The added worker effect over the business cycle: evidence from urban Mexico

Abstract

This paper investigates whether the added worker effect is an important determinant of the increased labour force participation by women and whether the magnitude of the added worker effect differs between the peak and trough of the business cycle. Use is made of repeated observations from spouses in urban Mexico, collected during the Peso crisis (1994:4–1995:4) and during the period of economic prosperity (1998:4–1999:4). Significant added worker effects are found in both periods. The magnitude of the added worker effect during the crisis period is found to be twice as large as that observed during the period of economic prosperity.

Notes

 Cullen and Gruber (2000) for example, estimate that hours of work would be roughly 30% higher during the husband's spell of unemployment if there were no unemployment insurance benefits and that the non-employment rate of wives with unemployed husbands would drop by almost 45%.

 In the USA literature based on aggregate data there is a general consensus that during economic downturns, the discouraged worker effect dominates the AWE (Mitchell, 1979).

 In a related paper we use a subsample of the same panel of households to examine family adjustments in work and schooling during a crisis period (Skoufias and Parker, 2003).

 Although we did not impose the restriction that the head of the household and his spouse need to be married, approximately 98% of the couples reported that they were married or in a liberal union. The set of variables we used to identify a family included the variable identifying whether the family has moved out of the residence (hogar mudado). Thus a family that moved into the same residence previously occupied by a family that moved out was identified as a new family.

 We have also estimated a random effects probit model, after rearranging our data appropriately, using the labour force status of the spouse in the current period as a dependent variable and the current unemployment status of her husband as an independent variable. Independently of the specification adopted for the vector X the estimates of the coefficient β were also significantly positive which suggests that much of the added worker effect is contemporaneous. This finding, similar to the one reported by Spletzer (1997) for women in the USA, is consistent with the interpretation that the husband's unemployment arrives unexpectedly.

 For each of the four specifications, we pooled the data from the two different periods into one probit equation allowing all coefficients to differ between the crisis period and the period of prosperity. In all cases, the hypothesis that the coefficient of the variable (headE →U ) was the same across the two periods was rejected at conventional levels of significance in favor of the alternative hypothesis that the AWE effect during the crisis period was greater.