Panel Paper: Conducting Return on Investment Analysis of Job Training Programs Using Linked Program and Administrative Data: Findings and Lessons Learned

Thursday, November 2, 2017
Gold Coast (Hyatt Regency Chicago)

*Names in bold indicate Presenter

Joseph Gasper, Westat


Despite evidence on the effectiveness of occupational training in workforce development, little is known about the return on investment (ROI) of these programs. Administrative data is an important resource for understanding the cost-benefit of job training programs. The purpose of this study is to examine the ROI of four NYC job-training programs: (1) Individual Training Grants (ITGs) are vouchers that cover the cost of occupation skills training; (2) Sectoral employment trainings include cohort trainings in high-growth industries including transportation, construction, advanced manufacturing, healthcare, and information technology; (3) The On-the-Job Training (OJT) Program provides employers with grants to cover 70 percent of up to 280 hours of training for new employees; (4) The Customized Training Program funds most of the cost for employers of providing training to low-wage incumbent workers.

We linked program data on 10,000 participants to their administrative data on UI wages, UI payments, welfare benefits, food stamp benefits, Medicaid enrollment, and criminal justice involvement. To create a comparison group, the study accessed program data on over half a million individuals who received workforce services through Workforce1 Career Centers, the City’s branding of One Stop Career Centers, over a five-year period. The administrative data were provided for several years before and after enrollment for participants.

The study uses propensity score matching (PSM) technique to estimate the net impacts of the training programs. The differences in earnings, UI payments, welfare benefits, food stamp benefits, and Medicaid enrollment were compared between participants who received training and a similar group that only received job search assistance. In terms of program costs, we work with each program to obtain their specific cost data, whether at the participant level or for the program as a whole. When the data was provided at the program level, we computed an average cost per person and used that in the ROI analysis which was conducted from the taxpayer and societal perspectives.

This paper will: (1) highlight the substantive findings from the job training ROI study and (2) discuss the challenges and solutions of using linked administrative data to conducted ROI analysis of local job training programs. This study is unique in that it had access to rich administrative data for a large number of program participants and a comparison pool and includes data over two years before and up to 3 years after participants enrolled in the program. In addition to presenting study results, we will discuss challenges that arose in accessing the data, the availability of key variables needed for propensity score matching, data cleaning and data quality, and the study timeline.