Application of Granger-Sims causality tests to monthly fertility data, 1958?1984

Abstract

When applied to monthly age specific data, Granger-Sims causality tests provide a useful technique for identifying the effective lag between business cycles and fertility in the United States. Male and female monthly age specific unemployment rates are used as a proxy for the business cycle, and test results are presented for first and higher order birth rates, as well as total age-specific monthly fertility rates. The period is subdivided (January 1958 – May 1973 and June 1973 – December 1984) in order to identify possible trends. Four results hold in all cases studied, with respect to the relationship between unemployment and fertility. (1) Noncausality is rejected in the direction from unemployment to fertility, and no ‘feedback’ effect is indicated; thus the relationship is one of simple causality. (2) In the ‘critical’ decision period from 9–16 months prior to realized fertility rates, the sign of the effect of unemployment on fertility is negative: this holds for both male and female unemployment rates. (3) There appears to have been a shortening of the effective lag between unemployment and fertility, of perhaps 2 – 3 months, between periods 1 (1959 – 1973) and 2 (1973 – 1984). (5) The strength of the (negative) relationship between unemployment and fertility appears to have increased from period 1 to period 2.