n.V1-7.10 | SECTORAL EMPLOYMENT PROGRAMS
01| The best-evidenced employment model in this book part does something the previous chapter's programmes do not. It picks the industry first.
02| The WorkAdvance demonstration screened applicants up front, trained them in the occupational and soft skills of a specific higher-wage sector, and wrapped support services around them through placement and into the job. The earnings gains were substantial and they persisted — in the range of 12 to 34 percent following training, which is unusual in a literature full of nulls.
03| Then the finding that matters more than the effect size. The gains did not come from raising employment rates, and they did not come from the wraparound support. They came from moving participants into higher-wage jobs in higher-earning industries and occupations. The mechanism was the destination.
04| This is a quiet demolition of a claim that appears everywhere in the sector: that what makes training work is the coordination and support around it — the case manager, the follow-up, the help with transport and childcare and confidence. In the best-documented programme of its kind, all of that was present, and something else did the work. Support services were an enabling condition. Sector targeting was the cause.
05| MDRC's longer-run analysis adds the necessary caution: not all sectoral programmes produce these gains. Choosing an industry is not sufficient; choosing an industry that is actually hiring at higher wages, in a labour market the participants can reach, with employers willing to take them, is a much harder operational problem than running a course. The programmes that fail are the ones that pick the sector from a policy document rather than from the local economy.
06| For anyone reading this as a design lesson, the honest version is uncomfortable. It says the valuable capability is not pedagogic but commercial: knowing which employers are expanding, what they will accept, and what they pay. That is labour-market intelligence and placement work. Most training organisations are structured, staffed and funded to deliver courses instead, and are measured on completions.
07| The distributional question is the one the literature leaves alone. Sectoral programmes move participants into better jobs, plausibly ahead of other applicants for the same jobs. If the higher-wage sector has a fixed number of openings, the programme has redistributed access rather than created it, and the correct evaluation would need to measure what happened to the people who did not get in. The trials measure participants against a control group of non-participants, which cannot detect displacement of that kind.
08| Open, and directly answerable: a factorial design — sector targeting with and without wraparound support, in both arms — would separate the two ingredients and price them. It has not been run, so the relative contribution is inferred from a mechanism analysis rather than measured.
