In many states, local teacher unions play an important role in the policy decisions of local school districts.  District policies on everything from salary schedules and the length of the work year to class sizes and evaluation systems can require the consent of the local teacher union during the collective bargaining process.   Yet, the role of teacher unions is not fully understood in the academic literature.  While several studies have looked at the average effect of unionization on educational inputs and outputs (e.g. Hoxby 1996, Lovenheim 2009), few have investigated the heterogeneity in union effects across unionized districts.   In the many states where nearly all districts have been unionized for decades, understanding why different local unions obtain different policy results may be of more interest to state policymakers than understanding the average effect of the formation of a union in a district.   There are two primary mechanisms for union effects to vary: differences in union bargaining power and differences in union composition.  I focus primarily on the latter mechanism, where unions, representing their membership, differ in their bargaining goals due to differences in the values of their memberships.  Specifically, I assume that preferences regarding the construction of district salary schedules vary with a teacher’s experience level, and explore the effect of the experience distribution of a union’s teachers on the salary schedule that union negotiates with the district.

I develop a simple voting model of a union’s decision to allocate a given average salary along the district’s salary schedule, and find that under some conditions, an increase in the proportion of teachers with high-levels of experience leads salary raises to be increasingly focused at higher experience levels (“back-loading”).  I test this prediction using an 11-year panel dataset (1999-2000 to 2009-2010) of complete district-level salary schedules for districts representing 97% of California students. In my primary specification, I estimate the effect of within district changes in the proportion of teachers with high-levels of experience on the difference between the average salary raises early-career and late-career (a measure of salary back-loading).  I use district and year fixed effects,  include the average outcome of other districts in the county to account for geographic relationships between districts, and alleviate the endogeneity of the teacher experience distribution by using 15 and 12 year lagged enrollment growth rates as instruments.  I find that a 10 percentage point increase in the proportion of teachers with 21+ years of experience leads to a statistically significant $150 increase in difference between the average yearly raise from 10-20 years experience and 0-10 years experience (in other words, back-loading increases). This corresponds to a 13% increase for the average district in 1999.  In further analysis, I will interact measures of the teacher experience distribution with teacher union bargaining power to see if these results are stronger in districts with more powerful unions.  I will use the number of school board members whose professions are listed as “Educator” in the California Elections Data Archive as a proxy for union power.