Abstract The synthesis fertility model emphasizes biological “supply” and tastes. But testing is difficult because many critical variables are unobserved. This study uses Nicaraguan data to estimate a latent variable system of reduced-form demands for births, infant mortality, contraception, nutrition and breastfeeding. The estimates support some of the synthesis extensions to the standard fertility model, such as the concurrent increase of contraception, health, nutrition and fertility and decline in breastfeeding with income increases from initial low levels. The estimated impact of tastes is limited. Income and mother's schooling have strong effects even with controls for tastes and community endowments. Resource demands of high population growth make understanding the fertility response to a variety of changes that occur during development clearly important. Increasing education, particularly for females, is the one such change that is (in versely) associated with fertility. According to the standard economic model of fertility (e.g., Becker 1960, 1981; Becker and Lewis 1978; Willis 1973), women's schooling increases household productivity and potential wages, thereby decreasing the price of child quality relative to child quantity. Within this model, child quantity and quality are determined simultaneously with parental consumption and depend on full income and prices. Preferences are assumed to be identical (at least up to a random term) across households. An alternative“synthesis” fertility model ($ Easterlin^{1} $) extends the standard one-period economic model of feritlity to incorporate concerns from the sociological and demographic literatures such as “potential family size” and “endogenous tastes”. Such concerns are perceived to be particularly important in developing countries. Potential family size refers to the incorporation of supply of biological factors into the analysis of fertility and child mortality since persons (families) may be unable to have as many children as they desire. In the process of development, improved health and nutrition may increase potential family size due to higher fecundity and lower child mortality. Endogenous tastes refer to the determination of preferences of a particular household by the decisions of other households; this includes intertemporal and intergenerational influences such as the conditioning of adults' taste norms by their childhood experience, as well as intratemporal influences of the norms of their peers (Easterlin et al. 1980). While such extensions are attractive to many, verification is desirable because of the important analytical and policy implications. But testing the synthesis model is difficult since critical variables — such as potential family size, taste norms and many prices — are not observed directly. In an earlier paper (Behrman and Wolfe 1984c) we presented evidence consistent with the expanded model by dividing the population into four groups based on some biological and taste variables. Such a test is limited, however, since corner solutions, rather than the differences emphasized by the synthesis model, may yield the estimated differences among the four groups. In the present paper our contribution is to explore a number of components of the synthesis model within an integrated, latent variable system, with empirical application to Nicaragua. z In addition to births, infant and child mortality and contraceptive use, we include nutrition and breastfeeding in the system of outcomes because they may be integrally linked to fertility and mortality through biological supply channels. Through these estimates we try to gain insight into the validity of the synthesis model in a particular developing country context. If our estimates suggest that tastes or factors such as $ nutrition^{3} $ and breastfeeding that relate to biological supply are important in a fertility system, they support the synthesis extension of the economic model of fertility. We undertake our extimates only for the population with knowledge of modern contraceptives because we do not want to confound the question of whether women have such knowledge with the question of whether the synthesis features are important. This also permits us to address a particular issue for developing countries: in the initial stages of development might increasing contraceptive use be associated with increasing fertility due to greater fecundity, as Easterlin and Crimmins (1985) have claimed. In the process of investigating the synthesis model, we also explore two other related issues: (1) is schooling in part a proxy for tastes and other endowments?4 and (2) is income important in such a system?5