Head Start Impact Study
Introduction
Head Start is the largest early childhood education program in the United States. It offers education, health, and nutrition services to disadvantaged children and their families. The program started in 1965 and by 2013 enrolled about 900,000 three and four-year-old children at a cost of $7.6 billion.
Kline and Walters (2016) evaluate the Head Start Impact Study (HSIS), a randomized control trial conducted between 2002 and 2006 that assigned a random sample of roughly 5,000 three- and four-year-old children into Head Start programs across the United States. One of the main findings of the paper is that the program’s net impact is significantly increased when considering that around one third of its participants are drawn from other competing preschool programs. This crowd-out reduces the net costs to the government (given that many of the competing programs also receive government funds), and is an important factor when constructing net costs of the policy.
The authors estimate the impact of Head Start on test scores, and use a projection method to translate this impact into lifetime earnings. The authors use this along with their upfront cost calculation to form the policy’s MVPF, which they estimate to be 1.85 in their baseline specification.
MVPF = 1.8
Net Cost
To compute the net cost of Head Start, the authors consider three components: i) the direct costs of the Head Start program to the government, ii) savings to the government due to substitution away from competing publicly funded preschool programs, and iii) increase in government income tax revenues due to increased lifetime earnings among program participants.
To compute direct program costs (i) authors use data from the U.S. Department of Health and Human Services, which estimated average Head Start provision costs at $8,000 per pupil per year in 2013. To estimate ii) Kline and Walters use the estimate that 33.8 percent of Head Start enrollees substitute away from other programs and assume that these other programs have a marginal enrollment cost that is 75 percent of Head Start’s. Together these imply $1,564 in savings to the government from lower enrollment in other childcare programs. To estimate iii) the authors rely on the method described above to calculate impacts on lifetime earnings. The authors then multiply the increase in lifetime earnings by a tax rate of 34 percent, obtained from the U.S. Congressional Budget Office to arrive at the increase in tax revenue caused by program participation. The final preferred net cost estimate is $2,991.
Willingness to Pay
To compute the willingness to pay for Head Start participation, the Kline and Walters (2016) assume that the children in Head Start value the increase in their lifetime earnings on a dollar-for-dollar basis. To calculate the impact on lifetime earnings, the authors rely on their own estimates of the causal impact of Head Start attendance on standardized test scores. To translate the program’s impacts on standardized test scores into impacts on lifetime earnings, the authors rely on a review of the relevant literature and in particular estimate from Chetty et al. (2011) to calibrate that a one standard deviation increase in early-childhood test scores leads to a 10 percent increase in lifetime earnings. Finally, the authors multiply the increase in earnings by a net-of-tax rate of 66% to arrive at the post-taxes earnings increase caused by the program. The final preferred WTP estimate is $5,513.
MVPF Summary
Dividing the WTP by the net cost, the authors arrive at an MVPF of 1.84 (s.e. 0.47) with their preferred set of assumptions. They also show under more pessimistic assumptions the MVPF is 1.5 (s.e. 0.34). Under the assumption that competing public preschools are rationed, the MVPF rises to 2.02.
In a corrigendum to the original paper, Kline and Walters (2023) report an error in the Online Appendix of the original paper. This error causes the MVPF under the assumption that competing public preschools are rationed to increased from 2.02 to 3.60.
In subsequent work that builds on Kline and Walters (2016), Hendren and Sprung-Keyser (2020) also use the estimates from Kline and Walters (2016) to compute the MVPF of the Head Start Impact Study. In that case, Hendren and Sprung-Keyser (2020) include one additional component into the willingness to pay for the policy. The adjust willingness to pay upward upward because Head Start results in an out-of-pocket savings for parents who would have sent their children to private preschool. If that additional benefit is incorporated, the MVPF rises from the 1.85 to 2.41.
References
Chetty, Raj et al. (2011) “How Does Your Kindergarten Classroom Affect Your Earnings? Evidence from Project STAR.” The Quarterly Journal of Economics 126(4): 1593-1660.
https://academic.oup.com/qje/article-pdf/126/4/1593/17089543/qjr041.pdf
Hendren, Nathaniel and Ben Sprung-Keyser (2020). “A Unified Welfare Analysis of Government Policies.” The Quarterly Journal of Economics, 135(3): 1209–1318. DOI: https://doi.org/10.1093/qje/qjaa006
Kline, Patrick and Christopher R Walters (2016). “Evaluating Public Programs with Close Substitutes: The Case of Head Start.” The Quarterly Journal of Economics, 131(4), 1795-1848. DOI: https://doi.org/10.1093/qje/qjw027
Kline, Patrick and Christopher R Walters (2023). Corrigendum to “Evaluating Public Programs with Close Substitutes: The Case of Head Start.” https://drive.google.com/file/d/1MdRQfashVSLYCL5djUrneZaUCVFxRvnj/view
Policy Features
- Category
- Education
- Sub-Category
- Child Education
- Beneficiary Type(s)
- Children and Youth, Children under five, Students
- Average Age
- 4
- Country of Implementation
- United States
- Year of Implementation
- 2002
- Empirical Method
- RCT
- Research Type
- Primary
- Peer Reviewed
- Yes
- MVPF Publication Link
- academic.oup.com/qje/article/131/4/1795/2468877