In 2011, the federal Child Care and Development Fund (CCDF) spent $5 billion on child care subsidies for low income families; an enormous investment to support employment of low-income parents. Despite clear program objectives to support the work efforts of low-income parents, estimates of the impact of child care subsidies on employment are inconsistent across studies. Much of the literature inadequately resolves the econometric difficulties that arise due to the joint nature of employment and child care decisions.

This paper investigates the relationship between employment and child care subsidies while correcting for endogeneity. The data were obtained from a three-year longitudinal parent survey with a sample of low-income families with young children from Minnesota. We use maximum likelihood estimation of a bivariate probit endogenous switching model as our estimation strategy, including equations for employment and for subsidy receipt. We estimate models for a binary employment indicator and with an ordinal measure of full-time, part-time, and no employment. We expect that unobserved factors that affect the likelihood of receiving a child care subsidy are correlated with unobserved characteristics that impact employment decisions. Ignoring this unobserved heterogeneity will result in biased and inconsistent estimates.

The endogenous variable of interest is child care subsidy receipt for the focal child. In order to identify the effect of this variable, we test several instrumental variables in the subsidy receipt equation that were excluded from the employment equation. The key instrument is an innovative variable measuring parental preferences with regards to the characteristics of child care arrangements. Respondents were asked a series of questions about how important they considered different characteristics in a child care setting, such as a warm environment, educated staff, or the availability of books and learning materials. For each characteristic, the majority of respondents stated that the characteristic was extremely important. Through a factor analysis, we identified a factor that was influential in responses to questions about the importance of an educated staff, tracking the child’s learning and the presence of tools for teaching.   We labeled this the ‘parental value of education’ factor, because it indicates variation in the importance parents place on the educational features of a care setting. This factor predicted seeking a child care subsidy, but was otherwise unrelated to employment decisions.

The results demonstrate that, after correcting for the endogeneity of subsidy receipt, subsidy increases the probability of employment, and also the probability of full time employment substantially. Subsidy receipt increased the probability of employment, primarily full time employment, by approximately 30 percentage points. We also find a significant relationship between the unmeasured factors that are associated with subsidy and employment decisions. Given the importance of employment in U.S. welfare policy, the results have important policy implications for the role of child care subsidies in supporting employment of low-income parents.