Health Insurance and Taxes: The Impact of Proposed Changes in Current Federal Policy
The National Coalition on Health Care (NCHC)
October 18, 1999
The Lewin Group, Inc.
TABLE OF CONTENTS
TECHNICAL APPENDIX: ESTIMATING THE IMPACT OF TAX CREDITS ON INSURANCE COVERAGE
The various federal tax exemptions and deductions for health insurance premiums and health services provide Americans with about $125.6 billion in tax subsidies per year. These subsidies derive primarily from the fact that most spending for employer-sponsored health benefits is not counted as taxable income to the individual in computing either income tax or Social Security tax, even though these benefits are compensation to the employee. This tax-exempt treatment of health benefits has encouraged the development of employer-sponsored plans, which now provide coverage to about 158 million workers and their dependents. These tax subsidies derive from the following:
- Employer contributions for health benefits are not taxable to the individual;
- Employee contributions for health benefits in firms with a Section 125 plan (i.e., a “cafeteria plan”) are tax-exempt;
- Out-of-pocket expenditures for health services can be paid in pre-tax income in firms with tax-exempt reimbursement accounts; and
- Expenditures for health insurance and health services in excess of 7.5 percent of adjusted gross income (AGI) are tax-deductible.
We estimate that tax expenditures under the federal income tax (i.e. ñ the amount not collected in taxes due to the above exclusions) will be $84.9 billion in 2000 (Figure ES – 1). This includes tax expenditures resulting from the health benefits exclusions, including the exclusion for employer contributions and employee contributions in Section 125 plans ($74.5 billion); the exclusion for reimbursement accounts ($5.6 billion); and the deduction for out-of-pocket spending over 7.5 percent of AGI ($4.8 billion). The exclusion of health benefits from Social Security and Medicare HI taxes will account for tax expenditures of $31.9 billion and $8.8 billion, respectively. The average tax subsidy for families with a family head under age 65 will be $1,155 in 2000 (Figure ES – 2).
The structure of the tax exemption raises important equity issues. Employer-provided health benefits are tax-exempt to the worker while individuals who must purchase coverage in the individual market receive no subsidy. Moreover, the value of the tax exemption is greatest among higher-income individuals who are more likely to be insured and who are in higher tax brackets (i.e., the tax expenditure is equal to the cost of coverage that is tax-exempt multiplied by the familyís marginal tax rate). For example, the average tax subsidy per family varies from just $79 for families with incomes of less than $15,000 to $2,638 for families with incomes in excess of $100,000 per year (Figure ES – 2).
In fact, 68.7 percent of federal health benefits tax subsidies under the current tax system are going to families with incomes of $50,000 or more, even though this group accounts for only 36 percent of the population. Redirecting some of these subsidies to lower-income groups could help many of the uninsured obtain coverage. With 43.4 million uninsured persons in the United States, many of whom are in relatively low-income groups, it is important for policy makers to question whether it is appropriate for the majority of health benefits tax subsidies to go to the highest-income groups.
Figure ES – 1
Federal Tax Expenditures for Health Benefits in 2000 (in billions)
Source: Lewin Group estimates using the Health Benefits Simulation Model (HBSM).
Figure ES – 2
Average Federal Health Benefits Tax Expenditure by Income Level in 2000 a/
a/ Estimates for families with a family head under age 65.
Source: Lewin Group estimates using the Health Benefits Simulation Model (HBSM).
Concern over these issues has led to a series of proposals that are designed to provide a more equitable distribution of tax subsidies across families and assist uninsured persons in obtaining health insurance. These reforms range from incremental changes that extend a tax deduction to persons with non-group coverage to sweeping reforms that replace the current tax exemption for health benefits with a refundable tax credit to all privately insured persons. Some proposals are also specifically designed to encourage individuals to moderate their use of health services.
One proposal is to extend a health insurance tax deduction or credit to persons who purchase non-group coverage in the individual market. We estimate that a tax deduction for purchases of non-group coverage would induce about 3.9 million uninsured persons to purchase coverage at a cost to the federal government of $6.3 billion (Table ES – 1). However, because the value of the tax deduction is greatest for persons at high income levels, about 61 percent of these newly insured persons would be in families with incomes of $50,000 or more, with little change in coverage for lower-income groups.
Table ES – 1
Summary of Cost and Coverage Impacts Under Alternative Changes in Subsidies for Health in 2000
|Net Federal Cost|
|Reduction in Uninsured|
|Federal Cost per Newly Insured Person|
|Tax Deduction for Non-Group Coverage||$6.3||3.9||$1,599|
|Tax Credit for Non-Group Coverage ($500 single, $1,000 family)||$5.0||4.0||$1,247|
|30 Percent Tax Credit to Low-Income Workers||$3.3||1.5||$2,121|
|30 Percent Tax Credit to All Low-Income Families||$11.3||4.5||$2,530|
|Replace Tax Exemption with Flat-Dollar Credit ($800 single, $2,400 family)||$48.6||4.6||$10,541|
|Credit/Exemption Model (maximum of credit or current expenditure)||$53.2||9.8||$5,429|
|Heritage Foundation Proposal||$55.3||43.4 a/||$1,274|
a/ Estimates of the number of uninsured vary across data sources. For illustrative purposes, we based our analysis on the Bureau of the Census estimate of the number of uninsured persons in 1997 (43.4 million).
Source: Lewin Group estimates.
By comparison, we estimate that a refundable tax credit for purchasers of non-group coverage ($500 for single coverage and $1,000 for family coverage) would result in roughly the same number of persons taking the coverage (4.0 million) at a cost of $5.0 billion. But the tax credit would tend to increase coverage among lower-income groups because the credit amount for these groups would be substantially greater than the value of the deduction among persons with lower incomes. In fact, 61 percent of the persons who would become covered as a result of the tax credit would have incomes of less than $50,000 per year. Thus, the design of the tax subsidy can make a substantial difference in the types of population groups that would take coverage.
Another approach is to create a targeted tax credit for lower-income families. One proposal would provide a 30 percent health insurance tax credit for workers who do not have access to employer coverage and who have incomes below $35,000 for single individuals and $50,000 for families. There would be about 12.2 million uninsured workers and dependents who would be eligible for such a credit. However, only about 1.5 million uninsured persons would be induced to obtain coverage by the credit, at a cost of $3.3 billion. The reason for this low level of participation is that the credit covers only 30 percent of the premium, leaving the individuals to pay the remaining 70 percent, which many lower-income families would still find unaffordable.
The impact of such a 30% tax credit could be increased by extending it to non-workers with individual coverage, or by increasing the percentage of the premium covered by the credit. For example, the 30 percent credit could be extended to all low-income persons rather than just low-income worker families. The number of uninsured who would take coverage under this variation would be 4.5 million persons, at a cost of $11.3 billion.
Some proposals would eliminate the income tax exclusion for health benefits (i.e., the Social Security tax exemption would remain) and replace it with a flat-dollar tax credit. One proposal establishes a tax credit of $800 for single individuals and $1,600 for married couples, with an additional $400 per child up to a maximum of $2,400 per family. The cost of such a tax credit ($117.1 billion) would be partly offset by the amount of taxes collected as a result of eliminating the income tax exclusion for employer-provided benefits ($68.5 billion), reducing the net federal cost of this proposal to $48.6 billion. We estimate that 9.8 million uninsured persons would take coverage as a result of such a flat-dollar tax credit. However, many higher-income persons would find that the credit is lower than the tax subsidy they received under current law and we estimate that some (5.2 million) would actually drop their current coverage. Thus, this flat-dollar tax credit would result in a net increase in coverage of 4.6 million persons (i.e., 9.8 ñ 5.2).
To prevent any drop in coverage, some flat-dollar tax credit proposals include a provision where tax filers can choose between the tax credit or the existing tax exclusion to ensure that all families do at least as well as they would under current tax policy. This provision would increase the net federal cost of the flat-dollar tax credit proposal from $48.6 billion to $53.2 billion, but would reduce the number of uninsured by 9.8 million persons.
The Heritage Foundation plan is the most sweeping reform plan we examined. The Heritage plan achieves universal coverage by requiring all persons to purchase at least a high-deductible insurance package specified in the plan. Employers who now provide coverage are required to cash-out their health benefit by increasing wages by the amount of the cost of the coverage that they have been providing. Individuals would then use this income to purchase insurance in the individual market. The plan would provide a refundable tax credit for insurance premiums and out-of-pocket payments for health services, which would cover a percentage of these costs. The percentage of costs covered would vary by the level of health spending as a percentage of income. The tax credit would cost about $171.4 billion in 2000, which would be partly offset by $116.1 billion in increased tax paymentsñboth income and FICA taxesñon the increase in wages resulting from the cash-out of employee health benefits and the elimination of the health expense deduction. Thus, the net federal cost of the program would be $55.3 billion (i.e., $171.4 ñ$116.1).
The Heritage plan is unique in that individuals are given the cash that their employerís are now spending for health benefits, which they can use to purchase the coverage of their choosing. This creates a financial incentive for individuals to enroll in less expensive managed care health plans or high-deductible plans so that they can retain as much of the cash-out as possible for other uses. We estimate that the resulting moderation in health services utilization would reduce health spending by about $20.3 billion. However, these savings would be more than offset by an increase in utilization for newly insured persons ($37.8 billion) and an increase in insurance administrative costs ($3.9 billion), for a net increase in national health spending of $21.4 billion in 2000.
Overall, our analysis shows that refundable tax credits would tend to shift federal tax subsidies from higher-income groups to lower-income groups. This would enable many uninsured individuals to purchase coverage, resulting in a decline in the number of uninsured. However, unless there is a mandate for individuals to have coverage, there still would be 30 million or more persons without health insurance under the tax credit proposals that Congress is now considering. This is because many individuals would find that the cost of insurance is greater than they feel that they can afford, even with the credit.
While the impact of tax credits on coverage would be limited (with the exception of the Heritage plan), the emerging debate over tax credits provides an opportunity to rethink the role of tax policy in insurance coverage. Historically, the tax-exempt status of health benefits has played an important role in encouraging employers to provide health benefits. However, it is important to critically evaluate the equity of the existing system and its role in encouraging employers to offer comprehensive coverage that in turn encourages increased use of health services. Well-designed changes in tax policy could lead to an expansion in coverage together with improved equity and
greater emphasis on efficient health care delivery.
The current tax code will provide about $125.6 billion in tax subsidies for the purchase of health insurance and health services in 2000. These subsidies derive primarily from the fact that most spending for employer-sponsored health benefits is exempt from taxation under both the income tax and Social Security tax, even though these benefits are compensation to the employee. This tax-exempt treatment of health benefits has encouraged the development of employer-sponsored plans, which now provide coverage to about 158 million workers and dependents.
Under current tax law, health insurance premiums are largely tax exempt if the insurance is provided through an employer, but generally are not deductible when an individual purchases it directly. The share of the premium paid by the employer is not counted as income to the employee under the federal income and Social Security payroll taxes, and the employeeís share of the premium can be tax-exempt in firms with Section 125 flexible spending plans. Out-of-pocket health expenditures in excess of 7.5 percent of adjusted gross income (AGI) are tax deductible for all individuals. Also, many employees have access to a reimbursement account under their employerís flexible spending plan through which out-of-pocket health expenditures may be paid in pre-tax dollars.
While the current tax treatment of health benefits has encouraged employers to offer coverage, it has been criticized as inequitable and a major contributor to health care cost inflation. A major source of inequity in the current system is that it provides substantial tax benefits to persons with employer-sponsored coverage while persons who must purchase coverage in the individual market receive no tax subsidy. Also, the value of the health benefits tax exemption is greatest among higher-income workers, who are more likely to have coverage and face higher tax rates (i.e., the value of the exemption is equal to the cost of exempt benefits multiplied by the taxpayerís marginal tax rate). In addition, critics of the current system argue that the tax subsidy for health benefits tends to encourage increased use of health services by artificially lowering the net cost of health insurance to the individual. This encourages individuals to enroll in relatively comprehensive health plans that can encourage individuals to use more services.
Another criticism of the employer benefits tax exclusion is that it tends to lock workers into the coverage offered by their employer. Workers must take the coverage offered by their employer in order to take advantage of the tax exemption. Workers who prefer some other form of coverage cannot apply the tax exemption to the cost of purchasing an alternative health plan in the individual market. This tends to lock workers into the coverage decisions made by employers concerning managed care, covered services, and the physician network. This can be particularly restrictive for workers in firms that offer only one coverage option, which make up about 44 percent of all workers.
It also has been argued that employees would be better off if the value of health benefits were cashed-out as wages so that they could purchase health insurance policies that best meet their needs. In particular, this would provide individuals with the option of purchasing a lower-cost health plan and using the unspent portion of the cash-out for other uses. This approach provides a powerful cash incentive for individuals to enroll in plans that control costs through managed care or higher deductibles and co-payments that encourage individuals to moderate their use of health services. Enrollment in these types of plans could slow health care cost growth.
Concern over these issues has led to several proposals to reform the current tax treatment of employer-sponsored health benefits. These proposals range from relatively modest reforms, such as extending a tax deduction or tax credit to persons with non-group coverage, to sweeping reforms that would replace the existing tax exclusion with a tax credit for purchases of health insurance. To varying degrees all of these proposals are designed to increase insurance coverage by providing tax subsidies to individuals who are not receiving subsidies under current law.
Other proposals have emerged that would use tax credits to expand coverage among lower-income persons. For example, one proposal would provide a refundable 30 percent tax credit for purchases of health insurance for lower income families. The credit would be available only to workers with incomes below $35,000 ($50,000 for families) who do not have access to employer coverage (excluding Medicare and Medicaid recipients). However, this credit could be made generally available to all persons with incomes below these levels, including non-workers and workers who have access to employer coverage (credit for employee contribution only). Making these subsidies available to persons who have access to employer coverage could have a significant impact on coverage because, as discussed below, there are about 10.2 million uninsured workers and dependents who have declined the coverage that is available to them through employment.
Some proposals would create a flat-dollar tax credit that would replace the current exemption for employer-provided coverage. These proposals would effectively shift tax subsidies away from higher-income groups to lower-income groups. For instance, one proposal creates a refundable flat-dollar tax credit of $800 for single coverage or $1,600 for couples, plus $400 per child up to a maximum of $2,400. This credit would be substantially greater than the tax subsidy now going to lower income persons under the current tax exemption for employer-sponsored coverage. However, the credit would be less than the value of the exemption for some higher-income tax filers, which could result in some of these individuals dropping coverage. Moreover, the flat-dollar credit effectively places a limit on the subsidies a family receives, which will encourage some persons to enroll in more cost-effective health plans.
To eliminate the reduction in tax subsidies that would occur for some taxpayers under a flat-dollar tax credit, the National Association of Health Underwriters (NAHU) has developed a similar flat-dollar tax credit program where individuals can choose between the credit and the existing exclusion. This would result in a general increase in the subsidies available to lower-income tax filers while ensuring that no family will see a net reduction in tax subsidies under the plan.
The Heritage Foundation has developed a proposal to reform the tax treatment of health benefits that would cover all Americans while changing the tax subsidy in a way that encourages more efficient use of the health care system. The Heritage plan would eliminate the current tax exemption for health benefits and replace it with a refundable tax credit for premium and health care expenses for all persons, including workers and non-workers (Medicare and Medicaid beneficiaries would not be eligible for the credit). Employers would be required to cash-out their health benefits in the form of wages that would be subject to federal income and Social Security (i.e., FICA) taxes. All individuals would be required to purchase coverage, which would generally be provided through the individual market. The plan establishes a minimum standard benefits package with high deductibles that all insurers must offer, but permits individuals to purchase more comprehensive coverage if they wish. The insurance market would also be reformed to ensure guaranteed issue of coverage regardless of health status, with rates that vary only with age, sex and geography. This approach is designed to encourage individuals to enroll in efficient managed care plans and/or high-deductible plans that encourage moderation of the use of health services.
In this study we estimate the amount of tax subsidies provided under the current system and estimate the impact of proposed changes in the tax treatment of health benefits on federal costs and coverage. We then analyze the impact of the individual health benefits tax reform proposals. A discussion of the data and methods used to develop these estimates is presented in Appendix A. Our analysis is presented in the following sections:
- Health Benefits Tax Expenditures in 1998;
- Providing Tax Subsidies for Individually Purchased Non-Group Coverage;
- Low-Income Tax Credit Proposals;
- Flat-Dollar Tax Credit Proposal; and
- The Heritage Foundation Plan.
These various tax preferences for health benefits have effectively subsidized the purchase of employer-based insurance, which has encouraged employers to offer coverage. However, it is also widely believed that this tax subsidy has led some employers to provide relatively comprehensive insurance coverage with broad benefits and low deductibles and co-payments that encourage greater consumption of health care services than otherwise would be the case. Moreover, these tax preferences result in substantial reductions in tax revenues that tend to favor higher-income groups who are most likely to have employer-sponsored coverage. The amount of the tax revenues forgone by the government because of these tax preferences is called the “tax expenditure” for health benefits.
In this section, we present estimates of the magnitude of the “tax expenditure” resulting from the special tax treatment of health insurance premiums and of health care expenditures under the current tax system.
The tax expenditure for health benefits is heavily skewed toward high-income groups. We estimate that the average tax expenditure per family (including all families) will be $1,155 in 2000 (Figure 2). However, the average tax expenditure will be $2,638 for families with incomes in excess of $100,000 per year. By comparison, the average tax expenditure for families with incomes of less than $15,000 per year will be only $79 per family in 2000.
This reflects the fact that families with relatively higher incomes are in higher tax brackets and face a higher marginal tax rate, resulting in a large tax expenditure. It also reflects the fact that higher-income workers are more likely to have employer-sponsored coverage.
Average Federal Health Benefits Tax Expenditure by Income Level in 2000 a/
a/ Estimates for families with a family head under age 65.
Source: Lewin Group estimates using the Health Benefits Simulation Model (HBSM).
We estimate that about 23.6 percent of all federal tax expenditures in 2000 will be attributed to families with incomes of $100,000 or more per year even though this group will account for only 10 percent of the population (Figure 3). In fact, 68.7 percent of tax expenditures will be for families with incomes of $50,000 or more per year, who will make up only about 36 percent of the population. Only 31.3 percent of all tax expenditures will go to families with incomes below $50,000, even though this group will make up 64 percent of all families in the United States.
Families headed by a married individual on average will receive over twice as much tax benefit as families headed by single individuals. We estimate that the average tax expenditure in 2000 will be $1,585 in married-couple families compared with an average of $719 per single-head family (Figure 4). This reflects the fact that a disproportionate share of single individuals will be uninsured. It also reflects the fact that married-couple families tend to have a larger average family size than single-head families and therefore tend to have more family coverage policies, which are more costly. However, these estimates reflect our assumption that the value of the health benefit to an individual will vary by whether the worker has a single or a family policy. These numbers would change substantially if a uniform benefit value for all workers in firms had been used.
Distribution of Federal Health Benefits Tax Expenditures by Family Income in 2000 (in billions)
Source: Lewin Group estimates using the Health Benefits Simulation Model (HBSM).
Average Federal Health Benefits Tax Expenditures by Marital Status and Age in 2000
Source: Lewin Group estimates using the Health Benefits Simulation Model (HBSM).
The average tax expenditure amount varies with the age of the family head. We estimate that the average tax expenditure for families with a family head under the age of 25 will be $846 in 2000, compared with an average of $1,630 per family in families headed by a person aged 45 to 54 (Figure 4). This reflects the fact that younger individuals are more likely to be uninsured and are disproportionately single, resulting in less family coverage. It also reflects the fact that older workers tend to have relatively higher incomes where the marginal tax rates are higher. The average tax expenditure starts to fall among families with a head aged 55 to 64 as workers begin to retire. The average tax expenditure is estimated to be $396 per family with a head aged 65 and older even though these individuals are generally eligible for Medicare. This reflects the tax expenditure for retiree coverage and the tax expenditure for families with a head over age 65 who has a working spouse with employer coverage.
One of the greatest criticisms of the existing tax treatment of health benefits is that employer health benefits expenditures for workers are tax-exempt while persons who must purchase coverage in the individual insurance market receive no tax subsidy. In 2000, we estimate that there will be about 209.8 million persons without Medicare or Medicaid coverage. Of these 209.8 million persons, 158.1 million would be workers and dependents covered under an employer plan, all of whom would benefit from the tax exclusion for employer-provided health benefits. This would leave 51.7 million persons (i.e., 209.8 ñ 158.1) in families who would receive no tax subsidy for the purchase of insurance, including 8.3 million persons who would purchase individual non-group insurance and 43.4 million uninsured persons.
This inequity in the tax treatment of health benefits could be addressed by permitting individuals who do not have employer-sponsored coverage to deduct the full amount of the cost of individually purchased health insurance in determining their income tax. This deduction would be available only to individuals who purchase insurance in the individual market (excluding persons with Medigap coverage), including both workers whose employer does not provide health benefits and non-workers. Also, the availability of this deduction would encourage uninsured individuals to purchase non-group coverage by lowering the after-tax cost of insurance to the individual.
This approach would result in roughly the same federal income tax subsidies at a given income level for persons with and without employer-based coverage. However, the deduction would expand upon another perceived inequity in the current system. Under the current system, the value of the tax exemption generally increases with income. The reason for this is that the value of the deduction to the individual is greatest at higher income levels where marginal tax rates are highest (the value of the tax deduction is equal to the exempt amount of health benefits costs multiplied by the marginal tax rate, which increases with income). Thus, a tax deduction for non-group coverage would favor primarily higher-income persons while providing relatively little tax benefit to lower-income persons. Moreover, the tax benefit under the deduction can not be larger than the amount of taxes paid, which means that very low-income persons with little or no tax liability would receive little or no assistance under the deduction.
An alternative approach is to provide a flat-dollar tax credit to all tax filers purchasing non-group coverage regardless of their incomes. The tax credit would be $500 for single individuals and $1,000 for families. It would be refundable so that the credit amount could exceed a tax filerís tax liability for the year. This ensures that all persons purchasing non-group insurance will have access to substantial tax subsidies regardless of their incomes. Moreover, because subsidies for low-income persons under the tax credit are greater than they would be under the tax deduction, lower-income persons are more likely to respond to the credit by purchasing coverage for themselves and their families.
In this section, we estimate the impact of the tax deduction for persons purchasing individual non-group coverage on federal revenues and the number of persons with health insurance coverage. We then estimate the cost and coverage impacts of a bill that, instead of creating a tax deduction, provides a tax credit for tax filers purchasing non-group insurance in the individual market. This is the same group targeted under the tax deduction proposal. We then compare the costs and characteristics of uninsured persons who are induced to purchase coverage under these two proposals. In addition, we examine ways of funding these new tax subsidies by limiting the amount of premium contributions that are exempt from taxation for persons covered under employer plans.
As discussed above, the tax deduction would be available for the full amount of premiums paid for individually purchased non-group coverage. The tax credit would be available to tax filers for non-group premiums paid in months in which they were not covered by an employer plan, Medicare, or Medicaid. Excluding Medicare recipients effectively ensures that the deduction is not available for premiums paid for Medicare supplemental coverage (i.e., Medigap policies). Under the tax deduction model, premium payments for non-group coverage would no longer be counted in determining the deduction for health expenses in excess of 7.5 percent of AGI because these already will have been deducted.
In this scenario, we assumed that the deduction would take the form of an offset in calculating AGI rather than an itemized deduction. This means that the full amount of premium payments for non-group coverage is subtracted from income for all tax filers, including those who do not itemize deductions. This ensures that all eligible tax filers would benefit from the deduction even if they do not have enough deductions to itemize. Conceptually, this is the approach that is most comparable to the employee benefits exclusion, which effectively excludes the exempt amount of health benefits from AGI. Thus, the income tax deduction for non-group coverage would effectively mimic the effect of the exclusion for employer benefits under the current system.
About 51.7 million persons would be eligible for the tax deduction in 2000, including 8.3 million persons who would already be purchasing non-group coverage and about 43.4 million persons who would be uninsured. We assume that all of those persons who would be purchasing non-group coverage would take the deduction. We also estimate that about 3.9 million uninsured persons will be induced to purchase coverage as a result of the reduction in the net after-tax cost of coverage resulting from the deduction (Table 1). As discussed above, this estimate is based upon a prior Lewin Group study indicating that a 1.0 percent reduction in the net cost of insurance to the worker results in roughly a 0.2 percent increase in the percentage of workers and dependents with coverage.
The Impact of a Health Insurance Deduction for Persons Without Employer-Sponsored Insurance on Coverage and Federal Costs in 2000 a/
|Persons in Families Eligible for Deduction (1,000s)||Persons in Families Who Take Deduction (1,000s)||Percentage Taking Deduction||Reduction in Uninsured (1,000s)||Tax Revenue Loss (millions)||Cost Per Recipient||Cost Per Newly Insured Person|
|Currently Purchasing Non-Group Coverage||8,281||8,281||100.0%||– –||$3,555||$429||– –|
a/ Estimates assume that persons who are not covered under an employer-sponsored plan are permitted to deduct from AGI the amount of their non-Medigap premium payments for individually purchased non-group coverage.
b/ Equals the tax revenue loss ($6.3 billion) divided by the number of newly insured persons (3.9 million).
Source: Lewin Group estimates using the Health Benefits Simulation Model (HBSM).
The tax deduction would result in a loss of tax revenues of about $6.3 billion in 2000. Of this $6.3 billion in tax subsidies, about $2.7 billion would go to the newly insured persons while the remaining $3.6 billion would be attributed to persons who are already purchasing insurance. The average annual federal cost per newly insured person would be $1,599.
- Tax Credit for Non-Group Insurance Premiums
- The Tax Deduction and the Tax Credit Compared
- Placing Limits on the Tax-Exempt Amount of Employee Benefits
Several members of Congress are reportedly considering legislation that would provide a refundable tax credit for the health insurance premium payments by workers for themselves and their dependents. For example, one proposal calls for a tax credit that would be equal to 30 percent of premium payments and would be available to persons with incomes below specified levels. Such a program would provide subsidies to lower income workers and their dependents for the purchase of insurance, which would help reduce the number of uninsured. It also relieves tax inequities arising from the fact that workers with employer-sponsored coverage benefit from the health benefits exclusion while workers in firms that do not offer employer-sponsored benefits receive no tax benefits for the insurance that they can obtain only in the individual market.
In this analysis, we estimate the impact of a 30 percent tax credit for health insurance premium payments for persons who are not already receiving tax subsidies through the benefits exclusion. The amounts subject to the credit do not include Medicare Part-B premiums and premiums for a Medicare supplemental policy (i.e., Medigap), which effectively excludes Medicare beneficiaries from eligibility. The credit would be equal to 30 percent of the cost of insurance for working taxpayers who do not have access to an employer-sponsored plan. The tax credit would be refundable, which means that the tax credit amount can exceed the amount paid in taxes. We assume that the amount of the credit is phased out for taxpayers above specified income thresholds as follows:
- The credit would be equal to 30 percent of allowable insurance payments for single individuals with AGI below $25,000. The credit would be phased out for single persons with AGI between $25,000 and $35,000.
- Married couples with AGI below $40,000 would receive the full 30 percent credit, while the credit is phased out for those with AGI between $40,000 and $50,000.
While the low-income tax credits that have been proposed limit eligibility to workers without access to employer coverage, they could be extended to two other groups as well. First, the tax credit could be provided for the employee contribution for employer-sponsored coverage (except in cases where the workerís contribution is already tax exempt through a Section 125 program). This would assist those who have declined employer coverage for themselves or their dependents (estimated to be 10.2 million) in taking the coverage that is available to them through employment. Second, the tax credit could be made available for insurance purchases for non-workers who currently receive no tax benefits for purchasing insurance. Our analysis is presented in the following sections:
- Tax Credit for Workers Without Access to Employer Coverage;
- Tax Credit for All Workers;
- Tax Credit for Workers and Non-Workers;
- Varying the Size of the Tax Credit Amounts; and
- Implementation Issues.
- Tax Credit for Workers Without Access to Employer Coverage
- Tax Credit for All Workers
- Tax Credit for Workers and Non-Workers
- Varying the Size of the Tax Credit Amounts
- Implementation Issues
A central issue with these refundable tax credits is administrative feasibility. For the program to be effective, there would have to be a way of getting the tax credit to individuals at the time they are purchasing coverage rather than waiting until the following spring to get the credit in a refund from the federal Treasury. This is particularly true for low-income persons who cannot afford to “front” the cost of insurance until tax refunds are distributed in the following year.
This problem could be remedied under a tax credit where the tax credit is available to all individuals regardless of income. Under such a system, the insurer could collect the credit from the US Treasury on a monthly basis as partial payment for coverage based on their enrollment ledgers. However, it is unlikely that this approach would be used under a tax credit where eligibility or the amount of the credit varies with income. This is because employers and insurers do not have the information required to determine income eligibility. Moreover, insurers seeking to maximize enrollment would have a conflict of interest in determining whether individuals are eligible for the credit.
Advance payments of the tax credit could be arranged through the employer withholding system, as is currently done with the Earned Income Tax Credit (EITC). Under this advance payment system, the expected amount of the credit is offset against the individualís expected tax payments to allow eligible individuals to receive a greater portion of their gross income in their paychecks. In cases where the refundable EITC amount is greater than the expected tax payment, the difference is added to the individualís pay check (i.e., a “negative withholding”) so that the money is available to the individual when needed. A similar approach could be used to disburse refundable health insurance tax credits.
However, experience with the existing EITC advance payment system suggests that it may not be a very effective means of disbursing advance payments of a health insurance tax credit. The EITC advance payment system is sufficiently complex for workers and employers to use that only a small fraction of those who are eligible for the EITC use it. It is doubtful that the advance payment system would be any more effective where health insurance tax credits are concerned. Moreover, the employer income tax withholding system would not offer a solution for persons who are not working. Thus, a major effort would be required to develop an effective mechanism for distributing means-tested tax credits on a timely basis before this approach could be expected to have a significant impact on the number of uninsured.
Several proposals have emerged that would replace the employee health benefits exemption with a flat-dollar tax credit to all eligible tax filers. Under the flat-dollar tax credit, each family is provided with a fixed dollar amount of subsidy in the form of a tax credit that varies only by family size. The amount of the subsidy will not vary with income or source of private coverage as it does under the current system. The credit is refundable so that the amount of the credit can exceed the amount of taxes owed by the individual. Also, the credit applies to health insurance premiums for non-group insurance purchased in the individual market as well as the premium for employer-sponsored health insurance and provides a more uniform distribution of tax subsidies across income and insurance status groups. The current income tax exemptions for health benefits would be eliminated, which means that individuals would pay income tax on employer contributions for health benefits. The flat-dollar tax credit has been proposed as a more equitable means of providing tax subsidies for the purchase of health insurance.
Under the flat-dollar tax credit proposal, the current tax exemption for the cost of employer-sponsored health benefits is eliminated. This includes the exemption for the employer share of health benefits, the exemption for employee contribution for benefits in Section 125 plans, and the deduction for self-employed persons. Thus, workers would pay income tax on the income that is currently tax-exempt. However, the tax exemption would be retained for retirees who are excluded from the credit as a result of their participation in Medicare or Medicaid. For illustrative purposes, we assume that this income would not be subject to Medicare FICA taxes. In addition, we assume that the amount of any tax credit received is to be subtracted from premium payments when computing the deduction for health spending in excess of 7.5 percent of AGI.
For illustrative purposes, we simulated the impact of adopting a flat-dollar tax credit for the United States. We assume that the tax credit would be $800 per adult, plus $400 per child up to a maximum of $2,400 per family. The amount of the credit is limited not to exceed the amount of the premium payment. The credit would be available to all except those covered under Medicare or Medicaid.
Some tax filers would find that the tax credit amount is less than the tax that they would pay on the value of employer-provided health benefits. This would increase the after-tax cost of insurance for these individuals, which could result in a reduction in coverage among this group. Consequently, some of the flat-dollar tax credit proposals now under consideration would permit the taxpayer to choose between the tax credit and the current exemption, depending upon which option is most beneficial to the individual tax filer. To show the effects of this provision, we have also analyzed a scenario in which individuals are permitted to maximize their own personal tax benefits by choosing between the credit and the exemption. This is similar to the tax credit proposal developed by the NAHU.
- Coverage and Cost Impacts of Flat-Dollar Tax Credit
- Credit/Exemption Alternative
- Implementation Issues
- Impact on Health Expenditures
The flat-dollar tax credit has the potential to reduce the inflationary impact that tax subsidies have on health care costs. As discussed above, the current tax exclusion effectively reduces the price of health coverage to the individual, which induces persons to purchase more costly coverage that results in increased health services utilization and higher health care spending. Replacing the exemption with a flat-dollar tax credit effectively limits the tax subsidy available to an individual to the amount of the credit. This increases the incentive to enroll in more efficient health plans with somewhat less comprehensive benefits. This could have the effect of slowing health care inflation, at least for those who now have the highest-cost health plans. Of course, the cost-containment potential of the flat-dollar credit, is negated if individuals are permitted to choose between the exemption and the credit as it would be under several tax credit proposals.
Conversely, the flat-dollar tax credit would also create an incentive for uninsured persons to obtain insurance, resulting in an increase in utilization for this population. This increase in utilization is expected for newly insured persons as they make more appropriate use of the health care system, particularly for those who are currently forgoing needed care because of a lack of coverage. However, the flat-dollar tax credit would still encourage persons to join more efficient health plans because of the fact that it would limit the tax subsidy to the amount of the credit. This would encourage economic discipline in the use of health services among newly insured persons just as it would for persons who are already covered.
The tax credit proposals discussed above focus on a more equitable means of distributing tax subsidies across families by extending tax subsidies to persons purchasing non-group health insurance, and by providing refundable tax credits so that low-income persons can receive a substantial tax subsidy even if they have little or no tax liability. While some of these proposals would encourage some individuals to enroll in more cost-efficient health plans, none of these proposals are designed to relieve the increase in health services utilization that is believed to be attributed to the favorable tax treatment of health benefits under the current system. Arguably, the various tax credit proposals will contribute to higher health care costs by encouraging persons who are newly eligible for tax subsidies to purchase more comprehensive health coverage, at least up to the amount of the credit.
The Heritage Foundation has developed a tax credit proposal that is designed to encourage individuals to conserve their use of health services. The Heritage plan would require all employers to cash-out their benefits and would require individuals to make their own choice of insurance coverage from the individual market. The proposal also requires all individuals to have insurance. The existing health benefits exclusion would be replaced with a refundable health insurance tax credit that would be available to all persons except those already covered by Medicare and Medicaid. The federal cost of the tax credit would be financed with the increases in income and Social Security tax payments on the wages paid to workers under the cash-out. The tax credit would assist persons in all income and employment status groups.
This cash incentive would set in motion a new dynamic in employer-based coverage that could help reduce the inflationary pressures that tax subsidies have had on health spending. Workers who receive a health benefits cash-out automatically would have an incentive to select a lower-cost health plan so that they could retain as much of the cash-out as possible for uses other than health care. This would encourage enrollment in lower-cost health plans using managed care and/or high-deductible policies, which provide incentives to minimize the use of unnecessary and ineffective treatments.
Another proposal now being considered would permit employers to cash-out their health benefits without losing the exemption. As discussed above, this proposal would provide a tax credit for insurance purchases by persons who do not have employer coverage. The proposal would also give employers the option to cash-out their health benefits by discontinuing coverage and giving employees a cash benefit equal to the value of the health benefit without loss of the exclusion for employer-provided benefits (i.e., the cash-out amount would be tax – exempt). Individuals could then select the coverage that they prefer in the individual market. While the purpose of this provision is to provide greater choice of health coverage alternatives (about 44 percent of workers do not have a choice of health plans), it would create incentives among affected workers to enter more efficient plans to retain as much of the cash-out as possible, resulting in lower health care costs. However, the proposal does not actually require employers to cash-out their benefits, which substantially diminishes potential health care savings under the proposal.
In this section, we estimate the financial impact of the Heritage health care reform proposal on federal costs, health spending, coverage, and families in various income groups. We begin with a description of the key features of the Heritage Foundation plan.
- Program Design
- Cost and Coverage Impacts
- Distributional Impacts
- Impact on Total Health Spending
- Adequacy of Insurance Market Reforms
One of the greatest concerns with proposals that replace employer-sponsored coverage with an expanded individual insurance market is adverse selection and its associated consequences. Under a typical employer group plan, workers of assorted health status take the employer-sponsored coverage. This has the effect of pooling higher-cost people with lower-cost people. However, under an individual insurance model, individuals can join a wide range of alternative health plans, which substantially eliminates the pooling of risk that occurs in employer groups. Insurers that can attract a disproportionate share of healthier individuals will have high profits and may be able to lower premiums for the standard plan to attract even more lower-cost persons. Conversely, other plans with more comprehensive benefits will accumulate higher-cost persons, resulting in losses and increases in premiums for the individuals they cover.
These selection effects can substantially undermine insurance markets by encouraging insurers to adopt marketing practices that attract only healthy individuals rather than developing care management technologies designed to provide high quality care at the lowest cost. In fact, many insurers may decide to terminate their comprehensive health plans and offer only the low-cost standard plan to discourage high-cost individuals from enrolling with them at all. This would drive up premiums in the market for the more comprehensive plans that less healthy individuals prefer and could lead to a general reduction in access to coverage for this population. Such a problem could be addressed by requiring all insurers to offer two or more standard plans, including the minimum standard benefits plan and a more comprehensive coverage alternative. However, even with such a requirement, insurers would still be able to adopt marketing practices that would result in selection problems.
Under the Heritage Foundation plan, the potential for selection is greatly reduced by requiring all individuals to have insurance. This means that healthy individuals who expect few or no health expenditures will no longer be able to opt out of the broader insurance pool by declining to obtain coverage. Covering these individuals would increase the numbers of healthier persons purchasing insurance, thus permitting a broader pooling of risk for individual insurers. However, the degree of selection and its impact on access and competition would need to be carefully monitored and met with additional market reforms if necessary.
Our analysis raises several important equity issues concerning the current distribution of tax benefits. With over 43 million uninsured persons in the United States, most of whom are in relatively low-income groups, it is important to ask whether it is appropriate that 68.7 percent of federal health benefits tax expenditures are going to the 36 percent of the population with the highest incomes (i.e., $50,000 or more). Moreover, these tax expenditures should be reevaluated in terms of their tendency to encourage overuse of the health care system.
These concerns have led to proposals that would allocate health insurance tax benefits more equitably across families. For example, the various tax credit proposals would replace the current tax preferences with a refundable tax credit to families, which more equitably distributes tax benefits across all families that purchase insurance, including those who purchase coverage in the individual market. In fact, these tax credits can be designed in a way that targets the assistance to lower-income families by computing the credit based on health expenditures as a percentage of family income. These tax credits could be limited, however, so that all families have an incentive to enroll in efficient health plans and conserve their use of health services.
The tax credit model provides an alternative to an expansion of Medicaid as a means of reducing the number of uninsured Americans. Under the tax credit, individuals select their health policies from among the variety of private coverage options available in the market. This permits individuals to use the credit to purchase the coverage that best meets their needs. This differs from Medicaid, where the state and federal governments specify a uniform benefits package and managed care enrollment is often mandatory.
However, compared with Medicaid, the tax credit provides fewer assurances of the adequacy of coverage for eligible persons. For example, private health plans will often impose pre-existing condition limitations and point-of-service cost sharing, while Medicaid does not (except for nominal co-payment amounts for some services). These cost-sharing requirements can become barriers to access for lower-income persons and are likely to discourage many potentially eligible persons from enrolling. Moreover, unlike the Medicaid program, the tax credit program provides no mechanism for monitoring and influencing the quality of care provided to beneficiaries.
Despite these limitations, tax credits may be a more appropriate means of expanding coverage among lower- and -middle-income people. Arguably, persons who have incomes above the Medicaid income eligibility level do not need a plan that is as comprehensive as that provided to the lowest-income groups now covered by Medicaid. These individuals often will be able to afford making at least some contribution to the cost of their care. Thus, a program to partially subsidize the purchase of private insurance coverage could be an appropriate “next tier” of subsidized health coverage for lower- and- middle-income families. Moreover, the availability of the tax credit could enable those who have declined the coverage now available to them through employment to effectively combine the employerís contribution with the tax credit amount to acquire coverage under their employerís plan.
The success of the tax credit model will depend on its ability to provide premium subsidies to individuals at the time they are needed to purchase coverage. As discussed above, this is readily accomplished under tax credit models where the credit amount is uniform for each enrollee (i.e., flat-dollar amount or flat percentage of premium) by arranging for an electronic transfer of subsidy dollars to insurers based on a verified roster of individuals whom they cover.
However, tax credit administration becomes substantially more complex if the credit is targeted to lower-income persons. This is because a mechanism would be needed to determine an individualís eligibility for the subsidy at the time he or she purchase coverage rather than expecting the individual to “front” the full cost of insurance until tax returns are filed in the following spring. This would involve establishing an income eligibility determination process that may operate much like that used under Medicaid. In fact, building on the Medicaid eligibility determination process may be the most efficient way to administer advance payments of targeted income tax credits.
Our analysis shows that refundable tax credits would tend to shift federal tax subsidies from higher-income groups to lower-income groups. This would enable many uninsured individuals to purchase coverage, resulting in a decline in the number of uninsured. However, under most of the tax credit proposals now under consideration, there still would be 30 million or more persons without coverage because many individuals would find that the cost of insurance is greater than they feel that they can afford, even with the credit.
The emerging debate over tax credits provides the opportunity to rethink the role of tax policy in insurance coverage. Historically, the tax-exempt status of health benefits has played an important role in encouraging employers to provide health benefits. However, it is important to critically evaluate the equity of the existing system and its role in encouraging employers to offer comprehensive coverage that in turn encourages increased use of health services. Well-designed changes in tax policy could lead to an expansion in coverage together with improved equity and greater emphasis on efficient health care delivery.
Estimating the Impact of Tax Credits on Insurance Coverage
In this analysis, we estimated the impact of health insurance tax credits and other tax subsidies on the number of persons with insurance coverage. The principle is that these various tax subsidies effectively reduce the net cost of health insurance to the individual, which increases the proportion of persons purchasing coverage. Therefore, our analysis focused on measuring the change in coverage resulting from a given change in the net after-tax price of insurance. Our approach was to use The Lewin Group Health Benefits Simulation Model (HBSM) to estimate the prices faced in the market for uninsured persons, estimate the change in prices resulting from these tax subsidies, and estimate the number of uninsured persons who would take coverage based on estimates of how a change in the price of insurance affects the likelihood that an individual will take coverage
The key assumption in our analysis is the assumed price elasticity for demand for insurance. Price elasticity is defined as the percentage change in persons purchasing coverage given a 1.0 percent change in price. The elasticity estimate that we used in this analysis is based on an analysis of the impact of changes in the employee contribution amount in employer plans on the number of workers and dependents taking coverage conducted by The Lewin Group, Inc., in 1998. This study indicated a price elasticity of 0.2 percent, which means that on average a 1.0 percent real increase (i.e., increase after standardizing for price inflation) in premium (i.e., price) was associated with a 0.2 percent reduction in coverage. Weighted to national coverage numbers, this estimate indicates that a 1.0 percent increase in premiums results in a loss of coverage for about 300,000 persons.
In this appendix, we describe the data and methods used to develop this price elasticity estimate. We also explain how these data were used to develop estimates of coverage effects under the various tax subsidy proposals.
1. Data and Methods
Our analysis is based on the March Current Population Survey (CPS) data for 1989 through 1996. The CPS is a survey of households conducted by the Bureau of the Census. It includes information on employment, earnings, and sources of health insurance coverage. We pooled the CPS data for each year between 1989 and 1996 to create a pooled time-series, cross-sectional database. These data provide much of the information required to measure the impact of changes in demographic and economic factors on the level of employer coverage over time. For example, these data provide the information required to analyze how employer coverage has changed as a result if changes in earnings, industry of employment, and other employment and demographic characteristics of workers.
While the CPS data provide much of the information required to measure factors affecting coverage, they do not provide information on the price of insurance. To correct for this, we imputed the amount of the employee share of premium payments to workers in the CPS who indicated that they have employer coverage on their jobs. We did this based on the average employee share of premiums for single and family coverage reported in the National Medical Expenditures Survey (NMES) for workers with employer coverage. These data were adjusted over time based on the average rates of growth in employee spending as reported in the KPMG Peat Marwick employer surveys for 1991 through 1996 and the Health Insurance Association of America (HIAA) survey data for employers from 1988 through 1990., In addition, we adjusted the share of the premium paid by the worker based on the average percentage of premiums for employer coverage paid by the employee as reported in these employer surveys for 1988 through 1996.
The average premium for employer-sponsored health benefits has been increasing more rapidly for family coverage than for single coverage. Between 1988 and 1996, average premiums for family coverage increased by 111 percent, from $2,530 in 1988 to $5,349 by 1996. Premiums for single coverage increased by only 79 percent over that period, from $1,153 in 1988 to $2,059 in 1996 (Table A – 1). This may help explain much of the rapid decline in employer-sponsored coverage for children in recent years.
However, the overall average percentage of premiums paid by employees has increased more rapidly for single coverage than for family coverage. The reason for this is that while most firms have long required at least some contribution toward family coverage, many firms did not require a contribution for employee-only coverage until recently. For example, the Bureau of Labor Statistics (BLS) reports that the percentage of workers required to contribute to the cost of single coverage increased from 28 percent in 1980 to 63 percent by 1993. By comparison, the percentage of workers required to contribute to family coverage increased from 49 percent in 1980 to 79 percent in 1993. Thus, the overall average percentage of the premium paid by the worker increased more rapidly for employee-only coverage than for family coverage over the 1988 through 1996 period.
Over the 1988 through 1996 period, average employee contributions for health benefits increased by 283.9 percent for employee-only coverage and 145.6 percent for family coverage. Adjusting for inflation, the real increase in average employee premium contributions over the 1988 through 1996 period was 189.4 percent (14.2 percent annually) for employee-only coverage and 85.1 percent (8.0 percent annually) for family coverage. This reflects both increases in premiums and increases in the share of the premium paid by the worker. The premium contribution amounts that we imputed to the CPS data for the 1989 through 1996 period reflect these estimates of the differential growth in premium contributions for employee-only and family coverage.
Table A – 1
Growth in Employee Premium Share for Employer Coverage 1998 Through 1996 a/
|Employee-Only Coverage||Family Coverage|
|Average Premium||Percent Paid by Worker b/||Average Contribution c/||Real Growth in Employee Share d/||Average Premium||Percent Paid by Worker||Average Contribution c/||Real Growth in Employee Share d/|
|1988||$1,153||10.2%||$118||– –||$2,530||26.0%||$658||– –|
|1991||$1,738||13.0% e/||$226||-5.3%||$4,307||23.0% e/||$991||-5.3%|
|Average Annual Growth 1988 – 1996||7.5%||10.1%||18.3%||14.2%||9.8%||1.9%||11.9%||8.0%|
|Total Percent Growth 1988 – 1996||78.6%||115.7%||283.9%||189.4%||111.4%||16.2%||145.6%||85.1%|
a/ KPMG Peat Marwick, 1991-1996 and HIAA data for 1988-1990.
b/ This is the overall average percentage of the premium paid by the worker, including both covered workers who contribute to the cost of coverage and those who are not required to make an employee contribution.
c/ Estimate reflects the combined effect of premium price increases and increases in the percentage of the premium paid by the worker.
d/ Includes adjustment for inflation.
e/ There are differences in the survey methods used in the HIAA and the KPMG survey designs that make these data less than strictly comparable. This may be the reason for the abrupt drop in the percent of premium paid by workers between 1990 and 1991.
Source: Lewin Group estimates.
Using the CPS data for 1989 through 1996, we estimated three separate multivariate models of employer-sponsored health insurance coverage for workers, dependent spouses, and dependent children. The first multivariate model estimates the probability that a worker is covered by an employer plan. The explanatory variables include demographic characteristics that are correlated with coverage such as age, race, ethnicity, marital status, and whether the individual is the family head. The model also includes employment-related variables such as industry and occupation of the worker, the size of the employing firm, the full-time/part-time status of the worker, and worker earnings. We also included a variable indicating whether individuals are covered under Medicaid to measure the impact of expanded coverage under Medicaid on employer coverage levels. In addition, we included the imputed amount of the employee share of the premium, which over time reflects changes in both premium amounts and the percentage of the premium paid by the worker.
The second multivariate model estimates the likelihood that spouses of covered workers will have coverage as a dependent spouse. The explanatory variables used in the model include age, race, ethnicity, family income, and an estimate of the incremental cost of electing the family coverage option. The incremental cost of coverage was calculated by taking the difference between the average family premium and the average employee-only coverage premium for a given firm size/industry group. The third model, which is similar to the model of spousal coverage, estimates the likelihood that children of parents who have employer coverage will be covered as dependents.
These multivariate models were estimated using a logit estimation methodology, which is ideally suited to estimate models where the dependent variable is bounded between zero and one. These models provide a basis for measuring the impact of the price of insurance and various economic and demographic factors on the level of coverage for workers and dependents over the 1989 through 1996 period, given the level of employment in these years. They also provide a basis for projecting coverage levels in future years under alternative assumptions concerning premium growth, employee contribution shares, and other economic factors in future years.
- Multivariate Analyses
As discussed above, we developed multivariate models that show how the proportion of persons with employer coverage changes as demographic and economic factors change over time. We did this by estimating logistic functions of the form where p is the proportion of persons with employer coverage, and z represents the sum of the products of the estimated coefficients and the corresponding values of the explanatory variables (i.e., earnings, age, etc.). This approach has the unique feature of bounding the modelís estimates of the proportion of persons with employer coverage to between 0.0 and 1.0. In general, the explanatory variables that we included in these employer coverage models were statistically significant at the 99.5 percent confidence level.
The estimated coefficients for the logit model are difficult to interpret because the logit function is essentially non-linear. However, the direction of effects can be interpreted based on the sign (positive or negative) of the estimated coefficients. For example, the workers equation generally indicates that Blacks, Hispanics and Asians are less likely to have coverage than is the average population (Figure A – 1). These estimates also show that coverage levels go down as the employee contribution amounts increase and that coverage increases as income rises. In general, the direction of the effects estimated for the various explanatory variables is as expected. However, it is difficult to discern the magnitude of these effects from the coefficients.
To measure the magnitude of the effects of these variables, we solve the estimated equations under selected variations in the explanatory variables. Solving the equation simply means computing the proportion of persons with coverage by use of the estimated coefficients and various assumptions on the mean values of the explanatory variables. For example, we can obtain the average coverage levels in 1996 by solving these equations for that year using the actual means for the explanatory variables in that year. We can then test the sensitivity of estimated coverage levels to changes in the employee premium contribution amount by varying the assumed premium level from the 1996 value and calculating the change in the estimated coverage level. Similarly, the sensitivity of coverage to changes in other explanatory variables can be estimated using this method. In this analysis, we use this approach to measure the magnitude of the effect that changes in the various explanatory variables have had on coverage since 1989.
b. Coverage Levels
In 1996, there were 144.7 million workers and dependents covered by employer health plans (Table A – 2). This includes 72.6 million workers aged 19 through 64, 27.6 million persons covered as a dependent spouse, and 44.5 million children of these workers who are covered as dependents. This compares 137.0 million workers and dependents covered by employer plans in 1989. However, the percentage of workers and dependents who had coverage actually declined over the 1989 through 1996 period. The percentage of workers with coverage on their own job declined from 60.6 percent in 1989 to 60.0 percent by 1996, with the percentage of workers with coverage reaching a low of 58.4 percent in 1993.
Figure A – 1
Estimated Parameters for Logistic Health Insurance Coverage Equations a/
|Worker Coverage Equation b/||Dependent Spouse Equation g/||Dependent Children Equation g/|
|Variable||Parameter Estimate||Variable||Parameter Estimate||Variable||Parameter Estimate|
|Married Family Head||0.0324||*||Asian||-0.3743||*||Asian||-0.3106||*|
|Spouse of Family Head||-0.4789||*||Family Income h/||1.4446||*||Family Income h/||1.4545||*|
|Age||5.1666||*||Premium Amount i/||-0.4978||*||Premium Amount i/||-0.3005||*|
|Age Squared||-4.3975||*||Covered by Medicaid||-1.2692||*||Covered by Medicaid||-1.5798||*|
|Full-Time Worker||1.2653||*||Time Squared||0.1067||*||Time Squared||0.3098||*|
|Premium Amount d/||-0.7579||*|
|Covered by Medicaid||-1.1174||*||R-Squared|
|Covered by Medicare||-0.5279||*||Worker Equation||0.4161|
|High-Coverage Occupation e/||0.4636||*||Dependent Spouse Equation||0.0713|
|High-Coverage Industry f/||0.3112||*||Dependent Children Equation||0.1146|
|Firms with Fewer than 25 Workers||-1.7485||*|
|25 – 99 Workers||-0.6366||*|
|100 or More Workers||-0.2090||*|
* Significant at the 99.5 percent level.
a/ Note that the logit is a non-linear estimation technique. The parameters are not directly interpreted as derivatives. If p is proportion covered, the derivative with respect to a continuous variable xi is bi(p)(1-p), where bi is the parameter, and the elasticity is bi(1-p)xi.
b/ The equation estimates the likelihood that an employed person has employer-based coverage.
c/ Includes annual earnings reported by the worker in 1996 dollars.
d/ The employee share of premiums for covered workers was imputed to the CPS based upon individual’s reported industry, firm size and state of residence. Amounts in 1996 dollars. Reflects both price increases and increases in the percentage of the premium paid by the employee.
e/ Identifies workers employed in a high-coverage occupation. A high-coverage occupation is defined as one where the average percentage of workers with coverage is greater than the overall average percentage of workers with coverage.
f/ Identifies workers employed in a high-coverage industry. A high-coverage industry is defined as one where the average percentage of workers with coverage is greater than the overall average percentage of workers with coverage.
g/ The universe of persons included in the analysis is dependents of workers who do not have coverage on their own jobs.
h/ Total income of all family members in 1996 dollars.
i/ Includes the incremental cost to the employee of electing the family coverage option in 1996 dollars. Reflects both premium price increases and increases in the share of the premiums paid by the employee.
Source: Lewin Group estimates using a pooled cross-section of individuals from the March Current Population Surveys for 1988 through 1996.
Table A – 2
Number and Percent of Workers and Dependents with Employer Coverage
|Workers||Dependent Spouses||Dependent Children||All Workers and Dependents|
|Number a/||Percent Covered||Number Covered||Number b/||Percent Covered||Number Covered||Number c/||Percent Covered||Number Covered||Number d/||Percent Covered||Number Covered|
a/ Includes persons aged 19 to 64 who worked sometime in the year.
b/ Includes all non-working spouses of workers and working spouses who did not take coverage on their own job. These working spouses were also counted above as workers.
c/ Includes all children with a working parent.
d/ Includes all workers age 19 to 64, non-working spouses, and children of workers. To avoid double counting, working spouses are counted as workers.
e/ An unknown portion of the increase in covered children in 1995 and 1996 is attributed to changes in the survey questionnaire introduced in these years.
Source: Lewin Group estimates using the CPS data for 1989 through 1996.
The percentage of dependent spouses and children with employer-sponsored coverage also declined over this period. The percentage of dependent spouses with employer-sponsored coverage declined from 89.0 percent in 1989 to 88.3 percent by 1996 (Table A – 2). Among dependent children of workers, the percentage with employer-sponsored coverage declined from 76.7 percent in 1989 to 71.6 percent by 1994. The CPS data report an increase in the percentage of dependent children with employer-sponsored coverage to 75.0 percent by 1995 and 1996, but this is still less than the coverage rate of 76.7 percent reported in 1989. While there were improvements in the economy in 1995 and 1996 that may have resulted in increased coverage, much of the increase is due simply to revisions in the CPS insurance questionnaire starting in 1995 that are specifically designed to identify additional children who have employer coverage. Thus, we do not know how much of the increase in childrenís coverage in 1995 is attributed to a genuine improvement in childrenís coverage and how much is due to the change in the questionnaire for childrenís coverage.
Overall, the percentage of workers and their dependents who have employer-sponsored coverage declined from 76.1 percent in 1989 to 71.8 percent in 1994, and then leveled off at about 75 percent in 1995 and 1996. This is an overall measure of coverage that includes workers, spouses, and dependent children. However, as discussed above, at least a portion of the abrupt increase in the percent of workers and dependents with coverage in 1995 is attributable to the use of a different insurance questionnaire starting in that year. Thus, these data may understate the actual loss of coverage between 1989 and 1996.
These data also indicate that many covered workers with families have not elected the family coverage option. For example, in 1996, there were 3.7 million children who did not have employer-sponsored coverage even though one or both of their parents had employer-sponsored coverage on their own job. Similarly, there were 2.1 million spouses of covered workers who were not covered as a dependent on an employer plan in 1996 even though their husband or wife had coverage on his or her own job (excludes spouses with coverage on own job). Overall, about 91.2 percent of all children with a parent who has employer coverage are covered as dependents. Since the available employer survey data indicated that there are virtually no employer plans where the family coverage option was not available, these appear to be cases where the workers have accepted coverage for themselves but have not accepted the family coverage option even though they have a spouse and/or children.
Figure A – 2 presents the sources of insurance coverage for the 3.7 million children who do not have employer coverage even though one or both of their parents is covered by an employer plan. About 1.9 million (51.5 percent) of these children are uninsured. These children may be in families where the parents found that they could not afford the monthly employee contribution amount required to obtain coverage. Also, these data indicate that 1.2 million (31.6 percent) of these children were covered by Medicaid.
Source: Lewin Group estimates using the 1996 Current Population Survey (CPS) data.
Of the 2.1 million spouses who do not have coverage as a dependent spouse on their husbandís or wifeís employer-sponsored health plan (excluding spouses covered on own job), about 1.3 million (63.8 percent) were uninsured and 164,000 (7.9 percent) were covered under Medicaid (Figure A – 2). Those spouses who are left uninsured include instances where covered workers do not take the family option to cover their spouses because they cannot afford the additional premium contribution for this coverage. Some of the Medicaid coverage also may be attributed to “crowd out” in cases where these spouses became eligible for Medicaid coverage during a pregnancy under recent expansions in Medicaid eligibility for pregnant women. About 392,000 (18.9 percent) had coverage under an individually purchased private insurance plan and 197,000 (9.4 percent) were covered by Medicare or some other form of public coverage.
c. Reasons for the Decline in Coverage
As discussed above, the percentage of workers and their dependents with employer-sponsored coverage declined from 76.1 percent in 1989 to 75.1 percent by 1996. These figures are difficult to compare, however, because they mask changes in the demographic composition of workers and their dependents over time that have significant impacts on coverage. For example, the average age of workers and their spouses has increased since 1989, which is important since coverage levels generally increase with age. Also, the racial, ethnic, and marital status composition of workers and their dependents also has changed over this period. Thus, to compute the loss of coverage over the 1989 through 1996 period, we need to adjust the reported coverage level for 1989 to reflect these changes in demographic characteristics between 1989 and 1996 to provide a consistently defined basis for comparison with 1996 coverage levels.
We did this by using the multivariate model to estimate what the percentage of workers and dependents with coverage would have been in 1989 if the demographic composition of the population in that year had been the same as it was in 1996. The resulting coverage rate (79.3 percent, see Figure A – 3) is comparable to the actual coverage level (75.1 percent) in 1996 because both figures reflect the demographic profile of the 1996 population. The number of workers and dependents losing coverage over this period is computed by calculating the difference between the expected coverage level for 1996 (79.3 percent) and the actual coverage rate (75.1 percent) and multiplying the difference (4.2 percent) by the total number of workers and dependents in 1996. Using this approach, we estimate that about 8.0 million workers and dependents lost coverage between 1989 and 1996. This figure probably understates the loss of coverage over this period because the coverage levels reported in the CPS data for 1995 and 1996 are not fully comparable to the data for earlier years because of a change in the CPS questionnaire beginning in 1995.
With standardization for demographic differences between 1989 and 1996, all of the coverage loss (8.0 million persons) is attributed to changes in key economic factors over this period such as real income growth/decline and premium cost growth. We can measure the impact of these factors on coverage by gradually relaxing the multivariate models to reflect the actual change in these economic variables over the 1989 through 1996 period. For example, over this period, there was a gradual shift of workers out of industries where most workers have coverage, such as manufacturing and mining, into lower-coverage industries such as services. Also, there was a shift in employment toward lower-coverage occupations and a small increase in the percentage of workers employed in small firms, where health insurance coverage rates are lowest. In addition, there has been a gradual shift from full-time employment to part-time employment, which could lower coverage rates since part-time workers are often ineligible for the health benefits offered by the employer. When we account for these changes over the 1989 through 1996 period, the expected coverage level for 1996 drops from 79.3 percent to 78.9 percent.
Average weekly earnings declined between 1989 and 1992 and have remained flat since that time. We accounted for these changes in income by re-estimating the expected level of coverage where the 1996 average earnings and the income amounts for 1989 are replaced in the multivariate model with the actual levels of real earnings and income over this period. This lowers the expected coverage level for 1996 to 78.5 percent.
The growth in the employee premium contribution amount over the 1989 through 1996 period accounts for most of our estimated coverage loss in 1996. The employee premium contribution amount reflects both increases in the premium over time and changes in the percentage of the premium paid by the worker. As discussed above, employee contribution amounts have increased at rates up to 15.0 percent per year between 1989 and 1996. When we replace the 1989
a/ Includes only workers age 19 to 64 and their dependents. The abrupt increase in the percentage of workers and dependents covered by employer-sponsored insurance in 1995 and 1996 is at least partially due to a change in the survey methodology that was implemented in these years and does not necessarily represent an actual increase in coverage in these years.
b/ Includes changes in industry, occupation, firm size and full-time/part-time status.
c/ Reflects reductions in real income.
d/ Includes impact of real growth in premiums and the percentage paid by the worker.
e/ Reflects increases in Medicaid eligibility over the 1988 through 1996 period.
Source: Lewin Group estimates using the CPS data for 1988 through 1996.
employee contribution amounts in the multivariate model with actual employee contribution levels in these years, the expected coverage level drops to 75.3 percent. The overall coverage level further declined to 75.1 percent, which is the actual coverage rate in 1996, when we replaced the mean percentage covered by Medicaid in 1989 with the actual percentage covered in the 1990 through 1996 period. Most of the effect of expanded Medicaid coverage concentrated among children, where the greatest expansions in Medicaid eligibility have occurred.
Of the 8.0 million persons who we estimate lost coverage by 1996, about 48.7 percent were workers, 31.7 percent were dependent children, and 19.6 percent were dependent spouses. Overall, about 76.4 percent of this coverage loss is attributed to the growth in the employee share of the premium over time. About 10.7 percent is attributed to employment shifts by industry, occupation, firm size, and full-time/part-time status. The decline in real earnings accounts for about 7.8 percent of the coverage lost, while the growth in Medicaid enrollment accounts for only about 5.1 percent of this coverage loss.
Our analysis is consistent with other studies showing that 10 to 17 percent of coverage loss has been attributable to industry shifts. However, we attribute less of the coverage loss to declining real wages than do other studies. For example, while we attributed 10.7 percent of the coverage loss to industry shifts and 7.8 percent to real wage declines, Fronstin and Snyder attributed about 10 percent of coverage loss to industry shifts, 7 percent to part-time work, and 23 percent to changes in real wages. Acs attributed 17 percent of coverage loss to industry shifts and 38 percent to changes in family income. Long and Rodgers also attributed less than 15 percent of coverage loss to industry shifts. However, we know of no other estimates of the impact that the employee contribution amount has on coverage with which we can compare our estimates. Some of the difference between our estimates and those of other researchers may be explained by the fact that we are studying a more recent time period than was considered in the other studies. Also, it is possible that by including the employee contribution amount in our multivariate models, our price variable is explaining a portion of the variation that might otherwise have been attributed to other variables.
- Price and Income Elasticities
These multivariate models provide a basis for estimating the price and income elasticities for employer coverage (i.e., the sensitivity of employer coverage to changes in price and income). This is done by solving the models described above by use if the mean value for each explanatory variable while varying price by 1.0 percent. Our analysis indicates a price elasticity of -0.203. That is, each 1.0 percent increase in the employee premium share is associated with a 0.203 percent reduction in coverage for workers. The model also shows an income elasticity of 0.367. This means that each 1.0 percent increase in average income will result in a 0.367 percent increase in the number of workers and dependents with coverage.
In this analysis, we estimated the change in the number of persons with health insurance resulting from alternative tax subsidy schemes using the Lewin Group Health Benefits Simulation Model (HBSM). HBSM is a microsimulation model of the US health care system, which is based on a representative sample of the US population that provides information on the income, demographic, and employment characteristics for each family member. These data also provide information on health coverage, service utilization, health expenditures, and premium payments. The model uses these data to identify persons who are potentially eligible for various subsidy programs, estimates the number of persons taking the subsidy, and estimates the impact on government costs and health expenditures.
The model uses the most recent population data available from the Bureau of the Census and health expenditure data from the Health Care Financing Administration (HCFA). The basic database used in the model is the 1987 National Medical Expenditures Survey (NMES), which is the most recent data available that provides information on coverage, expenditures, income, and employment characteristics. These data are adjusted to replicate the most recent data available on the number of persons by income, employment status, industry, age, sex and coverage status as reported in the most recent Bureau of the Census Current Population Survey (CPS) data (1998). This is done in a large, multistage, iterative proportional fitting process applied to the sample weights in which we simultaneously control the data to approximate 450 classifications of population data from the CPS. These data are statistically matched to the Internal Revenue Service (IRS) Statistics of Income (SOI) data to provide information on tax deductions and marginal tax rates.
The health expenditure data are also adjusted to replicate health expenditure amounts reported in the National Health Accounts data. Medicaid eligibility simulations were used to reflect expansions in childrenís enrollment under the program. Recent coverage data from the Medical Expenditures Panel Survey (MEPS) data also have been used as controls to model estimates of the number of persons with access to employer-based insurance who have declined coverage.
In this analysis, we used the data in the HBSM to identify persons who are potentially eligible for the various tax subsidies available under each proposal. We then computed the price of insurance that uninsured individuals would face in the individual market based on HBSM estimates of the cost of a given benefits package by age of policy holder and family composition. We then computed the after-tax cost of that insurance under current policy and for the tax subsidies provided under each proposal.
In this analysis, we used the HBSM model and data to identify individuals with employer-provided health coverage and the employer share of the premium, which is tax-exempt. We then estimated the proportion of this population that is covered under Section 125 cafeteria plans where the employee share of the premium is also tax-exempt. We then used the tax data to calculate the tax subsidy resulting from these exemptions based on their marginal tax rates and the Social Security FICA tax rate. This provided estimates of the tax subsidies provided for health benefits under the current system.
We also used the HBSM to identify persons who are potentially eligible for the tax subsidies that would be available under each of the proposals that we analyzed. This permitted us to estimate the cost of adopting these policies and to estimate the revenues collected by limiting or eliminating existing tax subsidies. The model provides these estimates by income group so that we are able to estimate the change in the distribution of tax subsidies under these proposals. In addition, the model estimates the change in tax subsidies for individual families, which enables us to estimate the distribution of families by the amount of the increase or decrease in tax subsidies under the proposal (i.e., “winners and losers”).
A key element of this analysis was estimating the number of persons who would take coverage under the various tax credit proposals. We developed these estimates by identifying uninsured persons in the HBSM model who would qualify for these tax credits. We then estimated the cost of a typical health insurance plan for these individuals in the individual market by age and sex based upon the health spending data recorded in the HBSM. This enabled us to calculate the change in the after-tax cost of this insurance (i.e., reflecting the credit) to estimate the percent change in the cost of insurance to the individual as a result of the credit.
Using these data, we estimated the number of persons eligible for the new tax subsidies who take coverage based on a Lewin Group study of the impact of changes in premiums on the proportion of person who take coverage. The study showed that a 1.0 percent decrease in premiums on average results in a 0.2 percent increase in the number of persons who take coverage (i.e., an elasticity of -0.2). We then estimated the number of persons who are induced to purchase coverage as a result of these credits based on this estimated relationship between premium level (i.e., price) and the number of persons with coverage.
The price elasticity for insurance coverage varies by income level. Figure A – 4 shows the percentage increase in coverage resulting from a percent reduction in premiums as estimated using the equations estimated above. These data indicate that lower income individuals are more sensitive to price changes than are higher-income individuals. For example, holding all other factors constant, a 10 percent reduction in premiums results in a 0.34 percent increase in coverage among persons with income of $10,000. However, the percentage increase in coverage resulting from a 1.0 percent declines with income to only about 0.04 percent among persons with incomes of $100,000. This reflects the fact that higher income individuals can better absorb a price increases in insurance than can individuals with very limited incomes.
Figure A – 4
Percentage Increase in Coverage Resulting from a One-Percent Reduction in Premiums by Income Level (in percentages)
Source: Lewin Group estimates.
The price elasticity for insurance also varies with age and other demographic factors. For example, Figure A – 5 presents the estimated percentage increase in coverage resulting from a one-percent reduction in premiums. These data indicate that the percentage increase in coverage resulting from a one-percent reduction in premiums is about 0.27 percent for persons age 20, declining to 0.18 percent for persons age 60. The price response declines with age because older individuals are typically in poorer health than younger individuals and are more likely to maintain their coverage irrespective of price. Other differences in the price elasticity of coverage are evident by sex, race/ethnicity, and various employment status indicators.
Figure A – 5
Percentage Increase in Coverage Resulting from a One-Percent Reduction in Premiums by Age ((in percentages)
Source: Lewin Group estimates.
More from 1999PolicyStudies
The National Coalition on Health Care (NCHC)Final ReportOctober 18, 1999 John Sheils,Paul Hogan,andRandall Haught,The Lewin Group, Inc. TABLE OF CONTENTS EXECUTIVE…Read more
Final ReportPrepared for:The National Coalition on Health Care January 26, 1999 By:John F. SheilsMolly J. Coye, MDGenny B. JacksRichard E. Dixon,…Read more
Elliot K. Wicks, Ph.D.Jack A. Meyer, Ph.D.With assistance from Sharon Silow-Carroll, M.B.A., M.S.W.New Directions for PolicyMay 1999 TABLE OF CONTENTS EXECUTIVE…Read more