There is a huge need for job creation in the U.S. Given the length and severity of the recession, we are short around 12 million jobs compared to what would be needed to be at pre-recession labor market conditions.

I’ve been asked in recent days by several journalists about Georgia Works, because there are rumors that there is some interest by the White House in this program as a model for helping the long-term unemployed. Georgia Works is a program under which workers receiving unemployment benefits are provided the opportunity of on-the-job-training/work experience at participating local employers.  The OJT/work experience is 24 hours per week for 6 weeks. The worker receives some extra stipend, intended to reimburse work-related expenses; this stipend over the program’s history has varied from $40 to $100 per week. The employer is not required to hire the worker at the end of the 6 weeks, although the employer is supposed to have a job vacancy. Other states have similar programs, for example New Hampshire has a program called “Return to Work””, although that program does not pay workers a stipend.

The question is: are programs such as Georgia Works a good way to create better job opportunities for the unemployed? I am probably being asked such questions because I have written about job creation programs for the unemployed. In particular, I have praised the MEED program, which Minnesota used in the 1980s, which also subsidizes employers to help the unemployed.

Under MEED, employers received up to a $10 per hour wage subsidy (adjusted to today’s price level) to hire targeted disadvantaged workers for six months for a full-time job slot. The targeted workers under MEED were any unemployed worker who was NOT receiving unemployment benefits.  The subsidized jobs had to be newly created jobs, and the employers had to pay the usual wage they paid for that type of job. (For example, if the job paid $9, the employer would pay $9 and receive a wage subsidy of $9; if the job usually paid $12 per hour, the employer would pay $12 and receive a wage subsidy capped at $10 per hour.) Employers had to make a good faith effort to retain the subsidized worker for one year after the six-month subsidy period. In theory, if employers did not make such a good faith effort, they could be required to repay the MEED subsidy they received; in practice, employers either were asked to hire another MEED worker, or were excluded from future participation in the program.

My summary comment is: In current economic circumstances, when the U.S. economy is short of jobs, programs such as Georgia Works or New Hampshire’s Return to Work do not do enough to deal with the key problem, which is lack of jobs.  I believe we would be better off with programs such as MEED, which cost a lot more because they provide much bigger subsidies to employers, but also expect employers to do a lot more, in particular they expect employers to create jobs, which is what we should be targeting at present. Because Georgia Works and New Hampshire’s Return to Work do not create new job slots, at best they redistribute who fills currently available job opportunities. Reshuffling scarce job opportunities among the unemployed has relatively modest benefits in current economic circumstances.  We need to induce employers to create new jobs, while targeting some of these new job opportunities to the long-term unemployed, and MEED targets these goals.

More detailed comments:

1. In general, the idea of subsidizing employers to create new job opportunities for the long-term unemployed makes sense. We need to create new jobs in the current U.S. economy. And the long-term unemployed are particularly damaged by the current recession. The longer a spell of unemployment goes on, the more likely that spell will lead to permanently reduced earnings prospects for the worker.
2. Georgia Works or New Hampshire’s Return to Work are far smaller subsidies to employers than was Minnesota’s MEED. The subsidy under MEED was up to $10,000 per hired worker ($10 per hour times 1000 work hours for those 6 months). The subsidy under the Georgia and New Hampshire programs is the value of the free labor. The employer gets 144 hours of free labor (24 hours per week times 6 weeks). The subsidy is reduced to the extent the employer provides training in excess of what they ordinarily would have provided. It is hard to imagine that the typical value of the free labor exceeds $10 per hour, so the employer subsidy is at most $1,440 per worker, reduced by the cost of any extra training.
3. But Minnesota’s MEED program also expects a lot more. In particular, MEED expected employers to create a new job slot in addition to already existing jobs, whereas Georgia Works and New Hampshire’s Return to Work only expect the employer to consider a subsidized worker for an existing vacancy. The MEED program also has stronger pressure for the employer to continue employing the worker after the subsidy period.
4. Of course, the fact that MEED jobs had to be new jobs doesn’t mean that they would not have been created “but for” the MEED program. We don’t have rigorous evidence from random assignment experiments of MEED’s experience in actually creating jobs. But surveys of private employers participating in MEED suggest 60% would not have expanded but for the MEED program subsidies. This percentage increases to 68% for the smallest private employers. There are good reasons why MEED might particularly affect job expansion decisions of smaller employers. When asked why the MEED program triggered an employment expansion, employers particularly referenced cash flow problems, which would be a bigger issue for smaller employers.
5. MEED also seems to have much greater retention of subsidized workers. 78% of MEED workers completed the six month subsidy period plus at least 60 additional days.  Various reports cite figures ranging from 24% to 50% of workers participating in and completing Georgia Works were hired by the employer they were placed at. Because the Georgia Works figures are based on those who completed the program, the percentage hired by the subsidized employer would be even lower if calculated as a percentage of those starting the program, which would be a more comparable figure to the MEED statistics.
6. MEED targets a needier group than Georgia Works or New Hampshire’s Return to Work. MEED targeted unemployed workers who either were not qualified for unemployment benefits, or who had exhausted such benefits. The Georgia and New Hampshire programs target unemployed workers receiving unemployment benefits.
7. All of these programs to some extent reshuffle available job opportunities among the unemployed. But based on the survey evidence, 60% of the MEED subsidies also increase the total number of jobs.  I have calculated that MEED subsidies create new jobs at a cost of about $34,000 per job. This is about one-third of the typical $100,000 per job-year cost of traditional fiscal stimulus. In contrast, it is hard to see how Georgia Works or New Hampshire Return to Work create ANY jobs. These Georgia and New Hampshire programs may not cost much, but their cost per job created is probably infinite given that they don’t create any jobs.
8. MEED also redistributes job opportunities to a needier group, that is those unemployed who are not receiving unemployment benefits. In the long-run, this redistribution may boost the overall employability of the labor force, as this needier group is more likely to have their skills adversely affected by the recession if something is not done to help retain those skills. After the economy fully recovers from this recession, this greater long-run employability may boost overall U.S. employment by boosting the effective labor supply.
9. In contrast, it is hard to see what benefits there are from any success that Georgia Works and New Hampshire’s Return to Work have in increasing the employment of their clients. For example, as mentioned, 24% to 50% of the unemployed who entered and completed Georgia Works ended up being hired by that employer.  Reportedly, more than 60% of those in Georgia Works found work somewhere. Suppose for a moment that none of these Georgia Works clients would have found a job without the program, which is an extreme and unrealistic assumption. Even under this unrealistic assumption, because Georgia Works has not expanded the total number of jobs created, all that has happened is that available jobs have been reshuffled towards Georgia Work’s clients, and away from other workers. Because Georgia Works clients do not seem likely to be particularly needy compared to the average unemployed person, it is difficult to see large social benefits from this redistribution of available jobs.

For all these reasons, if we are going to have an employer subsidy to create job opportunities for the unemployed, I prefer a program similar to MEED over programs similar to Georgia Works.  As outlined in a previous blog post, I also think other job creation policies should be used.

One second-best option for helping the long-term unemployed is providing a grant program to states that would leave it to state discretion how to design the employer subsidies. If Minnesota wants to do a program similar to MEED, and Georgia wants to expand Georgia Works, each state would have that option. These state experiments with helping the long-term unemployed should then be rigorously evaluated.  I personally think we already know enough to suspect that the MEED approach works better than the Georgia Works approach, but obviously we could always use better evidence.