Enhancing Wage Subsidies: Evidence from a Randomized Controlled Trial

We evaluate two new services delivered alongside Hungary’s youth wage-subsidy program.

(1) The employer-youth information package, or Performance and Skills Information Package (PSIP) is a structured goal-setting and feedback program, intended to clarify expectations and document progress during the subsidized job. It includes a jointly developed performance plan, a mid-term check-in and final employer evaluation, an automatically generated skills certificate summarizing competencies gained, and an anonymous youth satisfaction survey.

(2) Mentoring is a light-touch mentoring support package to facilitate workplace integration during the subsidy period. It includes two small-group sessions (around five youths) near the start and end of the subsidy, individual follow-up calls for all participants, and a Mentor Helpdesk providing on-demand support to youth and employers.

2026-02-02

2026-11-30

The primary outcomes capture job retention and post-program labor-market status and earnings. We will estimate impacts on completion of the wage-subsidy spell, employment status, and number of employed days, unemployment (registered jobseeker), inactivity, and monthly labor income.

Wage-subsidy completion is a binary indicator for completing the full 4- or 6-month subsidy period versus exiting earlier, measured at program exit at the youth level. Employment, number of employment spell days, unemployment, inactivity, and monthly labor income are measured at exit and at 3, 6, 12, and 24 months after exit, at the youth level.

Secondary outcomes capture additional dimensions of job quality and workplace experience for participants and employer perceptions. Because we test multiple outcomes, we organize outcomes into three pre-specified families:

(1) Skills and soft skills: self-esteem, motivation, perceived technical skills, perceived soft skills.

(2) Job quality and work experience: contract type, perceived employer support, employer perceived employer support, perceived job quality and reservation wages.

(3) Employer-perceived youth-level outcomes: perceived technical skills, perceived workplace integration, employer stereotypes.

Details of the exact measurement will be made available at the end of the trial period.

(1) Skills and soft skills outcome family:
(a) Self-esteem: measured at exit, 4-item scale constructed as the mean of items, 5-point Likert scale, standardized mean
(b) Motivation: measured at exit, 1-item question, 5-point Likert scale; we will also construct a binary indicator Agree = 1 if response is 4–5, 0 otherwise
(c) Perceived technical skills: measured at exit, 1-item question, 5-point Likert scale; we will also construct a binary indicator Agree = 1 if response is 4–5, 0 otherwise
(d) Work self-efficacy: measured at exit, 3-item scale constructed as the mean of items, 5-point Likert scale
(e) Reservation wages

(2) Job quality and work experience outcome family:
(a) Contract type: measured at exit, at 3, 6, 12, and 24 months
(b) Perceived employer support: measured at exit, 4-item scale constructed as the mean of items, 5-point Likert scale
(c) Perceived job quality: measured at exit, 3-item scale constructed as the mean of items, 5-point Likert scale

(3) Employer-perceived youth-level outcome family:
(a) Perceived youth technical skills: measured at exit, 1-item question, 5-point Likert scale; we will also construct a binary indicator Agree = 1 if response is 4–5, 0 otherwise
(b) Perceived youth workplace integration: measured at exit, 1-item question5-point Likert scale; we will also construct a binary indicator Agree = 1 if response is 4–5, 0 otherwise
(c) Employer support: measured at exit, 6-item scale constructed as the mean of items, response scale 0–3 (0 = Never; 1 = Once a month; 2 = Twice a month; 3 = Once a week or more often.)
(d) Employer stereotypes: measured at exit, 6-item scale constructed as the mean of items, 5-point Likert

Details of the exact questions will be made available at the end of the trial period.

We will run a cluster-randomized controlled trial in Hungary. Clusters are firms: all eligible youths hired by the same firm receive the same assignment. Randomization is implemented strictly after a PES caseworker matches a youth to a firm and is carried out continuously as matches occur. Firms are assigned to one of four arms in a 2×2 factorial design: Control, PSIP only, Mentoring only, and both PSIP and Mentoring.

Not available

Assignment to the 2×2 factorial treatment arms is implemented in the Public Employment Service IT system and is conducted automatically when an employer–youth match is finalized. Randomization occurs at the employer level within PES office strata using a unique employer identifier, so that each employer is assigned to one of the four study arms and retains that assignment for any subsequent eligible matches during the pilot period (i.e., employers are not re-randomized).

Randomization clusters are firms.

Yes

1900 employers

3000 workers

475 firms control, 475 firms mentoring only, 475 firms PSIP only, 475 firms both treatment.

Power calculations are based on a clustered 2×2 factorial design with randomization at the firm level and outcomes measured at the youth level. We assume 3,000 youths in 1,900 firms (1–2 youths per firm), equal allocation across the four arms, 80% power, two-sided α = 0.05, and ICC = 0.10–0.30. Calculations use historical administrative data for wage-subsidy completion, employment, unemployment, inactivity, and monthly labor income. Main effects (PSIP and Mentoring) - Wage-subsidy completion (baseline 0.86, variance = 0.12): 3.5–3.8 pp - Employment at 6-month follow-up (baseline 0.36, variance = 0.23): 5.1–5.4 pp - Unemployment (baseline 0.16 among labor force, variance = 0.13): 3.6–3.8 pp - Inactivity (baseline 0.055 among labor force, variance = 0.05): 2.7–2.8 pp - Monthly labor income (mean HUF 233,265; SD = HUF 151,796): 0.10–0.11 SD (≈ HUF 15,800–16,800, EUR 41–43) Interaction effect (PSIP × Mentoring) With equal allocation, the interaction contrast is less precisely estimated than main effects. The study is powered to detect interaction effects of approximately: - Completion: 7–8 pp - Employment: 10–11 pp - Unemployment: 7–8 pp - Inactivity: 5–6 pp - Income: ~0.20 SD (≈ HUF 32,000–34,000, EUR 83–88)