The goal of this project is to provide evidence on the long-run impact of publicly-subsidized child care in the U.S. It uses as the policy "laboratory" child care subsidies funded through the Lanham Act of 1940. The Lanham Act was passed in response to the growing need for child care among women called into the labor force during World War II. The federal government provided grants to local communities in designated "war impact" areas to construct and operate child care centers during the war.

The Act was in operation starting in mid-1943, and was abruptly terminated in early-/mid-1946. During this 4-year period nearly $100 million (in 1940s $) was spent, making it the most expansive child care assistance program in U.S. history. The subsidy was generous—parents were required to pay only $0.50 per child per day. There was no means-testing to determine eligibility or to calculate the parent co-payment.

The Act provided subsidies for preschool- and school-aged child care centers. The focus of my analysis will be on the preschool program. There were fairly strict requirements that the Act not provide subsidies for children ages 0 to 1; only preschool children ages 2-5 were allowed to enter child care.

To identify the impact of the Lanham Act, I take advantage of the timing of the Act (1943-1946) and its target age range (ages 2-5) to define a set of treatment and comparison cohorts. The comparison cohorts were born between 1933-1937, and thus were never in the 2-5 age range when the program was in operation (ie., pre-reform cohorts). The treatment cohorts were born between 1938-1942, and thus were exposed to varying degrees to the program (ie., post-reform cohorts).

The other piece of information I have is state-level Lanham Act (child care) expenditures. These are based on local applications to the federal government requesting money to build/operate the centers. There is substantial variation across states in how much funding was received, and I will exploit this variation to proxy one’s exposure to the program, ie., to create treatment and comparison areas.

The estimating equation is:

Yics = Ac + b1SPENDING + b2TREATED + b3(SPENDING * TREATED) + bX + bZ + e

where:
Y is some education/labor market outcome for person i, in birth cohort c, and state of birth s
A are cohort fixed effects, or some control for age of person i
SPENDING is state-level Lanham Act expenditures (in i's state of birth)
TREATED=1 if person i is in one of the treatment cohorts
X is a set of individual characteristics
Z is a set of 1940 (pre-war) state characteristics

The model will be estimated on 1970, 1980, and 1990 U.S. Decennial Census data, allowing me to look at outcomes for treated cohorts during their 30's, 40's, and 50's. For example, treatment cohorts are late-20's to early-30's as of the 1970 Census.

The coefficient of interest is b3, which provides the differential impact (or diff-in-diff estimate) of Lanham Act spending across treatment and comparison cohorts.