The JOBSTART Demonstration, which ran between 1985 and 1988, was patterned after the Job Corps vocational education program but implemented in a nonresidential setting. The program was developed by the Manpower Demonstration Research Corporation (MDRC) to target low-skill, young school dropouts, and its operating costs were primarily funded through the Job Training Partnership Act of 1982. Hendren and Sprung-Keyser (2020) estimate the MVPF for JOBSTART based on the final report of the MDRC evaluation of the demonstration, Cave et al. (1993).
MVPF = 0.2
Cave et al. (1993) report that the operating costs of JOBSTART were $4,548 (in 1986 dollars) for each participant, and that individuals in the treatment and control groups incurred similar levels of costs from non-JOBSTART services. Cave et al. (1993) estimate that about 15 percent of the $4,548 program costs went to support services for JOBSTART participants such as child care and bus passes. The impacts of JOBSTART on AFDC ($73), food stamp (-$25), and General Assistance ($28) payments were reported for four years following the program, but none of these differences in transfer receipt are statistically significant in the aggregated sample. The final net cost amount is $4,620 (95% CI: [$3,648, $5,648]).
Cave et al. (1993) report statistically significant earnings losses in the first year ($499), but earnings gains in the last three years of observed earnings, summing to a $213 increase in earnings over the four years of observed earnings (this estimate is not statistically distinguishable from $0). In their baseline estimates, Hendren and Sprung-Keyser (2020) calculate tax rates from Congressional Budget Office estimates (0.7%) and assume that participants value JOBSTART as the sum of (a) the program’s post-tax earnings and transfers impacts and (b) support services provided to participants, discounted back to the start of the program at a 3 percent discount rate. The resulting WTP estimate is $911.65 (95% CI: [$255.56, $1,531.42]).
The resulting MVPF is 0.20 (95% CI: [0.04, 0.42]). The imprecision of the estimates for the observed four-year horizon is compounded when Hendren and Sprung-Keyser (2020) attempt to forecast these gains into the future. At a 21-year projection length, Hendren and Sprung-Keyser (2020) obtain a point estimate of 10.31 but Hendren and Sprung-Keyser (2020) cannot reject an MVPF of either 0 or \infty at a 95-percent confidence level.
If Hendren and Sprung-Keyser (2020) instead assume that participants valued JOBSTART at cost, the resulting MVPF would be 0.98 (95% CI: [0.80,1.24]). Projecting earnings effects to 21 years under this alternative assumption yields an MVPF of 3.20 (though Hendren and Sprung-Keyser (2020) cannot reject an MVPF of either 0.70 or \infty).
Cave, George, Hans Bos, Fred Doolittle and Cyril Toussaint (1993). “Jobstart: Final Report on a Program for School Dropouts.” New York: Manpower Demonstration Research Corporation. https://www.mdrc.org/sites/default/files/full_416.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