To specify the vector of regulatory variables included in the model we draw on previous work by Chipty & Witte (1994), Chipty (1995) and Chipty & Witte (1997). Specifically, we control for the following center regulations: whether liability insurance is required (INSURE), whether pre or in-servicing training is required for the staff (TRAIN), the minimum square feet of indoor space required per child (SQFEET), the maximum group size (GRSZ) by age, and the minimum staff-child ratio (SCRAT) by age.
Since center behavior was found to be affected by the nature of regulations for family child care homes as well as by center regulations, we also control for the regulations imposed on family child care homes. Specifically, we control for the following family child care home regulations: maximum group size (FGRSZ) allowed, and whether pre or in-service training is required (FTRAIN). Finally, to reflect enforcement of regulations we include the number of inspections required per year for centers (INSPECT) and family child care homes (FINSPECT).
To complete the specification of our model, we needed to obtain information on whether or not an R&R was in operation in each of our markets. We approached a number of individuals prominent in the R&R world. A member of the executive board of the national R&R association, NACCRA, compiled data that allowed us to discern if an R&R existed in each of the 100 counties in 1990 when the PCS data were collected. We used these data to construct RANDR, a binary variable indicating whether or not an R&R existed in the local area in 1990.
Descriptive statistics are presented in Table 1. Table 1(a) contains summary statistics for most demand-side and supply-side characteristics. Table 1(b) contains summary statistics for the tax deductibility and regulatory variables.