The debate on how teacher unions and collective bargaining influence education systems across the United States is a passionate one, but the passions ignited are not commensurate with the quality of empirical evidence on how unions and bargaining actually shape education. Scholars generally agree that collective bargaining leads to increases in teachers’ salaries, reductions in classroom size, and increases in education expenditure. However, this consensus is based mostly on studies that compare the salaries, class sizes and expenditures of districts or states that have collective bargaining, and districts or states that do not, at a single point in time, without accounting for long-standing differences between these states or districts that may explain both the emergence of collective bargaining and the characteristics of their education systems.

My work uncovers important historical differences between states that eventually introduced collective bargaining rights and those that did not. Using a new state-level annual dataset going back to 1959 –before any state had introduced the legal right to collective bargaining– and up to 1990, I show that the cross-sectional differences in teacher salaries, class sizes and education expenditure that existed between states with and without collective bargaining rights in 1990 were already present in 1959 and even earlier, highlighting that those states that eventually introduced bargaining rights have been historically inclined to pay teachers more, have smaller classrooms, and spend more on education. I also show that there are historically rooted differences between these groups of states that transcend the education sector –even within the South-, differences which are difficult to account for in cross-sectional analyses.

To assess the effect of collective bargaining laws, which are established at the state level, I employ comparative interrupted time series (CITS) models that exploit the idiosyncratic variation in the timing of introduction of bargaining rights, assessing whether introducing such rights affected salaries, class sizes and expenditures within a state. In addition, I examine whether states that have bargaining rights adjust differently than those that do not when confronted by external shocks, focusing specifically on state responses to the decline in student enrollment that occurred during the 1970s. The CITS estimates suggest, in line with prior research, that the introduction of collective bargaining rights led to an increase in salaries and per-pupil expenditure, although the magnitude of the estimated effects is smaller than that suggested by prior research. Further, in contrast to previous studies, I find no evidence to support the claim that the introduction of bargaining rights led to a reduction in class sizes, but I do find that during the 1970s, when student enrollment declined, states with collective bargaining rights reduced class sizes more than states where the right to collective bargaining did not exist. Taken together, this evidence suggests that unions have been concerned about the salaries and job stability of current members, but have not been particularly interested in expanding the size of their membership. These findings have important implications for current policy debates and for theories about the role of public sector unions.