Cost and Coverage Impacts of the President’s Health Care Reform Proposal and a Congressional Tax Credit Proposal. Key Modeling Assumptions for Tax Deduction and Tax Credit

05/19/2008

We modeled the effect of the President’s tax deduction proposal and the Congressional tax credit proposal on coverage using a uniform methodology.  We estimated the impact of these tax incentive methodologies in HBSM based upon a multivariate model of how the likelihood of taking coverage is affected by changes in the net cost of insurance to the individual. We assume that the value of the tax deduction or credit is seen by individuals as a reduction in the price of insurance. Our use of a uniform methodology assures that the difference in estimates for the tax deduction and tax credit proposals are due to differences in the design of these proposal rather than mere inconsistencies in assumptions.

Individual Insurance Premiums.  We simulated the premiums that individuals would pay for coverage in the non-group market using the HBSM individual insurance market sub-model under current law and under each proposal.  This part of the model estimates premiums for each person in the individual market, which includes people now purchasing non-group policies and the uninsured.  We simulated premiums for a uniform benefits package under the individual market rating rules now used in each state.  For illustrative purposes, we use as our uniform benefits package the Blue Cross/Blue Shield standard benefits option, which is estimated to be in the 75th percentile of all health plans on an actuarial value basis.

Neither proposal alters the premium rating methods used in the individual market.  Thus, under the tax credit proposal, the gross (i.e., before-tax) premium is the same as under current law.  For the President’s proposal, we adjusted the estimated premium under the proposal to reflect the elimination of state mandatory benefits requirements.  The model also adjusts premiums to reflect reduced provider cost-shifting for uncompensated care as the number of uninsured is reduced.

Multivariate Model of the Likelihood of Taking Coverage.  We estimated the impact of these proposals on coverage based upon a multivariate analysis of how the likelihood that an individual would take coverage varies with the amount of the premium. This estimate is based upon a pooled, time-series, cross-section analysis of private employer coverage reported in the Current Population Survey for the 1987 through 1997 period. 5 These analyses indicate an overall price elasticity of -0.34 percent, which means that on average, a one percent real (i.e., inflation adjusted) reduction in private employer premiums would result in a 0.34 percentage increase in the number of people with insurance.6

We estimated price elasticities by age, income and other demographic characteristics.  For example, the percentage increase in coverage resulting from a one percent reduction in premiums ranges from a high of 0.55 percent among people with incomes of $10,000 to 0.09 percent among people with incomes of $100,000 (Figure 4) (i.e. a price elasticity of –0.55 to –0.09).  Similarly, the percentage increase in coverage resulting from a one percent reduction in premiums ranges from 0.46 percent for people 20 years of age to 0.30 percent among people 60 years of age (Figure 5) (i.e. a price elasticity of –0.46 to –0.30).  Thus, the model shows that older people and people in higher income groups are less sensitive to changes in price than other population groups.


Figure 4 Percentage Change in Coverage Resulting from a One-Percent Reduction in Premiums by Income Level (in percentages) a/

Figure 4 Percentage Change in Coverage Resulting from a One-Percent Reduction in Premiums by Income Level (in percentages)

a/ Indicates a price elasticity ranging between –0.55 to -0.09 by income.

Source: Lewin Group estimates.


Figure 5 Percentage Change in Coverage Resulting from a One-Percent Reduction in Premiums by Age (in percentages) a/
Figure 5 Percentage Change in Coverage Resulting from a One-Percent Reduction in Premiums by Age (in percentages) 

a/  Indicates a price elasticity ranging between –0.46 and –0.30 by age.

Source: Lewin Group estimates.


Changes in Worker Enrollment.  We also used HBSM to simulate increases in the number of workers and dependents taking employer coverage when offered.  Up to 20 percent of uninsured workers are actually offered coverage through their job but have declined the coverage.  Because workers with ESI are eligible for the deduction and the tax credit, many of these individuals would take the coverage offered under one of the proposals.  For many, this has the effect of reducing the after-tax cost of insurance to the worker, which would result in an increase in the number of people taking coverage when offered.  We simulated this increase in take-up of employer coverage based upon the change in the after-tax cost of insurance to these individuals using the multivariate model discussed above.

 

Pre-emption of State Mandatory Benefits Law.  The President’s proposal would effectively pre-empt state mandated benefits laws in order for a state to offer an Affordable Choices benefit package similar to the one described above. Many states require coverage of selected services for all insurance policies sold in the state. The Affordable Choices benefits plan would require a reduced set of benefits and possible limits on benefits in to meet the definition of affordable (6 percent of state household median income). Thus, state mandatory benefit requirements would need to be eliminated for insurers and health plans to offer this product.

 

Employer Coverage Decision.  For each employer, we use the multivariate model to estimate the probability that an employer would offer coverage given the employer’s characteristics and the amount of the premium they would pay under current law. We then estimated the probability of offering coverage under the premiums they would pay under the proposal and simulated changes in employer coverage based upon the change in the probability of offering coverage.

We used the 1997 RWJF Survey of Employers which provides data on a representative sample of establishments.  These data include information on the size of the firm, industry and workforce characteristics of establishments.  Data include both firms that offer insurance and those that do not.  It also provides information on the characteristics of the health plans offered by each employer including premium costs and the share of the premium paid by the employer.

These data were used to develop a multivariate model to estimate price elasticities showing how the likelihood that a firm will offer coverage varies with wage level, workforce composition, firm size, industry, other firm characteristics and the price of health insurance.7 For example, the implicit price elasticity for firms with fewer than ten employees is -0.87.  This means that for each 1.0 percent reduction in price, there is an increase of 0.87 percent in the number of firms offering insurance.  The implicit price elasticity declines as firm size increases to -0.41 for firms with 10 to 20 workers and -0.22 for firms with 1,000 or more workers (Figure 6 ).


Figure 6: Employer Health Insurance Price Elasticity Estimates by Firm Size a/

Figure 6: Employer Health Insurance Price Elasticity Estimates by Firm Size

Firm Size

a/ Based upon multivariate analysis of the 1997 Robert Wood Johnson Foundation (RWJF) Survey of Employer Characteristics. “Health Benefits Simulation Model (HBSM),” The Lewin Group, August 2003.

Source: Lewin Group estimates using the Health Benefits Simulation Model (HBSM)


The model simulates the effect of employer premium subsidies using this multivariate model of the employer decision to offer coverage.  For each non-insuring employer in the data, we estimate the change in the price of insurance resulting from the premium subsidies.  The model then simulates the decisions to offer coverage based upon the predicted price elasticity for the employer.

The model reflects variations in firm price elasticity depending upon the characteristics of the firm and its workforce.  For example, the model shows that the firm price elasticity tends to decline as workers’ age and income rise, as shown in Figures 7 and 8. This results in a lower estimated price elasticity among currently insuring firms -- averaging about -0.56 for firms with 10 or fewer workers -- because the employers that offer coverage tend to have older and more highly compensated workers.


Figure 7 : Employer Health Insurance Price Elasticity Estimates for Firms with Under 10 Workers by Average Wages and Salaries per Worker a/

Figure 7 : Employer Health Insurance Price Elasticity Estimates for Firms with Under 10 Workers by Average Wages and Salaries per Worker

a/ Based upon multivariate analysis of the 1997 Robert Wood Johnson Foundation (RWJF) Survey of Employer Characteristics. “Health Benefits Simulation Model (HBSM),” The Lewin Group, August 2003.

Source: Lewin Group estimates using the Health Benefits Simulation Model (HBSM).


Figure 8: Employer Health Insurance Price Elasticity Estimates for Firms with Under 10 Workers by Age of Workers a/

bar chart

a/ Based upon multivariate analysis of the 1997 Robert Wood Johnson Foundation (RWJF) Survey of Employer  Characteristics. “Health Benefits Simulation Model (HBSM),” The Lewin Group, August 2003.

Source: Lewin Group estimates using the Health Benefits Simulation Model (HBSM).


Employer Premium Contribution.  We developed multivariate models predicting the percentage of the premium paid by the worker using the RWJF employer data.  These equations measure how premium shares vary with the characteristics of the firm, their workforce characteristics and the amount of the total premium.  These amounts are used to estimate the cost of insurance for workers in each firm selected to offer coverage in response to the program.

Worker Enrollment Decision.  Once firms are selected to offer coverage, we simulate enrollment among workers assigned to these plans.  The enrollment decision is simulated with a multivariate model of the likelihood that eligible workers would take the coverage offered to them based upon data reported in the 1996 MEPS data for people offered coverage through an employer. The model measures how take-up varies with the characteristics of the individual as well as the employee premium contribution required by the employer.

Impact of Individual Tax Deductions and Credits on Employer Coverage.   For both proposals, we used HBSM to estimate the number of employers who would discontinue coverage, as tax deductions and credits become available for non-group health insurance.  Using the synthetic firm database described above, we estimated the cost of covering each firm’s workforce under ESI with the tax credit or deduction and the cost to their workforce of purchasing coverage in the non-group market with the tax credit and deduction. In some of those firms that currently offer insurance, the after-tax cost of non-group coverage for their workforce would be less than the after-tax cost of continuing to provide ESI. We estimate that some portion of these firms would discontinue their ESI.

The underlying assumption in our analysis is that employers offer coverage because they need to in order to attract and retain workers. Thus, we do not expect employers to discontinue their coverage en-masse solely because the tax credits for non-group coverage become available. We assume that employers will make the decision to discontinue coverage only if it is more cost-effective for their workers to obtain the coverage on their own with the tax credit.

We simulated the employer decision to discontinue coverage using the synthetic firm data described above. These data include the family income and tax data required to determine the non-group premium and tax deductible and credit amounts for each worker in each firm.  Using these data, we are able to estimate the after-tax cost of coverage for each group under their current employer health plan, and the after-tax cost of coverage for these individuals if they all purchase coverage in the non-group market.

In firms where coverage in the individual market would be less costly, we assume that some portion would discontinue their health plans.  We simulated the employer decision to discontinue coverage as a shift to the less costly coverage alternative for the group (in this case non-group coverage).  To do this, we relied upon a study by Stombom et al. of the likelihood of shifting to another plan when a lower priced alternative is introduced.8 This study indicates that a 1.0 percent decrease in the price of an alternative source of coverage was on average associated with a 2.47 percent migration of enrollees to the lower cost health plan (i.e., a cross-price elasticity of -2.47). The likelihood of changing plans varies with age and health status as shown in Figure 9 .

Figure 9: Plan Switching Price Elasticity Estimates Used in HBSM

Age of Participant Low Risk High Risk a/
Under 31 -3.50 -2.78
31 to 45 -2.54 -2.54
Over 45 -2.07 -1.38

a/ People in the 90th percentile of health spending.

Source: Stombom, B., Buchmueller, T., Feldstein, P. “Switching Costs, Price Sensitivity and Health Plan Choice,” Journal of Health Economics, 21 (2002), 89-116.

To model the decision to offer or drop coverage, we calculated a “composite” plan-change price elasticity for each employer group based upon the average plan-change price elasticity for each group member.9 Firms were then simulated to discontinue coverage in proportion to the composite plan-change elasticity.

Tax Simulation Data.  HBSM is used to simulate changes in federal income and payroll taxes resulting from changes in the tax treatment of health benefits.  The CPS data provide information on tax payments and marginal income tax rates.  These data are used to impute average and marginal tax rates for households in MEPS, estimate the tax expenditure for health benefits, and estimate the value of tax deductions for health benefits.

Based upon an analysis of the CPS data on tax filings, we estimate that about 40 percent of all uninsured have no tax liability and are not required to file a tax return.  However, about half of these people file even though not required to do so, presumably so that they can obtain any refund they are entitled to (Figure 10).

Figure 10: Distribution of Insured and Uninsured Tax Filers by Marginal Tax rate in 2004
  All Tax Filing Units in the US Uninsured Tax Filing units in US
  With Earnings Without Earnings Total With Earnings Without Earnings Total
Total Potential Filers 119,981 39,367 159,348 23,004 5,016 28,020
Non-Filers 9,451 20,377 29,828 2,848 3,330 6,178

 

  All Filers by Marginal Tax Rate Uninsured Filers by Marginal Tax Rate
Figure 10: Distribution of Insured and Uninsured Tax Filers by Marginal Tax rate in 2004(continued)
0 18,855 11,203 30,068 5,982 648 6,630
10 15,679 2,470 18,149 4,992 354 5,346
15 43,914 3,447 47,361 7,389 484 7,873
27 25,537 1,394 26,931 1,424 140 1,564
30 4,437 359 4,796 242 43 285
35 870 60 930 60 9 69
39 1,235 54 1,289 67 7 74
Total Filers 110,530 18,990 129,520 20,156 1,686 21,842
Source: Lewin Group Estimates Using the 2005 Current Population Survey (CPS) Data.

 

View full report

Preview
Download

"report.pdf" (pdf, 417.83Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®