The University of Michigan
555 South Forest Street
Third Floor
Ann Arbor, MI 48104-2531

T 734-936-9842
F 734-998-6341
/


HIGHER TAX RATES INCREASE EMPLOYEE HEALTH INSURANCE CONTRIBUTIONS

Between 1982 and 1998 there was a dramatic increase in employee contributions to health insurance coverage. The rise in employee contributions is an important reason why employees turn down offers of coverage and remain uninsured. A study by Jonathan Gruber and Robin McKnight, funded by the Economic Research Initiative on the Uninsured (ERIU), examines the factors that have led to rising employee contributions. Specifically, they consider the effects of falling tax rates, rising eligibility through public programs or spousal coverage, and a troubled economy on employee contributions. They find that these factors explain about one-quarter of the change in employer contributions over the 1982 to 1996 period.

FINDINGS
Effect of Tax Price on Employee Health Contributions
Tax rates may lead employers to alter employee contributions because insurance is generally purchased with pre-tax dollars and higher tax rates increase the generosity of the tax subsidy. For each 10 percentage point increase in the tax price, which is approximately a 10 percentage point decrease in the tax rate,:

  • The odds of an employer paying all of the costs of health insurance drop by 3.67% (SE 0.098).
  • The odds of an employer paying some of the costs of insurance drop by 1.27% (SE 0.077).
  • Employment-based health insurance coverage drops by 5.54% (SE 0.091).
  • There is a shift of between 1.7% and 3.7% from employers paying all of the cost of insurance to employers paying some or none (using conservative assumptions).


Effect of Spousal Labor Supply on Employee Health Contributions
Employers may increase employee premium contributions as the percentage of working spouses rises to encourage workers to receive coverage from a spouse’s employer, and to discourage spouses who may have other insurance options from receiving coverage through their firm. For each 10 percentage point increase in the odds of having a working spouse:

  • The odds of an employer paying all of the costs of health insurance drop by 1% (SE 0.026).
  • The odds of an employer paying some of the costs of insurance drop by 1.2% (SE 0.030).
  • Employment-based health insurance coverage drops by 2.2% (SE 0.011).
  • There is a small shift from employers paying all of the cost of insurance to employers paying some or none.


Effect of Medicaid Eligibility on Employee Health Contributions

Employers may increase employee premium contributions following Medicaid expansions to encourage take up of public coverage. For each 10 percentage point increase in the share of medical spending eligible for Medicaid coverage:

  • The odds of an employer paying all of the costs of health insurance drop by 1.7% (SE 0.034).
  • The odds of an employer paying some of the costs of insurance drop by 0.08% (SE 0.047).
  • Employment-based health insurance coverage drops by 0.64% (SE 0.045).
  • There is a 1.53% shift from employers paying all of the cost of insurance to employers paying some or none (using conservative assumptions).


Effect of Managed Care Penetration on Employee Health Contributions

Rising managed care penetration has a small negative effect on employee contributions, though this effect is not statistically significant. This suggests that only a small amount of the trend in premium sharing is explained by employers trying to induce employees to choose the more cost effective plans.

POLICY IMPLICATIONS
Employers respond to changes in tax incentives, Medicaid eligibility, and spousal coverage both in terms of their offers of health insurance and whether they expect employees to contribute to the premium. Other factors not included in the models, however, explain most of the time trend in employee premium sharing.

CAVEATS
The findings are based on an instrumental variable technique that adjusts for the potential endogeneity of key variables. Spousal labor supply is instrumented for using characteristics of the spouse. Following what has become the accepted approach, tax rates are instrumented for using simulated tax rates, and a national cohort is used to instrument for Medicaid eligibility. In both cases the instruments are identified by interaction terms because the main effects are included in the main model. The validity of the findings hinges on the performance of the instruments. Though conceptually acceptable, few diagnostics of the instrumental variables approach are provided.

There are several technical reasons why the results may be biased toward zero. First, the analysis uses a linear probability model as opposed to a logit or probit approach. This provides for consistency in the instrumental variables estimates but may lead to some attenuation bias. Second, much of the analysis is based on state level data. Many firms serve multiple states. It is not clear how employer contribution policies are affected by variation in key variables across states, but this measurement issue could lead to attenuation bias. Third, the cost data is measured at the state level, as opposed to the market level. This will create measurement error and may diminish the observed effects of costs on coverage.

Finally, employee premium sharing may vary for single vs. family coverage, but the analysis does not distinguish among the groups.

DATA SOURCE
Current Population Surveys (CPS), March 1983 – 1997 Annual Demographic Files. The National Medical Expenditure Survey (NMES), 1987, is used for health spending data. The National Bureau of Economic Research (NBER) TAXSIM model is used for marginal tax rate of workers. Statistics of Income (SOI) data are used to impute the odds of itemization and the amount itemized by state and family earnings. All insured workers age 21 to 64. The self-employed and federal employees are excluded.

METHODOLOGY
Medicaid eligibility is simulated based on demographic information. Medicaid eligible medical spending is estimated by age group using NMES data. Instrumental variables are used to reduce noise and to address omitted variables bias for spousal labor supply, Medicaid eligibility, and tax subsidy. Linear probability models are estimated with managed care penetration, spousal labor supply, tax price, state/year medical spending, unemployment rate, and a set of individual covariates as independent variables.

CITATION
Why Did Employee Health Insurance Contributions Rise?
Jonathan Gruber and Robin McKnight, Massachusetts Institute of Technology

Conference paper presented at ERIU Research Conference, July 2002

ERIU Working Paper #9 (Adobe PDF)

Back to top


Funded by The Robert Wood Johnson Foundation, ERIU is a five-year program shedding new light on the causes and consequences of lack of coverage, and the crucial role that health insurance plays in shaping the U.S. labor market. The Foundation does not endorse the findings of this or other independent research projects.