The effect of hospital unions on nurse staffing and wages

Presenter: Joanne Spetz, University of California, San Francisco

Abstract

In the past decade, there has been resurgence in union activity in the health care industry in the United States, particularly in hospitals. Membership in nursing unions in the United States has grown 18 percent in less than three years and more than 80 percent since 1983. Of 2.4 million registered nurses employed in the United States in 2006, unions represented about 21 percent. In the 1970s and 1980s, a number of studies found that hospital unions increased the wages and fringe benefits of workers represented by the unions, and there was a corresponding increase in total hospital costs. Unions claim to have obtained significant concessions from hospitals in recent years, both in wage increases and in increasing nurse-to-patient staffing ratios, “mandatory” overtime, and other non-wage issues.

This paper examines the impact of unionization on two important labor market outcomes: nurse wages and nurse staffing in hospitals. We use a unique dataset from California, which merges data collected from all hospitals in the state about their unions with Annual Hospital Disclosure Reports data from the California Office of Statewide Health Planning and Development. We estimate multivariate regression equations to learn whether unions have a statistically significant effect on hospital wages. The dependent variable in these equations is the logarithm of wage of each type of hospital worker. Explanatory variables include patient acuity, service mix, profit status, system affiliation, HMO penetration, market concentration measured with a Herfindahl index, and other market conditions. We allow for both direct and indirect effects (i.e., union threat effects) of unions by considering whether nearby hospitals are unionized. We also will explore whether the ownership and system affiliation of a hospital interacts with union status to affect wages.

We next estimate the effect of unionization on labor demand using the same dataset. The dependent variable is the logarithm of staffing levels, with staffing measured as productive hours. Explanatory variables will include wages, wages of complementary/substitute employees, patient volumes (measured as patient days and/or patient discharges), casemix, service mix, profit status, system affiliation, market concentration, and HMO penetration. As before, we allow for both direct and indirect effects (i.e., union threat effects) of unions by considering whether nearby hospitals are unionized. One concern for this analysis is that demand is endogenous with wages; that is, wage increases that result from union activity will affect labor demand. We can address this endogeneity several ways. One possibility is to aggregate wages to the market level, thus measuring market wages. We also can use instrumental variables techniques, although recent research on hospital demand for nurses has not provided satisfactory instruments. Two potential instruments are community unemployment rates and the age distribution of the local supply of nurses.

Preliminary equations indicate that unionized hospitals pay higher wages, and that some union threat effect exists.