25
Defaulting New Hires into Health Insurance
Three main challenges of this type of analysis are to 1) identify who is affected, 2) estimate the fraction of workers who newly take-up insurance and 3) identify spillover effects.
To identify who would be affected by auto-enrollment, we use household surveys to learn about worker characteristics and an employer survey to learn about firm characteristics and insurance plan offerings at those firms. We estimate the number (and characteristics) of workers with offers of health insurance who are uninsured from the Medical Expenditure Panel Survey-Household Component (MEPS-HC) and the National Health Interview Survey (NHIS). The MEPS Insurance Component (MEPS-IC) provides data that allows us to determine the firm characteristics associated with employers who are likely to auto-enroll their new employees, the average generosity (actuarial value) of the plans that are being offered, and the average portion of the total premiums contributed by these employers.
To estimate the fraction who newly take-up insurance, we rely in part on estimates of the effect of auto-enrollment into 401k plans on 401k participation rates and adjust those rates for three factors specific to health insurance participation: biased assessments of risk, differential expected benefit of being insured, and larger reductions in wages required by the default. Studies have shown that individuals often have a poor understanding of risk—underestimating the risk of high probability events and overestimating the risk of low probability events. These biases in risk perception can influence individual take-up decisions in health insurance where individuals have difficulties assessing future health states. Individuals may also intrinsically place different values on a dollar invested towards a future retirement product versus a health insurance product, and be more likely to opt out of a default when the cost of remaining in the default option is higher (e.g. if the employee’s default premium contribution is higher than the default 401k contribution).
Auto-enrollment may result in spillover effects on other parts of the healthcare system. On average, individuals newly gaining insurance through auto-enrollment are younger and have fewer chronic conditions than the existing ESI enrollees; their enrollment may, therefore, decrease ESI premiums for other workers in the risk pool. If they would have been Medicaid eligible, autoenrollment may reduce Medicaid enrollment of those whose health later caused them to seek medical services.