What the Research Says About Personal Reemployment Accounts -- Key Components of the Proposal Fail the Long-Term Unemploye
(February 2003)
by National Employment Law Project

As Congress debates the Bush Administration’s proposal to create a new national system of “personal reemployment accounts” (PRA), policy makers should evaluate whether the initiative effectively responds to the severity of the long-term jobless situation.  The long-term jobless now account for 20% of the unemployed (or about 1.7 million workers), and record numbers of workers are also running out of state unemployment and federal extended benefits.

Based on the Administration’s proposal, House Republicans have introduced the “Back to Work Incentive Act of 2003“ (H.R. 444) which would provide $3.6 billion over two years for states to offer a maximum of $3,000 for jobless workers to pay for a variety of employment services.  If a qualifying worker finds a job within a fixed period of time, the individual would be allowed to cash in his or her remaining grant as a “reemployment bonus.”  The proposal also relies upon a system of “worker profiling” to screen workers to determine if they are likely to exhaust their unemployment benefits and thus qualify for the PRA program.

What follows is a summary of the research evaluating key components of the PRA proposal, including reemployment bonuses and worker profiling.  It was prepared in response to the position of PRA proponents that the research results support expanding these initiatives.  As described below, the demonstration studies have consistently found that core elements of the PRA initiative have failed to produce substantial benefits to the long-term jobless and those living in areas hardest hit by unemployment.  Thus, while not evaluating the PRA program as a whole, prior studies raise serious questions about key features of the program especially when long-term unemployment continues to rise and limited jobs are being created.

Reemployment Bonuses
Proponents of the PRA initiative argue that reemployment bonuses have been proven in state demonstration programs to reduce a worker’s stay on unemployment because of the cash incentive provided to find a job sooner.  In the four demonstration states where the program was tested during the late 1980s, workers were provided with extra cash of about $500 to $1,700 if they found work within a specific period while collecting unemployment benefits. Under H.R. 444, the PRA recipient may cash in the remaining grant if he or she finds full-time work within 13 weeks.  However, 40% of the balance must be retained until the worker remains on the job for at least six months. The amount of the PRA and the bonus would be established by the states, but it cannot exceed $3,000.

• In three of the four demonstration states, the overall effects of reemployment bonuses were found to be extremely limited, reducing unemployment on the part of the participants by just one-half of one week.

Evaluating the results of the New Jersey, Pennslyvania and Washington demonstration programs, Northwestern University Professor Bruce Meyer concluded that the overall impact of the reemployment bonus initiative was minimal:

“In New Jersey, the job-search assistance (JSA) and bonus treatment has a large estimated effect of almost one week, and this difference from the control group is statistically significant.  However, when one compares this treatment to the JSA only treatment, one sees that the estimated additional effect of the bonus is much smaller and only marginally significant.  In both Pennslyvania and Washington the estimated effects of the different bonuses on weeks of benefits are mostly small and of weak statistical significance.  To the put the magnitude of the estimated effects in perspective, the average weeks of UI received in the benefit year by the control group range from 15 to 20 weeks in the different experiments.  The effects of the bonuses cluster around one-half of a week, or about three percent of the average duration of receipt.” (1)

• The impact of reemployment bonuses was consistently found to be most limited in areas hardest hit by unemployment and as applied to those workers who were long-term unemployed.

According to the evaluation of reemployment bonuses conducted in Washington State in 1988, “claimants in areas experiencing moderate or high [total unemployment rates] responded weakly or not at all” to the program. (2)  In Washington, unemployment insurance payments were reduced by $201.45 (per claim) over one year in those low unemployment areas of the state that participated in the program.  However, in those areas of the state with moderate or high unemployment, unemployment insurance payments actually increased by $17.50, thus indicating no overall effect of the reemployment bonus program. (3)

An evaluation of the New Jersey demonstration program found that, “the bonus offer reduced UI receipt primarily among claimants who otherwise would have collected 24 weeks or less of UI benefits. The bonus offer does not appear to have had much of an impact on the UI spells of the long-term unemployed who were likely to exhaust their benefits in their first UI spell.”    (4)

Finally, according to the Pennsylvania evaluation, “most of the bonus offers reduced UI receipt more among relatively short-term claimants than among long-term claimants, who were likely to exhaust their UI benefits.” (5)   The report concluded:  “much of the reduction in UI receipt caused by the treatments occurred because the treatment effect was concentrated among claimants who faced relatively  short potential UI spells otherwise.  The treatments do not appear to have had much of an impact on the UI experience of long-term claimants who were likely to exhaust their benefits in their initial UI spell.”  (6) Although not considered by the study, the impact of the bonus would likely be even less significant if applied to workers who not only exhausted their state unemployment benefits, but their federal extended benefits as well.

• The impact of reemployment bonuses was found to be especially limited as applied to displaced workers and others who are more structurally unemployed, while primarily benefiting those who already possessed the most marketable employment skills.

Another evaluation of the New Jersey program focused on the limitations of the program as applied to those workers who faced the most challenges finding work.  According to the study:

“[A]n examination of the impacts of the treatments by population subgroup suggests that treatments were most successful promoting the reemployment of individuals who had marketable skills, such as clerical and other white collar workers.  The treatments were less successful for individuals who faced hard-core, structural unemployment problems, such as blue collar workers, workers from durable goods manufacturing industries and permanently separated workers. That is, the displaced workers with more severe reemployment problems may have been less affected by the demonstration treatments than were other workers who faced relatively more favorable reemployment prospects.” (7)

Specifically, the study found that those white collar and clerical workers who received job search assistance and a reemployment bonus increased their earnings by between $561 and $705 over two quarters, while manufacturing workers who participated in the program decreased their earnings by $18 to $163. (8)

Worker Profiling
The PRA initiative also requires the states to apply a system of “worker profiling” to determine whether most workers qualify for the program.  As a condition of receiving unemployment benefits, the states currently seek to determine through various individual and economic factors whether a worker is likely to exhaust his or her unemployment benefits.  If the worker  receives a high enough score based on the profiling screens (usually ranging from 1-20, with higher scores indicating workers with longer expected unemployment), then the individual is required to participate in additional reemployment services that are most often provided by the U.S. Employment Service.

• The latest worker profiling evaluation calls for additional research before the policy is extended to other programs, citing the program’s failure to serve as an effective screen for the purposes of delivering employment services to those who need the help most.

According to the most recent evaluation of worker profiling, there is “little evidence of any systematic relationship between the estimated impact of treatment and the profiling score.  This suggests that such profiling does not increase the efficiency of treatment allocation and indicates the potential value of further research on econometric methods of treatment allocation before extending profiling to other programs.”   (9)

• Not unlike reemployment bonuses, worker profiling is considered less effective when applied to workers who are long-term unemployed.

According to the latest study of the program, “the treatment has its largest impact on persons whose spells without treatment would be of moderate duration.” (10)  The study later elaborates on this finding, concluding that there was “little justification” for the assumption underlying the worker profiling program, that “those with the longest expected UI spells benefit the most from the profiling treatment.” (11)

The study was conducted in Kentucky during the period from 1994-1996 when the state’s economy was found to be “extremely strong.” (12)  As a result, most of those with high profiling scores (83%) were also able to receive reemployment services provided by the local offices.  The services included an initial  orientation followed by more intensive job search and job preparation activities.  About 14% of those who attended the orientation were referred to relatively more expensive education and training programs.  When unemployment rises and more people exhaust their unemployment benefits, far fewer workers receiving high profiling scores are likely to receive reemployment services.

Conclusion
The President’s PRA proposal raises a number of fundamental concerns not limited to those examined by the research described above.  For example, there is no guarantee in the legislation that the proposed funding for PRAs will not end up diverting resources devoted to the nation’s employment and training services (provided by a seriously under-funded Workforce Investment Act (WIA) program) or undermining efforts to extend unemployment benefits.  Moreover, the program denies many WIA services to PRA participants, and PRAs would not be available to most of the long-term unemployed.  However, as tested empirically in the states, it is fair to say that key features of the PRA proposal have failed to serve the needs of the long-term unemployed.  Thus, while not evaluating the PRA proposal as a whole, the research raises serious questions about these components of the initiative as applied to today’s growing problem of long-term joblessness.

References
(1) Meyer (1995) “Lessons from the U.S. Unemployment Insurance Experiments,”  Journal of Economic Literature, Volume 33:1 (March), 91-131, at p. 100
(2) Spiegelman, O’Leary and Kline, The Washington Re-Employment Bonus Experiment Final Report (W.E. Upjohn Institute, 1992), at p. xviii.
(3) Id. At  p. 144.
(4)Decker (1995) “The Impact of Re-employment Bonuses on Insured Unemployment in the New Jersey and Illinois Reemployment Experiments,”  Journal of Human Resources, 29:3, 718-741, at p. 733.
(5)Corson, Decker, Dunstan & Kerachsky, The Pennslyvania Reemployment Bonus Demonstration Final Report (Mathematica Policy Research, 1991), at p. xviii.
(6)Id. at p. 107.
(7)Corson, Decker, et al. (1989) The New Jersey Unemployment Insurance Reemployment Demonstration Project, Final Evaluation Report (Mathematica Policy Research), at p. xi.
(8)Id. at p. 297.
(9)Black, Smith, Berger & Noel, Is the Threat of Reemployment Services More Effective that the Services Themselves?  Experimental Evidence from the UI System,(September 2001), at pp. 4-5.
(10)Id. at p. 4.
(11)Id. at p. 28.
(12)Id. at p. 7.