HEALTH INSURANCE COVERAGE

The 1980s and 1990s were marked by two concurrent trends in employer-provided health insurance: a significant decrease in the fraction of workers receiving insurance through their employers and a sharp increase in the insurance premiums paid by workers. These trends and the possible link between them are explored in two new NBER studies. In Employee Costs and the Decline in Health Insurance Coverage (NBER Working Paper 9036), David Cutler finds that the fraction of non-elderly Americans receiving health insurance through employers fell from 71% in 1987 to 68% in 2000, despite a lengthy economic boom in the 1990s. Over this same time period, the fraction of the population that was uninsured rose by 3 percentage points, an increase of 7.2 million people.

Cutler notes that in order for a worker to have employment-based insurance, three things must happen: the worker must be employed by a firm that offers coverage, be eligible for coverage, and choose to take-up the coverage. Thus, changes in coverage can be explained by changes in the rate of offer, eligibility, or take-up over time. Figure 1 illustrates trends in these three factors for male full-time workers from 1988 to 2000. Interestingly, the share of firms offering insurance remained roughly constant at 80% over this period. The eligibility rate among these full-time workers was also relatively constant and quite high, at over 95%. By contrast, there was a significant drop in the fraction of workers electing to take up coverage, from 94% to 90%. Cutler calculates that 79% of the decrease in insurance coverage for these workers is due to reduced take-up. For female workers, there was also a significant decrease in eligibility associated with shifting to part-time work, so that for all workers, 61% of the decrease in employer-provided insurance between 1988 and 2000 is due to reduced take-up by workers.

A key question is why workers decline to take up benefits. Some workers may do so because they have access to insurance through a spouse. However, declining coverage is also common among those without access to other insurance - 20% of uninsured workers who are offered coverage decline it. When asked, over 60% of these workers list cost as the reason for declining coverage. In fact, employee premiums rose rapidly during this period. Between 1988 and 1993, the average annual employee premium for a family plan doubled from $814 to $1,656; this increase represents close to half of the total increase in the premium. Using data from several surveys of employers, Cutler estimates that each $10 increase in monthly employee premiums lowers the take-up rate by 0.4 percentage points. These results suggest that the increase in premiums from 1988 to 2000 can explain nearly all of the decrease in take-up of insurance coverage.

In "Why Did Employee Health Insurance Contributions Rise?" (NBER Working Paper 8878) , Jonathan Gruber and Robin McKnight explore why the share of premiums paid by workers increased during this period. As Figure 2 illustrates, the fraction of workers receiving employer-provided insurance who paid no premiums fell from 44% to 28% between 1982 and 1998. This low figure is surprising -- simple economic models suggest that workers should prefer to have firms to pay all premiums because the government subsidizes employer provision of insurance by exempting employer-paid premiums from income taxation, effectively lowering the price of health insurance if purchased by the firm. In this simple model, premium increases would be absorbed by the firm, with a corresponding decrease in wages, as workers understand that wages and insurance premiums are both part of the total compensation package.

The authors identify six reasons why employers might have shifted premium costs to employees and estimate the importance of each between 1982 and 1996 using the nationally-representative Current Population Survey. The first factor is increased managed care penetration. As managed care spread, firms began to offer multiple insurance plans; because the cost of the plans can vary significantly, firms may have increased the employee's share of premiums to encourage employees to choose the low-cost option. The authors find that increased managed care penetration is associated with a lower probability that a firm pays all premiums, but the effect is not statistically significant.

Next, the authors explore whether employees' increased access to other insurance through a spouse's employer or through Medicaid led firms to raise employee premiums to encourage them to switch. They find that a 10 percentage point increase in the share of the employee's health spending that is Medicaid-eligible is associated with a 1.7 percentage point decrease in the probability the firm pays all premiums. Similarly, a 10 percentage point increase in the probability that the employee's spouse works is associated with a 1 percentage point decrease in the probability the firm pays all premiums.

The authors also look at the effect of rising insurance premiums and cyclical downturns in the economy. If firms are unable to respond to these factors by decreasing wages (for example, because of contracts or workers' reluctance to accept pay cuts), firms may decrease their share of insurance premiums instead. A 10 percentage point increase in the unemployment rate is associated with a roughly 1.5 percentage point decrease in probability the firm pays all premiums, while rising insurance costs are not found to be a significant factor. Finally, the authors investigate the effect of declining income tax rates, which reduce the subsidy to employer-provided insurance, and find that this is also associated with a decreased probability of the firm paying all premiums. Overall, the factors examined by the authors can explain about one-quarter of the rise in employee premiums over the 1982-1996 period.

David Cutler acknowledges research support from the National Institute on Aging and the Robert Wood Johnson foundation. Jonathan Gruber acknowledges support from the National Institute on Aging and the National Science Foundation. This research was summarized by Courtney Coile.