DEMOGRAPHV©
Volume 15, Number 4 November 1978
AGE, BIOLOGICAL FACTORS, AND SOCIOECONOMIC
DETERMINANTS OF FERTILITY: A NEW MEASURE OF
CUMULATIVE FERTILITY FOR USE IN THE EMPIRICAL
ANALYSIS OF FAMILY SIZE
Bryan Boulier
Office of Population Research, Princeton University, Princeton, New Jersey 08540
Mark R. Rosenzweig
Department of Economics, Vale University, New Haven, Connecticut 06520
Abstract-To influence the number of children ever born to a woman, socioeco-
nomic variables must operate through behavioral and biological mechanisms
such as the age at marriage, the level of fertility in the absence of deliberate
fertility control, and the level of control exerted to reduce fertility within
marriage. In this paper, we propose a new measure of cumulative fertility
which is standardized for the age-fecundity relationship and for exposure to
the risk of conception associated with duration of marriage. A simple model
of fertility behavior which incorporates some of the mechanisms through
which socioeconomic factors may affect fertility is developed and applied to
data from the United States to demonstrate the properties of alternative
measures of family size. The results indicate that use of the new measure
allows more precise estimates of socioeconomic fertility relationships than
would be obtained with children ever born or by sample stratification.
INTRODUCTION
To influence the number of children
ever born to a woman, social and eco-
nomic variables must operate through be-
havioral and biological mechanisms such
as exposure to intercourse, fecundity, use
or nonuse of contraception, and others
(Davis and Blake, 1956; Ryder, 1959;
Yaukey, 1961; Easterlin, 1975; Freedman,
1975). All too often, however, many of the
constraints on fertility imposed by these
biological factors are ignored in both the-
oretical and empirical investigations of
fertility determination. For example, in
many economic models the demand for
own children of a newly married couple is
assumed to be influenced by variables
such as education and the price of the
woman's time and constrained by the level
of potential income (Willis, 1973; DeTray,
1973), but no account is taken of the bio-
logical constraints on the supply of own
children as part of the theoretical model.
A number of procedures have been em-
ployed to take into account some biologi-
cal factors in empirical work on the de-
terminants of cumulative fertility. A
common method consists of the inclusion
of variables such as age at marriage, dura-
tion of marriage, or some combination of
these variables in linear regressions of
children ever born on socioeconomic vari-
ables believed relevant to fertility determi-
nation. Thus, for example, Ben-Porath
(1973) includes age; Harman (1970) in-
cludes duration of marriage; Encarnacion
(1974) includes age at marriage, marriage
duration, and the square of marriage du-
ration; and Kelley (1976) includes age but
permits the intercept of the regression and
the coefficient of age to vary by age group.
We show below, however, that these mod-
487
Downloaded from http://read.dukeupress.edu/demography/article-pdf/15/4/487/907772/487boulier.pdf by guest on 12 January 2023