After early work in agricultural and international economics in developing countries, Behrman turned towards the analysis of households' economic decisions in the mid-1970s. In this area, together with Robert A. Pollak and Paul Taubman, Behrman developed a model for the analysis of parental allocations of resources among their children and finds evidence that U.S. American parents address inequality in their children's earnings by providing more resources to their less able children.[10] Studying whether and how children's order of birth affects how much resources their parents give them, Behrman and Taubman find that parents in the United States give more to firstborns, possibly because of endowment effects, even though they tend to favour later-borns. These differences are then reflected in the relationship between children's birth order and their (age-adjusted) schooling and earnings as young adults.[11] Studying the distribution of food among children within households in rural South India, Behrman finds that parents give preference to their sons when food is scarce, especially among lower caste households, and display only limited aversion for such inequality, thus suggesting that girls may be particularly vulnerable to malnutrition during famines.[12] Using data from rural Pakistan, Behrman, Harold Alderman, Victor Lavy and Rekha Menon analyse the relationship between child health and school enrollment while explicitly modeling child health as determined by household choices in response to unobserved factors such as children's health endowments and find child health (proxied by nutrition) is three times as important for enrollment under that specification.[13]
Using twin experiments, Behrman, Taubman and Mark Rosenzweig find that 27% of the variance in income and 42% of the variance in obesity between individuals in the United States can be explained by individuals' unique endowments. The effect of these individual endowments are further reinforced through schooling, though somewhat diluted as men with high individual endowments tend to marry less educated wives.[14] In another study using twins, Behrman and Rosenzweig find that increasing the schooling of women doesn't increase the schooling of their children once heritable ability and assortative matching are taken into account, and instead reduces the time mothers spent at home.[15]
The determinants and impacts of education, health and nutrition
Behrman's research on the determinants and impacts of education, health and nutrition (mostly in Latin America) began in the early 1980s. One early theme revolves around nutrition, what it is affected by, and how it in turn affects health, education and earnings.[16] For example, using evidence from south India, Behrman and Anil Deolalikar investigate the relationship between nutrition and income growth and find that the consumption of nutrients is essentially unresponsive to increases in income, even though food expenditure increases more or less in line with income; they conclude that as the poor grow richer, they tend to spend their income on more expensive nutrients rather than simply on more nutrients.[17] Behrman's work on the relationship between health and nutrition in developing countries also features prominently in his and Deolalikar's review of the corresponding literature in the Handbook of Development Economics.[18] However, though Behrman has called for the provision of public services to improve the health (and nutrition) of poor children based on the conclusion that child health likely has "considerable" effects on adult labour productivity by e.g. improving education, he repeatedly emphasized that the methodological weakness of the literature on the impact of health and nutrition on education (e.g. regarding reverse causality) prevented broader conclusions and called for more research on this issue (in the 1990s).[19] In line with this call, Behrman and Rosenzweig offer evidence that returns in terms of adult schooling attainment (and partly also in terms of higher earnings) to increasing birthweight are high, being underestimated by up to 50% if genetics and family background aren't controlled for. However, they also find that differences in birthweights don't play a large role in determining the world distribution of income.[20] Last, evaluating the impact of improving nutrition during early childhood on education among adults in Guatemala, Behrman, Maluccio, Hoddinott, Martorell, Quisumbing and Stein find that doing so increased the average number of grades completed by women by 1.2 grades as well as both male and female reading comprehension and non-verbal cognitive ability by one quarter standard deviation, thus suggesting that improvements to the nutrient intakes of very young children can have substantial long-term educational impacts.[21]
In a theme often related to nutrition and health, Behrman has conducted research on the determinants and impacts of education, especially in Latin America. One of the most cited examples of such research is his and Nancy Birdsall's critique of education economics' focus on the quantity of schooling at the detriment of schooling's quality. Using evidence on the returns to schooling for young Brazilian men, they find that accounting for differences in the quality of schooling reduces these returns by half, thus suggesting that the social returns to investments into schooling quality may even exceed those to investments into schooling quantity. Moreover, Behrman and Birdsall find that variation in schooling quality partly explains variation in returns to schooling over space and among individuals.[22] Finally, in a prominent programme evaluation, Behrman, Piyali Sengupta and Petra Todd find that the Mexican conditional cash transfer programme PROGRESA, wherein transfers are conditional upon children regularly attending school, effectively reduced dropout rates and facilitates grade progression, especially from primary to secondary school, and thereby increased schooling attainment by on average 0.7 years.[23]
