The Effect of Rising Health Insurance Premiums on Employment

Featured in print Bulletin on Aging & Health

Employer health insurance premiums have risen sharply in recent years - between 2000 and 2005, premiums rose by over fifty percent in real terms, according to the Kaiser Family Foundation.

Economic theory suggests that if workers fully value their health insurance benefits, they will bear the burden of higher premiums in the form of lower wages, with no change in the level of employment or total wage and benefit costs to employers. But if firms are limited in their ability to offset higher benefit costs through lower wages - for example, by minimum wage laws or union contracts - then rising premiums may cause employers to reduce their work force or shift employment to employees who do not receive benefits, such as part-time workers.

Understanding the relationship between health insurance premiums and employment is of growing policy relevance, since many proposals to cover the uninsured rely on "employer mandates" that would require employers to cover workers. Rising premiums provide an opportunity to observe the labor market effects of higher benefit costs.

NBER researchers Katherine Baicker and Amitabh Chandra explore this issue in The Labor Market Effects of Rising Health Insurance Premiums (NBER Working Paper 11160). Studying this question is difficult because most individual-level data sets have no information on employer premiums. Previous studies have tried to surmount this problem by using average premiums for the worker's industry, firm size, and family status. However, a second problem is that there may be unobservable characteristics, such as worker quality, that affect both health benefits and wages. If this is the case and quality is related to the characteristics used to impute premiums, then any estimated relationship between premiums and employment could be biased by the failure to control for worker quality.

The authors suggest a novel approach to overcome this problem - they use state-level per capita medical malpractice payments to predict health insurance premiums. The authors argue that malpractice costs will be reflected in health insurance premiums but are unlikely to be correlated with unobservable worker characteristics such as quality.

Using this approach, the authors find that the cost of increasing premiums is borne primarily by workers with employer-provided health insurance in the form of decreased wages. They also find effects on employment - rising premiums are associated with more unemployment and more part-time work. The authors find that when premiums rise, part-time workers receive higher wages but are less likely to receive benefits, which suggests that employers may be shifting employment to part-time employees with limited benefits in order to avoid premium increases.

These results are consistent with a model where workers do not value health insurance benefits at their full cost or firms are constrained in their ability to lower wages. Yet these findings differ from those in earlier studies, which typically concluded that employer mandates such as mandated maternity benefits are borne fully by workers through lower wages, with no employment effects. The authors suggest that one possible explanation for the difference is that workers may not fully value the recent increases in insurance premiums, since premium increases have typically come with little expansion in the scope of benefits.

The results of this study have important policy implications. They suggest that the cost of employer mandates is likely to be passed on to workers in the form of lower wages. They also suggest that if some groups of workers are exempt from an employer mandate, such as part-time workers or employees in small firms, then employers may increase their reliance on these workers, undermining the goal of the mandate. The authors conclude "more generally, rising health insurance premiums will place an increasing burden on workers and increase the ranks of both the uninsured and the unemployed."

The authors gratefully acknowledge funding from the National Institute on Aging (grant P01-AG19783-02).