The Impact of Unionization on Health Insurance Benefits
Fichtenbaum, Rudy, Olson, Paulette, Journal of Economic Issues
Within labor economics, it is a stylized fact that unions exert a positive influence on both wages and fringe benefits. Although most of the literature focuses on the union--non-union wage differential (Parsley 1980; Freeman and Medoff 1984; Lewis 1986; Hirsch and Addison 1986), more recently the interest has shifted to the union--non-union gap in fringe benefits (Belman and Heywood 1990, 1991; Even and MacPherson 1991, 1994). According to this literature, unionized employers are more likely to provide fringe benefits such as pensions and health insurance than their non-unionized counterparts.
During the 1980s, the topic of fringe benefits received more attention in the economic literature in part because money wages had declined as a proportion of total employee compensation. For instance, between 1966 and 1977, fringe benefits grew at a faster rate than wages in the private, non-farm economy. By 1977, fringe benefits represented 23.3 percent of employee compensation (Leibowitz 1983, 371). By 1994, it had reached a peak of 28.9 percent. Since then it has declined somewhat to 27.1 percent. (1)
Numerous studies have highlighted the impact of unionization on fringe benefits, but the data have been qualitative in nature. That is, economists have estimated the impact of unionization on the probability that pensions and health insurance benefits are provided by employers in the first place. Using Current Population Surveys data, Even and MacPherson (1991) found that one-half of the union--non-union gap in pension coverage and one-third to two-thirds of the gap in health insurance coverage is explained by the atypical labor market characteristics of unionized workers such as larger establishment size, higher wages, and longer workweeks. In a later study, Even and MacPherson (1994) found that after controlling for firm size, a unionized workplace significantly increases the probability of both pension and health insurance coverage. However, the probability of health insurance coverage declined between 1983 and 1988.
More recent studies concerned with the impact of unionization on fringe benefits have reexamined standard econometric methods, but they have not seriously challenged earlier findings. Belman and Heywood (1991), for instance, criticized previous studies for failing to take account of the possible simultaneity between wages and fringe benefits when using the log of wages in logit or probit equations to estimate the union impact on the probability of fringe benefit coverage. Nonetheless, they found that unionization increased the likelihood of both pension and health insurance coverage, although the size of the probit coefficients was reduced when they used a two-stage estimator.
As suggested earlier, most studies have focused on the relative probability of fringe benefits being provided by employers in the first place. Only a few studies have been concerned with whether unionization raises the expenditure levels of fringe benefits. For instance, using data from the Bureau of Labor Statistics Survey of Private Establishments, Fosu (1984) not only found that unionization increased the likelihood of both pension and health insurance coverage, he found that unionization increased the relative expenditures on health insurance, but not pension coverage. Another study by Lebowitz (1983) found that when expenditures on fringe benefits are included in a hedonic wage equation, the union-non-union wage differential increases. Although the results of the latter study are based on data that include employer expenditures on fringe benefits, the results are less than robust because of the small sample size. (2) Nevertheless, this study highlights the growing importance of including fringe benefits in estimating the impact of unionization on the union-non-union total compensation differential.
As table 1, column 1 shows, the unadjusted total compensation differential between unionized and non-unionized workers in private industry for March 2001 was 39. …