“Studies Show”: Or Do They?

In an article for the New York Times newsletter The Upshot, Claire Cain Miller and Jim Tankersley argue that mandatory parental leave is economically and socially beneficial. The authors review research that supposedly demonstrates that mandatory paid parental leave increases female employment, wages, and work hours without adversely affecting productivity and turnover. A closer look at the evidence, however, casts doubt upon these claims.

First, Cain Miller and Tankersley advocate a six-months paid leave policy proposed by California Governor Newsom. They reference economics literature that suggests that six months of parental leave is “ideal.” To support this claim, the authors cite an NBER working paper from 1995, “Parental Leave Policies in Europe and North America,” by Christopher Ruhm and Jackqueline Teague. The paper examines observational data from Europe spanning the years 1960-1989 and finds a modest correlation between the length of paid leave and outcomes such as female unemployment. The effects are small, however: a switch from a policy of no paid leave to 25 weeks of paid leave predicts a decrease in the female unemployment rate from 5.8% to 5.5% and an increase in the female labor force participation rate from 47.3% to 47.4%. Can these results from Europe in the 1960-1989 period be generalized to the contemporary United States? Europe differs from the United States along several dimensions, including the population’s age, race, educational attainment, household size, divorce rate, average marriage age, and unionization rate. Studies that do not control for these factors should be referenced with caution.

Next, the authors claim that mandatory parental leave would cause an increase in female workforce participation:

The absence of paid leave and other family-friendly policies has been found to be the major reason that more women aren’t working in the United States.

If policies like paid parental leave and subsidized child care enabled more mothers to work, the United States could add five million prime-age workers to its labor force, according to a new economic letter from the Federal Reserve Bank of San Francisco. 

This language suggests that there is a causal relationship between paid leave and female labor force participation, but neither of the referenced studies attempts to estimate causal effects.

The first study, a 2013 NBER working paper by Blau and Kahn, claims that an array of social policies might be responsible for as much as 29% of the difference in female labor force participation between the United States and other countries. For parental leave specifically, however, the paper finds no significant effect on the female labor force participation rate. The latter study, a 2018 FRBSF economic letter, does not even test for a statistically significant correlation between parental leave policies and the female labor force participation rate. It merely notes that if the U.S. female labor force participation rate increased to match that of Canada, there would be five million more prime-age workers in the U.S. labor force. The letter does not argue that a more generous parental leave policy is the cause of the higher Canadian labor force participation rate. In fact, it notes:

[R]esearch on whether such policies have affected the labor participation decisions of mothers has found little evidence of their effectiveness.

Next, Cain Miller and Tankersley write:

Spending on child care makes the biggest difference in female employment, earnings and fertility, found a recent paper.

The authors cite a 2017 paper by Olivetti and Petrongolo, which does not claim to estimate causal effects. The paper does find that, among the examined policies, spending on early childhood education and care was most consistently associated with improvements in female employment, fertility, the male-female employment gap, and the earnings gap. Cain Miller and Tankersley fail to mention, however, that the paper also examines the effect of the “percentage of total leave that was paid” on these outcomes and finds that this variable was significantly associated with a decrease in female employment, an increase in the male-female employment gap, no effect on earnings, and a very small positive effect on fertility.

Next, Cain Miller and Tankersley write:

Even though California companies opposed the original policy, surveys have found either no effect or a positive effect on productivity and turnover. Paid leave increases mothers’ wages and work hours, research shows, and improves the health of babies and mothers.

First, to corroborate the supposed positive effect of paid leave on wages, the authors cite a 2012 paper by Rossin-Slater, Ruhm, and Waldfogel. This paper simply claims that paid leave (as opposed to unpaid leave) increases the likelihood that women will take leave. This is not surprising. Furthermore, the paper reports no statistically significant relationship between paid leave and wage income that is independent of the relationship between paid leave and work hours:

PFL is also associated with a growth in wage income during the previous year that roughly corresponds to that expected given the higher work hours; however, these estimates are imprecise.

In addition, this statistically significant increase in hours worked is conditional on employment. The paper also separately tests employment itself, finding a negative (though insignificant) relationship between paid leave and employment for two of three measured subgroups. 

Next, the authors cite a California Employment Development Department study to substantiate the supposed positive association between paid leave and work hours. This study generates causal estimates but does not directly test for the effect of paid leave on hours worked. Rather, the study examines the likelihood of being employed four quarters after taking leave and finds no significant relationship between this variable and paid leave.

In a nod to the potential economic tradeoffs of mandating paid parental leave, the Upshot reassures us that:

 surveys have found either no effect or a positive effect on productivity and turnover.

The cited survey suggests that the overwhelming majority of employers find that paid leave has no effect or a positive effect on productivity, profitability, turnover, and morale. If this is true, why is mandated paid leave necessary? Are employers unable to recognize when a policy is in their own interest?

Cain and Miller survey the extensive labor economics literature and present what they consider to be the most convincing empirical case that mandated paid parental leave would be beneficial for California. Given the foregoing caveats, however, this case is unconvincing.

Please read my colleague Vanessa Brown Calder’s policy analysis. Therein you will find a sophisticated discussion of the economic tradeoffs that mandatory paid leave entails: higher unemployment for certain groups, intra-employee redistribution, non-leave margin adjustments, etc.