PREFERENCES INFLUENCE HEALTH INSURANCE DECISIONS

Economic theory holds that individual preferences influence choices about whether or not to have health insurance. A study by Alan Monheit and Jessica Primoff Vistnes, funded by the Economic Research Initiative on the Uninsured (ERIU), looks at preferences by examining how health insurance, risk, and medical care relate to health insurance status. They find some evidence that preferences are related to health insurance status.

FINDINGS
The study relies on survey responses to four statements that relate to health insurance. Two statements are termed direct measures of an individual’s health insurance preferences ("I'm healthy enough that I really don't need health insurance" and "Health insurance is not worth the money it costs") and two indirect ("I'm more likely to take risks than the average person" and "I can overcome illness without help from a medically trained person.") The authors aggregate the five-level response scale into three types of preferences for health insurance: "strong" (somewhat and strongly disagree), "uncertain" and "weak" (somewhat and strongly agree.)

Depending on the preference measure, from 15% to one-third of the U.S. non-elderly adult population has weak or uncertain preferences for health insurance

• Younger persons are more likely to report weak preferences. Weak preferences are more prevalent among men and those in (self-reported) excellent health, compared to those in (self-reported) poor health.
• For all preference measures except 'overcome illness,' the prevalence of weak preferences declines as educational attainment increases.

• For example, among single workers, 17.8% have strong preferences for health insurance but do not have jobs that offer coverage and 12.2% have weak or uncertain preferences for health insurance but are offered coverage anyway.

• Among those who begin the year without health insurance, more of those with strong direct preferences become insured over the year (by roughly 10 percentage points) than those with weak preferences. For the two indirect preference measures, the difference in gaining health insurance was not statistically significant.

• Individuals who were uninsured all year were between 5 and 13 percentage points more likely to indicate “weak preferences” for coverage than those insured for the full year.
• The fraction of those uninsured all year who express weak preferences for coverage rises with income, suggesting uninsurance among higher income persons may reflect choice rather than monetary constraint.

POLICY IMPLICATIONS
When health insurance is voluntary, preferences can influence who has health insurance. T hose with weaker preferences may be less likely to respond to initiatives to expand health insurance.

Reliance on voluntary approaches to expand health insurance through premium support programs may be more effective if accompanied by educational efforts to inform targeted uninsured groups about the merits of health insurance. Such educational efforts should not be restricted to information about the presence of a new program, but also provide more generic information regarding the purpose of health insurance, its costs, and its likely impact on improving timely access to services and quality and continuity of care.

CAVEATS
Responses to statements about health, health insurance, and risk reflect many dimensions of preferences. It is difficult to assess which dimension, or aggregation of dimensions, is most salient. For example, the "health insurance is not worth the money it costs" statement reflects preferences for non-medical goods and services as well as preferences for coverage. Views of the "not worth the money it costs" statement may also reflect perceptions about access to charity care for those who do not have health insurance.

The statements "I'm healthy enough that I really don't need health insurance" and "I can overcome illness without help from a medically trained person" reflect both preferences for coverage and perceptions about risk of illness and effectiveness of medical care. Respondents' views about their risk of illness and the effectiveness of care may not be accurate; their responses might be different if they were fully informed about their risk and the effectiveness of care. In addition, some responses may be an ex post rationalization of health insurance status rather than a direct measure of preferences. However, since the data on preferences were obtained in an independent component of the survey from questions on health insurance status, this problem may be mitigated.

The multivariate models reported in the paper are at an early stage of development and may change in future versions of the paper.

DATA SOURCE
Preference data are from the 2000 Medical Expenditure Panel Survey (MEPS.) The study sample includes 9545 individuals (workers and non-workers, excluding full-time students and the self-employed) aged 18 to 64 who responded to a Self Administered Questionnaire (SAQ), excluding proxy respondents; models are estimated from samples of 2613 unmarried wage earners and 1706 couples where both spouses responded to the SAQ and at least one was employed as a wage earner. Covariates were constructed from data from a variety of sources, including the MEPS Insurance

METHODOLOGY
The descriptive analysis involves tabulations of survey responses and cross tabulations of preference measures and individual characteristics. The multivariate analysis employs logit models to estimate characteristics associated with several health insurance-related outcomes and three of the preference measures. Separate estimates are produced for single workers and for married couples with one and with two workers.

CITATION
Health Insurance Enrollment Decisions: Understanding the Role of Preferences for Coverage
Alan Monheit, University of Medicine and Dentistry of New Jersey, and Jessica Primoff Vistnes, Agency for Healthcare Research and Quality.

Conference paper presented at ERIU Research Conference, July 2004

ERIU Working Paper #31 (Adobe PDF)

Back to top