A new look at the easterlin “synthesis” framework

Abstract

Easterlin's synthesis of the behavioral and biological factors affecting fertility in developing countries has proven enormously influential. It has served perhaps most notably as the organizing framework for the National Academy of Sciences' report on fertility determinants (Bulatao and Lee, 1983) and continues to inform the thinking of demographers and economists. The framework consists of three central concepts. These are the demand for children (or for births); the potential supply of children, a product of exposure to the risk of conception and fecundity; and the monetary and psychic costs of contraception. Reproductive and contraceptive behavior are succinctly described: couples whose potential supply exceeds demand consider contraception, taking account of contraceptive costs in choosing among methods. The synthesis is at once simple and attractive. Of course, as a static or one-period model, Easterlin's framework cannot address many interesting dynamic issues. Nevertheless, it provides an integrated explanation for contraception and cumulative fertility over a reproductive lifetime, taking important constraints on childbearing into account. The notion of demand used in the Easterlin model is very much a hypothetical notion: it has to do with the number of children or births desired if the costs of contraception were entirely absent. Such a formulation leaves economists uneasy (Schultz, 1986). The demand for any good is poorly defined when its attributes or key prices are assumed away. But this particular confusion is unnecessary. It is a simple matter to collapse Easterlin's three categories into two: the demand for children, taking all costs into account, and a ceiling or constraint on childbearing. The payoff to refashioning the model lies in implications for empirical work. A body of literature in economics has examined the estimation of demand equations for goods in the presence of rationing constraints; a moment's thought shows the Easterlin model to be formally equivalent, with the "supply" equation of the model representing a ration or constraint on lifetime fertility. Fertility data for some individuals are drawn from the demand function; for others, who are supply constrained, the data are generated by the position of the constraint. These sorts of models are called "switching regressions" models in economics. This paper presents an application of the switching regressions methodology to World Fertility Survey (WFS) data. The aim here is primarily methodological. I use, as did Easterlin and Crimmins (1985), the WFS data from Sri Lanka and Colombia. These data are supplemented with micro-samples drawn from the 1971 Sri Lankan and 1973 Colombian censuses. The use of census data to construct measures of child survival addresses one of the key criticisms of the Easterlin-Crimmins work (Schultz, 1986). The paper is in four parts. The first part restates the Easterlin framework in terms of a single demand equation and a constraint. The second describes the econometric methodology appropriate for the problem. Results are presented in the third part, and the paper concludes in the fourth.