In February 2005, Illinois became the first U.S. state to grant home-based child-care providers (HBCPs) the right to form a labor union in order to bargain collectively with the state government. This policy inspired similar efforts across the country and represents a potentially important direction for child-care policy. In other industries, labor unions have helped workers negotiate higher wages and a stronger voice in determining workplace practices. Among public-sector workers, unions have also been responsible for expanding the size of the unionized public-sector labor force. To date, the implications of labor unions for the cost, type and availability of subsidized childcare has not been evaluated empirically.

In this study, I examine the impact of granting Illinois home-based childcare providers the right to form a labor union on 1: the per-child cost of subsidized childcare for infants and toddlers, 2: the type of childcare (home-based vs. center-based/ licensed vs. unlicensed) used by subsidy-receiving Illinois infants and toddlers, and 3: the percentage of Illinois infants and toddlers who use childcare subsidies.

To conduct these analyses, I combine data from the Current Population Survey with Childcare and Development Fund administrative records on U.S. infants and toddlers (children from birth to age three) whose families received childcare subsidies during the period from 2002-2008. I employ a recently developed strategy for conducting comparative case studies by comparing the treated group to a “synthetic” control group generated using data on states that have not implemented child-care unionization (Abadie, Diamond, & Hainmueller, 2010). 

The synthetic control group approach improves on traditional comparative case studies by providing a transparent, empirical approach for constructing the counterfactual. Rather than choosing a single comparison unit (in this project, the state), the synthetic control-group approach uses information on outcome indicators and selected covariates to generate a weighted combination of multiple comparison states that best approximates the outcome data observed in the treatment state, prior to treatment. These weights are then applied to data on post-intervention outcomes from the untreated states.  In so doing, the approach documents the degree to which each comparison unit contributes to the synthetically created control group and makes explicit the degree to which the synthetic control-group is, or is not, similar to the treated unit on pre-intervention measures of the outcome as well as on other selected characteristics. 

I find that, relative to the synthetic control group, subsidy-receiving Illinois infants and toddlers spent an average of 7 percentage points more care hours in licensed settings, as compared to unlicensed settings, in the three years following childcare unionization. I also find that relative to the synthetic control group, 1.1 percentage points fewer Illinois infants and toddlers used child-care subsidies during this period. I find no effects on the percentage of care provided in unionized settings nor in the per-child amount paid to providers to care for subsidy–receiving infants and toddlers.