The effect of the right-to-work law on Oklahoma: A difference-in-differences study

In the past five years, four states with a longstanding union tradition — Indiana, Michigan, West Virginia and Wisconsin — passed right-to-work laws. Before then, Oklahoma was the last state to pass a right-to-work law in 2001. Right-to-work laws prohibit collective bargaining contracts between firms and unions that require all workers pay union dues — even those workers not belonging to the union. Proponents of such laws argue they foster a business-friendly environment that attracts new firms, leads to job creation and promotes economic growth. Detractors of such laws say the presence of free riders dilutes union power, lower wages and lower living standards. Studies of the impact these laws on these four states have been hampered by the limited number of annual, quarterly and monthly observations since the policy change.

One way to infer the possible effects of the law on the economies of these states is to examine what happened to Oklahoma’s economy after the right-to-work law was enacted in 2001. Using annual, quarterly and monthly data, this study employs the econometric methods used in the difference-in-differences technique to assess the passage of a right-to-work law on Oklahoma’s economy. The study estimates how passage of the law affected a number of economic variables including state GDP, job creation, labor force levels, population, employment, unemployment rates, the number of business establishments, poverty rates, union membership, the percentage of workers covered by collective bargaining agreements and the level of wages.

Difference-in-differences studies require observations from both a treatment group and a control group. The state that experienced the regime change—Oklahoma—is defined as the treatment group. The control group consists of observations from surrounding states including Arkansas, Colorado, Kansas, Missouri, New Mexico and Texas. Arkansas, Kansas and Texas are right-to-work states while Colorado, Missouri and New Mexico are union-shop states. Binary variables are used to identify Oklahoma and the other control states. Another binary variable identifies the time periods before and after the policy change. The regression coefficient associated with the interaction of these two variables is an estimate of the difference-in-differences. The t-statistic of this coefficient provides evidence whether passage of the right-to-work laws had an effect that was statistically significant. Given the presence of both right-to-work states and union-shop states in the data, the paper also considers the more complicated difference-in-difference-in-differences estimate.