage is less important than female age in determining the biological supply of chlildren. Unfortunately, the studies relating education to age of mar- riage do not contain enough rural studies to determine if there are urban-rural differences in the effect of education on the age of mar- riage. However, the Stycos and Weller study of Turkey seems to indicate that employment, education, and residencc may have impor- tant interaction effects on age of marriage whlich need to be studied further. llhe size of the effect of education seems to be greater at the secondary than at the primary school level, but there is also wide variation from country to country in thcse differences. Education and Fecundity 'hlle biological supply of births depends not only on the number of years of marriage and thus the age of marriage, but also on the fecundity of thosc years. It can bc assumed that the more educated are in better health and arc thus more fecund.'2 Unfortunately, it is difficult to acquire the kind of data neccssary to test this hypothesis. A few pieces of information do exist, howvcer. Several studies give insight into the ways in wlhichi education might be related to fecundity. In the first study, Jain (1969) estimated fecundity by the length of timc between marriage and first birth for 2,190 women who had not been prcmaritally pregnant and wlvo had not uscd contraception during this interval. Multiple classification analysis showed that, after controlling for the wife's age at marriage, those of the highest educa- tion group (husband's or wife's education) had higher probabilities of conception than the least educated, but the rclation was not always uniform for adjacent educational groups. Howevcr, it appcars that fecundity depends more on the husbanid's education than on the 12. See Butz and Habicht (1976) and Baird (1965) for discussions of the link bctwcen hcalth status and fecundity. Thc Butz-Ilabicht dcfinition of fccun- dity as the capacity of a woman to have childreni is followed here. Fertility repre- sents the actual number of clhildren born rather thani the capacity to bear them. 92 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? wife's, and, in fact, the husband's education is the strongest predictor of fecundity. Thus, another reason for the wcaker negative effect of male education on completed fertility is its positive relation with fecundity probably acting through income. Another study in Taiwan (Jain and others, 1970) deals with factors affecting fecundity after the birth of a child rather than fecundity before the first birth. This study examined the effect of cducation, residence, age, and number of births on lactation and postpartum amenorrhea. In this study, the authors found that "better education or urban residence goes with younger age and a shorter period of lactation and that these in turn shorten the period of amenorrhea. In this way modernization may act to increase the period of child- bearing risk." Yaukey (1963) found that, in urban areas, cducated women nursed their children for a shortcr time than the uneducated in Lebanon. Adegbola and others (1977) also found this to be true in two urban samples in Nigeria. In their preliminary work on Malaysia, Butz and DaVanzo (1978) found that when other variables were controlled in multiple regression, the mother's education was only significantly related to length of partial brcast feeding for those who breastfed and not to probability of breast feeding, length of feeding, or length of full feeding. With respect to partial breast feeding, the effect of education was negative as expccted. In the Malaysian World Fertility Survey (Chandler and Palan, 1977) both mean months of brcast feeding and proportion feeding were inversely related to education of women and also to income of the husband. The Nepal World Fertility Survey (1977) shows that if either hus- band or wife had some schooling, the woman was less likely to breast feed for 2 ycars or more. Other evidence is related to observance of postpartum abstinence. Hull and Hull (1977), using Indonesian data, related the lcngth of voluntary postpartum abstinencc to income level and found a fairly uniform pattern of shorter periods of abstinence among higlher in- come groups of each age. Adegbola and others related such abstinence in urban Nigeria to education and found the higher the level of education, the shorter the duration of abstinence. Thus, education appears to increase the potential supply of births througlh biological and behavioral factors-through health as meas- ured by the probability of conception, through breast feeding prac- tices, and through observance of postpartum abstinence. In all cases, the relations suggest higher natural fertility among the wealthier and TIIE BIOLOGICAL SUPPLY OF CIIILDREN 93 the more educated.1- Whether such results can be generalized to other populations is uncertain. Even if the fecundity in each year of marriage werc unaffected by education, it is still likely that the supply of surviving children would be affected by education bccause of the rclation of education to infant and child mortality. Education and Infant and Child Mortality The supply of surviving children depends not only on the supply of births but also on the proportion of thesc births that survive. It can be assumed that the children of the morc educated have higher survival rates. Therc is also some empirical support for such a relation. In this section a number of cross-national, cross-regional, and cross- individual studies of this relation are discusscd. Table 3.3 shows the results of these studies. Cross-national patterns Shin (1975) and Stockwell and Hutchinson (1975) have conducted cross-national studics of the rclation betwcen infant mortality rates and cducation. Stockwell and Hutchinson found that adult literacy was significantly negatively correlated with infant and post-neonatal mortality rates. In addition, thc corrclation cocfficients were second in importance only to percent of calories from animal protein. Hlow- ever, when the sample was split on the basis of the dependent vari- able, litcracy did not have a significant effcct in countries with higlh infant mortality. In addition, in countrics at lower lcvels of mortality only post-neonatal mortality was related significantly to literacy.'4 Shin, using data for 1958 and 1968, found that ncwspaper circula- tion had significantly negative correlations with infant mortality for developing countries in both time periods. However, in multiple re- 13 Anotlher way education may increase fcrtility is by reducing adult mor- tality, thus increasinig the number of years of exposure to pregnancy. Data on this relaton were not surveyed. 14. Post-neonatal mortality covers the period from 1 month to 1 year of age. Neonatal rates are calculated for deaths in the first month of life Post-neonatal rates are more influenced by environmental factors and less by congenital factors than neonatal or total infant mortality rates. 94 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? Table 3.3. Evidence on Education and Infant and Child Mortality Study Location - Direction (date published) (sample sige) Method of relation Cross-nattonal Shin 1960 and correlation inverse' (1975) 1970 censuses multiple mverse' regression Stockwell and censuscs correlation inverse' Hutchinson late 1960s high mortality inverse (1975) low mortality inverse' moderate inverse' mortality Cross-regional Heller Malaysia multiple (F) infant-invcrse' (1976) census-1947, regression toddler-inverse' 57, 70 (1970) (1947, 57, (F) infant-inverse' 70 pooled) toddler-mixed Sloan Costa Rica multiple (F) infant-inverse' (1971) regression preschool-inverse' Mexico multiple (F) infant-inverse' regression direct (1960) preschool-inverse East Pakistan multiple (F) infant-direct' /LTh(1971) regression prcschool-inverse Sloan Puerto Rico multiple (F) infant-mixed (1971) regression Cross-individual Butz and 1976-77 survey multiple (F) inverse' DaVanzo (4,067) regression (1978) Cochrane and Nepal multiple (M) mixed (not others 1976 survey regression significant (1977) (144) Hull and lull Indonesia cross-tabular (F) urban-inverse (1977) 1971 census (F) rural-inverse Kelley Kenya multiple (M) inverse (1976) urban survey regression (401) Khalifa Egypt cross-tabular (F) rural-inverse' (1976) 1973 survey (1,234) Khan and Pakistan multiple (M) inverse Sirageldin 1968-69 survey regression (F) direct (1975) (2,910) THE BIOLOGICAL SUPPLY OF CHILDREN 95 Table 3.3 (continued) - Study Location Direction (date published) (sample size) Method of relatzon Knowles and Kenya multiple (F) inverse' Anker survey regression (1975) (1,074) Olusanya Nigeria cross-tabular (F) inverse-perceived (1971) 1966 survey mortality (4,408) Singh India cross-tabular (F) working-inverse (1974) survey nonworking-no (311) relation Stycos Peru cross-tabular (F) Lima-inverse (1968) survey (perceived mortality) (1,995) Note (F) - education of women; (M) = education of men a. Statistically significant Sources: For complete references, see the sources for this chapter. gression equations including measures of urbanization, population for each physician, and a measure of development, newspaper circulation was only marginally significant in developing countries but was highly significant in developed countries. These two studies indicate that in the poorest of environments, there are perhaps fewer opportunities for parents to affect their chil- dren's health through the improved knowledge of, or the better access to, medical care associated with education. In addition, the Stockwell and Hutchinson study indicates that measures of mortality that are unrelated to the immediate circumstance of birth are more influenced by cnvironmental factors and thus by variables such as parents' edu- cation. Sloan's cross-regional multiple regression studies in Costa Rica, Mexico, East Pakistan, and Puerto Rico (1971) and Heller's cross- regional study of West Malaysia (1976) shed light on these issues. Cross-regional patterns Of these countries, Puerto Rico has the lowest infant mortality rate for the periods studied. In the Puerto Rican results, few of the mor- tality differences can be explained by the various factors, and female literacy is not significant and in fact fluctuates in sign. Perhaps this 96 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNow? indicates that in countries with very low mortality, parents' education has little effect, at least when aggregates are being analyzed. 'llc infant mortality rate in Puerto Rico was 42.1 for every 1,000 birtlhs, but East Pakistan, with a rate of 171.4, had the highlest rate of the four countries. Again, in this high infant mortality country, female literacy was not statistically significant, and its coefficient was positive rather than negative. The results for the countries with middle mortality levels-Mexico and Costa Rica-are in striking contrast to the countries with high and low mortality. In Costa Rica, with an infant mortality rate of 53.9, femalc literacy lad a significantly negative regression coefficient in equations for both infant mortality and preschool mortality. Mexico, witi higher infant mortality rates than Costa Rica (90 in 1950 and 70 in 1960), showed no significant relation between infant niortality and female literacy but did show significantly negative re- grcssion coefficients in the preschool mortality equation. In addition, in both Costa Rica and Mexico preschool mortality was much better explained by literacy and other environmental variables than was infant niortalitv. Heller's cross-regional study of West Malaysia gives varying results depending on the period studied.l" For 1970, when tile infant death rates was fairly low at 39 for cvery 1,000, female literacy had a signifi- cantly negative regression coefficient for infant but not toddler mor- tality. In the cross-regional data pooled for 1947, 1957, and 1970, the total female literacy rate is significantly negative for infant mortality. However in the toddler mortality equations, the percentages of Clli- nese and Indian wvomen who were literate tended to have significantly positive coefficients although the percentage of literate Malay women had significantly negative coefficients. These results are somewvliat difficult to interpret, however, since they combine ethnicity and edu- cation. From these results infant mortality appears to be inversely related to female cducation in both data sets, but toddler mortality is less well behaved. This tends to contradict Sloan's findings that pre- school mortality is more explainable. 15. [eller's regression model is quite different from Sloan's since fertility and mortality are entered as endogenous variables in Heller's study, that is, mortality and fertility are assumed to be determined simultaneously rather than for one to determine the other. THE BIOLOGICAL SUPPLY OF CHILDREN 97 Cross-zindividual patterns As mentioned in earlier parts of this study, aggregate relations need not hold at the individual level, and it is the relation between parent's education and child mortality that is relevant for this model. Thus the existing evidence on the individual relation is also pre- sented in Table 3.3. The cross-tabular evidence is rather sparse. The Indonesian data (Hull and Hull, 1977) are the best and show inverse relations between education and infant mortality in urban and rural areas. The study of rural Egypt (Khalifa, 1976) and the small Indian study (Singh, 1974) also show inverse relations except for the nonworking women in the Indian study. In Indonesia, rural Egypt, and for the working women in India, the mortality of offspring for the most educated women is less than half that of the least educated. The Nigcrian (Olusanya, 1971) and Peruvian (Stycos, 1968) studies also show that the more educated are more likely to perceive that mortality has been falling over time. The multiple regression results are mixed. The two Kenya studies (Knowles and Anker, 1975; Kelley, 1976) show inverse relations that are statistically significant-in one case for male and in one case for female education. The Pakistan study (Khan and Sirageldin, 1975) enters both the husband's and the wife's education in the same equa- tion. Nonc of these variables have significant coefficicnts, but three of the four are inverse. In each case, income was inversely related to child mortality, and in one case it was significant. Income was not significant in either study of Kenya, although its coefficient showed an inverse association in all cases. In the small-scale Nepalese pretest (Cochrane, 1977) the six measures of male education and educational outcomes (literacy, numeracy, and so forth) fluctuated in sign. Only years of school completed, wvhich was inverse, approached significance. In all cases income was inversely related to child mortality, but in no case was it significant. The Butz-DaVanzo study of Malaysia (1978) showed that female education was inversely related' to infant mortality, but this was only significant if measures of housing quality were not included in the equation. Thus, it appears the association of educa- tion with higher income and better living conditions is responsible for the association with lower infant and child mortality, but this is not a universal pattem. 98 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? Summary The biological supply of children depends on the years of exposure to pregnancy, which is determined by the age of marriage, the poten- tial fertility (fecundity) of those years, and the survival of offspring. There is good reason to believe that education affects all these factors through its effect on a wide range of opportunities. Education pro- vides men and women with more choices about their lives. This affects their decisions to marry and at what age. Education also increases the possibilities for improved health for parent and child by providing better knowledge and higher income. The effect of education on the decision to marry is less predictable than its effect on health because circumstances and values about the desirability of marriage differ widely, but improved health is generally highly valued. Education is not universally associated with improved health, probably because certain environments are either so healthy or so unhealthy that individuals at all educational levels can either have good access to health care or no access whatsoever and are fully exposed to the same environmental circumstances. The effect of education on marriage varies widely depending on the kind of society and on the sex of the individual. In the developed countries of the West, the effect of literacy seems fairly unimportant. In the less developed countries, female education has a fairly strong direct relation with age of marriage, but there is some evidence that this effect may be second to, and probably acts through, female labor participation prior to marriage. For males, the effect of education on age of marriage is less uniform and less strong. Female education in developing countries is usually inversely related to the proportion of women who marry before the end of the childbearing period. The effect on the proportion of males ever marrying is unclear. Thus, education tends to decrease fertility by reducing the exposure to pregnancy for women through increasing the age of marriage and reducing the proportion married. However, education increases the potential number of surviving children for each year of marriage. What little evidence exists indi- cates that fecundity, as measured by the potential supply of births in each year of marriage, is greater because of better health, less rigid observance of sexual postpartum taboos, and shorter lactation periods. THE BIOLOGICAL SUPPLY OF CHILDREN 99 The survival of infants is also more probable for the more educated, but this effect is not equally strong everywhere and seems to be strongest at middle levels of mortality.16 Thus education has positive and negative effects on the unregulated supply of children; it seems to lower the years spent in marriage but raises the biological supply of living children for each year of marriage. T'he effects of education on the demand for children are expected to be more generally negative. 16. The biological effect of improved survival is to increase the number of surviving children. However, such an effect may generate behavioral adjustments to reduce the number of children ever born by acting through the demand for an additional child and fertility regulation. 4 Education and the Demand for Children In the previous chapter, the relations between education and several factors determining the biological supply of children- age of marriage, fecundity, infant mortality, and practices of lactation and postpartum abstinence-were examined. In this section the rela- tions between education and factors determining the demand for children-perceived costs and benefits of children and ideal family size-are discussed. Factors Determining the Demand for Children The demand for an additional child depends directly on the de- sired family size (demand for children) and the current family size. The latter depends on previous fertility and on infant and child mortality. Since it was shown in the previous chapter that education is generally inversely related to such mortality, it is expected that education, through this variable, also reduces the demand for an additional birth and therefore the total demand for births. Its effect 100 THE DEMAND FOR CIIILDREN 101 on the demand for surviving children is less certain since rcduced mortality lowers the "price" of surviving chilclrcn. Sincc the evidence strongly supports a positive relation between infant and child mor- tality and births, it scems apparent that reduced mortality incrcases the demand for surviving chlildren less thain it increases the biological supply of surviving childrcn. Thc factors that determinc desired family size (other than mortality) for whiclh adequatc data exist are dis- cussed below. In terms of the model developed in Figure 2.2, desired family size (thc substitute for the static demand for childrcn) dcpends on prefer- ences and perccptions of the various costs and bencfits of children as well as on the ability to afford children. 'The data on these vari- ables are relatively scarce, in part becausc the precise meaning of the terms used in the theorctical model are not clarified. This scems to be a major problem with thc concept of income. Current incomc is oftcn used rather than some lifetimc mcasure of potential income. Part of the ambiguity in the rclation between in- comc and fcrtility results from the use of improper measures of income. Tlhus, although it might be intcresting to determine the relation between husband's education and incoimle in developing countrics, this information tells littlc about the fertility-cducation relation since the income-fertility link is not known.' Thus, no at- tempt will bc madc to invenitory the evidence on the cducation- income link. The ability to afford childrcn depcnds on incomc and thc "price" of children. No such pricc is, of course, observablc, although theo- retical models of shadow prices have been developed. Rather than trying to specify a conccpt of price, it is easier to usc the conccpt of the costs of children.2 The cconomist is intcrcsted in separating these costs into moncy cost and time cost, because such a distinction is vital to understandinig the interaction of female education, market 1. Chapter 1 proidces somc data on this relation In the thlirty-one regression equations reported, incomc was significanit in onc-thlird of the cases. Half of the significant coefficients were negative, and half were positive. 2 The price-cost issue is quite complcx, since many of the expcnditurcs made oi clhildrcn are not requLired by the market The market dictates the price of rice, but the cost of feeding a child depends on how much of what items it is fed. These discretionary expenditures have been labeled chilld quality by Becker (1960). Required expenditures will not be separated from discretionary ex- penditures in thlis discussion. 102 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? work, and fertility. Unfortunately, too little data have been collected in developing countries to allow the time costs of children and their relation to female education to be estimated.3 There is more evidence on the monetary costs, since several studies have been conducted on the perceived monetary costs of children of people of various char- acteristics. There is also data from the same studies on the perceived benefits of children in terms of child labor and old age security. Thle relation between level of education and preferences for children can best be estimated by examining data on ideal family size. Al- though ideal family size is not a measure of relative preferences, it does give some insight into family size prcfercnces unconstraincd by economic considerations. Therefore, the effect of education on the demand for children will first be examined by looking at the relation between education and famlly size prefercnces and between education and the perceived costs and benefits of children. The net effect of education, through these variables, on the demand for children can be evaluated by looking at the relation betwcen education and a measure of the constrained preferences for children, desired family size.4 Preferences for Children (Ideal Family Size) A number of studies of ideal family size are summarized in Table 4.1. Urban Latin America is well represented by the CELADE studies. Asia is represented by studies of 'Taiwan, Korea, rllailand, and India. Continental Africa is represented by three studies of Nigeria and one each for Sierra Leone and Egypt. Most of the evidence shows inverse relations between education and ideal family size. In Latin America, 3 Exceptions arc thc study of Thailand by Maurer and others (1973) and Rosenzweig's study of rural India (1976), which use areal aggregate data and estimate the relation between female education and the wages of women. In Thailand female cducation had the expected effect. In the Indian study, the pcrcentage of woIEnCIl with primary schooljing did not have a significant coefficient in the equation for the female wagc rate. 4. Thlis distinction between the ideal or uniconstrained and desired or con- straincd prefcrences may seem somewlhat artificial, but evidence shows that there are systematic differences in responses to questions about ideal family size in general and questions about the ideal family size for a person in the circum- stances of the respondent See Freedman, R., Goldberg, D1, and Sharp, H. (1955), and Ware (1974) for discussions of these issues. THE DENIIAND FOR CHILDREN 103 however, Buenos Aires and Rio de Janeiro fail to show inverse corre- lations for female education (CELADE, 1972). In fact Buenos Aires shows significant positive partial correlations between education and ideal family sizc when other variables are controlled. Korea shows a very slight upturn in ideal family size among women who have gone beyond high school (Chung, 1972). Males in Taichung City, Taiwan (Freedman and others, 1962), and females under 45 in Cairo (Khalifa, 1973) showed irregular association between education and ideal family size. Slight direct relations were found in two groups in the India study, but these were not significant (Pareek and Kothandapani, 1969). Tlhe India study showed that the inverse relation between education and ideal fertility was much more likely to be significant for urban and rural industrial workers than for cultivators or non- industrial urban workers. Differenccs in ideal family size by educational groups Some of the cross-tabular studies also give evidence about the size of the differences in ideal family size by educational group. Differ- ences in ideal family sizes between the most educated and the least educated in these cross-tabular comparisons range from 0.03 to 1.75 children. The smallest difference is for young xvomien in Cairo (Khalifa, 1973). The largest difference is observed in Korea (Chung, 1972). In Taiwan, the educational differences slightly increased be- tween 1965 and 1973 despite overall drops in ideal family sizes for all groups (Freedman and others, 1974). In Thailand, the educational differences in urban areas were almost twice those in rural areas, perhaps explaiimng why differences in completed family size are smaller in rural areas (Knodel and Pitakepsombati, 1973). In Sierra Leone, the differences in ideal family size by education were largest in the capital of Freetown, followed by the differences in towns (Dow, 1971). In the villages the differences were less than half as large as those in Freetown.5 5. Thie differences in ideal family size by educational level may over- or under- estimate the true impact on family size preferences. Two women may state ideal family sizes of two, but if given a second choice one might say one and the other three, and on third choice one may say zero and the other four. These two women would have very different preferences for children despite identical ideal (first choice) family size preferences. Lolagene Coombs (1975) has de- veloped a technique for estimating these underlying preferences. 104 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? Table 4.1. Evidence on Education and Family Size Preferences (Ideal Family Size) Study Locatiorn Direction (date published) (sample size) Method of relation CELADr 9 Latin American Zero-order (F) inverse except (1972) cities correlations Buenos Aires and (see Table 1.5) Rio de Janeiro Partial (F) inverse' in correlation Mexico City, Quito, and Guayaquil, dircct' in Buenos Aires Chung Korea cross-tabular (F) inverse through (1972) (1,883) senior high school then slight uptum Dow Sierra Leone cross-tabular (F) literary inverse, (1971) (5,952) sclhooling inverse Freedman and Taichung City, cross-tabular (M) curvilinear others Taiwan (1962) (1963) (241 couples) Freedman and Taiwan cross-tabular (F) inverse others (1965-73) (1974) (14,920) Khalifa Cairo, Egypt cross-tabular (F) over 45-inverse (1973) (569) (F) under 45- curvilinear (M) inverse Knodel and Thailand cross-tabular (F) urban-inverse Pitakepsombati (2,000 urban, (F) rural-inverse (1973) 1,500 rural) Additional differences are found in Sierra Leone. Dow showed fertility diffcrences both by literacy and by whether the woomen had ever attended school. For the country as a whole and for the capital and various towns and villages, the diffcrences for literacy werc some- what smaller than for school attendance, indicating that the act of attending school may affect family size preferences beyond providing literacy itself. Perhaps one of the most interesting results was that found by Olusanya in wcstern Nigeria (1971). The proportion of women who felt that fertility was "up to God" decreased from over 50 percent for those with no schooling to under 10 perccnt for those who went beyond primary school. 'T'his is perhaps one of the most important THE DEMAND FOR CHILDREN 105 Table 4.1 (continued) Study Location Direction (date published) (samisple size) Method of relation Ohadike Lagos, Nigeria Zero-order (F) inverse (1969) (596) correlation (M) inverse Olusanya western Nigeria cross-tabular (F) inverse (1971) (4,408) Pareek and India cross-tabular (M) significantly Kothandapanx (1,500) inverse-generally (1969) for urban and rural factory workers (M) Generally not signrficantly inverse for cultivators and nonindustrial urban work Ware western Nigeria cross-tabular (M) inverse (1975) (1,495 men; (F) inverse 1,495 women) Yaukey Lebanon cross-tabular (F) inverse but (1963) (613) differences are minute Note: (M) education of men; (F) = education of women a. Statistically significant. Sources. For complete references, see the sources for this chapter. effects of education. 'nose who feel fertility is solely up to God end up with family sizes determined by supply factors. Only when this position is abandoned can demand factors operate, which may lead to fertility limitation. Yaukey (1963) found that in urban Lebanon more educated women were more likely to express explicit opinions about ideal family size than were uneducated women. However, the differences in ideal family size of those who expressed opinions by education were too small to be significant. Preference for sons In the entire discussion so far the issue of preference for sons has been ignored. The more sons desired, the greater the family size needed to obtain the ideal number of sons. Thus, if preference for sons decreases with increased education, the desired family size will Table 4.2. Mean Ideal Number of Sons for Married Women 20 to 39 Years Old, by Wife's Education and Urban-Rural Residence Mean ideal number of sons Rural East West West Philip- Hun- United Calcutta Delhi Indza Java Korea Malaysia Malaysia pines Taiwan Belgium gary States Item 1970 1968-69 1970 1972 1971 1966-67 1970 1968 1967 1966 1966 1970 Wife's education No formal 2.1 1.9 2.3 2.1 2.5 2.6 2.5 3.5 2.2 b 2.4 Primary 1.9 1 8 2.2 1.9 2.2 2 5 2.3 2.8 2.1 1 1 1.1 Junior high 1.7 1.7 - 1.6 2 0 2.3 2.0 2.6 1.9 1.2 0.9 1.7 Senior high and over 1.4 1.5 1.9 2.2 1.8 2.0 1.9 2.4 1.8 1.3 0.9 1.4 Urban-rural residence Large city -^ - 2.0 -s 2.0 2.2 2 1 2.5 2 0 1.1 0.8 1.5 Small city - - 2.0 - 2.2 2.4 2.3 2.4 2.1 1.2 1.0 1.5 Urban township - - 2.1 -g 2.1 - -' 2.9 2.2 1.2 1.0 1.5 Rural township - - 2.3 - 2.4 2.7 2.4 2.8 2.2 1.3 1.1 1 6 Total 1.8 1.8 2.2 2.0 2.2 2 5 2.4 2.7 2 1 1.2 1.0 1.5 Sample size 947 5,242 10,246 411 1,620 4,242 11,918 28,632' 4,300 2,566 5,208 4,685 Note: See Appendix 1, items 1, 6, and 8 of source for definitions. a. Less than 20 cases in the category. b. No such category. c. Frequency weighted as follows: urban respondents X 4 and rural respondents x 12. Source: R Freedman and L. Coombs, Cross-Cultural Comparisons: Data on Two Factors in Fertility Behavior (New York: The Population Council, 1974). THE DEMAND FOR CHILDREN 107 also decrease, all other things being equal. Freedman and Coombs (1974) have reported on ideal number of sons by the wife's education in several Asian societies as well as for three developed countries. Their results are shown in Table 4.2. Tlhe only exceptions to an inverse relation between education and ideal number of sons occurs in rural East Java and Belgium. In developing countries the differ- ences across education classes range from 0.4 in Taiwan and India to 1.1 in the Philippines. These differences partially reflect changes in overall family size that occur with increases in cducation but also reflect reductions in the proportion of sons desired. Perceived Costs and Benefits of Children Many of the explanations of the decline in fertility that accom- panies cconomic development center on discussions of the changing costs and benefits of children brought about by the developmcnt process (Leibenstein, 1957). A numbcr of studies on the value of children have attempted to documcnt the relation between cer- tain variables and these perceptions. Most of these studies have bcen conducted in Asian countries, and thcse results may not apply else- where. Studies relating these perceptions to the level of education are examined here. The value of childreni study This ambitious cross-national survey of the costs and benefits of children has becn carried out at the East-WVest Center of the Uni- versity of Hawaii. Tablc 4.3 reproduces the zero-order correlations between the individual's education and perceptions of five aspects of the economic costs and benefits of children (Arnold, 1975). The most consistent pattern associatcd with these variables is the inverse rela- tion between the parents' education and their expectations of eco- nomic help from thcir children. Although the size of the correlations varied in rclation to this bencfit of childrcn, thcy were all significant at the 1 percent level. The relation between education of parents and their concern for the economic burden of cducating children is inverse in all cases and significant in every case except Japan. This does not mean that the more educated spend less on their childrcn's education, but rather 1 o8 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? Table 4.3. Zero-Order Correlations between Education and Perceived Costs and Benefits of Children Country Cost or benefit Korea Taiwan Japan Philippines T hailand Economic burden of education -0.44' -0.42' -0 05 -0 32^ -0.36' Financial case of large family' 0 02 0.16' 0.15- -0.02 -0.04 Expected economic help from children -0 65n -0.59- -0.22' -0.34' -0.71' Decreased utility of children 0.08b 0.20- -0 06 0.26' 0 16- Economic benefits of large families -0.07 -0.17- 0 10 -0.22' -0.17' Sample size 378 432 412 389 360 a. Statistically significant at the 1 percent level. b. Statistically significant at the 5 percent level. c. Ability to afford a large family. Source- Fred Arnold, and others, The Value of Children: A Cross-National Study, Volume I (Honolulu: The East-West Center, 1975), table 5 5. that they feel such expenditures to be less of a burden than do the less educated. In Taiwan and Japan the more educated feel less finan- cial difficulty in raising a large family than the less educated. In Korea, Philippines, and Thailand the correlations between education and the ability to afford children are not significantly different from zero. In all countries except Japan, the more educated generally perceive that the utility of children has been declining6 to a greater extent than do the less educated. In addition, in all countries except Korea and Japan the more educated are less likely to perceive economic benefits from large families. Thus on the three measures of the utility of and economic benefits from children, the more educated perceive less return from children than do the less educated. However, the more 6 This was measured by the perception that the current generation is less wlling than the previous one to have parents live with them, to give part of wages to parents in old age, and to help in the house, farm, or business. TIIE DEMAND FOR CIIILDREN 109 educated in these studies also appear to feel bctter able to afford children, which may offset some of the estimatcd effcct of education on the perceived bencfits of chlildren. The World Fertility Survey in Thailand also found that thei more educated cxpect less cconomic benefits from chlildren and also showed that the most educated felt better able to afford children than the least cducated. Thosc with a small amount of education felt the greatest cconomic burden from having four or fewer children (WFS, 1977). Thle net effect of these positive and negative cffects of education on the demand for chlildren cannot be determined from this study. However, Bulatao (1975) has used the Philippincs data to construct regressions of ideal family sizc on several demographic variables and measures of costs and benefits. For urban and rural samples, marital duration was significantly positively rclated to idcal family size. In all samples, the number of childrcn perceived as imposing a heavy finan- cial burden was significantly positively related to ideals. In rural areas, the help expected from children had a positive rclation with ideal family size. In urban areas those who felt there wcre benefits to large families had significantly larger ideal family sizes. In those regressions, education itself failed to bc significantly related to ideals whcn other demographic variables and values were con- trolled. Tfhus education appears to act through these perceived costs and benefits in determining ideal family size. Other studies on the costs and benefits of childrcn The simplest of such studies are those likc Olusanya's (1971) which ask whether more chlildren tend to raise or lower the family's standard of living. In Westcrn Nigeria, Olusanya found that the more edu- cated were lcss likely to believe that having more children raised the standard of living. Another intcresting study on perceivcd costs of children is that reported by Eva Mucller (1972). This work was based on a 1969 survey of over 2,000 husbands in Taiwan conducted by the Population Studies Center of the University of Michigan. Mueller found that the perceived utility of children declined with education and with income. Husband's education, income for each adult, and wife's edu- cation rank first, second, and third, respectively, in importance in explaining the husband's perception of the utility of children. The husband's education and number of living children were the most 110 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? important in explaining the husband's sensitivity to the cost of children. Mueller also constructed a series of regression equations and meas- ured the relative importance of cost sensitivity, perceived utility, in- come for each adult, and age of wife in explaining desired family size.' The results differed substantially depending on whether the wife was under or over 30 years old. For husbands with wives under 30, cost sensitivity, perceived utility, and husband's education were first, second, and third, respectively, in explaining desired family size. Thus husband's education has an impact above and beyond its effect on attitudes. For wives over 30, wife's education, cost sensitivity, age of wife, and perceived utility ranked in that order in explaining desired family size. One problem, of course, is that older couples are more likely to rationalize their own behavior by reporting desires quite close to actual family size. 'Thus education and age of the wife may have their effect through actual fertility. In a later paper, Mueller and Cohn (1977) used path analysis to determine how several variables affect desired family size through the intervening variables of educational aspirations for children, perceived benefits of children, and consumption aspirations. Although educa- tion's effect through these variables is not discussed explicitly, it can be deduced by using the path coefficients given in figure 2 of the study The correlation of wife's education and desired family size is -0.26, which in absolute size excceds correlation of desired family size with income. TIhis correlation can be broken down to direct, indirect, and interaction effects. The direct effect is somewhat over half the total (-0.14). Of the indirect cffects, the effect of education through the perceived benefits of children is the largest, followed by its effect through consumer aspirations, and then by the effect through the father's aspirations for his child's education. The failure of educa- tion to have a stronger impact on fcrtility througlh educational aspira- tions for children is unexpected and may be peculiar to Taiwan. Another study on the benefits of children in Asia was conducted 7. Mueller refers to this as ideal family size, but in a later article the question used to elicit the infornation appcars to be more closely related to desired family size, since it iiplicitly incorporated an income constraint The actual question was, "If you werc lust getting married and could have )ust the number of chil- dren )ou want, how many would you like to have had when your wife is through having ehildren, about age 45?" THE DEMAND FOR CHILDREN 111 by Chang in Singapore (1976). Subjective measures of the benefits and the costs of children and the ability to afford children were ob- tained for 900 Chinese women who had their first, second, or third child in the spring of 1972. These measures of the benefits and costs were related to seven demographic and economic variables. The wife's education was the most important of these variables in explaining both benefits and costs. The type of education was the second most important variable explaining the benefits of children. Those women trained in English rather than Chinese schools had lower perceived utility for children. Wife's employment and husband's income were also usually important particularly after the second and third child, but husband's education was the least important explanatory variable. The dircction of these variables was as expected employed women had lower evaluations of children than other women, and higher husband's income was related to lower perceived benefits of children. For the costs of children, wife's education is followed by employ- ment and type of education (in different orders) for the second and third children and for the sample as a whole, but for women with their first child, husband's education is second in importance. The husband's income ranges from fourth to seventh in importance. As a whole, the demographic and economic variables explain the benefits of children better than they explain the costs. Despitc the importance of the wife's education in determining the subjective evaluation of the benefits and costs of children, the wife's education is unimportant in explaining the perceived ability to afford children. In fact, this ability is very poorly explained by the variables included, and ranking of variables by importance shifts greatly de- pending on whiclh child is considered. Thus, while the perccived costs and benefits of children depend greatly on the wife's education and employmcnt in this sample, the perceived ability to afford children does not. Whcn the three subjective variables were used to predict desired family size, ability to afford children was more important by far than costs or bencfits of children for all children. If household income is used rather than the perceived ability to afford children, benefits and costs become more important in explaining desired family size, particularly after the first child. However, the use of actual income rather than the perceived ability to afford children substantially re- duces the overall explanatory power of the model. ThIus, the effect of 112 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? education on desired fertility in this study must be assessed through its effect on the perceived costs of, benefits of, and ability to afford children. Desired Family Size As mentioned at the beginning of this chapter, preferences and perceived costs and benefits are considered to be factors that deter- mine desired family size. The net impact of these various channels through which education acts can be estimated by examining the relation between education and desired fertility. There is unfortu- nately a degree of bias in questions of desired family size. Since this question is usually asked of women who have already borne children, desired fertility may represent some rationalization of actual fertility, and those who have little ability to control fertility may have to do more rationalizing. Thus educational differences may be exaggerated. This was a less serious problem when ideal fertility was examined, since ideals are not as specific to the individual's situation. The studies of desired family size summarized in Table 4.4 are fairly limited in number. In Latin America, zero-order correlations bctween education and desired feitility are negative for women except in Buenos Aires and Bogota (CELADE, 1972). Partial correlations show female education has significantly negative coefficients in only two cities and a significantly positive coefficient in Buenos Aires.8 In an early, small study of Taiwan by Freedman and others (1963), the relation between education and desired fertility for malcs and females was generally inverse, but somewhlat irregular. The more recent survey of females by Speare and others (1973) found a regular inverse relation for urban and rural women with and without work experience. In addition, multiple classification analysis showed educa- tion to be the most important variable in explaining desired fertility. While Speare and others did not control for actual number of chil- dren, an age control was used. The Ghanaian survey (Pool, 1970) shows inverse relations for both urban and rural regions. The West Malaysian survey (Palmore and 8. As hypothesizcd in the previous paragraph, desired family size is closely related to current family size in the CELADE data. This correlation is positive and significant after controlling for other variables in every country. THE DEMAND FOR CHILDREN 113 Table 4.4. Evidence on Education and Desired Family Size Study Location Direction (date published) (sample size) Method of relation CELADE 9 Latin American zero-order (F) inverse except (1972) cities correlations Buenos Aires, Rio, (see Table 1.5) and Bogota Partial (F) inverse' for correlation Mexico City and Guayaquil; direct' for Buenos Aires Freedmnan and Taichung City cross-tabular (M) inverse but others (241 couples) irregular (1963) (F) inverse but irregular Palmore and West Malaysia cross-tabular (F) metropolitan- others (5,457) inverse (1969) nonmetropolitan urban-nonlinear rural-no difference Paydarfar Iran cross-tabular (M) urban-inverse' (1975) (1,062 urban) rural and tribal- (176 rural) not significant (146 tribal) Pool Ghana cross-tabular (F) urban-inverse (1970) (5,700) rural-inverse Rizk Jordan cross-tabular (F) inverse through (1977) (1972) secondary school (5,214) Speare and Taiwan cross-tabular (F) work experience others (3,579) away from home, (1973) urban and rural- inverse no work experience away, urban and rural-inverse Weekes- Cameroon cross-tabular (F) inverse' Vagliani (213) (1976) World Fertility Nepal cross-tabular (F) inverse Survey (5,665) (1977) World Fertility Thailand cross-tabular (F) inverse Survey (3,000) (1977) Note: (M) = education of men; (F) = education of women. a. Statistically significant. Sources: For complete references, see the sources for this chapter. 114 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? Ariffin, 1969), however, shows strictly inverse relations only in urban metropolitan areas; rural areas show no differences by education level, and nonmetropolitan urban areas show higher desired fertility for those with some schooling than for those with no schooling. The authors also found that completed fertility was higher for women with some primary education than for women with more or less education. The study of Iran (Paydarfar, 1975) shows desired family size that is negatively correlated with education at a significant level for urban males, but not for rural or tribal males. One problem with these results, however, is the small sample sizes in the rural and tribal areas. In the Rizk study of Jordan (1977), current family size was added to the additional number of children desired to get a substitute for total desired family size. An inverse relation was found through secondary school. The difference between illiterates and those with secondary schooling was more than four children. The Wcekes-Vagliani sample in Cameroon (1976) shows an in- verse relation between education and desired family size in an urban and rural sample. While a test of independence showed a significant relation between education and desired fertility for the total sample, the sample size was too small to control for urban-rural residence as well as education. The Nepali (1976) and Thai (1977) World Fer- tility Surveys also show inverse relations. Thus the overall inverse relation is observed between education and desired family size, but the relation is not as consistent as that for ideal family size. Summary Education affects the demand for children by altering preferences and by changing the perceived costs, benefits, and ability to afford children. In Latin America, Buenos Aires and Rio dc Janeiro tend to exhibit atypical effects. Although education appears to be inversely related to ideal family size, a few nonlinear relations were observed, and the strength of the effect of education varies. The inverse relation with respect to the ideal number of sons is verified in all the developing countries except one. There are not enough cases, however, to determine if there are significant urban-rural, male-female, or literacy level differences in the relation. THE DEMAND FOR CHILDREN 115 The effect of education on perceived benefits was more uniform than that on preferences. Education was associated with a higher perceived ability to afford children but also with fewer perceived benefits from, and higher perceived costs of, children. The net effect of these factors generates an inverse relation between education and desired family size that tends to reduce the demand for childrcn. Exceptions to this pattern occurred in Buenos Aires and nonmetro- pohtan West Malaysia. Despite the evidence cited above, the linkages between education and the demand factors are not well established, and several crucial linkages are missing. The relation between female education, poten- tial wages, and type of work available seem to be vital for testing the hypotheses set forward in the new home economics theory of fertility.9 The effect of aggregate levels of female cducation on the cost of child care substitutes for the mother's time necds to be established if the interaction of individual and aggregate levels of education are to be understood. The effect of husband's income on the demand for children has been investigated frequently without conclusive results. 'I'his relation needs to be determined before the effect of husband's education on fertility can be established firmly. 9. The new home economics theory of fertility has grown out of Becker's original article on the economics of fertility. The theory attempts to examine the household's production of various goods-suchi as and including children-in a framework similar to that of the business firm. The various inputs to produc- tion are the time of various household members, particularly the wife, and goods purchased in the market. See Becker (1960) and Robert J. Willis (1973). 5 Education and the Regulation of Fertility The biological supply of children and the demand for children determine whether there is a potential demand for fertility regulation. However, the actual, and particularly the effective, use of contraception depends on several other factors. Attitudes toward fertility regulation, knowledge of birth control methods, access to the means of fertility regulation, and communication between husband and wife about family size goals are essential for effective fertility regulation. Where spouses have different desired family sizes, the balance of power between husband and wife' is also important in determining fertility regulation, but since the concepts of power are controversial and since power is not necessarily independent of the means of regulating fertility, the role of power will not be discussed in this review. The concept of access to the means of fertility regula- tion is also problematic because it depends on laws about provision 1. Economists might prefer to think of power in terms of the relative weights of each spouse's preferences in the family preference functions. ii6 THE REGULATION OF FER'I'ILI ry 117 of birth control, income, and the availability of subsidized family planning services. Therefore, attitudes toward birth control, knowl- edge of contraception, and husband-wife communication are dis- cussed in that order. The net effect of education through these variables can then be assessed by looking at the relation bctween edu- cation and contraception use. At a later stage, access is incorporated into the discussion. Attitudes toward Birth Control Table 5.1 summarizes a number of studies showing the relation between education and favorable attitudes toward birth control. Ex- cept for evidence of irregularities in Knodel and Pitakepsombati's study of Thailand (1973) and Palmore's study of West Malaysia (1969), the relation is always direct, but it does vary in significance and strength. llhe uniformity of the results in Table 5.1 is in striking contrast with the lack of uniformity in the relation between educa- tion and complcted fertility reviewed earlier. Particularly in Latin America (CELADE, 1972) where a number of anomalous relations have been observed, there is no deviation from a direct relation between education and favorablc attitudes toward birth control.2 The study of India (Morrison, 1961) shows that education has some interaction with type of work. If a man is employed in factory work, education has less of an effect on attitude than otherwise, since most mcn in that environment have favorable attitudes regardless of their education. Knowledge of Contraception and Education Of course, a favorable attitude toward birth control is only the first stcp in successful practice. Knowledge of birth control methods is also essential.3 Table 5.2 summarizes several studies relating contra- 2. Buenos Aires exhibited atypical results not only for completed family size but also for ideal family size. 3. This does not imply that knowledge follows attitude in any causal sense. Some minimal knowledge may be necessary to favorable attitudes. Some psy- chologists even feel that attitude follows use, that is, people validate their use of contraception by having favorable attitudes. 118 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? Table 5.1. Evidence on Education and Attitudes toward Fertility Regulation Study Location Direction (date published) (sample size) Method of relatlon CELADE 9 Latin American zero-order (F) direct (1972) cities correlations (H) direct (1964-65) partial (F) direct in correlation Mexico City (see Table 1.5) (H) direct' m Mexico City Dow Sierra Leone cross-tabular (F) literacy-direct (1971) (5,952) (F) schooling- direct Freedman and Taichung City, cross-tabular (F) direct others Taiwan (1962) (M) direct (1963) (241 couples) Khalifa Egypt-rural cross-tabular (F) direct (1976) (1,234) Knodel and Thailand cross-tabular (F) rural--direct, Pitakepsombati (1969-70) except 1 to 3 years (1973) (3,500) approve less than no school urban-direct, except 4 years approve less than those with less schooling Morgan Nigeria cross-tabular direct (1975) (1964-68) (1,296) ceptive knowledge and education. Again there is very strong support for a direct relation. Tlhe major exception is again Thailand (Knodel and Pitakepsombati, 1973), where those with 1 to 3 years of schooling in rural areas are slightly less likely to know any method.4 But even in Thailand the general impact of education is as expected. Thlis can be seen from the fact that the education regression coefficient for Thailand in the Value of Childrcn Study (Arnold, 1975) is positive 4. The more recent 1976 World Fertility Survey in Thailand showed a uni- formly direct relation between education and contraceptive knowledge. A very rapid change in fertility and family planning knowledge and access occurred be- tween 1969-70 and 1976. THE REGULATION OF FERTILITY 119 Table 5.1 (continued) Study Location Direction (date published) (sample size) Method of relation Morrison India cross-tabular (M) industrial- (1961) (1953) direct' (290) village-direct' Palmore West Malaysia cross-tabular (F) metropolitan- (1969) (5,457) direct other urban- curvilinear rural-inverse Pareek and India cross-tabular (M) cultivators and Kothandapani (1,500) urban nonindustrial (1969) -direct urban and rural factory workers- direct Pool Ghana cross-tabular direct (1970) (1965-66) (5,700) Williamson India zero-order direct (1970) Israel correlation direct Pakistan direct Chile direct Nigeria direct Note. (M) = education of men; (F) = education of women; (H) = hus- band's education of women. a. Statistically significant. Sources For complete references, see the sources for this chapter. and significant. Likewise, although Palmore (1969) found atypical relations between education and attitudes toward contraception in nonmetropolitan West Malaysia, he found that the relation between education and knowledge of contraception was direct everywhere. In the study at Dacca by Roberts and others (1965) the only exception to a direct relation shows up in a group of women who have gone beyond the ninth gradc. Since there are only nineteen women in the sample, this reversal in the general pattern is not significant. The CELADE data (1972) indicates the relation between knowledge and education to be much stronger than that between attitude and education. The correlations for female education range from 0.331 to 0.47 for knowledge compared with 0.024 to 0.161 for attitude. The 120 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? Table 5.2. Evidence on Education and Knowledge of Contraception Study Location Direction (date published) (sample size) Method of relation Arnold and Korea Multiple (F) direct' others Taiwan regression (F) direct' (1975) Japan (F) direct' Plilippines (see Table 4.3) (F) direct' Thailand (F) direct' Caldwell and Nigcria cross-tabular (F, M) direct Igun (8,800) (1975) CELADE 9 Latin American zero-order (F) direct (1972) cities correlation (1964-65) partial (F) direct,' correlation all cities (see Table 1.5) Chung and Korea cross-tabular (F) direct' others (1971) (1972) (1,883) Khalifa Egypt-rural cross-tabular (F) direct (1976) (1,234) Knodel and h'liailand cross-tabular (F) urban-direct Pitakepsombati (1969-70) rural-direct except (1973) (3,500) 1 to 3 years know less than less educated Palmore \V'est Malaysia cross-tabular (F) metropolian- (1969) (5,457) direct otlher urban-direct rural-direct Korean (Chung and others, 1972) and Thai cross-tabular studies indicate the magnitude of association. Of the Korean women who bad no schooling, 78 percent had heard of at least one method. Of those who finished high school, 95 percent had heard of one. In Thailand the percent knowing a method ranged from 58.2 to 88.7 percent in urban areas, and in rural areas the range was 35.5 to 73.1. The difference in knowledge across educational classes in rural areas seemed greater than in urban areas. This pattern also appeared to be true in Morrison's comparison of industrial and village areas (1961). Caldwell and Igun (1975), however, found that while those with some schooling were more likely to know modem methods of contra- ception, the differences for urban areas were not necessarily smaller than in rural areas. THE REGULATION OF FERTILITY 121 'Table 5.2 (continued) Study Locatwn Direction (date published) (sample size) Method of relation Nepal Family Nepal cross-tabular (F) direct Planiing Survey (1975) (6,012) Pool Ghana cross-tabular direct (1970) (1965-66) (5,700) Roberts and Dacca, East cross-tabular (M) direct others Pakistan (F) direct through (1965) (1962) ninth grade (547M, 547F) LSmrmons and Rural Latin cross-tabular (F) direct de Jong America ((H) direct Costa Rica (1,273) Colombia (1,707) Mexico (1,971) Peru (1,731) World Fertility Nepal cross-tabular (F) direct Suirvey (5.665) (1976) Note (F)= education of womcn; (M) = education or incn, (H) = hus- band's education of women. a Statistically significant. Soturces- For complete rcferences, see the sources for this chapter. The imrportance of education in contraceptive knowledge What causes the effect of education on contraceptive knowledge to differ from place to place? If this question could be answered it might be possible to determine whether it is education itself or some other variables that cause differences in contraceptive knowledge- such as urban residence, mass media exposure, and income. Only if such questions can be answered can education be considered an appropriate policv instrument. Some light can be shed on this issue. Fortunately, both the CELADE and Value of Children studies include the many factors affecting contraceptive knowledge in the same analysis. 122 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? The Value of Children survey (Arnold, 1975) used several socio- economic variables and many value indexes to explain contraceptive knowledge. The socioeconomic variables of interest are age, income, education, urban experience, media exposure, and marriage duration. Education was significantly positively related to knowledge in all cases even when these other variables were controlled. However, its relative importance in explaining knowledge varied. In Hawaii, the Philippines, and Thailand it was the most important variable. In Korea education ranked below media exposure and age. In Taiwan it ranked below media exposure. In Japan it ranked below media ex- posure and urban experience. Media exposure thus seemed to be the major alternative factor explaining contraceptive knowledge and was significantly positive in all cases and second to education in the Philippines and Thailand. The mass media effect was stronger in countries with strong family planning programs, indicating that mass media can possibly be a substitute for formal education, but the latter continues to have a strong effect even in those circumstances.5 In the CELADE study (1972), the partial correlations for education were significant everywhere and rank first in all cities except Bogota, San Jose, and Caracas, where they ranked sixth, fifth, and fourth, respectively. The number of living children also had consistently significant correlations in all cities and ranked higher than education in Bogota, San Jose, and Caracas. Family expenditures, a substitute for family income, was significantly positive in all cities except Rio de Janeiro and ranked first or second in five of the cities. This indicates the importance of income in providing acccss to knowledge in coun- tries that had neither official nor private family planning programs. In the early 1960s none of these Latin American countries had such programs. Thus, education has a significant effect on contraceptive knowledge in countries with and without family planning programs even after controlling for many other variables. The major limitation on the information available is the small number of studies in Africa and the limited number of published studies of rural as opposed to urban populations. 5. Part of the effect of education may be disguised in these results, since edu- catbon probably increases media exposure-at least exposure to printed media. The zero-order correlations between education and knowledge were by far the largest of all variables except in Japan. THE REGULATION OF FERTILITY 12 3 The community level of education and contraceptive knowledge In a recent unpublished paper using the rural Latin American sample surveys, Simmons and de Jong have attempted to determine whether the level of cducation in the general community interacts with the wife's level of schooling to determine contraceptive knowl- edge. They found that the community's level of education was im- portant in determining the percentage of women who knew about contraception even when the individual's level of education was con- trolled. In particular, in areas where there was the least knowledge of contraccption (Peru), the community lcvel of cducation rivaled the individual's level of education in explaining contraceptive knowl- edge.6 This relation may explain in part the interaction between communal level of education and individual education in the educa- tion-completed fertility relation observed in Chapter 1. It would be useful to know whether this phenomena is peculiar to rural arcas and unique to countries that lack broad family planning programs. Communication between Husband and Wife There is some evidence that communication between husband and wifc increases contraceptive use and reduces fertility (Mitchell, 1972; Michel, 1967; Hill and others, 1959; Ramakumiiar and Gopal, 1972; and Simmons and Culagovski, 1973).7 Tllhere is also some evidence that education increascs the level of husband-wife communications, as summarized in Tablc 5.3. This evidence is not cxtensive, but what there is shows a direct relation between education and level of com- munication. Tllhc study by Mukherjee (1975) is the most detailed of those cited and contains data from three areas of India divided into urban and rural samples. Number of communication media used, frequency of radio listening, literacy, wife's education, and husband's education were all generally directly related to frequency of husband- wife communication, and the strength of the correlations was in the 6. It is not possible from the study cited to determine why the community level of education has the effect it does. 7. Mukherjee also showed that husband-wife coimmumcation was positively correlated in a significant manner with knowlcdge of and attitude toward family planning. 124 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? Table 5.3. Evidence on Education and Communication between Husband and Wife Study Location Direction (date published) (sample size) Method of relation Brody and Jamaica correlation (F) direct others (150) (1977) Khalifa Egypt-rural cross-tabular (F) direct (1976) (1973) (1,234) Mukherjee India zero-order (F) education- (1975) correlation direct' (H) education- direct' (F) literacy-direct' number of communi- cation media used- direct' frequency of radio listening-direct' Olusanya Western Nigeria cross-tabular (F) direct (1971) (1966) (M) direct (4,408) Oppong Accra, Ghana cross-tabular (F) direct (1970) (1967-68) (163) Pool Ghana cross-tabular (F) direct (1970) (1965-66) (5,700) Ramakumar India cross-tabular (F) direct and Gopal (534) (1972) Note: (M) = education of men; (F) = education of women, (H) lhus- band's education of women a. Statistically significant in one case. b Statistically significant in three cases. c. Statistically significant in five of six cases. d Statistically significant in all cases. Sources: For complete references, see the sources for this chapter. order listed. Thus husband-wife communication seems to be related to level of communication in general more than to education itself. Some unsolved issues related to this variable exist. The relative im- portance of the husband's and wife's education is perhaps the most critical unsolved question. THE REGULATION OF FERTILITY 125 Contraceptive Use Behavior-as opposed to attitudes, opinions, or knowlcdge-needs to bc considered in the context of the individual's cxternal circum- stances. Tllius the relation between actual fertility and education needs to be examined in the context of the age or length of marriage for the woman, although desired or ideal family size can at least thco- retically be considcred indcpcndently of thosc variables. Likewise, attitudes toward and knowledge of different mcthods of contracep- tion can be examined for women of any age or marital status, but whether a womani actually uses contraccption or has uscd it in the past depends so much on her age, her number of births, and the duration of her marriage that examination of usc that does not coI1- trol for at least onc of these factors is likely to generate highly spurious results. Thercfore, the relation between education and con- traceptive use will be examincd only wherc age, number of births, or marital duration arc used as controls. Tablc 5.4 summarizes the rcsults of studies in Latin Amcrica, Asia, and the Middle East. Unfortunately, no studies of contraceptive use with the appropriate age controls could be found for Africa. In Latin America both the CELADE (1972) and other studics show direct rela- tions between education of woomen and use of contraception even after controlling for other factors. In all cities except Panama City, the cocfficients of education are significant. The coefficicnt of the husband's cducation is significant only in Panama City and Bogota. 'lhus the wife's education scems to be more important than that of her husbanid in determining contraceptive use. Asia is represented by studies of Korca (Chung and others, 1972), Taiwan (Speare and othcrs, 1973), 'llailand (Knodcl and Pitakep- sombati, 1973), West Malaysia (Palmore, 1969), and Nepal (WFS, 1977). Tlhe Taiwani and Korean studies show direct relations as do the rural Thai and urban West Malaysian sample. In urban 'flhailand, however, women with 1 to 3 years of schooling exhibit the lowest contraceptive use rather than thosc with no education.8 In rural West 8 The 1976 World Fertility Survey in Thailand showed a uniformily direct relation between education and contraceptive use ovcrall, but for women under 25, those with no education had sligltly hligler previous usage thall thosc with 1 to 4 years of schooling. 126 FERTILITY ANfD EDUCATION: WHAT DO WE REALLY KNOW? Table 5.4. Evidence on Education and the Use of Birth Control Study Location Direction (date published) (sample size) Method of relation CELADE 9 Latin American multiple (F) direct'-except (1972) cities correlation Panama City (1964-65) (see Table 1.5) (H) direct-only Panama City and Bogota Chung and Korea (1971) cross-tabular (F) direct' others (1972) Khahfa Cairo, Egypt cross-tabular (F) direct (1976) (1970) (M) curvilinear (569) Knodel and Thailand cross-tabular (F) rural-direct Pitakepsombati (1969-70) urban-lower use (1973) (3,500) with 4 years than less educated Palmore West Malaysia cross-tabular (F) metropolitan- (1969) (1966-67) direct (5,457) other urban-direct rural-no difference Sear Cali, Colombia multiple (F) direct' (1975) (1973) regression (508) Speare and Taiwan multple (F) direct others (3,579) classification (1973) analysis World Fertility Nepal cross-tabular (F) direct Survey (5,665) (H) direct (1977) Yaukey Lebanon cross-tabular (F) urban-direct (1963) (613) Note. Age, parity, or marital duration are controlled for. (F) = education of women; (M) = education of men; (H) = husband's education of women. a. Statistically significant. Sources For complete references, see the sources for this chapter. Malaysia there is no difference in use by education. The only other deviation from a strictly direct relation is found for males in Egypt (Khalifa, 1976), where those with a middle level of education among wives over 45 have the highest contraceptive use. But given the relatively small sample size, these differcnces are of questionable significance. In Lebanon (Yaukey, 1963), education is directly re- THE REGULATION OF FERTILITY 127 lated to both past use of abortion and contraception among Moslems and Christians in urban areas. Thus, as is the case with the relation between education and the other fertility regulation variables, the relation for contraceptive use is almost uniformly direct. It is certainly more uniform than is the relation between actual fertility and educa- tion and appears to be even more consistent than the relation between education and age of marriage, since there are no differences such as the East-West patterns observed by Dixon. Details of the relation The size of the effect of education on contraceptive use can be seen in Table 5.5, which is taken from the Freedman and Coombs cross-cultural study (1974). The data come primarily from Asian countries, although one Latin American and one Middle Eastern country are represented. No African data are presented. Table 5.6 summarizes west African studies (Caldwell, 1975) on the relation between education and contraceptive use. With two exceptions these show a direct rclation between use and education. However, these studies do not control for age, number of births, or marital duration. In Table 5.5 there are two countries that do not show a uniformly direct relation between education and contraceptive usage-Korea and urban Thailand. In Table 5.4 urban Thailand showed curvilinear relation, with those with 4 years of schooling showing lowest use. This does not show up in Table 5.5, perhaps because of the aggrega- tion over all levels of primary school, but there is a slight irregularity in an othervise uniformly inverse relation for those who achieved senior high school. The Korean result conflicts with the data pre- sented by Chung and others (1972), cited in Table 5.4. The differ- ence in Korea probably results from the fact that Chung and others cited data on whether a woman had ever used contraception or abor- tion, but the Freedman and Coombs data are on current use of contraception. Since the most educated women in Korea have high abortion use, they would be expected to rely less on contraception.9 Table 5.5 also shows the wide variety of levels of contraceptive use. Among women 30 to 39 the use rate ranges from 11 percent in West 9. Of women with four or more live births who have at least high school edu- cation, 59 percent have used abortion. Table 5.5. Percentage Currently Using Contraception for Married Women 20 to 39 Years Old, by Wife's Age and Education Thai- Thai- Wife's age West West Mexico Philip- land land (years) and Anlkara' Calcutta Inidza Korea Malaysia Malaysia City pines Taiwan Tazwan' (urban) (rural) educationi 1966 1970 1970 1971 1966-67 1970 1971 1968 1967 1970 1970 1969 20to 29 42 38 10 10 7 17 23 16 16 26 34 9 No formal 18 18 5 12 2 11 2 9 12 21 18 0 Primary 38 38 18 10 7 17 20 14 16 22 33 10 Junior high 41 37 _d 6 12 26 41 20 26 48 43 Senior high 68 65 30 14 33 30 -' 29 36 49 37 and over 30 to 39 41 43 15 33 11 19 25 18 45 58 48 16 No formal 26 23 10 22 5 13 8 11 37 51 33 9 Primary 50 51 21 36 17 23 21 15 48 61 50 17 Junior high 61 60 __d 41 36 36 51 22 66 72 57 Senior high 62 66 46 39 49 47 -' 34 70 81 55 - and over 20 to 39 37 40 12 24 9 18 24 17 32 44 42 13 No formal 22 21 8 20 4 12 5 10 26 39 30 6 Primary 43 44 19 24 11 19 21 15 33 43 43 14 Junior high 53 47 __d 24 23 29 45 21 45 60 30 -0 Senior high 65 65 36 26 39 36 47 32 51 63 47 and over Sample size 552 947 10,246 1,620 4,242 13,449 486 25,604' 4,300 2,491 1,080 642 Note. See Appendix 1, iteni 6 of source for deCfmtions. Sterilization included as contraception except in Belgium and Great Britain. a Defined as, ever used contraception. b. Taiwan 1970 data refer to ages 22 to 39. c Lcss than 20 cases in the category. d No such category. e Frequency is weighted as follows: urban respondents x 4 and rural respondents X 12 Source. R Freedman and L. Coombs, Cross-Cultural Comparisons: Data on Twvo Factors in Fertility Behavior (New York: The Population Council, 1974) Table 5.6. Peicentage Using Contraception in West Africa, by Various Measures of Education I-lighest level of education reached No Middle Study Location Year Sex Information schooling Elementary school, Secondary Tertiary' Ohadike Lagos 1964 F ever used any 5 8 - 20 71 contraception ever used modern 1 3 - 17 50 contraception Okediji Ibadan 1965-6 F ever used any 11 17 10 36 85 contraception ever used modern 0 0 2 27 58 contraception Igun Irrua 1970 F ever used any 2 5 - 5 -¢ Olusanya, (rural contraception and Acsadi Nigena) Caldwell Lagos 1969 M and F ever used modern 10 22 and Igun contraception Caldwell Southem 1969 M and F ever used modern 2 14 and Igun Nigeria contraception (UTban)d Caldwell Northern 1969 M and F ever used modern 0.7' 1.3' and Igun Nigeria contraception (urban)' Caldwell Nigeria 1969 M and F ever used modern 0.04' 1.9' and Igun (rural)' contraception Dow Freetown 1969-70 F ever used any 12 29 contraception Dow Sierra Leone 1969-70 F ever used any 9 20 (other contraception urban) Dow Sierra Leone 1969-70 F ever used any 3 16 (rural) contraception Caldwell Ghana 1963 F ever used any 27 44 (urban contraception elite) currently using 20 36 contraception Caldwell Ghana 1963 M ever used any 24 41 (urban contraception elite) currently using 20 36 modern contraception joint education of husband and wife Husband Neither Wife only only with Both with with any with some some some education education education education Ohadike Lagos 1964 F ever used modern 0 0 2 9 contraception Dow Freetown 1969-70 F ever used any 11 24 10 31 contraception (Table continues on the following page) Table 5.6 (continued) Literacy Level of informationh Study Location Year Sex Information Illiterate Literate Lowv High Dow Sierra Leonc 1969-70 F ever used any 8 20 8 23 (other contraception urban) Dow Sierra Leone 1960-70 F ever used any 3 12 3 18 (rural) contraception Note These studies do iiot control for age or marital duration. a. In somc ex-Britisl colonies, this is a postprimary or postelementary school to improve educational standards for admission to secondary schools b. All further education requiring at least some secondary schloolilng or incorporating it c. NumLbers too low to be significaint. d WVestern and Kwara states e Kano f Decinal parts of percentages are given becausc of the very small numbers g 7Vestcrn, Kwara, and Kano states h. Measured by a standardized scries of questions on various subjects Source- Caldwell, Johli C (ed ), Population Growth and Socio-Economic Change in West Africa (New York Columbia Uni- versity Press, 1975), Table 3 6. Sources cited in this table can be found in Caldwell (1975) but not in the sources to this chapter. TIIE REGULA l'ION OF FERI ILITY 133 Malaysia in 1966-67 to 58 percent in Taiwan in 1970. The diffcrcnces across educational groups range from 8 percent in rural Thailand to 44 percent in Wcst Malaysia in 1966-67. Differences are somcwhat larger in cities and urban areas than for countries as a whole. This can be seen by comparing Calcutta and India in 1970. The differ- ences for Ankara and Mexico City are both fairly large. Differences in use also appear to narrow over time. T'hey narrowed substantially in West Malaysia between 1966-67 and 1970 and somewhat less in Taiwan between those years. It would be difficult to say what increment in education has the greatest impact on contraceptive use. In Ankara, Calcutta, Korca, and urban Tliailand, the step from no education to primary education had the greatest impact. In West Malaysia, Mexico City, and Taiwan in 1967 it was the step from primary to secondary school. In the Philippines it was the stcp from junior to senior high school which had the greatest impact. Rcasons for direct relation Contraceptive use should increasc with cducation for scveral rca- sons. First, the more educated appear to have greater natural fertility and generally higher rates of survival for their children. Second, ideal and desired family size tend to be inversely related to education. Third, the morc educated have better attitudes toward and knowlcdgc of contraception. Fourth, more cducated people tend to bc more rational in their behavior: that is, if they do not want more chlildrcn, they will do something to prevent more births. T'hc first three reasons have already been discussed. Freedman and Coombs (1974) give some evidence on the fourth point by showing the proportions of women in each educational group who do not want more children and who are not using contraception." These data are prcsentcd in Table 5.7. Thle proportion of women who do not want more children and do 10. Women who do not want more childreni may not use contraception for several reasons they may no longer be fertile, they may be able to rely on abor- tion; contraception may be unavailable or available at high cost. Alternatively where women have little freedom or control over their own lives, they may not have access to contraception because of husband's restrictions. See Goldberg and Litton (1967) on this concept of access. 134 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? not use contraception ranges from 29 percent in Delhi to 89 percent in Jakarta. This probably reflects the availability of contraception in Jakarta at the time of the survey. However, high rates (80 percent or more) of nonuse were also found in rural Thailand in 1969-70, the Philippines, West Malaysia, and India. The effect of education on nonuse for this group is inverse, as expected in all cases except Korea, where reliance on abortion is a popular alternative. The differences across educational groups range from a low of 12 percent in rural Thailand, where differences between only two education groups were studied, to a high over 50 percent in West Malaysia and Mexico City. As in the case of overall contraceptive usage, the level of education with the greatest impact on use differs from study to study. The difference in use between those with no schooling and those with primary schooling is greatest in Ankara, Calcutta, Delhi, Korea, and urban Thailand. The step from primary to junior high levels has the greatest impact in West Malaysia, Mexico City, and Taiwan. T'hese patterns are the same for current use. In Jakarta the largest difference occurs between junior and senior high schools. This information on contraceptive use is impressive in the uni- formity of its relation to education. Unfortunately, the variation in differences by educational levels is so great that statements cannot be made about which level of education should be emphasized. An- other problem is that it is impossible to determine the extent to which the effect of education on use results from its effect on moti- vation (through its effects on the supply of and demand for children) and how much results from its effect on access to contraception through more knowledge, high incomes, urban residence, and better medical care. Some insight into these problems can be obtained from Table 5.8, which shows the variables that were significant in explaining contra- ceptive use in multiple correlation analysis of the CELADE data. The variables included here are not only objective measures such as age, income, and education but also subjective variables such as desired family size, contraceptive attitudes, and so forth. The wife's education was not significant when these intervening knowledge, attitude, and motivationi variables were controlled. Knowledge of contraception was the most important explanatory variable. The desire for more childrcn was also significant in all cities. Level of motivation and attitude toward family planning were next in importance, followed by an index of religiousness and desired family size. It has been shown that THE REGULATION OF FERTILITY 135 education acts on knowledge, attitude, and desired family size. Ex- penditures by the family (which substitutes for income) and the husband's education are only significant in three and two countries, respectively. Thus these results indicate that the major effects of the wife's education are those acting through contraceptive knowledge and attitude and desired family size"" rather than through family income, even in an environment where there were few public family planning programs. Mueller and Cohn's study (1977) of males in Taiwan gives an example of the factors explaining contraceptive use in an environ- ment where they are extensively provided to the public. This shows that the relative importance of education on contraceptive use de- pended on which other variables were included in the analysis. The effect of education was reduced most dramatically when media exposure was included in the analysis. In that case, only the wife's age surpassed media exposure, and the effect of the wife's education tied with cost sensitivity and the perceived utility of children in importance. The husband's education was even less important. Summary The relation between education and all the fertility regulation variables is the most direct and consistent observed in this study. There are very few deviations from the expected pattern. In Latin America, even Buenos Aires has the expected relations although not all of them are statistically significant. Only Thailand appears to be anomalous, and even here contraceptive knowledge, although not monotonically related to education, does have a significantly positive regression coefficient. Nonmetropolitan West Malaysia was atypical with respect to attitudes, but not to knowledge. In general, education increases contraceptive knowledge in a stronger and more uniform manner than it improves attitudes toward contraception. In addition, scarce evidence shows that education uniformly increases communication between husband and wife. Edu- 11. Evidence on use in Table 5.4 is obtained from multiple correlations which did not include the intervening variables of knowledge, attitude, motivation, or desired family size. In those calculations education did have a significant effect. Table 5.7. Percentage Not Using Contraception of Those Who Want No More Children, for Married Womiien 20 to 39 Years Old, by Wife's Age and Education Thai- Thai- Wife's age \West Mexico Philip- land land (years) and Ankara' Calcutta Delhi India Jakarta Korea Malaysia City pines Taiwan Taiwan' (urban) (rural) education 1966 1970 1968-69 1970 1968 1971 1966-67 1971 1968 1967 1970 1970 1969 20 to29 64 48 33 80 89 73 87 68 84 56 45 49 87 No formal 79 59 56 89 100 78 92 -' 95 63 53 -' -2 Primary 53 -_ 23 67 92 66 87 73 87 56 48 51 86 Junior high -e 52 15 _d 78 85 -c _ 74 39 16 45 Senior high - 2 26 12 54 67 69 28 11 and over 30 to 39 53 49 27 80 89 58 80 75 82 45 33 39 88 No formal 65 68 44 35 95 72 89 92 96 54 42 54 94 Primary 45 44 19 77 89 53 72 77 85 42 31 36 87 Junior high 38 37 10 -_d 89 47 43 44 76 26 17 31 -c Senior high -' 23 7 46 45 45 32 -' 62 21 6 30 and over 20 to 39 58 49 29 80 89 60 82 72 83 47 36 41 88 No formal 71 65 48 86 97 73 89 93 95 56 44 56 96 Primary 49 45 20 73 91 55 77 75 86 45 35 40 87 Junior high 41 43 11 __d 80 56 54 39 75 30 17 36 Senior high 23 26 9 49 61 52 37 -' 64 23 8 23 and over Sample size' 310 511 3,009 5,195 367 795 1,302 234 13,776f 2,313 1,521 543 370 Note See Appendix 1, items 3 and 6 of source for definitions Sterilization included as contraception except in Belgium and Great Britain a Defined as of those wanting no more children, percent never used contraception b Taiwan 1970 data refer to ages 22 to 39 c. Less than 20 cases in the category d No such category e. Sample size based only on the women who want no more children f. Frequency is weighted as follows urban respondents X 4 and rural respondents X 12. Source R Freedman and L Coombs, Cross Cultural Comparisons: Data on Two Factors in Fertility Behavior (New York: The Population Council, 1974). Table 5.8. Multiple Correlations between the Index of the Use of Contraception and Sets of Explanatory Variables with Significant Partial Correlations from a Thirty-seven Variable Regression Analysis Location Ecuador Variable Buenos Rio de San Mexico Panama Guaya- Aires Janeiro Bogotd Jose City City Caracas Quito quit 00 Multiple correlation coefficients (R) Total, 37 variables, all categories 0.408 0.487 0.637 0.553 0.609 0.536 0.535 0.592 0.570 Sociodemographic variables (13) 0.213 0.285 0.450 0.372 0.376 0.333 0.362 0.387 0.367 Social psychological variables (7) 0.343 0.463 0.606 0.504 0.597 0.495 0.503 0.555 0.529 Ideals of reproduction variables (5) 0.144 0.101 0.145 0.131 0.167 0.070 0.174 0.072 0.106 Religiosity variables (2) 0.123 0.067 0.136 0.156 0.082 0.035 0061 0.155 0.060 Family structure variables (3) 0.040 0.056 0081 0.080 0.132 0.128 0.174 0.069 0.120 Modernization variables (6) 0.132 0.209 0.375 0.268 0.264 0.207 0.291 0.290 0.247 Significant partial correlations Index of contraceptive knowledge 0.263 0.340 0.407 0.334 0 445 0.386 0.330 0.412 0.402 Index: couple desire for more children -0.088 -0.069 -0.128 -0.112 -0.061 -0.163 -0.104 -0.063 -0.084 Motivation index -0.076 -0.105 -0.071 -0.107 -0.071 -0.076 -0.086 -0.094 Index attitude toward family planning 0.116 0.059 0.073 0.067 0.088 Frequency of communion 0.072 0.110 0.089 0.113 Ideal family size-self 0.059 0.086 0.094 -0.062 Expenditures of the family 0.058 0.083 0.130 Fear of child mortality 0.055 0.072 0.060 Ideal interval-marriage to first child 0.057 -0.070 0.053 Educational attainment of husband 0.051 0.055 Magazine reading 0.067 0.063 Authority structure in family 0.055 0.071 Index: companionship in marriage 0.051 Index: self-referral -0.075 0.081 Place of birth of husband -0.074 Ideal interval between births 0.062 Index. legitimacy (religious stand) with respect to family planning 0.098 Ideal age to have last child -0.074 Marital status -0.062 Social mobility aspiration for son 0.095 Frequency of attendance religious services 0.075 Employment status of respondent -0.058 Social mobility aspiration for daughter -0.054 Additional education of the couple -0.052 Source: CELADE and CFsc, Fertility and Family Planning in Metropolitan Latin America (Chicago: University of Chicago, 1972) p 249. 140 FERTILITY AND EDUCATION: WHAT DO WE REALLY KNOW? cation increases contraceptive use, but this relation is slightly less uniform than that for contraceptive knowledge. Both mass media exposure and education are important in explain- ing knowledge of contraceptive methods in countries with public family planning programs. Even in these cases, however, education continues to have strong effects, and where there are (or were) no such programs, it is even more important. The CELADE study indicates that the effect of education on contra- ceptive use acts through its effects on knowledge, attitude, and moti- vation and perhaps to a lesser extent on access to family planning through higher income in Latin American cities. It is these multiple channels of effects, all in the same direction, that explain why the relation between contraceptive use and education is so uniformly strong. The fact that education is uniformly related to contraceptive use, but not uniformly related to completed fertility may appear para- doxical. However, the literature does not show that contraceptive users systematically have smaller families. They often have larger families. In circumstances where desired family size exceeds natural fertility, education may be associated with higher natural fertility and higher actual fertility. At the same time, only those with higher edu- cation and fertility that equals or exceeds desired fertility may be motivated to use contraception. 6 Summary and Implications Both theoretical and empirical evidence indicate that edu- cation in the poorest regions may increase the ability to conceive and carry conceptions to successful live births. In the short run, this in- crease would tend to increase actual fertillty. In the long run, how- ever, the positive initial effect of education on fertility may become negative. This reversal occurs in part simply as a result of recognizing the increased ability to have live births and the survival of those births. Thlis rccognition and the adjustment to new conditions take time, and this time lag explains why fertility often rises before it begins to fall (Leibenstein, 1977). In this chapter the empirical cvidence and the theoretical model supporting the above conclusion are summarized, and the research and policy implications of the re- sults are discussed. Summary Several recent reviews on the determinants of fertility have con- cluded that the inverse relation between education and fertility is one of the most consistent and best documented in the literature. 141 142 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? Such statements tempt policymakers to use educational policies to speed the reduction of fertility. The fairly extensive review of the evidence in this study, however, shows that the relation between education and fertility is not always inverse. The earlier generaliza- tions about such a relation probably resulted from a scarcity of data in the poorest, least literate societies and in rural areas where the inverse relation is less likely to occur. Aggregate and individual studies Population policy is concerned with the rate of population growth and thus with aggregate levels of fertility as measured by crude birth rates. Therefore, from a policy perspective the relation between the aggregate level of education in a country and the crude birth rate is of greatest interest. Such aggregate relations are almost in- variably inverse when countries with different educational levels are compared. The factors causing such relations are, however, highly uncertain given the many other factors that are associated with vari- ous education levels. If the age structure of the population and the level of income are controlled, then statistically significant inverse relations are only observed in less than 60 percent of the cases. In addition, when cross-regional data from within developing countries are used and the age structure and extent of urbanization is con- trolled, there are statistically significant inverse relations in less than 60 percent of the cases, and in some cases significantly positive rela- tions are observed. These aggregate data tend to indicate that inverse relations are less likely in the least developed countries. Unfortunately, given the lack of well-developed national statistical systems in the poorest countries, it is impossible to find enough aggregate studies relating education and fertility in these countries. To explore the relation of education and fertility further, studies on the individual level can be examined. Such studies can be based on sample surveys that are too small or too localized to provide cross- regional comparisons. Two kinds of cross-individual studies of the re- lation in developing countries are reviewed: those simply comparing the fertility (adjusted for age) of individuals with different levels of education and multiple regression studies in which fertility is explained using education and income as two of the explanatory variables and introducing a control for age. The first kind of study shows an inverse relation (not necessarily significant) in 49 percent of the cases; the SUMMARY AND IMPLICATIONS 143 second type shows an inverse relation in about 58 percent of the cases. These aggregate percentages are somewhat irrelevant. The overall rate depends on what subgroups are studied because of great differ- ences revealed in both kinds of studies in the relation in different subgroups. Both kinds of studies show that: (a) female education is more likely to be inversely related than male education; (b) educa- tion in urban areas is more likely to be inversely related than in rural areas; and (c) education in countries with literacy rates above 40 percent is more likely to be inversely related than in less literate countries. Thus, the overall percentage of inverse relations depends very much on the distribution of studies in various categories. Thus the data seem to show a definite pattern of nonlinearity. In the least-developed countries, small amounts of education are asso- ciated with higher fertility, but larger amounts are associated with lower fertility. In general, however, there are much fewer studies of rural than urban areas, of males than females, and in less literate than more literate societies. In addition, the studies in the least literate societies are more likely to be based on very small sample sizes. It is expected that the World Fertility Survey will correct this imbalance and will provide a much firmer empirical base for generalizations about fertility at various levels of development. Unfortunately, these surveys, like many of those cited earlier, are restricted to women who have been married and thus may tend to understate the total effect of education on fertility if education increases the proportion of women who never marry. Maodel of education and fertlltyv / The simple replication of studies relating education and fertility in a wider variety of countries with fairly uniform data will help to clarify the situations in which education is inversely related to fertility. Such studies, however, will not clarify how education affects fertility. A model is needed to establish whether the observed association be- tween the variables is in fact causal. In addition, such a model must explain why education affects fertility differently in different situa- tions. An attempt was made to develop such a model in Chapter 2. That model was based on the premise that education does not affect fer- tility directly but acts through many variables that in turn determine fertility. First, the evidence relating these intervening variables to 144 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? fertility was reviewed, and then the relation of education to these intervening variables was reviewed. A simplified version of that model is presented in Figure 6.1. This model traces the impact of education on fertility through the intervening variablcs determining fertility. The evidence reviewed on the relation between education and the intervening variables is summarized in Table 6.1. Fertility is determined by three factors: the biological supply of children, the demand for children, and the regula61i0-hof-J-ility. Each of these factors is in turn influenced directly by many vari- ables, as shown in Figure 2.2, and indirectly by education, as shown in Figure 6.1.1 The model represents a situation in which the cur- rent number of living children is compared with the demand for children. If that number equals or exceeds demand, then it is possible that fertility will be regulated to limit further births. Whether regu- lation is in fact used, however, also depends on the desired family size of husbands and wives; whether they communicate with each other; the relative power of each spouse in decisionmaking; and atti- tudes toward, knowledge of, and access to contraception. Thus, the effect of education on fertility depends on how education affects the three factors that determine fertility: the supply of children, the demand for children, and fertility regulation. Given the multiple channels through which education affects fertility, it is hardly sur- prising that its effect is not uniformly inverse. Data on education and the deterinilianats of fertility Table 6.1 summarizes the data on the relation of education to the intervening variables and the consequent probable effect of education through these variables on fertility and completed family size. The indirect effects in the third column of Table 6.1 are shown in Figure 6.1. Table 6.1 and Figure 6.1 show that education tends to reduce tfehgaem-and for children, as measured by desired family size, by re- ducing preferences for and perceived benefits of children. Education, particularly the husband's education, however also tends to increase the perceived abilityto.aff6rdchildren. This tendency counters the negative effects to some extent buit Toes not outweigh them, since I This model shows only the major blocks of variables contained in Figure 2 2 but adds the education of husband and wife and its impact on the intervening variables to the fertility model of Chapter 2. Figure 6.1. Model of the Effect of Education on Fertility through the Major Intervening Variables I (-I'+) Husband's toc o education I(+,-) Wife's children of demand Aged tonf(-) and supply Husbo f \ / S HusE and's l~~~~ad t dean fors comparison reglaio Wie' children of demand educatio and supply H_11- 146 FERTILITY AND EDUCATION WHIAT DO WE REALLY KNow? Table 6.1. Evidence Supporting the Relation between Education and Fertility through the Intervening Variables Results Probable relation Not Relation of education Supporting supporting of education through the (number (number Variable and variable variable of cases) of cases) Potential supply of births - - Probability of marrying inverse - 6 5 Age of marriage direct - 59 12' Health direct + 2 0 Lactation inverse + 6 0 Postpartum abstinence inverse + 2 0 Infant or child mortality inverse - 16 7 Demand for children (desired family size) inverse - 17 8' Preference for children inverse Ideal family size - 20 7' Desired number ofsons - 8 1 Perceived benefits of children inverse - 17 2 Perceived costs of children direct - 2 0 Perceived ability to afford children direct + 9 3 Fertility regulation (contraceptive use) direct - 26 11 Attitudes toward birth control direct - 28 4 Knowledge of birth control direct - 28 1 Husband-wife communication direct - 9 0 a. Relation of male education to the vanable is much weaker than that of female education. SUMMARY AND IMPLICATIONS 147 desired family size is generally negatively related to fertility. Education also reduces the number of births needed to achievc a particular de- sired family size by lowering infant and child mortality. The evidence is also very strong that education increases contraceptive use by im- proving attitudes toward, and knowledge of, contraception. Education has multiple, partially offsetting effects on the potential biological supply of births or fecundity. I se trs with the poorest health and nutrition combined with traditional reliance on contraceptive practices such as lactation and postpartum abstinence, e le?Ti~Ti wouldybe most likely to have a strong poieffct on fecundity. - Tge1em-ot imediate factor countering this increase in fecundity is probably the effect of education on the age of marriage.-Eeimale- educatioQnseems to reduce the years marricd by raising the age of marriage and, at least in some countries bv reducing the probability o marriage. If the marriage age is very young, however, raising that age by a year or so may have little effect on fertility, since few very young women give birth because of thc high incidence of adolescent sterility. Only when the age of marriage is in the late teens can further delays in the age of marriage be expectcd to reduce fertility. The effcct of education on the fecundity of married women, how- ever, appears to operate in a different direction by improving the health of women sufficiently to increase their chances of conceiving and their ability to carrv births to term. In addition, more educated women tend to give up traditional behavior suchi as pro1ongec lacta- tion and postpartum abstinence which tend to suppress fertility. Increased fecundity can, of course, be offset by' increased contra- ceptive use, but therc arc very good reasons for expecting contracep- tion use to lag behind the increase in fecundity. The supply of births and the survival rate must first increasc morc thani the increase in demand resulting from a greater ability to afford children. Oncc this point has been reached, individuals will need time to adapt to the new condition or, even in some cases, to understand that the changes necessitate changes in behavior. Since the more cducated tend to per- ceive changes in their environment more accurately (for examlple, changes in child mortality), education should reduce the adjustment lag. Therefore, the effect of education on fertility through the inter- vening variables tends to be negative except for possible effects through natural fertility (through biological and behavioral factors) and the 148 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? ability to afford children. These positive effects appear to be stronger as a consequence of the husband's education than of the wife's edu- cation. L&ImPlications for Further Research This review suggests that several kinds of research are needed to clarify these issues. First, more studies need to be done relating edu- cahon and age-adjusted fertility in certain under-researched areas. Sccond, the effect of cducation on several vanables that are important in determining fertility has not been studied sufficiently. Third, little or no work has been done to determine which of the various aspects of the education process are most important in reducing fertility. Fourth, the model presented above needs to be tested in its entirety to determine the relative importance of the many channels through which education affects fertility. More work in under-researched areas There is not enough research relating the age-adjusted fertility and the education of individuals in the poorest countries and in rural areas within most countnes. In addition, not enough work has explored the relation of education and fertility for males. The World Fertility Survey should provide the data needed for such studies in the poorest countries and in rural areas. In addition it will provide data on the husband's level of education and its effect on actual fertility and the wife's attitudes, but only a few of the WFS studies will interview husbands about their own attitudes. Such studies will help to establish if the patterns described in this review are correct. Explanation of observed patterns More research is needed to understand why these different patterns are observed. The simplest kind of work should examine the effect of education on the intervening variables separately for men and women in urban and rural areas. The most important variables that need to be examined in this detail are age of marriage, ideal and desired family size, the biological supply of children, and contraceptive knowledge and attitude. It is expected that such research will result in several conclusions. SUMMARY AND IMPLICATIONS 149 (a) If a large part of the effect of female education on age of marriage results from the opcning up of attractive, well paying jobs that are not compatible with marriage, then female cducation will probably have little effect on age of marriage of those in the rural arcas where such jobs do not exist Education, however, may raise the overall age of marriage of the country, even in the rural areas, by encouraging migration of educatcd women to the urban arcas where such jobs are availablc. (b) Education may have different cffccts on ideal and desired family size in urban and rural areas because the costs and benefits of children may differ in these circumstances. (c) ilhe fertility regulation variables showed very consistent rela- tions with education, but these data were gencrally not broken down by urban or rural residence. The differcnces in contraceptive use be- tween those with various levels of education tended to bc larger for cities than for whole countries, but this is higlhly tentativc. (d) In addition to fairly simple comparisons of the Iinks of educa- tion to intervening variables, more complex forms of research are needed. The cffect of education on the biological supply of children is not well understood. More research is nceded on finding mcasures of fecundity and rclating these to cducation. Education of males and females may have different effects on health and thus on the potential supply of births and children. Education may improve health througlh access to medical care and better nutrition given by higher incomc. If this is the case, male education may have stronger effects on the supply of childrcn than female education. Alternatively, education may affect health by giving individuals better knowledge of good health and nutrition practices, here, female education may be more important. Different effects of factors For the relation between fertility and education to generate policy implications, it is necessary to know not only the extent to which other factors can substitute for the effcct of cducation, but also what charactenstics of education decrease fertility. It may be that education itself has no effect, but that the educational system selects out individuals with certain background characteristics-for example, intelligencc, ambition, and high socioeconomic status-and thcse charactenstics may lead to lower fertility even if higher education 150 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? were not obtained. Alternatively, education may provide explicit skills, such as literacy and numeracy, which result in lower fertility either througlh better job opportunities or through improved abilities to acquire new information and to use complicated technologies. Education may primarily change attitudes, resulting in more modem attitudes toward the control of one's life, the possibilities of social mobility, or the proper roles of men and women. Education may pro- vide explicit knowledge that will result in lower fertility. Finally, education, particularly secondary education, may serve as a simple alternative to early marriage in societies where there are very few alternatives. Thlese various effects of education were sketched in Figure 1.1. The most important relations that need to be explored are those between: (a) female education, market opportunities, and women's wages and fertility and age of marriage; and (b) education and fertility regulation. Although much work has been done on female labor participation and fertility, both positive and negative relations have been shown. In addition, it has been shown that while female labor participation incrcases the age of marriage and proportion of woinen never married in Eastern Europe, the Middle East, and Asia, it has no such effect in Westem Europe and English overseas areas.2 Both of these factors suggest that it is not simply labor participation that must be con- sidered, but the compatibility of work with marriage and childbearing and the wages in that work. Compatibility of work and marriage or childbearing depends on many factors: social definitions of appro- priate roles, location of work, rigidity of hours, cost of child care, availability of contraceptives, and so forth. Wages are important be- causc economic models suggest that fertility, marriage, and labor supply are jointly determined by the wife's market wage. Both the cffcct of education on wages and on the compatibility of work and family rolcs and the cffcct of thcse variables on fertility need to bc explored more fully. One possible explanation of the interaction be- tween the individual's level of education and the aggregate level of litcracy and fertility is that as long as ovcrall female literacy or edu- cation is low, substitutes for the mother's time are cheaply available, so that more educated women can enjoy the benefits of more educa- tion, such as market work or other alternative activities, without reducing fertility. 2. The small effect that exists is negative in thic Wcst. SUMMARY AND IMPLICATIONS 151 The strong relation between education and fertility regulation also needs to be explored more fully if meaningful policies are going to be developed with respect to education. Does this relation result simply from knowledge and attitude changes or from changes result- ing from status and market opportunities? Among the major ques- tions to be answered is the extent to whiclh, mass-media campaigns and free family planning services can substitute for education in in- creasing contraceptive knowledge and use. These issues are fairly complex. Macro relations between education, fertility regulation, and contraceptive availability and mass-media content nced to be estab- lished as well as micro relations that include data on community level family planning variables. Testing of a complete model \All-of-flYe -abo~vefa-c-trs pro~bably operate to some cxtent. If educa- tion is to be considered a policy instrument to reduce fertility, the Telative importance of the various aspects of education must be lknown. In addition, since the various characteristics of education robably operate differently on the various intervening variables, it is tecessary not only to know these different effects, but also to know fhe relative effect of the intervening variables on completed fertility. path analysis of the system presented in Figure 6.1 would provide uch information. Tle amount of data neccssary for such a model is, towevcr, enormous, and therefore such a grandiose model probably will not be tested in its entirety. Implications for Policy The rescarch needed to further explore the relation between educa- tion and fertility is quite substantial. 'lTherefore, the policy conclusions that can be drawn from the existing work must be fairly tentative. First, education canniot be cxpected to automatically reduce fertility in all circumstances. In particular, in the poorest and least-literate societies, small amounts of cducation may actually lead to higlher fertility initially. But there is tentative evidence that over time, cdu- cation ultimately will reduce fertility. Second, increasing female educa- tion will bc morc likely to reduce fertility than increasing male 152 FER'I ILITY AND EDUCATION : WIHAT DO WE REALLY KNOW? education. Tllird, education is more likely to reduce fertility in urban than in rural areas. The major policy dilemma posed by the evidence presented here is what policy should be pursued whlere education is unlikely to reduce fertility immediately and may in fact increase fertility in the short run. If, as hypotlhesizcd here, increases in education increase fcrtility by improving the health of women, increasing the perceived ability to afford children, and reducing adherence to traditional contraccp- tive practices such as prolonged lactation and postpartum abstinence, then it seems that even if education is not increased, fertility will rise from tlhese factors as a result of any program to improve the well-being (health and income) of individuals and by the very process of modernization, whichl causes traditional behavior to be abandoned. Since the tendency for fertility to increase from these causes seems to be the inevitable product of developmcnt, the appropriate policy should be to minimize the time lag bctween the factors increasing fertility and the countervailing forces whlicl tend to reduce it. Educa- tion seems to be one factor that might minimize such a lag. Once desired family sizc falls, it is quite evident from the litcrature that education enables pcople to bettcr achieve thcse smaller sizes. To design the best educational strategy, it is not cnough to say that education should be increased despite the possibility of its imme- diately incrcasing fertility in certain circumstances. It is also necessary to know what kind of education should be increased. Tlhc direct finding of this review is that female education should receive priority. It is unclear, however, whetlher a broadly based program of elementary education should be recommended or a narrower program of second- ary education. In the least literate countries secondary education has more immcdiate negativc effects on fertility, but these fertility reduc- tions will have a fairly high cost and will imply a more limited dis- tribution of educational benefits, whiclh is perhaps unacceptable. Policies that concentrate on mass literacy will result in initially higher individual fertility but will also raise aggregate literacy, which appears to shift the fertility increase to a fertility decrease in the long run. Sources INTRODUCTION AND SUMMARY Encatnaci6n, Jose, Jr. "Fertility and Labor Force Participation: Philip- pincs 1968 " The Philippine Review of Business and Economics, vol. 11 (December 1974). Mason, Karen, and others. Social and EconlomIic Correlates of Family Fertility: A Survey of the Evidence. Researchl Triangle Park, N.C. Research Triangle Institute, 1971. McGreevev, William P., and Nancv Birdsall. The Policy Relevance of Recent Social Research on Fertility. Washington, D.C.: Interdisci- plimary Communications Program, Smithsonian Institute, 1974. Simiion, Julian L. The Effects of Income on Fertility. Chapel Hill: North Carolina Population Center, 1974. CHAPTER 1. EDUCATION AND FERTILrITY EMPIRICAL EVIDENCE Adelman, Irma. "An Econometric Analvsis of Population Growtlh." Amer- ican Economic Reviev, vol. 53 (June 1963), pp. 314-39. Amani, M. Report on the Study of the Impact of Literacy and Education on Fertility and Family Planning. Tehran. Iran Institute for Social Studies and Research, University of Tehran, 1971. '53 154 FERTILITY AND EDUCATION : WHAr DO WE REALLY KNOW? Anker, Riclhard. "An Analysis of Fertility Differentials in Developing Countries." International Labor Review (ILO) Population and Em- ployment Working Paper, no. 16. April 1975. Anker, Riclhard. "Thle Effect of Group Level Vanables on Fertility in a Rural Indian Sample." Journal of Development Studies, vol. 14, no. 1 (October 1977), pp. 63-76. Becker, Gary. "A Tlheory of the Allocation of Timc." Economic Journal, vol. 75 (Scptember 1965), pp. 493-517. Becker, Gary. "An Economiiic Analysis of Fcrtility." Demographic and Econzomitic Chanige in Developed Countries. Princeton: National Bureau of Economic Research, 1960. Ben-Poratli, Voram. "Economic Analysis for Fertility in Israel: Point and Counterpoint." Journal of Political Economy, vol. 81 (Marchi/April supplement 2, 1973), S202-233. Bogue, Donald. Principles and Demography. New York: Wiley, 1969. Caldwell, Jolhn C. Population Growth and Family Chantge in Africa: The New Urban Elite in Ghana. Ncw York: Humanities Press, 1968. Carelton, Robert 0. "Labor Force Participation: A Stimulus to Fertility in Puerto Rico." Demography, vol. 2 (1965), pp. 233-39. CELADE (United Nations Regional Center for Dcmograpllic Training and Researclh in Latin America) and CFSC (Community and Family Study Center). Fertility and Faintly Planniniig in Metropolitan Latin Amnerica. Chicago. Unilvcrsity of Chicago, 1972. Chernichovskoy, D "F'Fertilitv Bclavior in Developing Economiiies: An Investmcnt Approacl." Paper presented at IUSSP Scminar on I-louse- hlold Models of Economic-Demograplhic Decision-Making, Novcm)bcr 4-6, 1976, Mexico Citv Proccssed. Cliemiclhovskv, D. "Somc Socio-Econoilnc Aspects of Fertility Behavior in Nortlhcast Brazil A Note." October 1976. Processed. Cliung, Bom Mo, and otlhers. Psychological Perspective: Family Planning in Korea. Scoul: Hollym Corporation, 1972. Cochlrane, Susan. "Thle Detenninants of Clhildren Ever Born in Cross Regional Data on Thailand." Spring, 1978. Processed. Cochrane, Susan, Bal Gopal Baidya, and Jcnnic I-lay. "Memo oni Fer- tility in the Parsa Pretcst in Rural Nepal." World Bank, Division on Population and hluman Resourecs, Julv 1977. DaVanzo, Julie. The Determinants of Family Formationz in Chile, 1960: An Econometric Study of Female Labor Force Participationz, Marriage, and Fertility Decisionis. No. R830-AID, Santa MoInica, Califonila: Rand Corporation, 1972 SOURCES 155 Davidson, Maria. "A Comparative Study of Fcrtility in Mexico City and Caracas." Social Biology, vol. 20 (December 1973), p. 460-96. Del Rio, A. G., W. W. Hicks, and S. R. Jolinson. "Socio-Economic Deteniinants of Fertility in Mexico- An Analysis of Change in Struc- tural Relationslips 1950-1970." Processed. Dow, Thlomas E., Jr. "Fertility and Family Planning in Sierra Leonc." Studies in Family Planning, vol. 2 (AAugust 1971), pp 153-65. Drakatos, Constantine. "Tlhe Dcteniiinants of Birthi Rate in Developing Countries: An Econoietric Studv of Greece." Econjomizic Development and Cultural Change, vol. 17 (July 1969), pp. 596-603. Ekaneim, Ita. "A Furthier Note on thlc Rclationslhip betwveen Economic Development and Fertility." Demography, vol. 9 (August 1972), pp. 383-98. Ekancimi, Ita. "Correlates of Fcrtility in Eastcrn Nigeria." Nigerian jour- nal of Econiomiiic and Social Studies (Ibadan), vol. 16 (Marclh 1974), pp.115-27. El-Badry, M. A., and Hlanna Rizk. "Rcgional Fertility Diffecrcnces anmong Socio-Economic Groups in the United Arab Rcpublic." WVorld lPopu- lation Conferenice, 1965, vol. IV. New Yorkl United Nations, 1967. Encarnaci6n, Jos6, Jr. "Fertility and Labor Forcc Participationi Plhilip- pincs 1968." 'T'he Philippine Review of Businiess and Econzomiilcs, vol. 11 (Decemlbcr 1974). Ewbank, Douglas. "Indicators of Fertility Lcvcls in Tanzania: Diffcrcn- tials and Trcnds in Reported Povcrtv and Ciildlcssncss." Papcr prc- sented at the Annual Mcting of the Population Association of America, April 21-23, 1977, St. Louis, Missouri. Proccsscd. Farooq, C. M., and B. Tuncer. "Fcrtilitv and Labor Force Participation: Plhilippines 1968 " 'li'he Philippine Reviewv of Business and Econoillics, vol. 11 (Deccmbcr 1974). Frcediman, Ronald, and Bernard Bcrcison. "Tlic Rccord of Family Planning Programs." Studies tin Family P'lanninig, vol. 7 (January 1977), pp.1-40. Friedlander, Stanlcv, and Morris Silvcr. "A Quantitative Study of the Determiniants of Fertility Bcihavior." Demilography, vol. 4, no. 1 (1967), pp.30-70. Gcindcl, Murray, and othicrs. "Fcrtility and Economiiic Activity bf Womncn in Guateiiiala City, 1964." Demography, vol. 7 (August 1970), pp. 273-86. Goldberg, David. "Rcsidential Location and Fcrtility." Population and Developmienit: The Search for Selected Itnterventionis. Edited by Ronald Ridkcr. Baltimore Johns hlopkiins Univcrsity' Press, 1976. 156 FERTILITY AND EDUCA'IION : WIIAT DO WE REALLY KNOW? Goldstein, Sidney. "The Influence of Labor Forcc Participation and Edu- cation on Fertility in Tlailand." Population Studies, vol. 26 (Novem- ber 1972), pp. 419-36. Gregory, Paul, John Campbell, and Benjamin Cheng. "Differcnccs in Fertility Determinants: Developed and Developing Countries." journal of Development Studies, vol. 9 (July 1973), pp. 233-41. Hammond, John L. "Two Sources of Error in Ecological Correlations." American Sociological Review, vol. 38 (December 1973), pp. 764-77. Heer, David. "Economic Development and Fertility." Demography, vol. 3, no. 2 (1966), pp. 423-44. Heller, Peter S. "Tlhc Interaction of Childlhood Mortality and Fertility in West Malaysia: 1947-1970." Center for Economic Dcvelopment, The University of Michigan, Discussion Paper no. 57, September 1976. Hicks, W. Whlitney. "Economic Devclopment and Fertility Change in Mcxico, 1950-1970." Demography, vol. 11 (August 1974), pp. 407-22. Hull, Tcrence, and Valerie Hull. "Thc Relation of Economic Class and Fertility: An Analysis of Some Indonesian Data." Population Studies, vol. 31 (March 1977), pp. 73-87. lutaka, S., E. W. Bock, and W. G. Varnes. "Factors Affccting Fer- tility of Natives and Migrants in Urban Brazil." Population Studies, vol. 25 (March 1971), pp. 55-62. Jaffe, A. J., and K. Azumi. "Thc Birth Ratc and Cottage Industries in Underdeveloped Countries." Economic Development and Cultural Change, vol. 9 (October 1960), pp. 52--63. Janowitz, Barbara. "An Analysis of the Impact of Education on Family Size." Demography, vol. 13 (May 1976), pp. 189-98. Kasarda, Johln D. "Economic Structure and Fcrtility: A Comparative Analysis." Demography, vol. 8 (August 1971), pp. 307-18. Kellcv, Allen. "Interaction witlh Economic and Dcmographic 1Household Behavior." Papcr prcscnted at National Bureau of Economic Rescarch Conference on Population and Economic Change in Less Developed Countries, September 30-October 2, 1976. Philadelphia, Pennsylvania. Processed. Khan, M. Ali, and Ismail Sirageldin. "Education, Incomc and Fertility in Pakistan." Paper presented at the Applied Research Institutc, Uni- versity of Karachi and Research Department of United Bank, Ltd., Scptember 30, 1975. Processed. Kirk, Dudley. "New Demographic Transition?" Rapid Population SOURCES 157 Growth. National Academy of Sciences. Baltimore: Johns Hopkins Press, 1971. Knodel, John, and Visid Prachuabmolb. T'he Fertility of T-hai Women. Bangkok; Institute of Population Studies, Chulalongkorn University, 1973. Knowles, James C., and Richard Anker. "Economic Determinants of Demographic Behavior in Kenya." Intemational Labor Review (ILO) Population and Employment Working Paper no. 28, Decembcr 1975. Kocher, James E. "Rural Development and Fertility Change in North- east Tanzania." Paper presented at the Annual Meeting of the Popu- lation Association of America, April 21-23, 1977, St. Louis, Missouri. Processed. Kogut, Edy Luiz. "The Economic Analysis of Fertility: A Study of Brazil." International Labor Organization (ILO) Population and Em- ployment Working Paper no. 7, September 1974. Lal, R. B. "Literacy and Population Growth." Population Review, vol. 12 (January-Dccember 1968), pp. 55-59. Li, W. L. "Temporal and Spatial Analysis of Fertility Decline in Taiwan." Population Studies, vol. 27 (March 1970), pp. 97-104. Maurer, K., and others. Marriage, Fertility and Labor Force Participation of Thai Women: An Econometric Study. No. R829-AID/RF. Santa Monica, California: Rand Corporation, 1973. McCabe, James L., and Mark R. Rosenzweig. "Female Employment Creation and Family Size." Population and Development. Edited by Ronald Ridker. Baltimore: Johns Hopkins Univcrsity Press, 1976. McGreevey, William P., and Nancy Birdsall. The Policy Relevance of Recent Social Research on Fertility. Washington, D.C.: Interdisci- plinary Communications Program, Smithsonian Institute, 1974. Merrick, Thomas. "Interregional Differences in Fertility in Brazil, 1950- 1970." Demography, vol 11 (August 1974), pp. 423-40. Miro, Carmen, and Walter Mertens. "Influences Affecting Fertility in Urban and Rural Latin America." Milbank Memorial Fund Quarterly, vol. 46, no. 3, part 2 (July 1968), pp 89-120. Ohadike, Patrick. "Tlhe Possibility of Fertility Change in Modem Africa: A West African Case." Africant Social Research, vol. 8 (December 1969), pp. 602-14. Palmore, James A., Jr. "The West Malaysian Family Survey, 1966-1967." Studies in Family Planning, vol. 40 (April 1969), pp. 11-20. Repetto, Robert. "The Relationship of the Size Distribution of Income to Fertility and the Implications for Development Policy." Population 158 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? Policies and Economic Development. World Bank Staff Report. Edited by Timothy King. Baltimore: Johns Hopkins University Press, 1974. Rizk, Hanna. "Trends in Fertility and Family Planning in Jordan." Studies in Family Planning, vol. 8 (April 1977), pp. 91-99. Robinson, W. S. "Ecological Correlations and the Behavior of Indi- viduals." American Sociological Review, vol. 15 (June 1950), pp. 351-57. Rosenzweig, Mark R. "Rural Wages, Labor Supply and Land Reform: A Theoretical and Empirical Analysis." Discussion Paper no. 20. Re- search Program in Developmcnt Studies, Princeton University, De- cember 1976. Rule, J. R. "Fertility Differentials in India: Evidence from Rural Back- ground." Milbank Memorial Fund Quarterly, vol. 41 (April 1963), pp. 183-99. Russett, B. M. and others. World I-Iandbook of Social and Political Indi- cators. New Haven: Yale University Press, 1964. Safilios-Rothschild, Constantina. "Sociopsychological Factors Affecting Fertility in Urban Greece: A Preliminary Report." Journal of Marriage and the Family, vol. 31 (August 1969), pp. 595-606. Schultz, T. Paul. "An Economic Model of Family Planning and Fer- tility." journal of Political Economy, vol. 77 (March/April 1969), pp. 153-80. Schultz, T. Paul. Disequilibrium and Variation in Birth Rates over Space and Time: A Study of Taiwan. Santa Monica: Rand Corporation, September 1972. Schultz, T. Paul. "Fertility Pattems and Their Determinants in the Arab Middle East." Economic Development and Population Growth in the Middle East. Edited by Charles A. Cooper and Sidney Alexander. New York: Elsevier, 1972. Siever, Daniel. "Comment on W. Whitney Hicks. Economic Develop- ment and Fcrtility Change in Mexico, 1950-1970." Demography, vol. 13 (February 1976), pp. 149-52. Simmons, Alan B., and Johanna de Jong. "Education and Contraception in Rural Latin America." Processed. Srinivasan, K. "A Prospective Study of the Fertility Behavior of a Group of Married Women in Rural India-Design and Findings of the First Round of Enquiry." Population Review, vol. 11 (July-December 1967), pp. 46-60. Stycos, J. Mayone. Hluman Fertility in Latin America. Ithaca: Comell University Press, 1968. Stycos, J., and Robert Weller. "Female Working Roles and Fertility." Demography, vol. 4, no. 1 (1967), pp. 210-17. SOURCES 159 Timur, Serim. "Demographic Correlates of Woman's Education: Fer- tility Age at Marriage, and the Family." Paper presented at 18th General Conference of the International Union for the Scientific Study of Population, August 8-13, 1977, Mexico City. Traina, Frank J., and Herman D. Bontrager. "Statistical Determinants of Fertility Decline in Costa Rica." Paper presented at the Annual Meeting of the Population Association of America, April 21-23, 1977, St. Louis, Missouri. Processed. Willis, Robert J. "A New Approach to the Economic Theory of Fertility Behavior." Journal of Political Economy, vol. 81 (March/April 1973), pp. 514-69. World Bank. "Population Policies and Economic Development." World Bank Staff Report. Baltimore: Johns Hopkins University Press, 1974. World Fertility Survey. "Nepal Fertility Survey, 1976, First Report." Kathmandu, 1977. World-Fertility Survey. "The Survey of Fertility in Thailand." Country Report, vol. 1. Institute of Population Studies. Bangkok, 1977. Yaukey, David. "Some Immediate Determinants of Fertility Differences in Lebanon." Marriage and Family Living, vol. 25 (February 1963), pp. 27-3 3. Zarate, Celvan. "Fertility in Urban Areas of Mexico- Implications for the Theory of Demographic Transition." Demography, vol. 4, no. 1 (1977), pp. 363-73. CHAPI ER 2. THE THEORETICAL DETERMINANTS OF FERTILITY Arnold, Fred, and others. The Value of Children: A Cross-National Survey. Vol. 1, Introduction and Comparative Analysis. Honolulu: The East-West Center, 1975. Baird, Dugald. "Variation in Fertility Associated with Changes in Health Status." Public Health and Population Change. Pittsburgh: University of Pittsburgh Press, 1965. Bean, Frank, Susan Cochrane, Howard Savage, and Charles Wood. "The Effect of Income on the Demand for and/or Supply of Children." Research in Population Economics. Edited by Julian Simon. Vol. 1. Greenwich, Conn.: JAI Press, 1976. Bindary, A., and others. "Urban-Rural Differences in the Relationship between Women's Employment and Fertility." journal of Biosocial Science, vol. 5 (1973), pp. 159-67 Bulatao, Rodolfo. The Value of Children: A Cross-National Survey. Vol. 2, The Philippines. Honolulu: The East-West Center, 1975. i6o FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? Butz, William P., and Jean-Pierre Habicht. "The Effects of Nutrition and Health on Fertility: Hypotheses, Evidence and Intervention." Population and Development: The Search for Selective Interventions. Edited by Ronald Ridker. Baltimore: Johns Hopkins University Press, 1976. Cain, Glen, and Adriana Weinenger. "Economic Determinants of Fer- tility: Results from Cross-Sectional Aggregate Data." Demography, vol. 10 (May 1973), pp. 205-21. Carelton, Robert 0. "Labor Force Participation: A Stimulus to Fertility in Puerto Rico." Demography, no. 2 (1965), pp. 233-39. Cho, Lee-Jay, and Robert D. Rutherford. "Comparative Analysis of Re- cent Fertility Trends in East Asia." Proceedings at the International Union for the Scientific Study of Population (I.U.S.S.P.), vol. 2, pp. 163-81. Liege, 1973. Cochranc, Susan, and James Cochrane. "Child, Mortality and the De- sired Number of Children and Births." Australian Economic Papers (June 1974). DaVanzo, Julie. The Determinants of Family Formation in Chile, 1960: T'he Econometric Study of Female Labor Force Participation, Marnage and Fertility Decisions. No. R830-AID. Santa Monica, California: Rand Corporation, 1972. Davidson, Maria. "A Comparative Study of Fertility in Mexico City and Caracas." Social Biology, vol. 20 (December 1973), pp. 460-96. Davis, Kingsley, and Judith Blake. "Social Structure and Fertility: An Analytical Framework." Economic Development and Cultural Change, vol. 4 (1956), pp. 211-35. Duza, M. Badrud, and C. Stephen Baldwin. "Nuptiality as an Area for Policy Intervention: A Comparative Perspective." Adapted from Nuptiality and Population Policy: An Investigation in Sn Lanka, Malaysia and Bangladesh. New York Population Council, 1974. Easterlin, Richard A. "An Economic Framework for Fertility Analysis." Studies in Family Planning, vol. 6 (March 3, 1975), pp. 54-63. Easterlin, Richard A. "Toward a Socio-Economic Theory of Fertility: Survey of Recent Research on Economic Factors in American Fer- tilitv." Fertility and Family Planning: A World View. Ann Arbor: University of Michligan Press, 1969. Easterlin, Richard A., Robert A. Pollack, and Michael L. Wachter. "To- ward A More General Economic Model of Fertility Determination: Endogenous Preferences and Natural Fertility." Presented at the Na- tional Bureau of Economic Research Conference on Economic and Demographic Change in Less Developed Countries, September 30- October 2, 1976, Philadelphia. Processed. SOURCES i61 Encarnaci6n, Jose. "Fertility and Labor Force Participation: Philippines 1968." The Philippine Review of Business and Economics, vol. 11 (December 1974). Freedman, R., and L. Coombs. Cross-Cultural Comparisons: Data on Two Factors in Fertility Behavior. New York: The Population Council, 1974. Freedman, R. and others. "Hong Kong's Fertility Decline, 1961-68." Population Index, vol. 36 (January-March, 1970). Galbraith, John Kenneth. "The Economics of the American Housewife." The Atlantic Monthly, vol. 232 (1973), pp. 78-83. Goldstein, Sidney. "The Influence of Labor Force Participation and Edu- cation on Fertility in Thailand." Population Studies, vol. 26 (No- vember 1972), pp. 419-36. Haas, Paula. "Wanted and Unwanted Pregnancies: A Fertility Decision- Making Model." Journal of Social Forces, vol. 30 (1974), pp. 125-65. Harmon, Alvin. Fertility and Economic Behavior of Families in the Philippines. Santa Monica, California Rand Corporation, September 1970. Heckman, James I., and Robert J. Willis. "Estimation of a Stochastic Model of Rcproduction- An Econometric Approach." Hlousehold Pro- duction and Consumption. Edited by Nestor Terleckyj. New York: National Bureau of Economic Research, 1976 Hill, Rueben and others. The Family and Population Control. Chapel Hill: University of North Carolina Press, 1959. Holsinger, Donald B , and John D. Kasarda. "Education and I-luman Fertility: Sociological Perspectives." Population and Development: The Search for Selective Intervention. Baltimore: Johns Hopkins Uni- versity Press, 1976. Kim, Mo-Im. "Age at Marriage, Family Planning Practices and Other Variables as Correlates of Fertility in Korea." Demography, vol. 1 (Novcmber 1977), pp. 413-28. Mason, Karen and others. Social and Economic Correlates of Family Fertility: A Survey of the Evidence. Research Triangle Park, N.C.: Re- scarch Trianglc Institute, 1971. Maurer, K. and others. Marriage, Fertility and Labor Force Participation of Th1ai Woment: An Econometric Study. No. R829-AID/RF. Santa Monica, California: Rand Corporation, 1973. Mazur, D. P. "Relation of Marriage and Education to Fertility in the USSR." Population Studies, vol. 27, no. 1 (March 1973), pp. 105-15. McCabe, James L., and Mark Rosenzweig. "Female Employment Crea- tion and Family Size." Population and Development. Edited by Ronald Ridker. Baltimore. Johns Hopkins University Press, 1976. 162 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? McGreevey, W. P., and Nancy Birdsall. The Policy Relevance of Recent Social Research on Fertility. Waslhington, D.C.: Interdisciplinary Com- munications Program, Smithsonian Institute, 1974. Michael, Robert T. "Education and the Derived Demand for Children." Journal of Political Economy, vol. 8 (March/April 1973), Supplement: S128-S164. Michael, Robert T., and R. J. Willis. "Contraception and Fertility: Household Production under Uncertainty." Household Production and Consumption. Edited by Nestor Terleckyj. New York: National Bureau of Economic Research, 1976. Michel, A. "Interaction and Family Planning in the French Urban Family." Demography, vol. 4, no. 2 (1967), pp. 615-25. Miro, Carmen, and Walter Mertens. "Influences Affecting Fertility in Urban and Rural Latin America." Milbank Memorial Fund Quarterly, vol. 46, no. 3, part 2 (July 1968), pp. 89-120. Mitchell, Robert E. "Husband-Wife Relations and Family Planning Practice in Urban Hong Kong." journal of Marriage and Family, vol. 34 (Fcbruary 1972), pp. 139-46. Mueller, Eva. "Economic Motivations for Family Limitation." Demog- raphy, vol. 26, no. 3 (November 1972), pp. 383-403. Namboodiri, N. Krishnan. "Some Observations on the Economic Frame- work for Fertility Analysis." Population Studies, vol. 26 (July 1972.), pp. 175-206. Nerlove, Marc. "Household and Economy toward A New Theory of Population and Economic Growth." journal of Political Economy, vol. 82 (March/April 1974), Supplement S200-S218. Nerlove, Marc, and Paul T. Schultz. "Love and Life between the Cen- suses: Model of Family Decision-Making in Puerto Rico, 1950-1960." No. RM-6322-AID. Santa Monica, California: Rand Corporation, 1970. Processed. Palmore, James A., and M. Ariffin. "Marriage Patterns and Cumulative Fertility in West Malaysia: 1966-1967." Demography, vol. 6 (No- vember 1969), pp. 383-401. Ramakumar, S. R., and S. V. S. Gopal. "Husband-Wife Communications and Fertility in a Suburban Communitv Exposed to Family Planning." journal of Family Welfare, vol. 18 (March 1972), pp. 30-36. Repetto, Robert. "Inequality and the Birth Rate in Puerto Rico: Evi- dence from Household Census Data." Research Paper no. 14. Harvard University Center for Population Studies. June 1976. Rosen, Bemard C., and Alan B. Simmons. "Industrialization, Familv and Fertility: A Structural-Psychological Analysis of the Brazilian Case." Demography, vol. 8 (February 1971), pp. 49-69. SOURCES 163 Rosenzweig, Mark R., and Robert Evenson. "Fertility, Schooling and the Economic Contribution of Chlildren in Rural India: An Econometric Analysis."Econometrica, vol. 45 (July 1977), pp. 1065-79. Samuelson, Paul A. "Social Indifference Curves." Quarterly Journal of Economics, vol.70 (February 1956), pp. 1-22. Schultz, T. Paul. "Fertility Patterns and Their Determinants in the Arab Middle East." Economic Development and Population Growth in the Middle East. Edited by Charles Cooper and Sidney Alexander. New York: Elsevier, 1972. Shields, Nwanganga. "Female Labor Force Participation and Fertility: Review of Empirical Evidenec." Draft for Population and Human Resources Division, Development Economics Department. World Bank, February 1977. Simon, Julian L. The Effects of Income on Fertility. Chapel Hill: North Carolina Population Centcr, 1974. Snyder, Donald W. "Tlhe Economic Determinants of Famlly Size in West Africa." Demography, vol. 11 (November 1974), pp. 613-28. Turchli, Boone. "Micro-Economic Theories of Fertility." Social Forces, vol. 54 (September 1975), pp. 107-25. United Nations. The Determinants and Consequences of Population Trends. New York: Unitcd Nations, 1973. Wcller, Robert. "The Employmcnit of \NVives, Role Incompatibility and Fertility. A Study among the Lower and Middlc Class Residents of San Juan, Pucrto Rico." Milbank Meemorzal Fund Quarterly, vol. 46 (October 1968), pp. 507-26. Williams, Annle D. "Rcview and Evaluation of the Literature." Popula- tion, Public Policy and Economic Development. Edited by Michael C. Kceby New York. Praeger, 1976. Yaukey, David, and Timm Thorsen. "Differcntial Femalc Days at First Marriage in Six Latin American Citics." journal of Marriage and thle Family (May 1972),pp. 375-79. CHAPTIER 3. EDUCAIION AND rHE BIOLOGICAL SUPPLY OF CHILDREN: EMIPIRICAL EVIDENCE Adegbola, HI. J. Page, and R. Lesthaeglhe. "Breastfeeding and Post Partum Abstinencc in Metropolitan Lagos." Paper presented at the Annual Meeting of the Population Association of America, April 21-23, 1977, St. Louis, Missouri. Processed. Amani, M. Report on the Study of the Impact of Library and Education on Fertility and Family Planning. Tehran: Institute for Social Sttidies and Research, University of Telran, 1971. 164 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? Baird, Dugald. "Variation in Fertility Associated with Changes in Health Status." Public Health and Population Change. Pittsburgh: University of Pittsburgh Press, 1965. Butz, William P., and Julie DaVanzo. "Contracepting, Breastfceding and Birthspacing in Malaysia: A Model of Decision-Making Subject to Economic and Biological Constraints." A preliminary paper. Rand Corporation, April 1978. Butz, William P., and Jean-Pierre Habicht. "The Effects of Nutrition and Health on Fertility: Hypotheses, Evidence and Intervention." Popula- tion and Development. The Search For Selective Interventions. Edited by Ronald Ridker. Baltimore Johns Hopkins University Press, 1976. Caldwell, John C. Population of Growth and Family Change in Afrrca: The New Urban Elite in Ghana. New York: Humanities Press, 1968. CELADE (United Nations Regional Center for Demographlic Training and Research in Latin America) and CFSC (Community and Family Study Center). Fertility and Family Planning in Metropolitan Latin America. Chlicago: University of Chicago, 1972. Chander, R., and V. T. Palan. Malaysian Fertility and Family Survey 1974: First Country Report. Kuala Lumpur Department of Statistics, 1977. Choinacka, Helena. "Nuptiality Patterns in an Agrarian Society." Popu- lation Studies, vol. 30 (Julv 1976), pp. 203-27. Chung, Bom Mo, and others. Psychological Perspectives: Family Planning in Korea. Seoul: Hollym Corporation, 1972. Cochrane, Susan, Bal Copal Baidya, and Jennie Hay. "Memo on Fer- tility in the Parsa Pretest in Rural Nepal." World Bank, Division on Population and Human Resources, July 1977. DaVanzo, Julie. The Determinants of Family Formation in Chile, 1960: An Econometric Studv of Female Labor Force Participation, Marriage, and Fertility Decisions. No. R830-AID. Santa Monica, California: Rand Corporation, 1972. Dixon, Ruth. "Explaining Cross Cultural Vanations in Age of Marriage and Proportion Never Marrying." Population Studies, vol. 25 (Julv 1971), pp. 215-33. Harmon, Alvin. Fertility and Economic Behavior of Fainilies in the Philippines. Santa Monica, California. Rand Corporation, September 1970. Hawley, Amos, and Visid Prachuabmoh. "Family Growth and Family Planning in a Rural District of Thailand." Family Planning and Popu- lation Programs. Edited by Bernard Berelson and others. Chicago: University of Chicago Press, 1966. SOURCES I65 Heller, Peter S. "Tlhe Interaction of Clhildhood Mortality and Fertility in West Malaysia: 1947-1970." Discussion Paper no. 57. Center for Economic Development, University of Michigan. Scptember 1976. Hull, Terence, and Valerie Hull. "The Relation of Economic Class and Fertility: An Analysis of Some Indonesian Data." Population Studies, vol. 31 (March 1977), pp. 73-87. Jain, Anrudh K. "Socio-Economic Corrclates of Fccundability in a Sample of Taiwanese Women." Demography, vol. 6 (February 1969), pp. 75-90. Jain, Anrudh K., T. C. IHsu, R. Freedman, and M. C. Clhang. "Demo- graphic Aspects of Lactation and Postpartum Amcnorrhca " Demog- raphy, vol. 7 (May 1970), pp. 255-71. Kelley, Allen. "Interaction of Economic and Dcmiograplhic Houschold Behavior." Paper presented at National Bureau of Economic Research Conference on Population and Economilc Chalgc in Less Developed Countries, September 30-October 2, 1976. Philadelplhia, Pennsylvania. Processed. Khalifa, Atef M. "The Influence of Wife's Education on Fertility in Rural Egypt." Journal of Biosocial Science, vol. 8 (January 1976), pp. 53-60. Khan, M. Ali, and Ismail Sirageldin. "Education, Income and Fertility in Pakistan." Paper presented at the Applied Research Institute, Uni- versity of Karachi and the Research Department of United Bank Ltd., September 30, 1975. Processed. Knowles, James C., and Riclhard Anker "Economic Determinants of Demographic Behavior in Kenya." International Labor Review (ILO) Population and Employment Working Paper no. 28, Dccember 1975. Kogut, Edy Luiz. "Thc Economic Analysis of Fertility. A Study of Brazil." International Labor Organization (ILO) Population and Em- ployment Working paper no. 7, September 1974. Lapierre-Adamcyk, E., and T. Burch. "Trends and Differentials in Age of Marriage in Korea." Studies in Family Planning, vol. 5 (August 1972),pp. 255-60. Maurer, K. and others. Marriage, Fertility and Labor Force Participation of Thai Women: An Econometric Study. No. R829-AID/RF. Santa Monica, California: Rand Corporation, 1973. Mott, Frank. "Socio-Economic Determinants of Fertility in a Nigerian Village." 1976. Processed. Nayar, P. K. B. "The Influence of Education on Fertility." Journal of Family Welfare (Bombay), vol. 20 (March 1974), pp. 28-36. 166 rERTILITY AND EDUCATION : WHIAT DO WE REALLY KNOW? Olusanya, P. 0. "Status Differentials in the Fertility Attitudes of Married Women in Two Communities in Westem Nigeria." Economic De- velopment and Cultural Change, vol. 19, no. 4 (July 1971), pp. 641-51. Palmore, James A., and M. Ariffin. "Marriage Patterns and Cumulatlve Fertility in West Malaysia: 1966-1967." Demography, vol. 6 (No- vember 1969), pp. 383-401. Paydurfar, Ali A. "Sociocultural Correlates of Fertility among Tribal, Rural and Urban Populations in Iran." Social Biology, vol. 22 (Sum- mer 1975), pp. 151-66. Shin, Eui Hang. "Economic and Social Correlates of Infant Mortality: A Cross Sectional and Longitudinal Analysis of 63 Selected Countries." Social Biology, vol. 22, no 4 (Winter 1975), pp. 315-25. Singh, K. P. "Child Mortality, Social Status and Fertility in India." Social Biology, vol. 21, no 4 (Winter 1974), pp. 385-88. Sloan, Frank. Survival of Progeny in Developing Countries: An Analysis of Evidence from Costa Rica, Mexico, East Pakistan and Puerto Rico. Santa Monica, California Rand Corporation, 1971. Smith, Peter C. "Tradition and Transition in South and Southeast and East Asian Nuptiality." Paper prcscnted at the 1977 Annual Meeting of the Population Association of Amcrica, April 21-23, 1977, St. Louis, Missouri. Processed. Speare, Alden, Jr. and others. "Urbanization, Non-Familial Work, Edu- cation and Fertility in Taiwan." Population Studies, vol. 27 (July 1973), pp. 323-34. Stockwell, E. G., and B. W. Hutchinson. "A Note on Mortality Correlate of Economic Development." Population Review, vol. 19 (January- December 1975), pp. 46-50. Stycos, J. Mayone. Human Fertility in Latin America. Ithaca: Cornell University Press, 1968. Stycos, J., and Robert Weller. "Female Working Roles and Fertility." Demography, vol. 4, no. 1 (1967), pp. 210-17. World Fertility Survey. "Nepal Fcrtility Survey, 1976, First Report." Kathmandu, 1977. Yaukey, David. "Some Immediate Determinants of Fertility Differences in Lebanon." Marriage and Family Living, vol. 25 (February 1963), pp. 27-33. Yaukey, David and Timm Thorsen. "Differential Female Age at First Marriage in Six Latin American Cities." journal of Marriage and the Family (May 1972), pp. 375-79. SOURCES 167 CHAPTER 4. EDUCATION AND THE DEMAND FOR CHILDREN: EMPIRICAL EVIDENCE Arnold, Fred, and others. The Value of Children: A Cross-National Study. Vol. 1, Introduction and Comparative Analysis. Honolulu: The East- West Center, 1975. Becker, Gary. "An Economic Analysis of Fertility." Demographic and Economic Change in Developed Countries. Princeton: National Bureau of Economic Research, 1960. Bulatao, Rodolfo. The Value of Children: A Cross-National Survey. Vol. 2, The Philippines. Honolulu: The East-West Center, 1975. CELADE (United Nations Center for Demographic Training and Research in Latin America) and CFSC (Community and Family Study Center). Fertility and Family Planning in Metropolitan Latin America. Chicago: University of Chicago, 1972. Chang, Cheng-Tung. "Desired Fertility, Income and the Valuation of Childrcn." International Labor Organization (ILO), Population and Employment Working Paper no. 36, March 1976. Chung, Bom Mo, and others. Psychological Perspectives: Family Planning in Korea. Seoul: Hollym Corporation, 1972. Coombs, Lolagene. Are Cross-Cultural Preference Comparisons Possible? A Measurement-Theoretic Approach. iussP Paper no. 5. Liege, Bel- gium International Union for the Scientific Study of Population, 1975. Dow, Thomas E., Jr. "Fertility and Family Planning in Sierra Leone." Studies in Family Planning, vol. 2 (August 1971), pp. 153-65. Freedman, R., and L. Coombs. Cross-Cultural Comparisons: Data on Two Factors in Fertility Behavior. New York: The Population Council, 1974. Freedman, R., L. Coombs, M. Chang, and T. H. Sun. "Trends in Fer- tility, Family Size Preferences and Practices of Family Planning: Taiwan, 1965-73." Studies in Family Planning, vol. 5 (September 1974), pp. 270-88. Freedman, R., D. Goldberg, and Sharp. "Ideals about Family Size in the Detroit Metropolitan Area, 1954." Milbank Memorial Fund Ouarterly, vol. 33 (1955), pp. 187-95. Freedman, R., J. Y. Peng, Y. Takeshito, and T. H. Sun. "Fertility Trends in Taiwan: Transition and Change." Population Studies, vol. 16 (March 1963), pp. 219-36. Klialifa, Atef M. "A Proposed Explanation of the Fertility Gap Differen- tials by Socio-Economic Status and Modernity: The Case of Egypt." Population Studies, vol. 27 (November 1973), pp. 431-42. i68 FERTILITY AND EDUCATION : WIIAT DO WE REALLY KNOW? Knodel, John, and Pichiit Pitakepsombati. "Thnailand: Fertility and Family Planning among Rural and Urban 'Womcn." Studies in Family Plan- ning, vol. 4 (September 1973), pp. 229-55. Leibenstein, Harvey. Economic Backwardness and Economic Growth. New York: Jolhn Wiley and Sons, 1957. Maurer, K., and others. Marriage, Fertility and Labor Force Participation of Thai Women- An Econonmic Study. No. R829-AID/RF. Santa Monica, California Rand Corporation, 1973. Mueller, Eva. "Economic Motivations for Family Limitation." Demog- raphy, vol. 26, no. 3 (November 1972), pp. 383-403. Mueller, Eva, and Ricnard Colhn. "Tlhe Relation of Income to Fertility Dccisions in Taiwan." Economic Development-and Cultural Change, vol. 25 (January 1977), pp. 325-47. Ohadike, Patrick. "Thle Possibility of Fcrtilitv Clhange in Modern Africa: A West African Case." African Social Research, vol. 8 (December 1969), pp. 602-14. Olusanva, P. 0. "Status Differentials in the Fertility Attitudes of Married Women in Two Communities in Western Nigeria." Economic De- velopment and Cultural Change, vol. 19, no. 4 (July 1971), pp. 641- 51. Palmore, James, and M. Ariffin. "Marriage Patterns and Cumulative Fer- tility in WVest Malaysia: 1966-1967." Demography (November 1969), pp. 383-401. Pareek, Udai, and V. Kotlhandapani. "Modernization and Attitude to- ward Family Size and Family Planning: Analysis of Some Data from India." Social Biology, vol. 16 (Marchi 1969), pp. 44-48. Paydarfar, Ali A. 'Sociocultural Correlates of Fertility among Tribal, Rural, and Urban Populatioiis in Iran " Social Biology, vol. 22 (Sum- mer 1975),pp. 151-66. Pool, D. I. "Social Clhange and Intcrest in Family Planning in Glhana. An Exploratory Analvsis." Canadian Joulrnal of African Studies, vol. 4 (Spring 1970), pp 209-27. Rizk, Hanna. "Trends in Fertilltv and Family Planning in Jordan." Studies in Family Planning, vol. 8 (April 1977), pp. 91-99. Rosenzweig, Mark R. "Rural Wages, Labor Supply and Land Reform: A Theoretical and Empirical Analysis." Discussion Paper no. 20. Re- search Program in Development Studies, Princeton Jniversity, De- cember 1976. Speare, Alden Jr., and otlhers. "Urbanization, Non-Familial Worlk, Edu- cation and Fertility in Taiwan." Population Studies, vol. 27 (Julv 1973), pp. 323-34. SOURCES 169 WVare, Helen. "Ideal Family Size." World Fertility Survey Occasional Paper no. 13. October 1974. Ware, Helen. "'Tlhe Limits of Acceptable Family Size in Western Ni- gena." journal of Biosocial Services, vol. 7 (1975), pp. 273-96. Weekes-Vaglioni, Winifred. Family Life and Structure in Southern Cameroon. Paris: Development Centre of Office of Economic Coopera- tion and Development, 1976. Willis, Robert J. "A New Approach to the Economic Theory of Fertility Behavior." Journal of Political Economy, vol. 81 (no 2, part II- March/April 1973), pp. 514-564. World Fertility Survey. "Nepal Fertility Survey, 1976. First Report." Kathmandu, 1977. World Fertility Survey. "The Survey of Fertility in Thailand: Country Report," vol. 1. Institute of Population Studies, Bangkok, 1977. Yaukey, David. "Some Immediate Determinants of Fertility Differences in Lebanon." Marriage and Family Living, vol. 25 (February 1963), pp. 27-33. CHAPTER 5. EDUCATION AND FERTILITY REGULATION: EMIPIRICAL Evi- DENCE Amold, Fred and others. The Value of Children: A Cross-National Study, Vol. 1, Introduction and Comparative Analysis. Honolulu. The East- WVest Center, 1975. Brodv, Eugene B, Frank Ottey, and Janet La Cranade. "Fertility-Related Behavior in Jamaica." Cultural Factors and Population in Developing Countries. Occasional Monograph no. 6. Intemational Communica- tions Program. WVashington, D.C.: Smithsonian Institute, 1977. Caldwell, John C. (ed.). Population Growth and Socio-Economic Change in West Africa. New York: Columbia University Press, 1975. Caldwell, John C., and A. Igun. "Anti-Natal Knowledge and Practice in Nigeria." 1'opulation Growth and Economic Developmenzt in Africa. Edited by S. H. Ominde and C. N. Eligu. London: Heineman, 1972, pp. 67-76. CELADE (United Nations Regional Center for Demographic Training and Research in Latin America) and CFSC (Community and Fertility Studv Center). Fertility and Family Planning in Metropolitan Latin America. Clhicago: University of Chicago, 1972. Chung, Bom Mo, and others. Psychological Perspectives: Family Planning in Korea. Seoul: Hollym Corporation, 1972. Dow, Thomas E., Jr. "Fertility and Family Planning in Sierra Leone." Studies in Family Planning. Vol. 2 (August 1971), pp. 153-65. 170 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? Freedman, R., and L. Coombs. Cross-Cultural Comparisons: Data on Two Factors in Fertility Behavior. New York: The Population Council, 1974. Freedman, R., J. Y. Peng, Y. Takeshito, and T. H. Sun. "Fertility Trends in Taiwan: Transition and Change." Population Studies, vol. 16 (March 1963),pp. 219-36. Goldberg, David, and Greer Litton. "Family Planning: Observations and an Interpretive Scheme." Turkish Demography: Proceedings of a Con- ference. Edited by Frederic S. Shorter and Bozkurt Guvenc Hacettepe University, 1967, pp. 214-48. Hill, Rcuben, and others. The Family and Population Control. Chapel Hill: University of North Carolina Press, 1959. Khalifa, Atef M. "A Proposed Explanation of the Fertility Gap Differen- tials by Socio-Economic Status and Modernity: The Case of Egypt." Population Studies, vol. 27 (November 1973), pp. 431-42. Khalifa, Atef M. "The Influence of Wife's Education on Fertility in Rural Egypt." journal of Biosocial Science, vol. 8 (January 1976), pp. 53-60. Knodel, John, and Pichit Pitakepsombati. "Thailand: Fertility and Family Planning among Rural and Urban Women." Studies in Family Plan- ning, vol. 4 (September 1973), pp. 229-55. Michel, A. "Interaction and Family Planning in the French Urban Family." Demography, vol. 4, no. 2 (1967), pp. 615-25. Mitchell, Robert E. "Husband-Wife Relations and Family Planning Practice in Urban Hong Kong." journal of Marriage and Family, vol. 34 (February 1972), pp. 139-46. Morgan, Robert W. "Fertility Levels and Fertility Change." Population Growth and Socioeconomic Change in West Africa. Edited by John C. Caldwell and others. New York: Columbia University Press, 1975, pp. 187-238. Morrison, William. "Family Planning Attitudes of Industrial Workers of Ambornath, A City in Western India." Population Studies, vol. 14 (March 1961), pp. 235-48. Mueller, Eva, and Richard Cohn. "Thie Relation of Income to Fertility Decisions in Taiwan." Economic Development and Cultural Change, vol. 25 (January 1977), pp. 325-47. Mukherjee, B. N. "The Role of Husband-Wife Communications in Fam- ily Planning." journal of Marriage and the Family, vol. 37 (August 1975), pp. 655-67. Nepal Family Planning. "Fertility, Familly Planning and Desire for Chil- SOURCES 171 drcn." Four District Baseline Survey Report no. 1. His Majesty's Gov- ernment Ministry of Health. Kathmandu, Nepal. Olusanya, P. 0. "Status Differentials in the Fertility Attitudes of Married Women in Two Communities in Western Nigeria." Economic De- velopment and Cultural Change, vol. 19, no. 4 (July 1971), pp. 641-51. Oppong, Christine. "Conjugal Power and Resources: An Urban African Example." journal of Marriage and the Family (November 1970), pp. 676-80. Palmore, James A., Jr. "The West Malaysian Family Survey, 1966-1967." Studies in Family Planning, vol. 40 (April 1969), pp. 11-20. Pareek, Udai, and V. Kotlhandapani. "Modernization and Attitude to- ward Family Size and Family Planning Analysis of Some Data from India." Social Biology, vol. 16 (Marclh 1969), pp. 44-48. Pool, D. I. "Social Change and Interest in Family Planning in Ghana: An Exploratory Analysis." Canadian journal of African Studies, vol. 4 (Spring 1970), pp. 209-27. Ramakumar, S. R., and S. V. S. Copal. "Husband-Wife Communica- tions and Fertility in a Suburban Community Exposed to Family Planning." Journal of Faamily Welfare, vol. 18 (Marclh 1972), pp. 30-36. Roberts, J. B., and others. "Family Planning Survey in Dacca, East Pakistan." Demography, vol. 2 (1965), pp. 74-96. Sear, Alan M. "Predictors of Contraceptive Practice for Low Income Women in Cali, Colombia." journal of Biosocial Science, vol. 7 (1975),pp 171-88. Simmons, Alan B, and Mauricio Culagovski. "If Thley Know, Wlhy Don't They Use? Sclected Factors Influencing Contraceptive Adoption in Rural Latin America." April, 1975. Processed. Simmons, Alan B., and Jolhanna de Jong. "Education and Contraception in Rural Latin America." Processed. Speare, Alden Jr., and otlhers. "Urbanization, Non-Familial Work, Edu- cation and Fertility in Taiwan." Population Studies, vol. 27 (July 1973), pp. 323-34. Williamson, John B. "Subiective Efficicncv and Ideal Family Size as Predictors of Favorability toward Birth Control." Demography, vol. 7 (August 1970), pp. 329-39. World Fertility Survey. "Nepal Fcrtility Survcy, 1976. First Report." Kathmandu, 1977. Yaukey, David. "Some Immediatc Detenrunants of Fcrtility Differcnces in Lcbanon." Marriage and Family Living, vol. 25 (February 1963), pp. 27-33. 172 FERTILITY AND EDUCATION : WHAT DO WE REALLY KNOW? CHAPTER 6. EDUCATION AND FERTILITY: SUMMARY AND IMPLICATIONS FOR RESEARCH AND POLICY Leibenstein, Harvey. "Beyond Economic Man: Economics, Politics, and the Population Problem." Population and Development Review, vol. 3 (September 1977), pp. 183-99. Author Index Adegbola, 11 J Page, 92 Chang, Cheng-Tung, 111 Adelman, Irma, 17, 19, 20, 22 Chernichovsky, D , 44, 46 Amani, M, 34, 84, 87 Cho, Lee-Jay, 77n Anker, Rilchard, 16, 17, 18, 20, 21, 31n, Chojnacka, Helena, 81, 82 46, 86, 89, 95 Chung, Bom Mo, 36, 84, 88, 103, 104, Anflin, M, 66, 85, 114 120,125,126, 127 Arnold, Fred, 67, 107, 108n, 118, 120, Cochrane, James, 64n, 94 122 Cochrane, Susan Hill, 16, 25, 27, 46, Azumili K, 33n, 42n 64n, 97 Cohni, Richard, 110, 135 Coombs, Lolagene, 58n, 103n, 106n, Baird, Dugald, 66, 91n 107, 127, 129, 133, 137 Baldwin, C Stcphen, 77 Culagovski, Mauncio, 123 Bean, Frank, 65 Becker, Gary, lOln, II5n DaVaiuzo, Julie, 67, 68, 81, 82, 92, 94, Ben-Poratlh, Vorami, 16, 25, 26 9 Berelson, Bernard, 17, 19, 20 Davidson, Maria, 44, 66 Bindarv, A , 67 Birdsall, Nancy, 3, 4, 65, 66. 67 Davis, Kingsley, 59 Blake, Juidith, 61 de Jonig, johannia, 31i, 121, 123 Bogue, Donald, 15, 16, 18 Del Rio, A C, 26 Bontrague, Henald, D5, 16, 25 Dixon, Ruth, 80, 81, 82, 83, 84, 127 Bor,trager, Hcian D , 16,25 Dow, Thomas E., Jr, 34, 103 104, 118, Brody, Eugene B, 124o91,13 Bulatao, Rodolfo, 67, 109 131, 132 Burcii, T, 84, 89 Drakatos, Constantine, 16, 22, 24 Butz, WVilhliai P , 66, 91n, 92, 94, 97 Duza, M Badrud, 77 Easterln, Richard A., 53, 55, 56, 57 Cain, Glen, 67 Ekanern, Ita, 15, 16, 18, 19, 20, 34 Caldwell, John C., 16, 17, 23, 24, 33n, El-Badry, M A., 16, 38 84, 87, 120, 127, 130, 131, 132 Encarnaci6n, Jose, Jr, 4, 44, 66 Carelton, Robert 0, 36, 55 Evenson, Robert, 67, 68 Chander, R, 84, 92 Ewbank, Douglas, 38 173 174 INDEX Farooq, G. M., 16 Kocher, James E., 46 Freedman, Ronald, 17, 19, 20, 58n, 77, Kogut, Edy Suiz, 44, 86, 90 102n, 103, 104, 106n, 107, 112, 113, Kotbandapani, V., 103, 105, 119 118, 127, 129, 133, 137 Friedlander, Stanley, 16, 17, 19, 20, 21, 22 Lal, R. B, 16 Laprerre-Adamcyk, E., 84, 88, 89 Leibenstein, Harvey, 107, 141 Galbraith, John Kenneth, 58 Li, W. L, 17, 24 Gendell, Murray, 37 Litton, Greer, 133n Goldberg, David, 31n, 102n, 133n Goldstein, Sidney, 34, 37, 67, 78 Gopal, S. V. S, 67, 123, 124 Mason, Karen, 3, 4, 65, 66, 67, 68 Gregory, Paul, 16, 18, 19, 20 Maurer, K., 66, 81, 82, 102n Mazur, D P, 66 McCabe, James L., 16, 18, 19, 20, 21, Haas, Paula, 56, 61 44 Habicht, Jean-Pierre, 91n McGreevey, William P, 3, 4, 65, 66, Hammond, John L, 32n 67 Harmon, Alvin, 67, 86, 90 Mernek, Thomas, 16, 22, 24 Hawley, Amos, 84, 87 Mertens, Walter, 32, 42n, 66 IHeckman, James I , 57 Michael, Robert T, 57 Heer, David, 15, 17, 18, 19, 20 Michel, A., 67, 123 Heller, Peter S, 16, 94, 95, 96 Miro, Carmen, 32, 42n, 66 Hicks, W Whitney, 16, 23, 24, 26 Mitchell, Robert E., 65, 67, 123 Hill, Reuben, 67, 123 Morgan, Robert W., 118 Holsinger, Donald B , 55, 61 Morrison, William, 117, 119, 120 Hull, Terence, 37, 92, 94, 97 Mott, Frank, 84, 87 Hull, Valerie, 92, 94, 97 Mueller, Eva, 58n, 67, 69, 109, 110, Hutchinson, B. W., 93, 94, 95 135 Mukherjee, B N, 123, 124 Igun, A., 120, 130 lutaka, S., 44 Namboodin, N Knshnan, 56, 61 Nayar, P. K B., 84, 88 Nerlove, Marc, 58, 66 Jaffe, A J, 33n, 42n Jam, Anrudh K, 91, 92 Janowitz, Barbara, 16, 17, 18, 19, 20, Ohadike, Patrick, 38, 105, 130, 131 21 Olusanya, P. 0, 85, 87, 95, 97, 104, 105, 109, 124, 130 Oppong, Christine, 124 Kasarda, John D., 15, 17, 18, 55, 61 Kelley, Allen, 46, 94 Khalifa, Atef M., 94, 97, 103, 104, 118, Palan, V. T, 84, 92 120, 124, 126 Palmore, James A., Jr., 37, 66, 85, 112, Khan,M.Ali,44 113, 117, 119, 120, 125, 126 Khan, Z, 94, 97 Pareek, Udai, 103, 105, 119 Kim, Mo-lm, 66 Paydarfar, Ali A., 86, 89, 113, 114 Kirk, Dudley, 15, 16, 17, 18, 21, 41 Pitakepsombati, Pichit, 103, 104, 117, Knodel, John, 36, 103, 104, 117, 118, 118, 120, 125, 126 120, 125, 126 Pool, D. I., 112, 113, 119, 121, 124 Knowles, James C., 46, 86, 90, 95 Prachuabmob, Visid, 36, 84, 87 INDEX 175 Ramakumar, S R., 67, 123, 124 Speare, Alden, Jr., 81, 82, 112, 113, Repetto, Robert, 15, 17, 18, 19, 20, 56 125, 126 Rizk, Hanna, 32, 38, 113, 114 Srinivasan K, 38 Roberts, J. B., 119, 121 Stockwell, E. G, 93, 94, 95 Robinson, W. S., 32n Stycos, J. Mayone, 16, 23, 24, 26, 34, Rosen, Bernard C , 61, 65 37, 39, 42n, 85, 87, 88, 89, 91, 95, Rosenzweig, Mark R., 16, 18, 19, 20, 97 21, 44, 67, 68, 102n Rule, J R., 34 Russett, B. M., 16, 18, 19, 20 Thorsen, Timm, 85, 87, 88, 90n Rutherford, Robert D, 77n Timur, Serim, 42n Traina, Frank J., 16, 25 Tuncer, B, 16 Safilos-Rothschild, Constantina, 14, 32 Turchi, Boone, 57n, 63n Samuelson, Paul A, 58 Schultz, T. Paul, 16, 17, 24, 25, 26, 66, 67 Ware, Helen, 102n, 105 Sear, Alan M, 126 Weekes-Vaglioni, Winifred, 113, 114 Sharp, H , 102n Weinenger, Adnana, 67 Shlields, Nwanganga, 69 Weller, Robert, 34, 42n, 65, 67, 85, Shin., Eui Hang, 93, 94 89, 91 Siever, Daniel, 17, 23, 25 Williams, Anne D., 65, 66, 67 Silver, Morris, 16, 17, 19, 20, 21, 22 Williaamson, John B , 119 Simmons, Alan B., 31n, 61, 65, 119, Willis, Robert J, 57 123 Simon, Julian L, 3, 4, 66 Singh, K P., 95, 97 Yaukey, David, 34, 66, 85, 87, 88, 90n, Sirageldin, Ismail, 44, 94, 97 92, 105, 126 Sloan, Frank, 94, 95, 96 Smith, Peter C, 83n, 84, 89 Snyder, Donald W., 66, 67, 68 Zarate, Celvan, 16, 23, 24 -