Skip Navigation

About 19 million workers will be unemployed at some time during 1998, and more than one-third of them will experience at least a month without health insurance while they remain jobless. Several proposals have been developed to fill that gap by providing temporary health insurance subsidies to unemployed workers. This Congressional Budget Office (CBO) memorandum, prepared at the request of the Ranking Minority Member of the Senate Committee on Labor and Human Resources, analyzes some of those options in terms of their cost and policy implications.

This memorandum considers two main topics related to the provision of health insurance subsidies. First, it discusses issues in designing such a program, including possible eligibility rules and formulas for deciding the amount of a subsidy. Second, the memorandum presents estimates of the federal costs and performance of a prototypical subsidy plan that is based on recent legislative proposals; it also evaluates how the costs and performance of the program would change as a result of modifying the eligibility rules or subsidy formulas. In accordance with CBO's mandate to provide objective and impartial analysis, this memorandum makes no recommendations.

James Baumgardner of CBO's Health and Human Resources Division prepared the memorandum under the direction of Joseph Antos and Linda Bilheimer. Carol Frost programmed the simulation model used to produce the estimates. Amy Weber provided research assistance, and Kathryn Rarick checked the paper for accuracy. A number of people provided helpful comments and suggestions, including Jan Acton, Jim Blum, Doug Hamilton, Christi Hawley Sadoti, Arlene Holen, Jeff Lemieux, Dave O'Neill, Murray Ross, Ralph Smith, Paul Van de Water, Bruce Vavrichek, and Bob Williams.

Leah Mazade edited the manuscript, and Ron Moore and Sharon Corbin-Jallow prepared the memorandum for publication.
 
 

INTRODUCTION

DESIGNING A HEALTH INSURANCE SUBSIDY PROGRAM FOR THE UNEMPLOYED

• Receipt of Unemployment Insurance Benefits
• Low Levels of Family Income
• Recent Employment-Based Health Insurance Coverage
• Excluding People with an Alternative Source of Insurance
• Accounting for the Business Cycle
• Administering the Subsidy Program

IMPACT ON COVERAGE AND COST

• The Basic Results
• Accounting for the Unusually Good Labor Market Conditions Expected in 1998
• Increases in Outlays for Unemployment Insurance

EFFECTS OF RULE CHANGES ON PROGRAM PERFORMANCE

• Changing the Previous-Coverage Requirement
• Changing the Relationship Between Family Income and the Subsidy Rate

HOW THE STATE OF THE ECONOMY WOULD AFFECT PROGRAM OUTLAYS

CONCLUSION

APPENDIX: THE SIMULATION MODEL
 

INTRODUCTION

Because private health insurance is generally obtained through employment, workers who lose their jobs face a significant risk of becoming uninsured. The Congress, in the Consolidated Omnibus Budget Reconciliation Act of 1985 (COBRA), addressed that issue by requiring firms with 20 or more employees to continue offering health coverage to workers who separate from the firm. However, firms may charge former employees slightly more than the full group premium for that continued coverage.⁽¹⁾ Given that employers typically contribute most of the premium for active workers as part of employee compensation, former employees may face a large increase in their out-of-pocket premiums if they lose their jobs. Some firms continue to pay their customary share of the health insurance premiums for workers for a short period of time after a layoff, but those employer-subsidized benefits last only for a limited time. Other firms do not provide such extensions. Sooner or later, accepting coverage under the COBRA provisions generally means a manyfold increase in the explicit premium faced by the worker.

The Congressional Budget Office (CBO) estimates that about 19 million workers will be unemployed for at least some time during fiscal year 1998. About 7.5 million of those workers will go at least a month without health insurance during a spell of unemployment. Of those 7.5 million workers, almost 2.5 million will have had insurance in their most recent month of work; the other 5 million will not have been insured even when they were working. Although many unemployed people will go through some period without insurance, an even larger proportion of them--about 11.5 million unemployed workers--will have health coverage throughout their spell of unemployment.

To target initiatives toward particular groups who may be prone to not having insurance, some policymakers have proposed that the federal government provide health insurance subsidies for the unemployed. Both a Senate bill (S. 2149 of the 104th Congress) and the Clinton Administration's budget proposal for fiscal year 1998 presented new programs to help unemployed people purchase temporary health insurance coverage. The sweeping proposals for health care reform considered during the 103rd Congress also included provisions for insurance for the unemployed.

This memorandum examines the issue of providing health insurance subsidies to unemployed people. The first section discusses the design features of policies aimed at assisting unemployed people to purchase insurance. The second section analyzes the effects of a prototypical policy that is based on recent proposals. Besides considering the cost to the federal government, the number of unemployed people who are likely to use the subsidies, and changes in the incidence of being uninsured under such a policy, the memorandum also explores how altering the eligibility and subsidy rules would affect the policy's outcomes.
 

DESIGNING A HEALTH INSURANCE SUBSIDY PROGRAM FOR THE UNEMPLOYED

Recent policy proposals to provide health insurance for the unemployed share several basic design features. The underlying approach in each initiative is to provide cash subsidies to jobless workers to enable them to pay for private insurance; that is, federal funds would be used primarily to subsidize the payment of COBRA premiums. In addition, the proposals would direct that aid to the same group of uninsured workers: those who had coverage while they were working but who may have stopped having coverage because of its high out-of-pocket cost when they became unemployed. Each proposal also limits the duration of the subsidy to a relatively short period. The basic purpose of the proposals is thus to help unemployed workers continue their health insurance coverage during brief periods of joblessness.

Within that structure, a number of policy issues arise. How, for example, can program spending be limited to workers who become unemployed involuntarily rather than those who choose not to work? Likewise, how can program funds be restricted to those who would not otherwise have had coverage? For instance, some jobless workers who are members of families that can afford the full health insurance premium would have purchased coverage anyway; other workers may be members of families in which the spouse has access to family coverage through his or her job.

A number of questions also arise about the incentives such a program would provide to employers and to state governments that might administer the program. Many employers currently continue health insurance coverage for their workers for a limited time during spells of temporary unemployment. Would employers respond to the establishment of a federal subsidy program by reducing or dropping that coverage? And in terms of administration, how could states be provided with sufficient incentives to administer the program efficiently, given that primarily federal funds would be spent?

The remainder of this section examines several potential eligibility criteria--including receipt of unemployment insurance benefits, a sufficiently low income, and previous insurance coverage--that could focus federal efforts on providing continuity of health insurance coverage to temporarily unemployed workers. However, in developing such subsidies, issues related to program administration and the feasibility of enforcing program rules must also be considered.

Receipt of Unemployment Insurance Benefits

Defining who is unemployed for the purposes of the health insurance subsidy would be the obvious first step in establishing eligibility criteria. Most people in their prime working years, ages 25 to 54, who are not working are not unemployed but are simply out of the labor force. The Bureau of Labor Statistics considers someone to be unemployed if (1) they are not working, are available for work, and have actively sought work within the past month; or (2) they have been laid off but expect to be recalled to work. In 1996, only 18 percent of people aged 25 to 54 who were not working were unemployed. The other 82 percent were out of the labor force.⁽²⁾

For workers to receive health subsidies merely because they met the criteria used to define a person as unemployed could set up substantial incentives for people to switch from the out-of-the-labor-force category to the unemployed category. To target benefits toward people who had demonstrated ties to the work force and, furthermore, to limit the costs of the program, policymakers might prefer to restrict eligibility to those who qualified for and were receiving unemployment insurance benefits.

Tying eligibility for a health subsidy to receipt of unemployment insurance benefits would limit both the number of jobless workers who received benefits and the duration of those benefits for a particular worker. At a given point in time, only about a third of the unemployed receive unemployment insurance benefits.⁽³⁾ Potential recipients can be disqualified if they left their old job voluntarily without good cause or were fired because of misconduct. Eligibility for unemployment insurance benefits differs from state to state, but rules include minimum work and earnings requirements. Workers must have worked in at least two calendar quarters of the so-called base year (the first four of the last five completed calendar quarters), and they must have earned more than a specified level of earnings in that year. Qualifying workers can receive benefits only for a limited time. In many states, the duration of benefits varies based on the amount of the claimant's base-year earnings, with duration rising as those earnings increase. All but two states allow a maximum duration of 26 weeks, but the Congress has typically passed legislation extending those benefits during significant economic downturns. Another reason to use receipt of unemployment insurance benefits as an eligibility criterion is the straightforward administration of such a rule.

Low Levels of Family Income

Policymakers may find it desirable to limit the cost of the health insurance program by ensuring that only low-income people receive a subsidy, based on the view that those with higher incomes are better able to pay for their own coverage. The simplest approach would be to reduce the amount of the subsidy as income rose, with no subsidy available if income exceeded a particular standard.

Another policy question is what the maximum subsidy should be. The maximum subsidy could be set at 100 percent of a worker's COBRA premium or at some other, lower percentage of it. (Similar rules could apply to an individual market policy.) Alternatively, policymakers might want to set a maximum limit on the monthly dollar amount of the subsidy to avoid subsidizing unusually generous health benefit packages.

To take into account both the multiple sources of a family's income and the different expenses that face households of varying sizes, the rules of the health subsidy program could tie benefits to family income relative to the federal poverty guideline. The use of family income, as opposed to the unemployed person's income, would recognize the ability of two-earner families to continue to finance their living expenses if one earner becomes unemployed. And because the poverty guidelines are calculated based on family size, the use of poverty levels, rather than absolute dollar amounts, would take account of the greater difficulty larger families have in maintaining a particular standard of living on a particular dollar amount of income.⁽⁴⁾ For example, the program could pay the maximum subsidy to workers with a family income below the poverty guideline. The subsidy rate would then fall on a sliding scale as family income rose, dropping to zero at some point, such as 200 percent or 300 percent of the poverty level. In that way, families with higher incomes would become ineligible for a subsidy.

Although policymakers might want to target subsidies toward lower-income families on equity grounds and as a way of limiting the size and costs of the program, paying lower subsidies as income increased would create work disincentives within recipient families. If a family's income fell into the range over which the subsidy rate was being reduced, then earning additional income would bring down the family's health insurance subsidy. From the family's point of view, tying the subsidy rate to income is like taxing that income at a higher rate than would otherwise have been the case.⁽⁵⁾ Those work disincentives are limited, however, because recipient families can qualify for subsidies for only a limited time.

Measuring family income involves both conceptual and administrative difficulties. Unemployed workers typically experience a substantial decrease in income when they lose their jobs. Should one use a short-term measure of income, such as monthly income, or a longer-term measure--for example, annual income? Tracking income over shorter periods might involve more administrative complexity than looking at annual income, and the results might be more difficult to verify. Using short-term income would more narrowly pinpoint when an unemployed worker's ability to pay his or her insurance premium was at its lowest, but that might result in subsidies going to families whose long-term income was relatively high. One might argue that such families should be able to finance their premiums out of savings. But the family may not have anticipated the job loss and, on finding itself with a low monthly income, might let its health insurance coverage lapse. Clearly, policymakers face a trade-off between encouraging people to save money for unexpected needs and actually meeting those perceived needs, should an unfortunate event such as job loss occur.

Recent Employment-Based Health Insurance Coverage

If continuity of insurance coverage is viewed as a principal goal of health policy initiatives for the unemployed, it would be reasonable to restrict the new subsidy to workers who had participated in employment-based insurance plans for some required length of time before becoming unemployed. The federal subsidy would offset the explicit premium increases that unemployed workers would otherwise face because they would no longer be receiving a premium contribution through their employer. Restricting eligibility in that way would also address the concern that some people might try to "game" the system by signing up for insurance coverage only when they saw an immediate need for health benefits. A previous-coverage requirement would restrict the new subsidies to people who had shown some commitment to staying insured.

Several variants of such a requirement are possible. For example, the rules could require unemployed workers to have had employment-based coverage throughout the most recent six months during which they were employed (that is, they could have a break in coverage during periods when they did not work) or for a strictly continuous period of six months ending with their most recent employed month. Either a longer or shorter period of previous coverage could be required. Another variant would tie the maximum length of eligibility for the subsidy to the number of months of previous employment-based coverage. For example, people who had health insurance for only their most recent two months of employment could be limited to two months of a health insurance subsidy while unemployed. Six months could be set as the maximum duration of subsidies for people who had been insured for at least their last six months of employment.

Restricting the health subsidies to workers who had been covered when they were last employed would also simplify the mechanics of retaining health insurance coverage. Because all workers in firms of 20 or more employees who participate in an existing employer-sponsored health plan are eligible to continue coverage under COBRA, those workers would find it convenient to use the subsidy to pay the COBRA premium and continue in their previous employer's plan. Workers from firms with fewer than 20 employees (representing about 20 percent of the people eligible for the proposals analyzed here) might have to explore other sources of insurance to purchase with their subsidies.

An issue to consider is whether people who had been purchasing their insurance through the individual market or through a group other than their employer should be eligible for the health insurance subsidies. Although workers in that situation would not face the loss of an employer subsidy (because they did not have one), they would nevertheless find insurance less affordable because of the drop in income associated with becoming unemployed. The cost of the program would, of course, increase as the requirement for previous employment-based coverage was relaxed.

Excluding People with an Alternative Source of Insurance

Policymakers may wish to make a subsidy program more efficient in providing new insurance by excluding people with other sources of coverage. As a practical matter, however, restricting subsidies to those people who would have had no insurance in the absence of the program is quite difficult. First, a program administrator may not have the information to adequately observe someone's potential to obtain other coverage. Second, if the subsidy was sufficiently generous, beneficiaries would have an incentive to conceal alternative sources of insurance. Third, a government subsidy program could supplant the provision of insurance through private markets because the program would be meeting a demand that would otherwise have existed and that private firms could have responded to.

Clearly, unemployed workers who would buy COBRA coverage without a subsidy would also buy it if some or all of the premium was subsidized. But no method exists for determining who those workers would be and then preventing them from taking a subsidy. In addition, some potential subsidy recipients might secure insurance from other sources if there was no subsidy program. Insurance through a spouse's employment-based plan, continued employer-subsidized insurance, and insurance through other government programs are possible alternative sources.

Policymakers may want to exclude from the subsidy program unemployed people who are receiving Medicaid or other government-provided insurance and those who could get employer-subsidized insurance through their working spouse. It may be relatively easy to check on whether a person is enrolled in a government-sponsored program such as Medicaid. However, administrators would have to establish a new reporting apparatus to successfully enforce a rule that excluded someone from a subsidy if insurance was available through his or her spouse.

Enacting the proposed health subsidies for the unemployed would weaken the incentives that firms have to continue subsidizing insurance for laid-off workers. As part of their benefits policies (sometimes reflecting collective bargaining agreements), some firms continue to pay their usual contribution for health insurance to recently laid-off workers. The actual number of months of such extensions of employer-subsidized coverage, if any, differs among firms and may depend on the length of time that a particular worker has been with the firm. If a government health insurance subsidy was available for unemployed workers, employees would place less value on working for an employer that extended employer contributions for insurance into layoff periods, and employers would face less pressure to make such extensions available.

Accounting for the Business Cycle

It might be desirable to design a health insurance subsidy program for the unemployed in a way that would automatically adjust the extent of the program in response to business cycle conditions. For example, during economic downturns, as rates and durations of unemployment increased, policymakers might want the program to subsidize more people and to lengthen the duration of benefits for the typical recipient. By tying eligibility to receipt of unemployment insurance payments in the previous month, the health subsidy program would build in some automatic flexibility for responding to general economic conditions. As the number of people collecting unemployment insurance grew during a recession, the number receiving health subsidies would grow likewise.

Since the Congress typically passes emergency benefits legislation during recessions, which increases the duration of unemployment insurance benefits allowed to recipients, the health subsidy program could also provide benefits for longer periods by simply piggybacking its benefits onto the extensions for unemployment insurance.

A rule that appears in recent proposals would restrict the flexibility of the program to general economic downturns. The rule sets a maximum limit of six months on the length of time for which an unemployed worker can collect a health subsidy. Although eliminating the rule would make the program more responsive to the perceived needs of workers during difficult economic times, that flexibility would increase the costs of the program and would amplify those costs during recessions to an even greater extent than would otherwise occur.

Administering the Subsidy Program

One way to administer the program would be through the states, especially if the subsidies are limited to people who receive unemployment insurance benefits. The states administer the unemployment insurance program; consequently, administrators could establish initial eligibility for the health subsidy program when a person applied for unemployment insurance benefits. At that time, officials could check other eligibility criteria, such as the requirement for previous insurance coverage, family income, and the unavailability of coverage from a spouse's employer or another source.

The administering agencies would have to build systems to verify the previous insurance coverage of potential recipients. Some of the required elements are beginning to be put into place as a result of the Health Insurance Portability and Accountability Act of 1996 (HIPAA), which requires health plans to certify the periods of coverage of participants. Administering an exclusion for potential coverage under a spouse's plan may be difficult, however, even though HIPAA has laid the groundwork for providing documentation that could be used to detect the existence of such coverage.⁽⁶⁾

Using monthly family income to both restrict benefits to workers whose income was below a particular percentage of the poverty level and calculate the level of the subsidy that an unemployed worker would receive would add further complexity to the new program's administration. Clearly, officials in the unemployment insurance office would know the recipient's income from those benefits; they might have greater difficulty checking the income of other family members or some sources of unearned income. Even if data on income from all sources and all family members could be found, such information would arrive only after a lag. Basing subsidies on the most recent month's actual income might not be administratively feasible. One could, however, set up a system that estimated a family's monthly income. With a system based on estimated income, decisions would have to be made about whether to require a later reconciliation period, such as occurs with the income tax system.

If federal funds were used to pay for a state-administered program, certain incentive problems might develop. States might not face the same incentives to enforce eligibility and other rules as strictly as they would if state money was involved. To encourage state compliance, the federal government might want to consider various approaches, which might include some type of financial incentive.
 

IMPACT ON COVERAGE AND COST

This section analyzes the consequences of a health insurance subsidy program for the unemployed based on a prototype proposal. CBO prepared the proposal not as a recommendation of a particular approach but to provide an example of a possible program design. The section also explores the sensitivity of costs and coverage to changes in the eligibility rules and subsidy formulas and to declines in the business cycle. Unless stated otherwise, all estimates refer to fiscal year 1998 and reflect the health insurance costs and unemployment levels projected in CBO's January 1998 baseline. The estimates assume that the prototype program would be fully operational for the entire year.

Under the prototype proposal, eligibility would be restricted to unemployed people who had received some unemployment insurance benefits in the previous month. In addition, a beneficiary could not collect a health subsidy for more than six months during a spell of unemployment. To receive an insurance subsidy, a potential beneficiary would have to have monthly family income below 240 percent of the federal poverty level. The subsidy formula in the prototype plan would allow a 100 percent subsidy for those with monthly family income below the poverty guideline. The subsidy percentage would then fall linearly with monthly family income, phasing down to zero at 240 percent of the poverty line.⁽⁷⁾ To be eligible for the program, people must have been covered by employment-based health insurance throughout their most recent six months of employment. The program would exclude unemployed people who could receive health insurance through their spouse's employer for which the employer's explicit contribution was more than half of the total premium. People who could qualify for Medicaid or Medicare would also be excluded from the subsidy program.

For the purposes of this analysis, the prototype is assumed to be an entitlement. Through the states, the federal government would make funds available to provide subsidies to anyone who qualified for the program and chose to enroll.⁽⁸⁾ The results reported in the next section are based on a simulation model that used data from the 1992 Survey of Income and Program Participation (SIPP), which tracked respondents from early 1992 to late 1994. CBO's analysis adjusted the data for a variety of factors that have changed since the time of the survey, including altered economic conditions and the expiration of emergency unemployment compensation rules. The appendix contains a detailed description of the model used to analyze the data.

The prototype proposal would allow beneficiaries to use the subsidy to pay for their COBRA premium, which under current federal law cannot exceed 102 percent of the full premium cost for a regular employee. The program would also allow beneficiaries to use the subsidies for insurance purchased through the individual market; however, the subsidy would be capped at the cost of a standard Blue Cross policy in the Federal Employees Health Benefits program. Given the relative convenience of extending coverage under COBRA, CBO expects that former employees of COBRA-eligible firms (those with 20 or more employees) would use the subsidies to help pay their COBRA premium. About 80 percent of recipients would use the subsidies on COBRA policies. CBO's cost analysis is based on estimates from employer surveys of health plan premiums--updated for recent premium growth.

The Basic Results

Based on the prototype proposal described above, an estimated 1.4 million unemployed workers would participate in the subsidy program sometime during 1998--or roughly 15 percent of the 9.5 million jobless workers who are expected to receive some unemployment insurance benefits this year (see Table 1).⁽⁹⁾ About two-thirds of unemployment insurance recipients would have a sufficiently low reported family income to qualify for some amount of subsidy. The requirement of employer-sponsored insurance during the last six months of employment would eliminate about 60 percent of them, leaving about 2.5 million jobless workers eligible for the program sometime during the year. Excluding people who have the option of coverage under a spouse's plan or who would be eligible for Medicaid or Medicare reduces the eligible pool another 15 percent to about 2.1 million unemployed workers who would be potentially eligible for subsidies.
 

Many potentially eligible people would not participate because they would only qualify for partial subsidies and would therefore have to pay a portion of their premium out of their own pocket to receive health insurance. Some of those who would be eligible for a full subsidy might not participate either. After accounting for people in those situations, CBO estimates that about 1.4 million workers (or about 2.8 million people including their insured family members) would receive a subsidy sometime during the year.

The prototype proposal is set up to assist unemployed workers with continuity of coverage during a period in which their income would be temporarily low. As a result, the monthly family income of those receiving a subsidy would not be representative of their income during "normal" periods, that is, when they are working. During the most recent six months of employment, the median subsidy recipient would have had a family income of around 250 percent of the poverty level, which is about the 40th percentile of the income distribution among all nonelderly families. During their months of unemployment, subsidy recipients would experience a substantial drop in income compared with the income they had when they were employed. CBO estimates that the median reported family income of those taking a subsidy would be about 110 percent of the poverty level.

The prototype plan would provide subsidies to pay for part or all of the premiums for 4.5 million months of health insurance at a cost of about $1.7 billion in fiscal year 1998. About a third of the subsidized months are months in which recipients would have received health insurance subsidized by their former employer if there had been no government program.⁽¹⁰⁾ An additional 20 percent of the subsidized months represent months in which recipients would have purchased either COBRA coverage entirely from their own funds or insurance through some other source. That means almost 50 percent of the subsidy months would be aiding the purchase of insurance coverage during months when recipients would have been uninsured if no program had existed.

Under the prototype proposal, about 550,000 unemployed workers (1.1 million people including insured dependents) would obtain subsidized coverage for at least one month during which they would otherwise have been uninsured in the course of their spell of unemployment. The proposal would decrease the number of uninsured people by about 350,000, out of about 40 million people who would otherwise be uninsured in 1998 (those figures represent estimates of the number of people who would be uninsured in a typical month). On average, recipients of the subsidy would spend between three and four months of the year participating in the program. Thus, even though 2.8 million people (including insured dependents) would get a subsidy sometime during the year, only a fraction of them would receive a subsidy in a particular month. Furthermore, about half of the months of subsidized coverage would be months during which the recipient would have been insured, even in the absence of the program. The program would spend about $4,800 annually for each person subtracted from the number of uninsured people. The size of that amount relative to the cost of a self-only policy (about $2,300 annually) primarily reflects the fact that about half of the subsidized months would be months when a recipient would have had coverage anyway.

Under the subsidy formula, different recipients would have different percentages of their premiums paid for by the program, depending on their monthly family income. CBO's analysis shows that the average subsidy rate for recipients would be almost 85 percent.⁽¹¹⁾ For a family of four participating in the subsidy program at that average subsidy rate, the dollar value of the health subsidy would equal about 25 percent of the family's reported monthly income.

Because the program rules call for subsidy rates to fall with income, recipient families with monthly income between the poverty level and 240 percent of the poverty level would face a disincentive to earn or report additional income while they were participating in the program. For a family of four, the drop in their subsidy rate for generating more income would be equivalent to having their income taxed at a rate that was about 25 percentage points higher than would otherwise apply.

Like all estimates, those presented here entail some degree of uncertainty. Although CBO's estimates of the program's take-up rate are based on empirical studies of the demand for health insurance, analysts cannot be sure about how a worker's demand for insurance would change when income declined with the onset of unemployment or whether workers would respond to a federal subsidy in the same way they responded to the explicit contribution their employers made to their insurance premiums. The strictness with which the program's eligibility rules would be enforced is also subject to uncertainty.

Accounting for the Unusually Good Labor Market Conditions Expected in 1998

Fiscal year 1998 is expected to have the lowest unemployment rate in more than 25 years. As a result, a more accurate perspective on the typical size of the health subsidy program can be gained by basing the estimates on the labor market conditions that CBO expects in 2002. Currently, CBO projects an average monthly unemployment rate of 4.7 percent in fiscal year 1998 and an average unemployment rate of 5.8 percent in 2002. The latter is quite close to the unemployment rate CBO anticipates over the longer term.

Based on the more typical labor market conditions expected in 2002 (but maintaining the original assumptions about 1998 costs for health insurance), CBO estimates that federal costs for the year for the health subsidy program would be $2.1 billion (see Table 1, column 2). About 1.7 million unemployed workers would participate in the subsidy program, and about 650,000 unemployed workers (1.3 million people including insured dependents) would obtain at least one month of new coverage as a result. The program would reduce the number of uninsured people by about 425,000 in the typical month. Thus, moving from the unusually good labor market conditions expected for 1998 to a more typical labor market increases estimates of the program's size by about 20 percent.

Increases in Outlays for Unemployment Insurance

The new health subsidy program would increase by about one week the average time that affected workers would collect unemployment insurance benefits. That increase would in turn boost unemployment insurance outlays by about $150 million in fiscal year 1998 (or by about $200 million, in 1998 dollars, under the labor market conditions expected in 2002). That estimate is based on a number of studies, which have shown that more generous unemployment insurance benefits cause people to remain unemployed--and collecting benefits--for longer periods compared with less generous benefits. One would expect affected workers to respond to the new health subsidies in a fashion similar to their response to greater unemployment insurance benefits.⁽¹²⁾ Ultimately, those higher outlays would result in higher unemployment insurance taxes charged to employers, which in turn would be passed on to employees in the form of lower pay or benefits than they would otherwise have received.
 

EFFECTS OF RULE CHANGES ON PROGRAM PERFORMANCE

Altering the eligibility rules and the rule for calculating subsidy rates would change the costs of the program and the number of people receiving a subsidy. Rule changes would also affect other outcomes, such as the program's effectiveness in reducing the incidence of being uninsured.

Changing the Previous-Coverage Requirement

Both the number of months of previous insurance required and the strictness of the continuity requirement for previous coverage could be changed.⁽¹³⁾ For example, increasing the required number of months of previous insurance to nine and requiring strict continuity of that period of prior coverage would reduce the estimated annual costs of the program to $1.3 billion and the number of recipient workers to 1.1 million (see Table 2). But the program would be less effective in providing new insurance than under the prototype rules. The number of workers and their insured family members who would gain at least a month of new insurance during a time when they would otherwise be uninsured would fall to 700,000. Thus, the number of people getting some new months of insurance would fall by a greater percentage than the drop in federal costs.
 

Changing the requirements in that way would reduce the monthly number of uninsured people by about 225,000 and increase the cost per newly insured person to $5,500 from $4,800 under the prototype proposal. The cost per newly insured person would be higher because a larger fraction of recipients would have had insurance anyway. Unemployed workers who met the stricter previous-coverage requirements would have been more likely to maintain insurance coverage while they were unemployed, even without a government subsidy program.

Under the stricter previous-coverage screen for eligibility, the program would spend fewer dollars for each person receiving a subsidy--about $575 versus $625 under the prototype rules. The amount spent per person (including insured dependents) is lower because workers who met the stricter screen would tend to remain unemployed for slightly shorter periods than workers who met the eligibility rules of the prototype proposal. Those findings suggest that workers with longer continuous periods of previous coverage may be in jobs with more generous fringe benefits, have more savings with which to purchase COBRA coverage on their own, or, for whatever reason, find it easier or more pressing to get back to work if they have been displaced from a job.

Alternatively, one could propose a less severe previous-coverage screen than in the prototype. The rules could, for example, allow people with fewer than six months of prior coverage to qualify for the program but only for the number of months equal to their period of prior coverage. That rule change would raise federal expenditures to $1.8 billion and increase the number of workers participating in the program at some time during the year to about 1.5 million (about 3 million people including insured family members). The program would be about as efficient in providing new insurance coverage as the prototype proposal. About 1.2 million workers and insured dependents would gain at least one new month of coverage sometime during the year.

Changing the Relationship Between Family Income and the Subsidy Rate

Changing income eligibility rules and the formula relating income to the subsidy rate would affect the costs and other performance measures of the program. A tougher means-testing rule would phase down the subsidy rate more quickly as family income rose. For example, a tougher rule might allow full subsidies for families whose income fell below the poverty level, with a declining subsidy rate as income rose and with no subsidy available to families with income above 185 percent of the poverty level--instead of 240 percent of the poverty level as in the prototype (see Figure 1). That stricter means-testing would reduce federal costs to about $1.5 billion for the year and result in about 1.2 million unemployed workers using the program during the course of the year (see Table 2).
 

Figure 1. Three Alternative Rules for Calculating the Subsidy Rate

A more generous means-testing rule would have different outcomes. For example, a rule that allowed a full subsidy to families with income of up to 150 percent of the poverty level and then a gradual phaseout of subsidies for income of up to 300 percent of the poverty level would cost considerably more than the prototype proposal and result in more unemployed workers using the program (see Figure 1). The program would cost about $2.4 billion and provide subsidies to about 1.8 million unemployed workers at some time during the year.

Several factors lead to those greater costs. Not only would more people be eligible for some subsidy but everyone who was eligible under the prototype proposal would have a subsidy rate that was at least as high, and in most cases higher, than under the original rules. At higher subsidy rates, more people would be willing to join the program because their out-of-pocket premiums would be lower.

Under the more generous means-testing rule, the absolute number of people receiving insurance for some period during which they would otherwise have been uninsured would increase; however, the program would be less efficient than the prototype proposal in providing new insurance. The number of unemployed workers and insured family members getting at least a month of new insurance would rise to 1.3 million from 1.1 million under the prototype rules, but the proportionate increase in federal costs would exceed the increase in people getting some new insurance. Similarly, although the monthly number of uninsured people would fall by a greater number under more generous means-testing, the federal cost per newly insured person would be substantially higher, rising to about $5,500 from $4,800 under the prototype. Two reasons underlie that change. First, subsidy rates would be higher, so the government would be paying more to recipients. Second, as the subsidy program became available to higher-income unemployed workers, the recipient pool would contain relatively more people who would have continued to be insured even in the absence of a subsidy program for the unemployed.
 

HOW THE STATE OF THE ECONOMY WOULD AFFECT PROGRAM OUTLAYS

Near-term growth in outlays for the subsidy program would be on the order of 5 percent to 10 percent annually because of the rising cost of health insurance premiums and growth in the size of the labor force. But recessions, with their sharp increases in unemployment, would cause the most dramatic increases in program costs. Based on the changes in the number of unemployment insurance recipients during the 1990-1991 recession, CBO would expect a recession of similar magnitude to result in increased outlays of between 40 percent and 50 percent over a two-year period. Deeper recessions would cause an even greater jump in costs. (The 1980-1981 recession had a peak unemployment rate that was 1.4 times higher than that during the 1990-1991 recession.) Furthermore, if the Congress responded to a recession by adding a program of emergency unemployment benefits, which it has done in recent recessions, that action would boost the costs of the health subsidy program by an additional 10 percent to 15 percent. (Those increases would occur because emergency unemployment insurance programs prolong the time that a person is eligible for benefits, thereby lengthening the period over which recipients would collect the health subsidies.)⁽¹⁴⁾
 

CONCLUSION

Any policy decision involves trade-offs, and decisions on the design of a program to provide health insurance subsidies to the unemployed are no exception. The estimates in this memorandum center around a prototype proposal for such subsidies. Policymakers might well consider other approaches, and CBO does not recommend any particular method for addressing the issue.

The prototype proposal considered here would assist in maintaining continuity of health insurance coverage for people who were covered by insurance before their spell of unemployment. The program would not make large changes in the number of uninsured people and would not affect those who were chronically uninsured. If the program was fully operational in fiscal year 1998, the number of uninsured people in a typical month would fall by about 350,000 from an estimated total of 40 million. About 1.1 million unemployed people (including insured dependents) would get at least a month of new insurance coverage sometime during the year as a result of the proposed program of subsidies.

The exceptionally low unemployment rate projected for 1998 may give a somewhat distorted sense of the typical size of the program. Using CBO's economic forecast about labor market conditions in 2002, a year in which unemployment is expected to be closer to its longer-term equilibrium level, provides a more representative estimate of the size of the proposed program. With the labor market conditions expected for 2002, the program would reduce the number of uninsured people in a typical month by about 425,000. About 1.3 million people (including insured dependents) would get at least a month of new insurance sometime during the year.
 

APPENDIX: THE SIMULATION MODEL

The estimates in this memorandum come from a model that simulates the effects of different eligibility and payment rules on various performance measures of a proposed program to subsidize health insurance for the short-term unemployed. The principal data source for the model is the 1992 panel of the Survey of Income and Program Participation (SIPP) conducted by the Bureau of the Census. That data source is a nationwide sample that followed respondents from approximately the beginning of calendar year 1992 through late 1994. SIPP included a number of questions about health insurance coverage, work status, income, and the receipt of unemployment insurance (UI) and benefits from other government-sponsored programs by the people in the sample's households. The survey interviewed household members three times a year and asked about the monthly status of many of the important variables related to eligibility for programs such as the health subsidy plans analyzed in this report.

The model identifies people in the SIPP sample who meet the eligibility criteria for the health subsidy and calculates subsidy rates for eligible unemployed workers based on the rules of the particular policy option being considered. Those rules would typically base the subsidy rate on family income relative to the poverty level. The model allocates a subsidized worker's coverage to either a self-only or family plan based on the type of coverage that the worker had when recently employed.

Using a longitudinal data file like SIPP for the policy simulation has two major advantages over the use of other data sets. First, because the eligibility rules have a longitudinal component (for example, the rule requiring employment-based insurance for the last six months of work), only a survey such as SIPP, which tracks people's income, employment status, insurance coverage, and receipt of unemployment insurance over time, allows direct estimation of the number of eligible people. Second, to the extent that people who meet the various eligibility rules are unobservably different from other unemployed people who receive unemployment insurance, analysts can capture those differences by simulating the policy with a longitudinal data base. For example, people who meet the six-month previous-coverage requirement may be people who feel a greater commitment to the workplace or have a greater desire to have health insurance. Because CBO's simulation follows those exact people through their period of unemployment, it can discern their shorter durations of unemployment or their greater likelihood of keeping insurance during their unemployed period (compared with other unemployed people). The anticipated result is more accurate projections of federal outlays and the effects of the proposed programs on the number of uninsured people than one could obtain from estimates based on repeated cross-sectional data sets like the Current Population Survey or tabulations on the UI program.

For people in the sample who qualify for a subsidy, one must calculate the probability that they would actually participate in the program. Many of the potential recipients would not receive a full subsidy to purchase health insurance while they were unemployed. Consequently, a number of qualifying people might choose to forego the subsidy because they would be unwilling to pay the remaining cost of health insurance on their own. (Even some of the people who could qualify for a full subsidy might not join the program for reasons other than monetary costs.)

The model predicts take-up rates based on empirical studies of the demand for health insurance and on the changes in income and the price of insurance that are faced by potential recipients. Workers who lose their jobs incur a substantial loss in income. Furthermore, workers who formerly had employment-based insurance no longer receive the employer's contribution to those premiums. For jobless workers who were eligible for the proposed subsidy program, the lack of an explicit employer contribution might be offset by the fact that workers would have to pay only part of the total premium.

The simulation model accounts for changes in both family income and in the net premium faced by a worker who becomes unemployed but is potentially eligible for the subsidy program. Lower-income people would get larger subsidies; that means that their out-of-pocket price for insurance would be lower, suggesting that they would be more likely to join the program than higher-income eligibles. At the same time, CBO expects that an unemployed worker, with less income available, would have a lower probability of purchasing insurance. In terms of take-up for the proposed program, the price effect dominates the income effect.⁽¹⁵⁾ On average, take-up rates among lower-income eligibles would be higher because those people would have more favorable subsidy rates. The average take-up rate among the potentially eligible population is estimated to be about 65 percent.

Because of the incentives that people would face to underreport monthly family income, CBO assumed some failure would occur in the reporting of earnings of family members other than the UI recipient. Those underreporting assumptions were built into the calculations that determined both eligibility and subsidy rates. The underreporting assumptions should be viewed as correcting for income-reporting problems per se, for more general difficulties in administering the other eligibility rules, and for behavioral changes that occur when subsidies are offered.

In constructing the model, CBO adjusted the SIPP data for several other factors. The UI rules have changed since the SIPP; in particular, laws that paid emergency unemployment compensation have now expired. Using administrative data from the Department of Labor, CBO calibrated the model to account for the fact that some periods of unemployment compensation that occurred during the SIPP sample period would not occur under current law because of the lack of emergency UI programs. Other adjustments included calibrating the SIPP data with administrative data on the number of UI claimants in a year (the unadjusted SIPP data would have resulted in about a 20 percent undercount of UI recipients). Working in the other direction, CBO calibrated the data to match employment conditions that are expected for fiscal year 1998. Given that the economy has improved since the SIPP survey, those corrections reduce the number of UI claimants expected in 1998. (CBO also provided estimates that were calibrated to labor market conditions expected in 2002.)

The model also took into account the behavioral response that was likely to result from the availability of free or reduced-price health insurance to participating UI recipients. In particular, the health subsidy would represent an increase in the generosity of UI benefits, which previous research shows would increase the amount of time that people remained unemployed and collected benefits. In contrast, the expiration of the emergency benefit program since the SIPP data were collected results in shorter UI spells, because a shorter potential duration of benefits has been shown to reduce the length of time that people stay on unemployment insurance.⁽¹⁶⁾

1. Firms are allowed to charge 102 percent of the full premium to those who elect COBRA continuation coverage. That additional 2 percent is intended to compensate for added administrative costs that arise from the firm's complying with COBRA continuation rules. Only those firms that sponsor employment-based health plans must make continuation coverage available to former employees and covered family members. Separated workers may continue COBRA coverage for 18 months. For a full discussion, see Congressional Research Service, Health Insurance Continuation Coverage Under COBRA, IB87182 (January 30, 1995).

2. In 1996, of the entire population aged 25 to 54, 80.2 percent were employed, 16.2 percent were out of the labor force, and 3.6 percent were unemployed.

3. See Daniel P. McMurrer and Amy B. Chasanov, "Trends in Unemployment Insurance Benefits," Monthly Labor Review (September 1995), pp. 30-39, for a discussion of trends in receipt of unemployment insurance payments.

4. The 1997 poverty guideline is $7,890 for a single person. It rises by $2,720 for each additional person in the family.

5. In the context of a health insurance subsidy for which receipt of unemployment insurance (UI) is an eligibility requirement, the disincentives created by an income-related phaseout of the subsidy appear to be more relevant to work by the unemployed worker's spouse or other family members than to work by the UI recipient. The UI recipient already faces work disincentives because of UI benefit formulas; however, those disincentives are temporary because of the limited duration of UI benefits.

6. Under the new law, employers who provide family insurance plans must allow an employee's spouse and other dependents to join the plan if the spouse has lost employer-subsidized coverage (as may happen if the spouse becomes unemployed) and the employee had previously indicated that the reason the dependents had not joined the plan was because they were insured by another group plan. The degree to which employees will make use of their right to reserve the possibility of adding their other family members to their employer-sponsored coverage is not known. Also to be considered is the potential for counterincentives. By reserving the right to sign up for coverage through a spouse's employer, a person could lose the option of qualifying for the insurance subsidy for the unemployed.

7. The 1997 poverty guideline is $658 per month (or $7,890 annually) for a single person and $1,338 per month (or $16,050 annually) for a family of four. Thus, under the rule for the prototype policy, no subsidy would be available for single people with monthly incomes greater than $1,578 ($18,936 at an annual rate) nor for families of four with monthly incomes exceeding $3,210 ($38,520 at an annual rate).

8. Two recent proposals differed in their funding mechanisms. The Administration's budget proposal for fiscal year 1998 provided a formula for determining the specific amounts to be appropriated for each fiscal year and a method to allocate the appropriated funds among the states. If the appropriations proved insufficient to cover all those who met the eligibility requirements and wished to participate in the program, a state would have to either shut down its program for the remainder of the year or reduce the subsidy amounts paid to beneficiaries. S. 2149 of the 104th Congress would have appropriated a fixed amount for the first year of the program and then authorized the Secretaries of Health and Human Services and Labor to adjust that appropriation over time based on changes in health insurance costs and the number of unemployed people.

9. CBO expects that about 7.1 million people will file new claims for unemployment insurance benefits during the year. If recipients who are carried over from the previous fiscal year are added, almost 9.5 million unemployed workers are expected to receive unemployment insurance (UI) benefits in fiscal year 1998. Those figures include a small number of workers (for example, former military personnel) who are not part of the federal/state UI system. (CBO projects 7.0 million new claimants in the federal/state system.)

10. In total, employers would spend around $375 million less on health insurance benefits for laid-off workers than they otherwise would. Competitive labor market pressures are likely to cause employers to pass those savings on to employees by increasing other forms of compensation.

11. The average subsidy rate for potentially eligible workers would be lower than 85 percent. However, because take-up rates would be higher among eligible workers who could receive higher subsidy rates, the average subsidy rate among recipients would be higher than the average subsidy rate available to eligibles.

12. A compendium of these results appears in Paul T. Decker, "Work Incentives and Disincentives," in Christopher J. O'Leary and Stephen A. Wandner, eds., Unemployment Insurance in the United States: Analysis of Policy Issues (Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research, 1997), pp. 285-320.

13. For example, an eligibility rule could require strict continuity. Such a rule would mean that the worker had to be continuously insured for the required number of months, ending when the worker was last employed. Any break in that prior coverage period would disqualify the worker, even if the break had occurred at a time when the person was not working. Alternatively, a rule might require that a person have the necessary months of coverage while employed but allow breaks in coverage as long as they occurred when the person was not employed. The prototype assumed the latter rule.

14. Under the regular rules for unemployment insurance, many recipients qualify for fewer than six months of benefits. An emergency unemployment compensation (EUC) program would add weeks of benefits for many recipients that were still within the six-month cap for the health program. Thus, EUC programs would lead to an increase in the costs of the health subsidy program even if the six-month limit was kept in place. If the six-month limit was dropped during recessions, costs would rise even higher.

15. CBO used elasticity estimates within the range found by M. Susan Marquis and Stephen H. Long, "Worker Demand for Health Insurance in the Non-Group Market," Journal of Health Economics, vol. 14 (May 1995), pp. 47-63. Marquis's and Long's results are within the range of elasticity estimates that have been reported for the health insurance market more generally.

16. See footnote 12 in the text.