Health Insurance For Workers Who Lose Jobs: Implications For Various Subsidy Schemes Subsidies for continuation coverage would benefit few of the uninsured; subsidies to all low-income people who leave a job would help more of them. by Kanika Kapur and M. Susan Marquis ABSTRACT: A number of proposals have been made to help laid-off workers purchase health insurance. We use data from the Medical Expenditure Panel Survey to profile the insurance status of workers who left a job. Our descriptive analysis suggests that it might be difficult to design policies that target those who would otherwise be uninsured and that large subsidies might be needed to help laid-off workers. About two-thirds of americans who become uninsured do so when they lose employer-sponsored coverage. In 000 the uninsurance rate among unemployed adults was percent, more than twice that of all adults. Under federal law, workers who leave a job with insurance benefits have the option to continue group coverage known as (Consolidated Omnibus Budget Reconciliation Act of ) coverage for up to eighteen months by paying 0 percent of the premium. The federal law applies to workers in firms with twenty or more workers, but thirty-eight states have enacted laws that also require continuation provisions for workers in smaller firms. Only 0 percent of those eligible to purchase coverage do so. The high cost of coverage, especially for those who have just lost a job, may be a deterrent for many. Estimates for 00 place the annual cost of family coverage at about $,000 $,000, or as much as two-thirds of a typical unemployment check. The economic downturn and rising unemployment have prompted proposals to help those who have lost their jobs to afford transitional insurance coverage. Such transitional subsidies can also help reduce job lock employees reluctance to change jobs because of the possibility of loss of insurance and thereby improve productivity. Some of the proposals have explicitly focused on subsidizing coverage. Kanika Kapur is an associate economist at RAND in Santa Monica, California. Susan Marquis is a senior economist at RAND in Arlington, Virginia. HEALTH AFFAIRS ~ Volume, Number 0 00 Project HOPE The People-to-People Health Foundation, Inc. Downloaded from HealthAffairs.org on February, 0.
DataWatch Others would link the subsidy with eligibility for unemployment insurance and would allow the subsidy to be applied to the purchase of individual coverage, as well as continuation coverage. The Trade Act of 00 provides a refundable tax credit for dislocated workers eligible to receive Trade Adjustment Assistance benefits that can be applied to the purchase of continuation coverage and certain other qualified health insurance coverage. Still other proposals, including the administration s proposal, would provide tax credits to the low-income unemployed and those whose employers do not provide coverage. To design and assess the cost of such proposals requires information about the number and characteristics of people who lose a job; eligibility and take-up rates; the number and characteristics of people who become unemployed and find other sources of coverage; and the duration of coverage under transitional policies and how it varies among subpopulations. Analysis of these issues has been limited for several reasons. First, answers require panel data to identify those who lose a job and to track their subsequent insurance coverage. Many studies have used cross-sectional data to determine how many people would be eligible for should they lose the job. However, job loss rates are likely to be higher for people in firms that do not offer insurance and for workers who are less likely to take up coverage; earlier studies have confirmed that jobs with insurance last longer than jobs without insurance do. Thus, we need to study people when they lose a job and observe their sources of insurance in the period following the job loss that is, we need longitudinal data. A panel study of the unemployed in concluded that only about percent of the unemployed lost insurance because of job loss and that job loss was the direct cause of uninsurance for only about 0 percent of the uninsured unemployed. If similar results are found today, policies that focus only on those who lose insurance, such as subsidies, would not help many of the unemployed uninsured. Second, few data sources provide direct information about coverage. We overcame these two problems by using data from a recent panel study that explicitly identified as a source. 0 Data And Methods Data. Our data come from the Medical Expenditure Panel Survey (MEPS), a stratified, nationally representative sample of about,00 noninstitutionalized people in 0,00 households. Each family participated in five rounds of data collection over a two-year period. During each round, information on family members employment and health insurance coverage was collected, along with a detailed set of variables on health, demographics, and health care use. The data from the study are released as a variety of public use files. Detailed information about coverage is available only for the year of the panel members participation. Because this information is important in evaluating new policies related to transition coverage, we focus our analysis on. However, to provide some information about the insurance status of the unem- 0 May/June 00 Downloaded from HealthAffairs.org on February, 0.
ployed over a longer period, we also discuss results for the two-year period (including and ) of the panel. Our analysis sample from the MEPS includes people under age sixty-five who reported leaving a job during the year. Our final analysis sample includes, job leavers. Our study year of was a time of low unemployment and economic prosperity, in contrast to the economic conditions of today. Therefore, we also used data from the and 00 February Current Population Survey (CPS) to profile the changes in the population of job leavers since the time of our study. We also used estimates of the newly unemployed in 00 to estimate the number of workers who might benefit from alternative policies to help them gain access to insurance. Study methods. Voluntary and involuntary job terminations. The objective of many proposals is to help laid-off workers. Some proposals for extending continuation health insurance coverage target them explicitly for example, by restricting eligibility to those receiving unemployment benefits. Therefore, we examined voluntary and involuntary job terminations separately. The latter are those who reported that they were laid off, the job ended, the business dissolved, or they were receiving unemployment insurance benefits during the year. Involuntary unemployment may be more prevalent in today s economic environment than it was in, so separating out voluntary and involuntary terminations is important for extrapolating findings from to today s economy. As noted above, we use data from the 00 CPS to help in this extrapolation. Insurance following job loss. We measure insurance following job loss by looking at reported coverage in the following month. Because workers can elect coverage at any time within sixty days after losing or leaving a job, coverage in the month following the job loss may understate participation rates. Therefore, we also look at reported coverage three and six months later. In addition, looking at multiple points in time provides some information about the duration of immediate post job loss insurance status. The sample size is reduced for these analyses, because it is limited to those who lost a job early enough in the year so that the three- or six-month follow-up period is during. We were able to observe a three-month follow-up period for, workers leaving or losing a job and a six-month follow-up for workers. Insurance status in the period after a job loss is classified as coverage from lost job, group coverage from a new or another employer of the policyholder, group coverage from the old employer, group coverage through another family member, other private coverage, public coverage, or uninsured. These categories were defined to be mutually exclusive by imposing an insurance hierarchy in the order specified above. Some employers provide continued subsidized coverage in the group plan after a worker is laid off, and so we distinguish between coverage in a new or old group plan to measure the extent of this practice. We also examined the length of time that employees who elect coverage continue with it. The cumulative loss rate for each month, t, following the pur- HEALTH AFFAIRS ~ Volume, Number 0 Downloaded from HealthAffairs.org on February, 0.
DataWatch chase of coverage is the proportion of enrollees who drop coverage by time t. To estimate this rate in each period t, we selected policyholders and dependents that purchased their coverage at least t + months prior to the end of and measure the proportion of them who have dropped coverage by time t. Statistical significance. Unless otherwise noted, all contrasts that we mention in the text are statistically significant. However, small sample sizes limit some of our contrasts, and so we also give the percent confidence interval for selected important rates; these are estimated taking into account the complex survey design of MEPS. All analyses were weighted by the sampling weights provided in MEPS. Study Findings Characteristics of job leavers. Most people who left a job did not have insurance from that job (Exhibit ). More than 0 percent of voluntary job leavers did not have coverage from the employer. The share of voluntary and involuntary job leavers in jobs without insurance benefits did not differ significantly. About three-quarters of those who left a job with insurance benefits were eligible for ( percent eligible for divided by percent leaving a job with insurance). Those who left such a job but were not eligible for ( percent of all job leavers) were employed in small businesses. EXHIBIT Characteristics Of People Leaving A Job, All job leavers Voluntary job leavers Involuntary job leavers Type of job loss Voluntary Involuntary % 00% 0 0% 00 Prior insurance and eligibility Job without insurance -eligible job Job with insurance, not eligible 0 Income and eligibility Low income Job without insurance -eligible job Job with insurance, not eligible High income Job without insurance -eligible job Job with insurance, not eligible 0,, SOURCE: Medical Expenditure Panel Survey,. NOTES: is Consolidated Omnibus Budget Reconciliation Act. Low income denotes family income at or below 00 percent of the federal poverty level. 0 May/June 00 Downloaded from HealthAffairs.org on February, 0.
Almost 0 percent of all job leavers were in low-income families (below 00 percent of the federal poverty level), even though such workers accounted for only percent of employees (not shown). Low-income job leavers were less likely to be eligible for than higher-income workers were, primarily because they were less likely to have left jobs with insurance. Only percent of low-income job leavers were in a job with insurance, in contrast to percent of high-income job leavers. Conditional on leaving a job with insurance, low-income workers were not significantly less likely to be eligible for. Job leavers in and 00. Involuntary unemployment is somewhat more prevalent today than in, but the characteristics of job leavers have remained fairly constant over time (Exhibit ). The share of all job leavers who left a job involuntarily increased about percent between and 00. In both time periods, involuntary job leavers had lower income and less education than voluntary job leavers had and were more likely to be nonwhite and less likely to have a working spouse. Because the characteristics of job leavers in 00 and are similar, our conclusions from the MEPS are likely to apply in today s environment. Insurance status following job loss. In the month after losing or leaving a job, percent ( percent CI ± ) of voluntary job leavers were uninsured (Exhibit ). However, percent of them were uninsured prior to the job loss; only percent of job leavers lost coverage because of the job loss. In contrast, percent ( per- EXHIBIT Profile Of People Who Left A Job, And 00 Voluntary Involuntary Characteristic 00 00 Percent of total % % % % Below 00% of poverty Above 00% of poverty 0 0 0 0 High school education or lower More than high school education 0 0 Spouse is working Spouse is not working White Nonwhite Age Less than years years years years years 0 0 SOURCE: Current Population Survey, February and 00. NOTES: Job leavers are all those reporting leaving or changing a job within four weeks of the interview. Involuntary job leavers are those reporting that the reason for unemployment was loss of job or end of temporary job. HEALTH AFFAIRS ~ Volume, Number 0 Downloaded from HealthAffairs.org on February, 0.
DataWatch EXHIBIT Insurance Following A Job Loss, By Type Of Coverage, Voluntary job loss Involuntary job loss Type of coverage All -eligible All -eligible One month following job loss Own employer: New/other job % % % % 0 0 a Newly uninsured Previously uninsured, 0 0 0 Three months following job loss Own employer: New/other job % % % % 0 0, a Six months following job loss Own employer: New/other job % % % % 0 0 0 SOURCE: Medical Expenditure Panel Survey,. NOTE: is Consolidated Omnibus Budget Reconciliation Act. a Less than 0. percent. cent CI ± ) of those who voluntarily left -eligible jobs lost coverage and became uninsured in the month following the job loss. The rate of uninsurance was about percent higher or percent ( percent CI ± ) among those who left a job involuntarily than among voluntary job leavers; percent of involuntary job leavers were uninsured prior to the job loss, and only percent became uninsured as a consequence of the job loss. Thirtyseven percent of involuntary job leavers in -eligible jobs became uninsured following the job loss. A substantial proportion of those who lost a job were uninsured even when 0 May/June 00 Downloaded from HealthAffairs.org on February, 0.
they had access to subsidized group coverage from another family member. Although half of involuntary job leavers who had another family member with group coverage were enrolled in that plan one month after the job loss, almost 0 percent of them were uninsured in that time frame. Most of the remaining 0 percent had coverage through their old or new job (data not shown). Eighteen percent ( percent CI ± ) of voluntary job leavers eligible for coverage enrolled within the first month following a job loss (Exhibit ). But an almost equal number continued to have subsidized coverage from their former employer. Among involuntary job leavers eligible for coverage, take-up was similar to take-up among voluntary leavers eligible for. However, slightly more of these former workers continued to have subsidized coverage from their prior employer. This coverage is a short-term benefit, and many fewer workers report this source of insurance six months later. Involuntary job leavers were somewhat less likely than voluntary job leavers were to move directly to a new job with insurance benefits ( percent versus percent). -eligible job leavers, who all left a job with insurance, were more likely than other job leavers were to move to a new job with insurance within a month. These differences persisted over time, and within eighteen months of a job loss, percent of workers who were eligible for benefits held employer-sponsored health insurance (data not shown). During the six months following a job loss, more workers arranged for coverage, and the uninsurance rate fell. Nonetheless, more than a third of voluntary job leavers and 0 percent of involuntary job leavers remained uninsured six months following a job loss. Eighteen months later, the uninsurance rate was percent for both voluntary and involuntary job leavers (data not shown). For voluntary job leavers this was a small decline in the uninsurance rate. However, for involuntary workers this constituted a nine-percentage-point drop in the uninsurance rate. Income differences in post-job insurance status. Low-income workers who left a job (those in families with an income of 00 percent of poverty or below) were more likely to be uninsured than higher-income workers were (Exhibit ). 0 More than half of low-income workers who voluntarily left a job were uninsured ( percent CI percent), and almost two-thirds of low-income involuntary job losers were uninsured ( percent CI 0 percent). However, most of these workers were uninsured prior to the job loss; those who lost insurance accounted for percent and percent of the population, respectively. The uninsurance rate for higher-income voluntary job leavers was percent ( percent CI ± ); for higherincome involuntary job leavers, percent ( percent CI ± ). take-up rates and the receipt of transitional coverage subsidized by a former employer were quite similar for low- and higher-income workers who left a job with insurance and were eligible for coverage. Nonetheless, almost half of these low-income workers became uninsured following a job loss ( percent CI percent), in contrast to percent ( percent CI ± ) of their HEALTH AFFAIRS ~ Volume, Number 0 Downloaded from HealthAffairs.org on February, 0.
DataWatch EXHIBIT Insurance Following A Job Loss, By Income And Type Of Coverage, Low-income workers Higher-income workers Type of coverage Voluntary Involuntary -eligible Voluntary Involuntary -eligible One month following job loss Own employer: New job % % % % % 0 % 0 0 a Newly uninsured Previously uninsured 0 0 0 Three months following job loss Own employer: New job % % % % % % 0 0 0 0 0 Six months following job loss Own employer: New job % % % % 0 % % 0 0 0 0 SOURCE: Medical Expenditure Panel Survey,. NOTES: Low-income workers are those with family incomes at or below 00 percent of the federal poverty level. is Consolidated Omnibus Budget Reconciliation Act. a Less than 0. percent. higher-income peers. Moreover, since few low-income workers who left a job were eligible for, as discussed earlier, low-income workers who left a job voluntarily or involuntarily were less likely than higher-income workers were to have or transitional coverage from a previous employer. Low-income workers who left a job were less likely than higher-income workers were to have insurance from another job and were less likely to get coverage through a family member in the month following a job loss. However, they were more likely to have public coverage. These differences persist over time. Estimated beneficiaries under alternative subsidy schemes. We used our results and the February 00 CPS data to estimate how many involuntary job leavers would benefit from various subsidy schemes. Based on the number of new lost 0 May/June 00 Downloaded from HealthAffairs.org on February, 0.
jobs in February 00, we estimated that about. million workers would be involuntarily terminated from a job during the year. coverage subsidies would be available to. million of them (or percent, Exhibit ). Only about 0. million of these would be workers with family income below 00 percent of poverty ( percent). In contrast, tax credits targeted to all low-income involuntary job leavers would be available to about. million workers (0 percent). If the credit were restricted to uninsured low-income workers ( percent of involuntary low-income leavers), the credit would be available to. million workers. Duration of coverage. As intended, coverage provides shortterm coverage for most who purchase it. About percent ( percent CI ± ) of participants held the policy for a month or less, about percent for three months or less, and about percent ( percent CI ± ) for fewer than six months. About percent of those who dropped coverage became uninsured. Discussion Most lost jobs do not carry insurance benefits. Our analysis does not identify the factors that account for this, but the result is consistent with job lock that is, a reluctance to leave a job when it would also result in loss of insurance. Alternatively, jobs without insurance may be more likely to be bad jobs with higher turnover. subsidies. Proposals to help laid-off workers with subsidies for the purchase of coverage or other continuation of benefits would benefit only a small fraction of involuntary job leavers and would not be target-efficient. About percent of -eligible involuntary job leavers purchase private insurance (Exhibit ). Some of these people would take advantage of the subsidy. Those who now purchase coverage (about percent of eligible involuntary job leavers) would be expected to take advantage of the subsidy. Those who obtain new group insurance on their own or through a family member might also take advantage of the public subsidy if it exceeded the group subsidy. The substitution of public money for private money among involuntary job leavers might be desirable on equity grounds, but it would increase the cost of the program. Subsidies to low-income involuntary job leavers. Tax credits or other subsidies restricted to low-income involuntary job leavers would be available to 0 percent of this group (Exhibit ) and would be more target-efficient. About percent of uninsured involuntary job leavers would qualify for such subsidies. Moreover, there would be little substitution of subsidies for currently purchased private insurance, even if available to all low-income involuntary job leavers, because only 0 percent of these workers purchase private insurance in the month after losing a job. Almost two-thirds of them are uninsured (Exhibit ). However, subsidies may have to be quite substantial to make private coverage affordable for these workers; expansion of public programs might reach more of them. This possibility is illustrated by the high rate of uninsurance among involuntary job leavers who have access to sub- HEALTH AFFAIRS ~ Volume, Number Downloaded from HealthAffairs.org on February, 0.
DataWatch sidized coverage through a family member. Duration of subsidy. The period of participation in transitional coverage policies may suggest that temporary subsidies need not be of long duration. On the other hand, the short duration of coverage may be in part a result of its high cost. Many who drop become uninsured. Moreover, the majority of -eligible beneficiaries obtained new jobs with insurance within a relatively short period, but this was not true of others who left or lost a job. Behavioral effects. Subsidies might alter workers behavior, leading to increased periods of unemployment. On the other hand, they might also increase mobility and improve productivity. These effects need to be factored into any estimates of the costs and benefits of alternative policies. Subsidies also might alter employers behavior. Almost 0 percent of involuntary job leavers in -eligible jobs have subsidized employer benefits that continue for several months after the job loss. But this could change with government subsidies. Earlier simulation analyses of subsidies to the unemployed have indicated that up to a third of subsidy benefits would be paid to those who otherwise would have had continued employer-subsidized benefits. Limitations. Our analysis presents new data about insurance coverage for laid-off workers, take-up, and the implications for current policy proposals. Our data, however, are limited by small sample sizes and limited follow-up duration. They also rely on self-reports of type of coverage that may be subject to misreporting especially coverage. Moreover, the data pertain to a period of economic good times, and the characteristics of workers losing a job and their patterns of insurance may differ during an economic downturn. Further exploration of these issues with larger samples and longer periods of time would help in formulating effective policy to help laid-off workers obtain affordable coverage. This study was supported by Contract no. J--P--00 from the Pension and Welfare Benefits Administration, U.S. Department of Labor (DoL). Any views expressed herein are solely those of the authors, and no endorsement by the DoL or RAND is intended or should be inferred. NOTES. S.A. Glied, Challenges and Options for Increasing the Number of Americans with Health Insurance, Inquiry, no. (00): 0 0. J.M. Lambrew, How the Slowing U.S. Economy Threatens Employer-Based Health Insurance, Pub. no. (New York: Commonwealth Fund, November 00).. P. Flynn, Qualifying Events and Elections,, Inquiry (Summer ): 0; and M.C. Berger et al., Health Insurance Coverage of the Unemployed: and the Potential Effects of Kassebaum-Kennedy, Journal of Policy Analysis and Management, no. (): 0.. G. Scandlen, Helping Laid-Off Workers Keep Insurance, Brief Analysis no. (Washington: National Center for Policy Analysis, October 00); and Lambrew, How the Slowing U.S. Economy Threatens Employer-Based Health Insurance.. J. Gruber, Transitional Subsidies for Health Insurance Coverage, Inquiry (Summer 00):.. See S. Dorn and J.A. Meyer, What Health Coverage Would Laid-Off Workers Obtain under Recent Tax Credit Proposals? (Washington: Economic and Social Research Institute, March 00).. S. Zuckerman, J. Haley, and M. Fragale, Could Subsidizing Health Insurance Coverage Help Most Low- May/June 00 Downloaded from HealthAffairs.org on February, 0.
Income Unemployed? Health Policy Online, no., October 00, www.urban.org/uploadedpdf/0_ HPOnline_.pdf ( March 00); and L. Duchon et al., Security Matters: How Instability in Health Insurance Puts U.S. Workers at Risk (New York: Commonwealth Fund, December 00).. J.A. Klerman, J.L. Buchanan, and A. Leibowitz, Labor Turnover and Health Insurance, in Health Benefits and the Workforce (Washington: U.S. Department of Labor, Pension and Welfare Benefits Administration, ),.. A. Monheit et al., Health Insurance for the Unemployed: Is Federal Legislation Needed? Health Affairs (Spring ): 0. 0. The Survey of Income and Program Participation (SIPP) is another panel data set that tracks changes in insurance. However, it does not explicitly identify coverage, so we chose to use MEPS.. If a person had more than one job turnover, we examined insurance status following the first job loss.. Other reasons given for leaving a job include retired, quit, took unpaid leave, or had illness or injury.. We compared the insurance status in the initial month following the job loss for the full sample and these two subsamples to ensure that the patterns were similar and that the change in status we report between the initial month and these later months was not attributable to sample differences.. These are based on self-reports of type of coverage. We define eligibility as holding a job with insurance from a firm with twenty or more employees. A categorical measure for firm size, which groups firms with ten to twenty-five employees in a single category, was used when the continuous measure for firm size was missing. In this case, eligibility was identified for those with jobs with insurance in firms with more than twenty-five workers. If a person has coverage and non- employer coverage from the same employer in the same round, it is not possible to distinguish these two types of coverage in the MEPS data. Therefore, we use the reported job end date to determine the start of coverage.. If the worker reported having coverage from an employer in the month after leaving a job but did not report coverage in the month, we assume that the employer continued to subsidize the insurance. Workers on temporary layoff are dropped from the analysis.. We use the CPS for both years of the contrast, rather than the MEPS data, to have consistent definitions over time. We defined job leavers in the CPS as all people who were not working but had worked in the past month, and those who had changed jobs within the past month. Involuntary job leavers are those reporting that the reason for unemployment was loss of a job or the end of a temporary job.. This is similar to the percent figure for. Monheit et al., Health Insurance for the Unemployed.. In some cases, this is a new job; in others, it is a previously held second job.. We calculated uninsurance rates for twelve and eighteen months after job loss by linking the and MEPS data. The MEPS data do not contain information on. 0. Family income is measured as total income for, so it does not necessarily reflect the family circumstances after the job loss.. This estimate is based on the number of people under age sixty-five who reported a lost job in the month preceding the February 00 CPS. The estimate does not include those who left a job voluntarily. It may also overcount the number of people displaced because we counted lost jobs in a month and multiplied them by ; that is, we did not correct for multiple. If subsidies were available to all job leavers including voluntary terminations, the number eligible for the subsidy would increase to 0. million.. The samples of low-income and involuntary job loss participants were too small to allow reliable estimates of durations for these groups.. J.R. Baumgardner, Providing Health Insurance to the Short-Term Unemployed, Inquiry, no. ():.. Gruber, Transitional Subsidies.. Baumgardner, Providing Health Insurance to the Short-Term Unemployed. HEALTH AFFAIRS ~ Volume, Number Downloaded from HealthAffairs.org on February, 0.