Grandchildren and Grandparents’ Labor Force Attachment

TitleGrandchildren and Grandparents’ Labor Force Attachment
Publication TypeConference Paper
Year of Publication2018
AuthorsAsquith, B
Conference NameAmerican Economic Association
KeywordsFamily, Labor

As workforces age and life expectancy grows, understanding what motivates workers to strengthen or weaken their labor force attachment is a matter of growing policy concern. This paper asks how grandparents change their labor force attachment when grandchildren arrive by first using a multigenerational sample from the Panel Study of Income Dynamics (PSID) to study individual-level responses, and then use Current Population Survey (CPS) data to study how grandparenthood trends change labor force participation rate of older male workers. Grandchildren's impact on age of retirement, hours worked, whether the grandfather is in the labor force, or the grandmother reports non-zero annual hours worked are estimated. Endogeneity between fertility timing and grandparent characteristics is instrumented for by exploiting exogenous state-by-year variation in access to reproductive technologies. I find that grandfathers work 339 fewer hours and become 19.5\% more likely to retire, while grandmothers respond to the marginal grandchild by becoming 10\% more likely to retire and working 132 fewer hours a year if non-retired. This paper shows evidence that the arrival of grandchildren does change grandparents' labor supply, but that trends in grandparenthood have only had a muted impact on trends in older men's labor force participation (LFP) rate. In a predictive exercise simulating labor force participation rates, the response to grandchildren is specification-sensitive, but interactions between grandchildren measures and Social Security benefits indicating that a 1 point increase in the fraction grandparent decreases the LFP rate by 0.18 points, and by 4.1 points with a 1 child rise in the average number of grandchildren. Collectively, across alternative fertility and grandchildren histories, trends in the simulated LFP rates do not meaningfully change from trends in the observed LFP rate, although the levels of participation would have been between 3-5 points higher between 1962-1994 if the Baby Boom had not occurred.

Citation Key10711