Gains in the education of Peruvian women, 1940 to 1980

Policy, Research, and External Affairln WORKING PAPERS Education and Employment Population and Human Resources Department The World Bank August 1990 WPS 472 Gains in the Education of Peruvian Women, 1940 to 1980 Elizabeth M. King and Rosemary Bellew What determines girls' educational attainment? School quality (measured by the number of textbooks and teazhers); changes in attitudes and bettereconomic opportunities foreducated women; parents' (especially mothers') years of schooling and occupations; and the opportunity cost of sending a girl to school - especially in rural families, or when mothers must hold jobs outside the home. The Policy, Research, and External Affairs Complex distributes PRE Working Papers todisseminatethe findings of work in progrcsa and to encourage the exchange of ideas among Bank staff and aU others interested in developmetnt issues. These papers carry the namcs of the authors, reflect only their views, and should bc used and cited accordingly. The findings, interpretations, and conclusions are the authors' own. They should not be attnbuted to the World Bank, its Hoard of Duectors, its managmernt. or any of its member countries.  - __ - - - - . -- ~1 IPolicy, Resear.h, and ExternaI Affaire l ~ l Education and Employment WPS 472 This paper- ajoint product of the Education and Employment Division and the Women in Development Division, Population and Human Resources Depariment - is part of a larger effort in PRE to determnine how to improve women's access to education in developing countries and if and how that education improves their productivity and family welfare. The paper will be part of a book on women and the economy of Peru to be published by thle Bank. Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington DC 20433. Please contact Cynthia Cristobal, room S6-033, extension 33640 (47 pages, including tables). Since the mid-1950s, Peru's education policies schools expanded and primary education became have been designed to raise skill levels and make more available. education available to more of the population. Those policies rested mainly on expanding the The relative effects of parents' education number of schools. As a result, school enroll- differed for boys and girls. In the adult sample, ment rates and attainment levels rose. But an both parents' education had a strong positive apparent parental preference to educate sons effect on daughters' education; for sons, the more than daughters meant that boys' schooling father's education had double the effect of the levels rose more quickly than girls.' Policies mother's education. In the youth sample, the were not enough to brings girls' schooling even mother's education had a stronger effect on the with boys,' especially in rurai areas. daughter's education. These differences reflect a preference on the part of fathers to send their School quality, measured crudely by the sons to school, which mothers partly counter- supply of textbooks and the number of teachers, balanced. appears to have improved the schooling of women. Girls who had a textbook for their own Peru's education policies have reduced the use attained more than half a year of schooling direct costs associated with going to school. But than those who did not. Changes in attitudes and time allocation patterns reveal that the opportu- bt>ter economic opportunities for educated nity cost to the family of school attendance could women also seem to have strengthened the be an effective barrier to further improvements demand for educating rural girls. in school enrollment and continuation rates. Even at a young age, girls- especially in rural Parents' years of schooling and occupations families - participate in the labor market and were significant deterrninants of educational contribute substantially to productive work at levels. The impact of these socioeconomic home. factors lessened over time as the number of The PRE Working Paper SCriCs dissemiiiates thc findings of work under way in the Bank's Policy, Rcsearch, and Extemal AffairsComplex. An objective of thescrics is to get thesc findings out quickly, even if presentations arc less than fully polished. Thc findings, interpretations, and conclusions in these papers de not necessarily reprcsent official Bank policy. Produced by the PRE Dissemination Centcr  Gains in the Education of Peruvian Women, 1940 to 1980 by Elizabetn M. King and Rosemary Bellew Table of Contents Trends in Education 1 A Household Model of Education with Gender Differences 6 Empirical Model 9 Empirical Results from the Adult Sample 15 Educational Profiles by Gender end Residence 15 Changes in Educational Levels 17 What Explains Educational Attainment? 18 Results from the Youth Sample 26 What Factors Explain School Enrollment? 30 Nonschool Activities of Females 32 Conclusions 37 References 40 Appendices 43  Over the past few decades, expanding and improving the quality of public education have been important components of the Peruvian government's plan to accelerate economic development and redistribute income. This paper addresses the following questions: How rapidly have educational opportunities and schooling levels changed over time? Have they become more equitably distributed betweern men and women? Have women benefited from recent educational policies as much as men? Whlat other factors (for instance, family background and community characteristics) explain variations in the levels of education between men and women? The paper is organized as follows. We first trace the expansion of education in Peru and discuss the patterns by gender. Section 2 presents an empirical model of schooling choice within the family and defines the variables used. Section 3 discusses the findings in light of educational change and development, and Section 4 describes gender differences in the education of the present generation. Trends In Education Educational opportunities in Peru have changed tremendously since 1940 as a result of major economic changes and reforms in education policy1'. In the early 1940s few Peruvians attended school. More than half of all 1' Due to a lack of data prior to 1950, most of the discussion pertains to 1950-1980. For these decades educational trends are traced by five-year periods to capture the effects of different administrations on levels of educational attainment. To explain variations in adult educational attainment, each five-year period is matched with the birth cohort whose educational experience corresponds to that period.  2 adults were illiterate (table 1)21; only one child in three was enrolled in primary or lower secondary schoo.1 (table 2). Moreover, raising school attendance was particularly challenging since two-thirds of all Peruvians were scattered in rural areas, and 35 percent of the population spoke only Quechua or Aymara--the two main Indian languages--while the language of the schools was Spanish (Government of Peru 1981). Table 1: Education of Males and Females, Aged 15 and Over, 1940-81 1940 1961 1972 1981 Percent Literate 42 61 73 82 Males 55 74 83 90 Females 31 48 62 75 Urban 82 88 92 Rural 41 49 62 Mean Years of Schooling 1.9 3.1 4.4 6.0 Males 2.4 3.8 5.1 6.7 Females 1.4 2.4 3.6 5.4 Hiehest Level of Education Attended (percentage) No school 58 39 27 16 Males 45 26 16 9 Females 69 52 37 23 Primary 34 48 48 42 Males 47 58 54 44 Females 27 38 42 40 Secondary 1 2 5 10 Males 6 14 24 35 Females 3 9 17 28 Postsecondary 1 2 5 10 Males 2 3 6 12 Females 0.3 1 3 9 Sources: Literacy Rates: Government of Peru 1981, Fernandez 1986. / Among Latin American countries only El Salvador, Nicaragua, Honduras, Bolivia, Guatemala, and the Dominican Republic had higher illiteracy rates than Peru in the early 1940s (Drysdale and Meyer 1975).  .3 Table 2: Percent of Children Aged 6-14 and 15-19 Enrolled in School, 1940-81 Age GrouRs 1940 1961 1972 1981 Ages 6 to 14 Total 30 58 78 90 Males 34 62 82 91 Females *26 53 75 88 Urban - - 90 96 Rural - - 63 79 Ages 15 to 19 Total 17 33 49 56 Males 23 41 57 61 Females 11 ,6 41 52 Urban - 54 63 Rural - - 17 24 Source: Government of Peru 1981, Fernandez 1986. Since that time educational opportunities have expanded considerably. Appendix figures 1A-3B show the rapid rise in enrollment and in gross enrollment ratios between 1950 and 1980. At the primary level the rise in girls' enro.lment came after the increase in boys' enrollment. In 1950 only 69 percent of all girls were enrolled compared to 100 percent of the  4 boys.V At ot'aer levels _here was less difference between males and females; educational opportunities were still very l'ited for both sexes. The data show that more girls who were of school age during the late 1950s and 1960s (that is, cohorts born between 1950 and 1964) enrolled in primary school than earlier cohorts. As a result enrollment ratios for girls rose from 65 percent in 1955 to 99 percent by 1970, narrowing the difference in enrollment between boys and girls. Males born betweer. 1955 and 1964 show the largest enrollment increases, but since men began the 1950s with already high rates of enrollment, their gains were far less dramatic than those achieved by women. Educational gains were not limited to primary education. In the 1960s more students went on co secondary school. Men and women born between 1950 and 1959 registered the largest proportional increases in secondar- school attendance. Enrollment in higher education began a strong upward trend in 1960 that continued throughout the subsequent decade. But at this level of education, the gap between the percentage of males and females enrolled widened. / Gross enrollment ratios are computed as the ratio of total enrollment to the population aged 6 to 11. When under- or over-age students are enrolled, owing to repetition, early or delayed entry, or re-entry, the ratio can exceed 100 percent. On the other hand, net enrollment ratios exclude over- and under-aged youths. They are computed as the ratio of 6- to 11-year olds enrolled in school to the 6- to 11-year-old population. To gauge how these two ratios differ, Peru's 1980 net enrollment ratio was 85 percent, compared to a gross enrollment ratio of 115 percent. Therefore, in 1980, approximately 30 percent of all students were either under- or over-age, but we do not know the share attributable to repetition, delayed or early entry, or re-entry.  5 Throughout the years of expansion, rural children were less fortunate than those in urban areas, where ma!.y of the new schools were concentrated. In 1972 only 63 percent of rural children aged 6 to 14 were enrolled in primary and lower secondary schools, compared to 90 percent of urban youths. Opportunities for upper secondary and tertiary education were rare in rural communities; only 17 percent of rural youths aged 12 to 17 were enrolled in such schools, compared to 54 percent of the same age group in urban areas (table 2). From 1970, however, an increasing number of rural parents sent their children to primary and lower secondary school. By 1981 the gap in school attendance between urban and rural children was closing. At the upper levels, though, rural residents made little progress. In 1981 only 24 percent of the relevant school-age population attended an upper secondary or tertiary institution, compared to 63 percent of their urban counterparts.i- Changes in education levels mirror these enrollment patterns. The average years of schooling of persons aged 15 and above has incrsased steadily over time and the proportion of adults who did not attend school ha: allen from 58 percent in 1940 to 16 percent in 1981. The strorg growth in female enrollment in primary school during the 60s and the trend toward secondary and highe- education are particularly evident. The proportion of adult women whose formal education stopped at primary school rose from 38 percent in 1960 to 42 percent in 1972. By that time 24 and 18 percent of all men and women, respectively, had attended secondary school, as against 6 and 3 percent in 1940. By 1981 these proportions rose by 10 percentage points, reflecting V See table A.I.  6 comparable enrollment growth for women and men. Higher education showed a similar pattern. We next examine the impact of school expansion policies on the educational levels of men and women in the context of family resources and preferences for schooling. These influences add to our understanding of the effectiveness of policy reforms. A Household Model of Education with Gender Differences The human capital theory identifies the principal benefit from education as raising productivity. In the workplace this increased productivity translates into higher earnings (Becker 1964, Mincer 1974); at home it means more efficient home production, such as child care (Gronau 1977). The hypotheses that more highly educated people learn (that is, produce even greater human capital) more effectively (Ben-Porath 1967), or are better able to deal with problems or "disequilibria" in their lives (Schultz 1975) are related to this model. The model assumes that the decision to begin or continue schooling is a function of returns and costs. Returns are usually measured as expected earnings in the labor market corresponding to given levels of education. Costs, which include both direct outlays and indirect (time) costs, are often measured by the availability of, or distance to, school. A few studies have estimated the effect of the opportunity cost of schooling on enrollment or attainment in developing countries and have found a negative effect (for instance, Rosenzweig and Evenson 1977).  7 Other factors may also influence enrollment and attainment decisions. A household choice model of schooling investments implies that family background Is an important determinant of enrollment and attainment not only because it may reflect the student's schooling preferences and income but also because it measures the support for education in the home. Studies of parental influence report strong positive effects. For example, Heyneman and Loxley (1983) found that tour family background variables (mother's and father's education, father's occupation, and books in the home) explained an average of 18 percent of the variance in student achievement in a study of nine developing countries, compared to the 24 percent that was explained by school characteristics. Although research in this area has shifted recently to exploring such questions as the relative effects on achievement of alternative inputs, material versus nonmaterial inputs, or administrative and teaching quality (Lockheed and Komenan 1987), sufficient data to support studies of this genre are harder to come by. Last, genetically determined ability also affects learning and educational attainment (and thus income), but due to limited data on cognitive abi.lity, the effect of this factor on income has been neglected in most studies.l This framework implies that schooling decrl.ions are influenced by a host of factors, including learning ability, wages in the labor market, proximity of the school, and school inputs. But do these factors have a W Griliches and Mason 1972 estimated that failure to control for the effect of ability overstates the estimated rate of return to education by between 7 and 15 percent. In a study on Tanzania and Kenya, Boissiere and others (1985) found that controlling for ability lowered the rate of return by about 60 percent.  8 different effect on men and women? What accounts for gender differences in the amount of schooling? In a household mnodel of schooling choice, gender can be introduced in several ways. One is to assume that parents do not necessarily have the same preferences for: their sons' and daughters' education. Sever-l studies have found that parents tend to favor sons in certain societies (Creenhalgh 1985; Rosenzweig and Schultz 1982). In an economic model this can be shown by representing the household utility as a function Q. two different commodities--the human capital stock of sons and daughters (Rosenzweig and Evenson 1977, Rosenzweig and Schultz 1980). While serious gender inequality is pernicious, this preference does not necessarily imply discrimination by parents. The unequal treatment of sons and daughters might simply be a rational or efficient response to family resource and technological constraints, and to market conditions, rather than a reflection of their own tastes or preferences. This distinction is helpful in formulating policy. The human capital model shows that where the labor market rewards the education of males mo e than that of females, parents may respond by giving daughters less education. Human capital theory also suggests that if the costs associated with schooling were reduced sufficiently, girls' educational levels would rise even without a correspondin7 increase in female wages. A government school-building program, for example, could yield such a result. Or, if the demand for male child labor increases, the opportunity cost of educating daughters may be sufficiently smaller than for sons (barring strong cultural prohibitions against girls' education).  9 In certain settings cttltural forces, such as norms proscribing women's economic and familial roles, influence parents by imposing a heavy cost (for instance, ostracism) on nonconformist behavior. With economic development and increasing work opportunities for women, ter.sion might build up between traditional social norms and the family's desire to benefit from changing conditions. Which families will respond to these changes, and when? Economic theory does not deal formally with the impact of sociocultural forces but it does predict behavioral adjustments to char.ges in prices and income. For example, we would expect that a rise in female wages that increases the returns to their education would tend to increase the parents' desire to irvest in their daughters' education. The magnitude and speec of the response depends on the acquisition of new information and the price and income elasticities of their demand for education. Empirical Modes 'rhere are several indicators of the amount of educational investment, including school enrollment and number of years of education. The framework above implies the following empirical model of demand for schooling: E - a'X + el (1) where E is the educational investment; X is a vector of explanatory variableai; and, el is a random disturbance term. Since parents may or may not invest as  10 much in the education of daughters as in sons, this equation should be estimated separately for males and females to allow the a coefficients to vary between the sexes. We estimate demand for schooling for two samples of the Peru Living Standards Survey (PLSS): a sample of adults aged 20-59,