An official website of the United States government
The Unemployment Insurance (UI) system—consisting of 53 UI programs ran by 50 states, the District of Columbia, Puerto Rico, and the Virgin Islands—provides temporary financial assistance to workers who lost their jobs involuntarily, with the objective to help them sustain their quality of life and make efficient job choices. Unemployed workers may file a UI claim and, depending on state-specific UI eligibility requirements, may qualify to collect a certain number of weekly benefit payments during the claim that expires 1 year after it was filed. Benefits are drawn from the UI Trust Fund, which in most states is exclusively financed through an employer tax. Although many recipients find jobs while collecting UI and remain employed for long periods, others are not able to do so and experience unemployment soon after their claim’s expiration. These recipients may return to the UI program and start collecting benefits on a new claim; this article refers to this phenomenon as “repeat use of UI.”
Repeat users are likely to be recipients with a less stable attachment to the workforce, either because of lack of necessary skills and job search resources to find a sustainable job or because of the cyclical nature of their work. Repeat use also may be a result of moral hazard—some recipients may not exert the job search effort required to find sustainable work, relying on the insurance provided by the program. An important consequence is that repeat users collect benefits on multiple claims, so they impose a potentially substantial burden on the UI Trust Fund. Although high prevalence of repeat use would have important implications for the effectiveness and solvency of state UI programs, previous research has largely ignored the issue.
This article fills this research gap by examining repeat use in seven states that represent a wide spectrum of state UI programs within the entire U.S. UI system: Pennsylvania, North Carolina, Oregon, Iowa, Montana, Hawaii, and New Hampshire. The analyses rely on state UI administrative data that provide information on all unemployed workers who started collecting benefits in 2003, a year when the unemployment rate was 6 percent and the insured unemployment rate was 2.8 percent, about the average rates in the United States in the past 20 years. These data provide the following information: recipient socioeconomic characteristics at program entry, benefit amounts collected until the expiration of the 2003 claim, and benefit amounts collected on a new UI claim that started within 12 months of the end of their 2003 claim. Using these data, this article examines the prevalence and costs of repeat use for two types of UI recipients—permanently displaced workers and temporarily laid-off workers.
The results show that more than half of recipients on temporary layoff and nearly a third of displaced recipients started a new claim within 12 months of the expiration of their 2003 claim. Further analysis shows that, among both temporarily laid-off and displaced recipients, repeat use was higher for recipients with relatively weak employment history, those usually employed in volatile sectors and low-skill jobs, low-education workers, and older workers. Repeat users collected substantial benefit amounts on their repeat claims; as a result, the benefit amounts collected by repeat users substantially exceeded those collected by nonrepeat users in all seven states. Finally, the article considers whether providing reemployment assistance to displaced recipients is an effective policy for reducing repeat use. Using Pennsylvania and Hawaii data, the article finds that displaced recipients who were referred to reemployment assistance at the start of their initial claims were much less likely to become repeat users, thus collected substantially lower benefit amounts on repeat claims than their peers. These results provide some evidence that reemployment assistance may be an effective strategy for reducing repeat use and alleviating its burden on the UI Trust Fund.
The U.S. UI system was established in 1935 with the passage of the Social Security Act, largely as a response to the Great Depression, with the objective to insure U.S. workers against temporary periods of involuntary joblessness. The program provides short-term wage replacement to workers who lost their jobs involuntarily in order to help them sustain their quality of life and make efficient job choices during periods of financial strain. Each state administers its own program and has a unique set of rules for determining whether unemployed workers who file a UI claim are eligible for benefits.1 In all states, only UI claimants who lost their job involuntarily are eligible for benefits—job leavers, those who lost their jobs for cause, and new labor force entrants are ineligible. Furthermore, to qualify for benefits, claimants are usually required to have positive earnings in at least two calendar quarters during the claim’s base period.2
Once a claimant is deemed eligible for UI benefits, program administrators use information on the claimant’s employment and earnings during the base period to determine the weekly benefit amount and the maximum number of weekly benefit payments the claimant is eligible to receive on the claim. Typically, recipients with stable employment throughout the base period are eligible to receive the maximum benefit amount allowed and up to 26 weeks of benefits on their claim, which expires 1 year after the date it was filed.3 Recipients with less stable work history may be eligible for lower benefit amounts and/or fewer weeks of benefits depending on state rules. At any time after the expiration of the claim, recipients may file for a new UI claim, in which similar criteria are used to determine eligibility.
Benefits are drawn from each state’s UI Trust Fund, which in all states—except Alaska, New Jersey, and Pennsylvania—is exclusively financed through an employer tax.4 From 2001 through 2010, more than 96 million new UI claims were started in the United States, of which approximately 24 million (25 percent) were started in 2008–2009, the midst of the most recent recession.5 During the same period, UI recipients collected $424 billion in benefits, of which $138 billion (33 percent) were collected in 2008–2009. These figures illustrate the importance of the UI program in the modern U.S. economy, particularly during recessions when demand for unemployment benefits is high.
Research on the effectiveness of the UI program would fill hundreds of volumes. Previous work shows that about one-third of unemployed workers in the United States receive UI benefits.6 Low UI receipt is attributed to several factors, including the fact that job leavers, those who lost their jobs for cause, and new labor force entrants are ineligible for benefits. Low UI receipt also is because of strict eligibility requirements in some states7 and because many unemployed workers do not know that they may be eligible for benefits.8 Nevertheless, research shows that UI benefits are sufficient to assist those who do receive benefits to avoid major drops in their consumption through periods of joblessness,9 providing substantial countercyclical stimulus for the U.S. economy.11 and has an adverse effect on the work search effort of unemployed workers and of employed workers who face an impending layoff.12 Another area of concern is maintaining the solvency of the UI Trust Fund without compromising the fundamental objective of the program.13 By retaining the solvency of their UI Trust Fund, states avoid seeking federal support or increasing employer taxes to cover a potential deficit. Previous work has discussed several strategies for controlling benefit payments and maintaining UI Trust Fund solvency, including implementing stricter eligibility requirements, adjusting benefit amounts and duration to reduce disincentive effects, imposing strict work search requirements for retaining eligibility, and providing reemployment and job search assistance to hard-to-employ recipients.15 In the other article, McCall finds more than one-third of participants in the 1979 National Longitudinal Survey of Youth had multiple UI claims through 2002.16 This research, however, is dated and provides little evidence on the prevalence of repeat use in the modern U.S. economy, the characteristics of repeat users, and the financial burden repeat use imposes on the UI Trust Fund.
In Canada, repeat use and its implications for the Canadian UI system have received significant attention from researchers and policymakers. In its efforts to assess repeat use in the Canadian UI program, Canada’s Department of Human Resources and Skills Development funded the Survey of Repeat Use of Employment Insurance, a nationally representative study of 30,000 individuals who received UI benefits in 1996. Statistics Canada conducted the survey and collected information on UI use for the period 1996–1998. Using the survey responses, two separate studies found that more than half of all individuals who collected UI in 1996 had at least one repeat claim during the study period.17 These studies also showed that men, older workers, workers with no high school diploma, and workers in construction and agriculture, fishing, and forestry were significantly more likely than average to become repeat users.
Another study examined repeat use by relying on Canadian UI administrative data for the period 1971–1989, showing that about 40 percent of male and female displaced workers who started a UI claim during the study period returned to the program within a year of the start of their initial claim.18 The same study showed that repeat use was more prevalent for younger workers; workers in construction, agriculture, forestry, and fishing; and individuals with an inconsistent employment history. A study by de Raaf et al. examined repeat use for workers employed in seasonal jobs using the 1993 and 1996 Survey of Labour and Income Dynamics, a longitudinal panel survey that followed a representative sample of Canadian households for a period of 3 years.19 The study showed 61 percent of workers employed in seasonal jobs had repeat UI claims during the study period; men, older workers, workers with no high school diploma, and married workers were significantly more likely than average to have repeat UI claims.
Studies show that certain groups of displaced workers are much more likely than others are to become frequent UI users.The results of these studies, although dated, show that repeat use is very common in the Canadian UI system, possibly suggesting that many unemployed workers lack the skills needed to secure sustainable employment or that the existence of the program provides some workers with the opportunity to hold temporary or seasonal jobs—instead of pursuing more permanent jobs—and use UI to replace a portion of their lost earnings when their employment ends. The studies also show that certain groups of displaced workers are much more likely than others are to become frequent UI users. These results raise concerns about the efficacy of the Canadian UI system and have important implications for future policy decisions to improve its effectiveness and financial viability. Similar research is essential to shed light on the prevalence and consequences of repeat use in the U.S. UI program; the objective of this article is to fill this research gap.
For the analyses, this article uses UI administrative data from seven states that represent a wide spectrum of UI programs: Pennsylvania, North Carolina, Oregon, Iowa, Montana, Hawaii, and New Hampshire. These data provide information on the universe of unemployed workers who started a UI claim in 2003, a period of moderate unemployment.20 The data provide information on recipient socioeconomic characteristics (e.g., gender, race, age, and education) and employment history (e.g., industry, occupation, and tenure with prior employer) at the start of their 2003 claim. The data also provide the benefit entitlements of recipients on their 2003 claim (weekly benefit amount and weeks of eligibility) and the total benefit weeks and amounts collected during the entire claim. Finally, the data report whether recipients started a new UI claim at any time within 12 months of the end of their 2003 claim,21 and the benefit weeks and amounts collected on the repeat claim.
Before any analyses are presented, it is useful to describe the extent to which the seven states in this study represent other state UI programs within the entire U.S. UI system. According to the U.S. Department of Labor’s Unemployment Insurance Data Summary,22 unemployed workers started 9.9 million new UI claims in 2003. During the same year, UI recipients collected nearly 21 million benefit weeks and a total of $5.5 billion in benefits, while state UI Trust Funds had $3.6 billion in revenue from employer taxes. According to the same data, the seven states in this study accounted for about 13 percent of the new UI claims, total benefit weeks and benefit amounts collected, and UI Trust Fund revenue. In fact, UI recipients in the seven study states collected on the average UI claim 16.2 weeks of benefits with a $259 weekly benefit amount, which are very similar to the 16.4 weeks and $253 weekly benefit amount collected in the entire U.S. UI system. As a result, the total benefit amounts collected per new UI claim in the seven states ($4,196) was quite similar to that of the entire system ($4,149). Further, the average UI Trust Fund revenue in the seven states ($2,780 per new UI claim) was nearly identical to the entire system average ($2,751 per new UI claim). Overall, the seven states examined in this study accounted for a large share of new UI claims in 2003 and, on average, represented the entire system in terms of weekly benefit amounts allowed, benefit weeks and amounts collected, and UI Trust Fund revenue.
In table 1, available recipient characteristics are summarized by state.23 The first row reports the total number of new UI recipients—in 2003, the number of new UI recipients was as few as 24,825 in New Hampshire and as many as 542,779 in Pennsylvania. In total, based on state administrative data, these seven states reported 1,209,456 new UI recipients in 2003. Table 1 also shows that a significant proportion of recipients on temporary layoff, that is, workers who lost their jobs but were expecting to be recalled by their previous employers. Of the seven study states, Iowa and Montana had the largest shares of workers on temporary layoff (60 percent and 58 percent, respectively), while Hawaii (28 percent) and New Hampshire (31 percent) had the lowest shares. These disparities might be attributed partly because workers on temporary layoff are typically employed in construction and in the other sectors category, which may account for larger shares of recipients in Iowa and Montana.24
Category | Pennsylvania | North Carolina | Oregon | Iowa | Montana | Hawaii | New Hampshire |
---|---|---|---|---|---|---|---|
Total | 542,779 | 333,186 | 163,396 | 89,374 | 28,406 | 27,490 | 24,825 |
Temporary layoff | 35 | 36 | 40 | 60 | 58 | 28 | 31 |
Prior UI claim | 46 | — | — | 53 | 48 | — | 34 |
Industry sector | |||||||
Manufacturing | 24 | — | 20 | — | 11 | — | — |
Trade | 12 | — | 17 | — | 12 | — | — |
Construction | 15 | — | 15 | — | 21 | — | — |
Services | 34 | — | 40 | — | 36 | — | — |
Other | 15 | — | 8 | — | 20 | — | — |
Occupational status | |||||||
White collar, high skill | — | — | 17 | — | 11 | 11 | — |
White collar, low skill | — | — | 27 | — | 20 | 27 | — |
Blue collar, high skill | — | — | 30 | — | 49 | 37 | — |
Blue collar, low skill | — | — | 26 | — | 20 | 25 | — |
Union hiring hall (1) | 7 | — | 5 | 9 | 10 | — | — |
Men | 60 | 55 | 60 | 66 | 66 | 60 | 59 |
White | 83 | 60 | 67 | 94 | 89 | 20 | 93 |
Hispanic | 4 | 4 | 9 | 4 | 2 | 5 | 3 |
Age | |||||||
Less than 25 | 10 | 9 | 12 | 12 | 10 | 8 | 8 |
25–54 | 75 | 76 | 75 | 73 | 74 | 76 | 76 |
55+ | 15 | 15 | 14 | 15 | 16 | 16 | 16 |
Education | |||||||
Less than high school diploma | 12 | 12 | — | — | — | 12 | 12 |
High school diploma | 60 | 58 | — | — | — | 57 | 63 |
Associate’s degree/some college | 16 | 22 | — | — | — | 17 | 8 |
College degree | 9 | 7 | — | — | — | 9 | 13 |
Postgraduate degree | 4 | 2 | — | — | — | 5 | 4 |
Disabled | 2 | 2 | 4 | 1 | 2 | 1 | 2 |
U.S. citizen | 98 | 98 | 94 | 98 | 99 | 92 | 96 |
Veteran | 11 | 9 | 9 | 5 | — | — | 12 |
Weeks of eligibility | |||||||
0–10 | — | — | 2 | 1 | 1 | — | — |
11–15 | — | 13 | 11 | 8 | 15 | — | — |
16–20 | 1 | 14 | 3 | 14 | 35 | — | — |
21–25 | — | 17 | 4 | 20 | 26 | — | — |
26 | 99 | 56 | 80 | 57 | 23 | 100 | 100 |
Log weekly benefit amount | $297 (126) | $256 (101) | $257 (119) | $260 (69) | $218 (79) | $321 (100) | $246 (99) |
Weeks received | 17.3 (10.1) | 13.4 (9.2) | 13.4 (12.2) | 12.5 (8.9) | 12.3 (7.0) | 15.4 (10.2) | 12.6 (9.7) |
Benefits received | $4,803 (3,700) | (2,973) | (2,567) | (2,278) | (1,909) | (3,751) | (2,748) |
Notes: Note: Reported are proportions of new UI recipients in 2003; in the last three rows, reported are means, with standard deviations in parentheses. Source: State UI administrative data. |
Analyses presented in this article, where available, find that significant proportions of recipients had started a prior UI claim within 2 years before the start of their 2003 claim. For example, 46 percent of new UI recipients in Pennsylvania had a prior UI claim, indicating that nearly half the 2003 new UI claims in Pennsylvania were started by returning UI recipients. Industry information is only available for Pennsylvania, Oregon, and Montana, where the majority of recipients were in the services sector. Manufacturing was the second largest sector in Pennsylvania (24 percent) and Oregon (20 percent), while in Montana, construction (21 percent) and other sectors (20 percent) were the largest sectors following the services sector. In the states where occupation is available (Oregon, Montana, and Hawaii), the majority of recipients were blue-collar workers.25 In addition, no more than 10 percent of recipients were conducting their job search through a union hiring hall in states where this information is available. Also reported are recipient distributions by gender, race, ethnicity, age, education, disability status, citizenship, and veteran status.
The seven states in this study represent a wide variety of programs in terms of average eligibility duration. New Hampshire and Hawaii were uniform eligibility states, where all recipients were entitled to 26 weeks of benefits. In Pennsylvania, UI eligibility duration was either 16 or 26 weeks, whereas eligibility varied from 8–26 weeks in Iowa and Montana, 13–26 weeks in North Carolina, and 3–26 weeks in Oregon.26 Furthermore, although the weekly benefit amount in all seven states was determined based on the claimant’s base-period wages, state differences in average wages and in the generosity of the UI program caused disparities in average entitlements. State disparities in eligibility duration and benefit entitlements are presented in table 1.
Finally, average weeks of benefits received on the entire 2003 claim ranged from 12.3 weeks in Montana to 17.3 weeks in Pennsylvania and average total benefit amounts collected on the 2003 claim ranged from $2,622 in Montana to $4,983 in Hawaii. These disparities are partly a result of state differences in average entitlements. For example, recipients in Montana were eligible for the fewer number of weeks and lower weekly benefit amounts than were recipients in any of the other states; thus, on average, Montana recipients collected fewer benefit weeks and amounts on their claim. In contrast, Hawaii and Pennsylvania had the highest entitlements, and recipients in these states collected higher-than-average total benefit amounts. Overall, the seven state UI programs served a substantial number of new recipients in 2003 and varied significantly in program size, as well as in recipient characteristics, employment history, benefit entitlements, and benefits received.
Using the data just described, this article examines the prevalence of repeat use in the U.S. UI system during the study period. A repeat user in this study is a new 2003 UI recipient who started a new claim within 12 months of the expiration of the 2003 claim. The analysis divides recipients into two groups: (1) displaced workers, who lost their jobs and did not expect a recall by their employers, and (2) temporarily laid-off workers, who lost their jobs and expected a recall by their employers. Separating the two groups is important because they have different unemployment experiences and are treated differently by state UI programs. Whereas most displaced workers actively look for a new job, most workers on temporary layoff collect benefits until they return to their prior employers.27 Displaced workers are, in fact, required to search for a new job or at least be willing and able to start a new job to retain their UI eligibility in the states examined. Displaced workers also may be required to receive reemployment assistance from available programs, including the Worker Profiling and Reemployment Services (WPRS) program, to retain UI eligibility. Workers on temporary layoff are exempt from work search and reemployment assistance requirements.30
The expectation is that the likelihood of returning to the UI program soon after the end of the initial claim would vary across key recipient characteristics that capture workforce attachment, job types, and human capital. For example, weak prior attachment to the workforce (prior UI claim, short tenure with prior employer) may indicate that the recipient is unable or, perhaps, unwilling to establish a consistent attachment to the workforce, thus more likely to become a repeat user. It also is likely that repeat use varies by job type—for example, recipients in volatile sectors such as construction and in blue-collar occupations may experience frequent unemployment spells, so they would be more likely to start a repeat claim. Similarly, recipients with low levels of human capital (low education, younger workers, etc.) may be less likely than are others to obtain stable employment while collecting benefits on their initial claim. To assess the relationship between repeat use and characteristics related to workforce attachment, job types, and human capital, the following model is used:
RUi = α + WORKi × β + JOBi × γ +Xi × δ + CONDi × ε + ui (1)
The dependent variable (RUi) is the likelihood of repeat use (equals 1 if recipient i was a repeat user, 0 else). Control variables include all available recipient characteristics, as reported in table 1: WORKi includes variables capturing prior workforce attachment indicators (prior UI claim and tenure), JOBi includes variables capturing recipient job type (industry, occupation, and union), and Xi includes 2003 claim benefit entitlements (weeks of eligibility and logarithm of weekly benefit amount) and available human capital characteristics (gender, race, ethnicity, education, age, etc.). The model also includes fixed effects for the recipient’s county of residence, the workforce area in which the claim was filed, and the month the 2003 claim was filed (CONDi) and a zero-mean disturbance term (ui). The fixed effects remove variation in the repeat use probability caused by local economic conditions or the timing of the start of the 2003 claim, providing more accurate estimates of the relationship between repeat use and recipient characteristics to be produced. This model is estimated separately for displaced recipients and for recipients on temporary layoff, by state, using a linear regression model with clustered standard errors based on the month the 2003 claim was started.
Table 3 presents the regression results for displaced recipients. Those with prior UI use were significantly more likely to become repeat users in states where this information is available. In Pennsylvania, for example, recipients with a prior UI claim were 22.5 percentage points more likely to become repeat users relative to those with no prior use of UI. Furthermore, tenure with the prior employer was negatively related to repeat use in states where this variable is available. An additional year of tenure with the prior employer is associated with a lower repeat use likelihood of 0.1 to 0.5 percentage points. These results suggest that displaced recipients with a less stable employment history are more likely to return to the program either because they cannot secure stable employment or by choice.
Category | Pennsylvania | North Carolina | Oregon | Iowa | Montana | Hawaii | New Hampshire |
---|---|---|---|---|---|---|---|
Prior UI claim | .225** (.002) | — | — | (.005) | .096** (.009) | .265** (.043) | .098** (.007) |
Tenure | (.000) | — | — | (.000) | –.001 (.001) | (.001) | (.001) |
Industry sector | |||||||
Manufacturing | –.176** (.003) | — | (.005) | — | (.019) | — | — |
Trade | –.204** (.002) | — | –.049** (.006) | — | (.017) | — | — |
Services | –.176** (.003) | — | –.038** (.004) | — | –.030* (.015) | — | — |
Other | –.152** (.003) | — | –.012* (.006) | — | –.016 (.016) | — | — |
Occupational status | |||||||
White collar, high skill | — | — | –.115** (.005) | — | –.064** (.014) | –.250** (.010) | — |
White collar, low skill | — | — | –.115** (.004) | — | –.038** (.014) | –.197** (.008) | — |
Blue collar, high skill | — | — | –.074** (.004) | — | –.020 (.012) | –.186** (.008) | — |
Union hiring hall | .200** (.003) | — | .281** (.027) | .329** (.008) | .327** (.016) | — | — |
Men | .012** (.002) | .012** (.002) | –.007* (.003) | .044** (.005) | .004 (.008) | .009 (.006) | .029** (.007) |
White | .004* (.002) | (.002) | –.219** (.003) | –.079** (.016) | .004 (.014) | .003 (.007) | –.044** (.012) |
Hispanic | .022** (.004) | –.024** (.006) | –.011 (.006) | –.020 (.018) | –.022 (.025) | .049** (.012) | .014 (.021) |
Age | |||||||
Less than 25 | –.021** (.002) | –.048** (.003) | –.050** (.004) | –.011 (.007) | –.045** (.012) | –.051** (.011) | –.033** (.012) |
55+ | .008** (.002) | .031** (.003) | .019** (.004) | .018** (.007) | .008 (.010) | .027** (.008) | .009 (.009) |
Education | |||||||
Less than high school diploma | .008** (.002) | .004 (.003) | — | — | — | –.001 (.010) | .002 (.011) |
Associate’s degree/some college | –.023** (.002) | –.084** (.003) | — | — | — | .012 (.007) | –.015 (.011) |
College degree | –.061** (.003) | –.119** (.004) | — | — | — | –.001 (.010) | –.049** (.009) |
Postgraduate degree | –.074** (.004) | –.130** (.007) | — | — | — | –.031** (.012) | –.060** (.002) |
Disabled | –.054** (.004) | .027** (.007) | .068** (.007) | –.044 (.031) | .049* (.023) | –.017 (.030) | –.004 (.023) |
U.S. citizen | .011* (.005) | –.010 (.007) | –.007 (.006) | –.014 (.019) | –.006 (.047) | .038** (.011) | –.115** (.017) |
Veteran | –.024** (.002) | –.012** (.003) | .040** (.005) | –.001 (.009) | — | — | –.019 (.011) |
Weeks of eligibility | |||||||
0–10 | — | — | –.065** (.010) | –.016 (.020) | .033 (.032) | — | — |
11–15 | — | –.057** (.003) | –.033** (.004) | .005 (.007) | .033** (.012) | — | — |
16–20 | –.025** (.006) | –.014** (.003) | –.029** (.009) | .016* (.007) | .030** (.010) | — | — |
21–25 | — | .005 (.003) | –.001 (.007) | .013** (.006) | .032** (.010) | — | — |
Log weekly benefit amount | .008** (.002) | .041** (.002) | –.036** (.003) | .026** (.006) | –.009 (.009) | .054** (.007) | .006 (.007) |
R-squared | .1512 | .068 | .1036 | .1221 | .1408 | .0894 | .0577 |
Number of observations | 354,851 | 212,692 | 98,816 | 35,441 | 11,989 | 19,769 | 17,038 |
Notes: *Statistically significant at 5 percent. Note: Reported are linear regression coefficients, with clustered standard errors in parentheses. All specifications include fixed effects for county of residence, workforce area in which the 2003 claim was filed, and the month the 2003 claim was filed. Source: State UI administrative data. |
Industry and occupation also have a strong relationship to repeat use. Regression results in table 3 show that Pennsylvania recipients in construction (the omitted category) were at least 15.2 percentage points more likely to become repeat users than those in manufacturing, trade, services, and other sectors. In Oregon, construction recipients were between 1.2 and 4.9 percentage points more likely to be repeat users than those in manufacturing, trade, and other sectors. In Montana, construction recipients were more likely to be repeat users than those in manufacturing (4.2 percentage points), trade (3.7 percentage points), and services (3.0 percentage points); no statistical significant difference was detected between construction and other sectors. These results show that repeat use was more prevalent for recipients in construction, a volatile and cyclical sector, in which workers are likely to experience frequent periods of unemployment.
The analyses also show that displaced recipients in blue-collar occupations were more likely than were their white-collar peers to become repeat users in Oregon, Montana, and Hawaii. In Oregon, white-collar high-skilled workers and white-collar low-skilled recipients were 11.5 percentage points less likely than were blue-collar low-skilled recipients to become repeat users. Further, blue-collar high-skilled recipients in Oregon and Hawaii were 7.4 and 18.6 percentage points less likely than were their blue-collar low-skilled peers to return to the program, respectively. These results partially reflect the fact that blue-collar jobs were less stable than white-collar jobs, as evidenced by unemployment rate differences between these two groups during the study period.33 The negative relationship between repeat use and education is partly attributable to that college-educated workers were less likely than were their peers to experience unemployment during the study period.38 Each week, based on available resources at each local workforce office, states refer recipients with the highest profiling scores (i.e., predicted likelihood of exhausting benefits) to WPRS.39 Displaced recipients referred to WPRS are required to receive those services, otherwise they forfeit their UI eligibility. Temporarily laid-off recipients are exempt from WPRS requirements.
Multiple researchers have shown that reemployment assistance provided by the public workforce development system and the WPRS program, in particular, is effective in reducing initial UI spells (i.e., number of UI weeks collected by recipients on their initial claim).40 Some of this research also has shown that reemployment assistance may have long-term impacts. For example, Bloom finds that reemployment services provided by the Texas Worker Adjustment Demonstration were effective in assisting displaced workers stay employed longer and reduce their future UI spells.41 Corson and Haimson find that displaced workers who were offered reemployment services by the New Jersey Job Search Assistance demonstration were more likely than were their peers to find sustainable jobs and avoid future unemployment spells.42 However, limited evidence exists on the effectiveness of reemployment services in reducing repeat use.
The Pennsylvania and Hawaii data used in this article report which recipients were referred to WPRS at the start of their 2003 UI claim.43 Using this information, this article produces evidence about the potential efficacy of referral to reemployment services in reducing repeat use and the benefit amounts received on repeat claims. As just indicated, the likelihood of referral to WPRS is based on the recipient’s profiling score and the available resources at the local workforce office where the recipient is assigned. Specifically, each week, recipients with the highest profiling scores in each local workforce office are referred to services. This selection procedure produces an implicit profiling score cutoff point—each week, recipients with profiling scores below the cutoff point have a zero probability of being referred to services, while those above the cutoff point have a high probability of being referred to services. Based on this referral mechanism, one can estimate the impact of WPRS referral on recipient repeat use outcomes through a regression discontinuity design, in which the outcomes of recipients with a profiling score marginally above the cutoff point are compared with the outcomes of recipients with a profiling score marginally below the cutoff point. Unfortunately, recipient profiling scores and their local workforce office were not reported in the data, so implementing this approach is not feasible.
Instead, this article assesses the potential impact of WPRS on repeat use by estimating two models for Pennsylvania and Hawaii, in which the dependent variables are the likelihood of repeat use and the benefit amounts collected on the repeat claim. These models include all recipient characteristics used in the repeat use models (as shown in table 3) and WPRS Referral, a dummy variable that equals 1 if the recipient was referred to WPRS and 0 else. Although recipient profiling score is unavailable, the specifications include all characteristics used to construct the profiling score in Pennsylvania (industry, tenure, benefit entitlement, education, county of residence, and month of claim) and in Hawaii (prior use of UI, occupation, tenure, education, benefit entitlement, county of residence, and month of claim). Furthermore, although the local workforce office in which the claim was filed is not reported in the data, the specifications include workforce area fixed effects (i.e., the workforce area that includes the local workforce office where the claim was filed) and county of residence fixed effects. The profiling score variables and the workforce area and county of residence fixed effects capture, to a great extent, the variation in the likelihood of services referral produced by the WPRS selection mechanism. Thus, although the parameters of WPRS Referral do not constitute unbiased estimates of the effect of reemployment services on repeat use outcomes, they do strongly indicate whether such services may be effective in reducing repeat use.
Table 7 reports the regression results. As shown, displaced recipients referred to WPRS in Pennsylvania and Hawaii were 10.5 and 8.0 percentage points less likely to become repeat users than their peers, respectively. When these figures are compared with the average repeat use rate for displaced recipients in each state, referral to WPRS services is associated with a 34 percent and 40 percent lower repeat use probability in Pennsylvania and Hawaii, respectively. Table 7 also shows that referral to WPRS services is associated with significantly lower benefit amounts collected on repeat claims. Specifically, displaced recipients referred to reemployment services in Pennsylvania and Hawaii collected, on average, $348 and $424 lower benefit amounts on repeat claims than their peers. Compared with the average repeat benefits received, WPRS referral is associated with 24 percent and 40 percent lower repeat benefits collected in Pennsylvania and in Hawaii, respectively.
Category | Likelihood of repeat use | Repeat benefit amounts collected | ||
---|---|---|---|---|
Pennsylvania | Hawaii | Pennsylvania | Hawaii | |
WPRS referral | –.105** (.003) | (.008) | –$348** (18) | –$424** (54) |
Prior unemployment insurance claim | .199** (.002) | .267** (.042) | 796** (10) | 1,117** (290) |
Tenure | –.001** (.000) | –.005** (.001) | –3** (1) | –31** (4) |
Industry sector | ||||
Manufacturing | –.170** (.003) | — | –1,365** (19) | — |
Trade | –.200** (.003) | — | –1,340** (20) | — |
Services | –.170** (.003) | — | –1,225** (18) | — |
Other | –.149** (.003) | — | –1,116** (20) | — |
Occupational status | ||||
White collar, high skill | — | –.235** (.010) | — | –1,352** (67) |
White collar, low skill | — | –.184** (.008) | — | –1,097** (58) |
Blue collar, high skill | — | –.172** (.008) | — | –995** (54) |
Union hiring hall | .198** (.003) | — | 2,350** (21) | — |
Men | .012** (.002) | .008 (.006) | 103** (10) | 114** (44) |
White | .005** (.002) | .005 (.007) | –54** (14) | 139** (47) |
Hispanic | .021** (.004) | .049** (.012) | 78** (25) | 268* (84) |
Age | ||||
Less than 25 | –.023** (.002) | –.051** (.010) | –131** (16) | –314** (71) |
55+ | –.007** (.002) | .028** (.008) | 75** (14) | 209** (52) |
Education | ||||
Less than high school diploma | .013** (.002) | –.001 (.010) | 12 (15) | –165** (66) |
Associate’s degree/some college | –.023** (.002) | .010 (.007) | –4 (13) | 168** (51) |
College degree | –.065** (.003) | –.008 (.009) | –178** (16) | –3 (65) |
Postgraduate degree | –.078** (.004) | .025* (.012) | –222** (23) | 162 (82) |
Disabled | –.052** (.004) | –.017 (.030) | –172** (28) | –182 (206) |
U.S. citizen | .013** (.005) | .040** (.011) | 175** (34) | 386** (76) |
Veteran | –.024** (.002) | — | –38** (16) | — |
Weeks of eligibility | ||||
0–10 | — | — | — | — |
11–15 | — | — | — | — |
16–20 | –.027** (.006) | — | 78* (40) | — |
21–25 | — | — | — | — |
Log weekly benefit amount | .009** (.002) | .066** (.007) | 637** (10) | 732 (48) |
R–squared | .1546 | .0941 | .177 | .0765 |
Number of observations | 354,851 | 19,769 | 354,851 | 19,769 |
Dependent variable mean | 31 percent | 20 percent | 1,460 (2,973) | 1,053 (2,703) |
Notes: *Statistically significant at 5 percent. Note: Reported are linear regression coefficients, with clustered standard errors in parentheses. All specifications include fixed effects for county of residence, workforce area in which the 2003 claim was filed, and the month the 2003 claim was filed. Source: State unemployment insurance administrative data. |
THIS ARTICLE PRESENTS EVIDENCE of the prevalence of repeat use of UI benefits during a period of moderate unemployment using administrative data from seven states that represent a wide spectrum of programs in the U.S. UI system. The results show that more than half of temporarily laid-off recipients and more than a quarter of displaced recipients in 2003 started a new UI claim within 12 months of the end of their initial claim. The high prevalence of repeat use for recipients on temporary layoff was perhaps not surprising, because they are typically employed in seasonal sectors and experience frequent short-term unemployment spells. Furthermore, because they are not required to search for a job or receive reemployment assistance to remain eligible for benefits, most of the recipients on temporary layoff are likely to return to their prior jobs, thus experience unemployment soon after their initial claim expires. In contrast, the high repeat use rates for displaced recipients are quite revealing and suggest that many displaced recipients are unable to find sustainable jobs and establish a consistent attachment to the workforce.
Multivariate regression analyses show that recipient prior workforce attachment, job types, and human capital characteristics strongly predict repeat use for both displaced and temporarily laid-off recipients. Workers with a weak prior workforce attachment, as captured by prior participation in the UI program and short tenure with their prior employer, were significantly more likely to become repeat users than were their peers. Repeat use also was higher for recipients usually employed in cyclical sectors, particularly construction, and in blue-collar jobs, particularly low-skill jobs. The most important human capital predictors of repeat use were education and age; recipients with a college education and younger recipients were less likely to return to the program. Overall, these results show that repeat use is more prevalent for workers with an inconsistent employment history, those employed in low-skill and/or cyclical jobs, workers with low education, and older workers. Therefore, state UI programs that have a large intake of such workers are likely to experience high repeat use rates.
This study also shows that repeat use is very costly for the UI Trust Fund. Repeat users collected substantial benefit amounts on their repeat claims and, as a result, collected much higher benefit amounts than nonrepeat users. The benefit amounts collected by repeat users on their repeat claim accounted for about a third of the amounts collected by all UI recipients in the seven states during the study period. These results have important policy implications. States use an experience-based system to determine employer UI tax rates, which presumably imposes higher rates on employers responsible for repeat use. If in fact these employers have higher UI tax rates and their contributions suffice to cover the repeat use burden, then repeat use is not a cause of deficit for the UI Trust Fund. On the other hand, if the repeat use burden is not covered by the contributions of employers responsible for repeat use, states may want to adjust their system to penalize these employers with higher UI tax rates.
Finally, this article examines whether providing reemployment assistance to displaced recipients is an effective policy for reducing repeat use. With Pennsylvania and Hawaii data used, it is evident that displaced recipients who were referred to reemployment services under the WPRS program were 34 percent and 40 percent less likely than their peers to become repeat users, respectively. As a result, displaced recipients who were referred to reemployment services collected substantially lower benefit amounts on their repeat claims relative to their peers. These results provide some evidence that providing reemployment assistance to displaced recipients early in their initial claims may be an effective policy for reducing the prevalence and financial burden of repeat use.
In conclusion, this article fills a conspicuous gap in the literature about the prevalence of repeat use and its implications for the effectiveness and solvency of the U.S. UI system. The findings highlight the importance of better understanding the causes and consequences of repeat use, including examining repeat use during periods of high unemployment, particularly during the most recent recession, and assessing the effect of repeat use on the solvency of the UI Trust Fund and employer UI taxes. Finally, further research is needed to assist policymakers and program administrators identify strategies, including reemployment assistance, to help unemployed workers establish a strong attachment to the workforce and avoid becoming repeat users of the UI system.
Marios Michaelides, "Repeat use in the U.S. unemployment insurance system," Monthly Labor Review, U.S. Bureau of Labor Statistics, September 2014, https://doi.org/10.21916/mlr.2014.30
1 For a detailed discussion of current eligibility rules by state, see “Comparison of state unemployment laws 2013,” Employment and Training Administration (U.S. Department of Labor, June 2014), https://www.dol.gov/agencies/eta/.
2 In most states, the base period includes the first four of the five calendar quarters preceding the date the claim was filed.
3 For example, a claim filed on May 1, 2003, would expire on April 30, 2004.
4 Alaska, New Jersey, and Pennsylvania are the only states that levy a UI tax on employees, in addition to the UI tax on employers.
5 “Monthly program and financial data,” Employment and Training Administration (U.S. Department of Labor, June 2014), http://www.ows.doleta.gov/unemploy/claimssum.asp.
6 See Walter Corson and Walter Nicholson, “An examination of declining UI claims during the 1980s,” Unemployment Insurance Occasional Paper 88-3, Employment and Training Administration (U.S. Department of Labor, 1988); David C. Wittenburg, Michael Fishman, David Stapleton, Scott Scrivner, and Adam Tucker, “Literature review and empirical analysis of unemployment insurance recipiency ratios: final report,” The Lewin Group (U.S. Department of Labor, 1999); Jonathan Gruber, “The wealth of the unemployed: adequacy and implications of unemployment insurance,” NBER Working Paper No. 7348 (Cambridge, MA: National Bureau of Economic Research, 1999); and Aisling Cleary, Joyce Kwok, and Rob Valletta, “New highs in unemployment insurance claims,” FRBSF Economic Letter, September 8, 2009, Series 2009-28.
7 See Harry Holzer, “Unemployment insurance and welfare recipients: what happens when the recession comes?” New federalism: issues and options for states, Series A-46 (Urban Institute, December 2001), http://www.urban.org/publications/310032.html.
8 See Stephen A. Wandner and Andrew Stettner, “Why are many jobless workers not applying for benefits?” Monthly Labor Review, June 2000, pp. 21–32; and Wayne Vroman, “Unemployment insurance recipients and nonrecipients in the CPS,” Monthly Labor Review, October 2009, pp. 44–53.
9 See Jonathan Gruber, “The consumption smoothing benefits of unemployment insurance,” American Economic Review, March 1997, pp. 192–205; and Jonathan Gruber, “The wealth of the unemployed: adequacy and implications of unemployment insurance,” NBER Working Paper No. 7348 (Cambridge, MA: National Bureau of Economic Research, October 2001).
10 See Lawrence Chimerine, Theodore S. Black, and Lester Coffey, “Unemployment insurance as an automatic stabilizer: evidence on the effectiveness over three decades,” Unemployment Insurance Occasional Paper 99-8, Martha K. Matze, ed. (U.S. Department of Labor, 1999); and Wayne Vroman, “The role of unemployment insurance as an automatic stabilizer during a recession,” ETA Occasional Paper 2010-10 (U.S. Department of Labor, November 2010).
11 See Bruce D. Meyer, “Unemployment insurance and unemployment spells,” Econometrica, July 1990, pp. 757–782; Elena G. F. Stancanelli, “Unemployment duration and the duration to unemployment benefits: an empirical study for Britain,” Applied Economics, 1999, pp. 1043–1051; and ŠtepánJurajda and Frederick J. Tannery, “Unemployment durations and extended unemployment benefits in local labor markets,” Industrial and Labor Relations Review, February 2003, pp. 324–348.
12 See James M. Poterba and Lawrence H. Summers, “Unemployment benefits and labor market transitions: a multinomial logit model with errors in classification,” Review of Economics and Statistics, May 1998, pp. 207–216; Štepán Jurajda, “Estimating the effect of unemployment insurance compensation on the labor market histories of displaced workers,” Journal of Econometrics, May 2001, pp. 227–252; and Chris Riddell and Peter J. Kuhn, “The long-term effects of unemployment insurance: evidence from New Brunswick and Maine,” 1940–1991, Industrial and Labor Relations Review, January 2010, pp. 183–204.
13 See Wayne Vroman, “The history of state UI financing,” in Topics in Unemployment Insurance Financing (Kalamazoo, MI: W.E. Upjohn Institute for Employment Research, 1998); Ernest Goss and James Knudsen, “Evaluation of solvency standards for state Unemployment Insurance Trust Funds,” Public Budgeting and Finance, Winter 1999, pp. 3–20; and Wayne Vroman, “The recession of 2001 and unemployment insurance financing,” Research Report (The Urban Institute, January 2005).
14 See Paul T. Decker, Robert B. Olsen, and Lance Freeman, “Assisting unemployment insurance claimants: the long-term impacts of the job search assistance demonstration,” Office of Workforce Security Occasional Paper 2000-02 (U.S. Department of Labor, February 2000); Gary Burtless and Wayne Vroman, “New policy opportunities for unemployment insurance,” IMPAQ International Paper, submitted to U.S. Department of Labor, 2006; and Walter Nicholson and Karen Needels, “Unemployment insurance: strengthening the relationship between theory and policy,” Journal of Economic Perspectives, Summer 2006, pp. 47–70.
15 Decker et al., “Assisting unemployment insurance claimants”; Burtless and Vroman, “New policy opportunities”; and Nicholson and Needels, “Unemployment insurance,” pp. 47–70.
16 See Brian P. McCall, “Repeat use of unemployment insurance in the United States: evidence from the NLSY79,” Working Paper (University of Minnesota, 2005).
17 See Saul Schwartz, Wendy Bancroft, David Gyarmati, and Claudia Nicholson, “The frequent use of unemployment insurance in Canada,” Earnings Supplement Project Report (Social Research and Demonstration, March 2001); Lori M. Stratychuk, “Repeat users of employment insurance,” Perspectives on Labour and Income, Summer 2001.
18 See Miles Corak, “Unemployment insurance once again: the incidence of repeat participation in the Canadian UI Program,” Canadian Public Policy, June 1993, pp. 162–176.
19 See Shawn de Raaf, Constantine Kapsalis, and Carole Vincent, “Seasonal work and employment insurance use,” MPRA Working Paper 25700 (University Library of Munich, Germany, September 2003).
20 In 2003, the U.S. economy was recovering from the early 2000s recession, which according to the National Bureau of Economic Research (NBER) lasted from March 2001 through November 2001. In 2003, the U.S. unemployment rate was 6.0 percent and dropped to 5.5 percent in 2004 and 5.1 percent in 2005. See “Unemployment,” Databases, Tables & Calculators by Subject (U.S. Bureau of Labor Statistics), https://www.bls.gov/data/#unemployment. Similarly, the insured unemployment rate (UI claims as a percentage of covered employment) was 2.8 percent in 2003 and dropped to 2.3 percent and 2.1 percent in 2004 and 2005, respectively. See “Unemployment Insurance Chartbook,” Employment and Training Administration (U.S. Department of Labor, February 2013), https://www.dol.gov/agencies/eta/). These rates are about average for the 1991–2011 period (5.9-percent unemployment and 2.5 percent insured unemployment) but are much lower than the rates for the 2008–2010 period (8.2-percent unemployment and 3.5-percent insured unemployment), which includes the most recent recession (December 2007–July 2009, according to NBER).
21 For example, consider a recipient who started a new UI claim on June 1, 2003—this claim expired on May 31, 2004. The data report whether the recipient started a new UI claim at any time from June 1, 2004 through May 31, 2005.
22 These quarterly data report estimates of key indicators for the UI system (overall and by state), including number of new UI claims started, total benefit amounts paid, total benefit weeks paid, average weekly benefit amounts, and total revenue from UI employer taxes. The data also report average weeks of benefits received (total benefit weeks paid divided by number of new UI claims started), average benefit amounts received (total benefit amounts paid divided by number of new UI claims started), and average UI Trust Fund revenue (total revenue divided by number of new UI claims started). For more details, see http://www.ows.doleta.gov/unemploy/content/data.asp.
23 Because of state differences in the information gathered for UI claimants during the application process, availability of recipient characteristics varies across the seven states in this study.
24 Industry information is only available in Pennsylvania, Oregon, and Montana. As shown in table 1, of these three states, Montana had the highest proportion of recipients in construction (21 percent) and in the other sectors category (20 percent), which includes agriculture, mining, and transportation.
25 For convenience, four occupation groups are defined: (1) white collar, high skilled, which includes high-wage, white-collar occupations (e.g., management; business and financial operations; legal and healthcare practitioners); (2) white collar, low skilled, which includes low-wage white-collar jobs (e.g., community and social service; education, and office and administrative support); (3) blue collar, high skilled, which includes high-wage blue-collar occupations (e.g., protective service; personal care and service; and installation, maintenance, and repair occupations); and (4) blue collar, low skilled, which includes low-wage blue-collar occupations (e.g., cleaning and maintenance; farming, fishing, and forestry; and laborers).
26 These seven states represent several states in the UI system in terms of eligibility duration in 2003, including Connecticut, Illinois, Maryland, and New York (26 weeks); Indiana (8–26 weeks); Arkansas (9–26 weeks); Massachusetts, Minnesota, and Texas (10–26 weeks); Arizona and Wisconsin (12–26 weeks); and California and Colorado (14–26 weeks).
27 See Alan B. Krueger and Andreas I. Mueller, “Job search and unemployment insurance: new evidence from time use data,” Journal of Public Economics, April 2010, pp. 298–307.
28 See Stephen A. Wandner, “Employment programs for recipients of unemployment insurance,” Monthly Labor Review, October 2008.
29 See Donald R. Deere, “Unemployment insurance and employment,” Journal of Labor Economics, October 1991, pp. 307–324; Patricia M. Anderson and Bruce D. Meyer, “The unemployment insurance payroll tax and interindustry and interfirm subsidies,” in Tax Policy and the Economy (Cambridge, MA, National Bureau of Economic Research, October 1993), pp. 111–144; and Patricia M. Anderson and Bruce D. Meyer, “Unemployment insurance in the United States: layoff incentives and cross subsidies, Part 2: U.S. and Canadian Income Maintenance Programs,” Journal of Labor Economics, January 1993, pp. S70–S95.
30 In 2004, seven states in this study reported about 1.1 million new UI claims. For details, see http://ows.doleta.gov/unemploy/content/data.asp.
31 Tabulations of the January 2004 Current Population Survey, Displaced Worker Supplement (CPS-DW) show 8.3 percent of blue-collar and 4.0 percent of white-collar workers experienced unemployment at least once during 2003. Furthermore, tabulations of the monthly CPS data show the average unemployment rate in 2004–2005 was 6.9 percent for blue-collar workers and 3.7 percent for white-collar workers.
32 See Brian P. McCall, “The effect of unions on the receipt of unemployment insurance benefits,” Industrial and Labor Relations Review, 1997, pp. 478–492; and McCall, “Repeat use of unemployment insurance in the United States.”
33 The relationship between education, industry, and occupation may affect the results since none of the seven states reports all three variables. For example, education attainment may be negatively correlated with blue-collar jobs; if so, the coefficients of college degree and postgraduate degree may be underestimated when occupation is not available. Data limitations do not allow us to disentangle these relationships; nevertheless, the results suggest that, overall, repeat use is higher for recipients with low educational attainment.
34 Tabulations of the January 2004 CPS-DW show 7.2 percent of workers with no college degree and 3.0 percent of college-educated workers experienced unemployment at least once in 2003. Also, monthly CPS data tabulations show the unemployment rate in 2004–2005 was 2.4 percent for college-educated workers and 5.6 percent for all other workers.
35 See “Unemployment insurance: information on benefit receipt,” GAO-05-291 (U.S. Government Accountability Office, March 2005); and Wayne Vroman, “An analysis of unemployment insurance non-filers: 2005 CPS supplement results,” ETA Occasional Paper 2009-7 (U.S. Department of Labor, 2005a).
36 “Unemployment insurance data summary,” Employment and Training Administration (U.S. Department of Labor, March 2004), http://ows.doleta.gov/unemploy/content/data.asp.
37 As indicated earlier in the text, recipients on temporary layoff are exempt from reemployment assistance requirements.
38 See William F. Sullivan, Lester Coffey, Lisa Kolovich, Charles W. McGlew, Douglas Sanford, and Richard Sullivan, “Worker profiling and reemployment services: evaluation of state worker profiling models,” ETA Occasional Paper 2007-15 (U.S. Department of Labor, 2007).
39 The median proportion of displaced recipients referred to WPRS services across all states is 15 percent. See Sullivan et al., “Worker profiling and reemployment services evaluation.”
40 See Howard S. Bloom, “The demonstration,” in Back to Work: Testing Reemployment Services for Displaced Workers,” Research Report 143 (Kalamazoo, MI: W.E. Upjohn Institute for Employment Research); Walter Corson, Joshua Haimson, “The New Jersey unemployment insurance reemployment demonstration project: six-year follow-up and summary report,” Unemployment Insurance Occasional Paper 95-2 (U.S. Department of Labor, January 1995); Decker et al., “Assisting unemployment insurance claimants”; and Dan A. Black, Jeffrey A. Smith, Mark C. Berger, Brett J. Noel, “Is the threat of reemployment services more effective than the services themselves? Evidence from Random Assignment in the UI System,” American Economic Review, September 2003, pp. 1313–1327.
41 Bloom, “The demonstration.”
42 See Corson and Haimson, “The New Jersey unemployment insurance.”
43 In 2003, Pennsylvania and Hawaii referred 27,059 (or 8 percent) and 3,289 (or 17 percent) of displaced recipients to reemployment services, respectively.