The severity of the Great Recession has propelled public sector pension health into the spotlight. Once considered a virtual guarantee, concerns over massive under-funding and the erosion of protections surrounding benefit levels has brought into question the retirement security for millions of state and local employees. Despite rhetoric that public sector unions are at least partially responsible for the crisis, very little concrete evidence exists to defend this.

We propose a new measure of pension generosity and use it to explore the extent to which variation in the strength of teachers' unions affects the generosity of their retirement systems. Unlike many margins over which unions may exert influence, pension benefits are set by statute rather than being negotiated through the collective bargaining process. We measure the strength of teachers' unions by their aggregate campaign contributions in statewide elections to candidates, parties, and other organizations.

Teachers' unions are actively engaged in the political process at both the federal and subnational levels. In the 2014 electoral cycle, the National Education Association (NEA) contributed more than $29 million dollars at the federal level to candidates, parties and their committees, and other groups and organizations. The dollar amount was large enough that the NEA was ranked the 2nd most generous organization in terms of total contributions (out of more than 17,000 organizations) by The Center for Responsive Politics. At the state level, teachers' unions ranked among the top 20 donors in 33 states in 2014 and were the single largest donor organization in two states.

We pursue an instrumental variables strategy for identification, relying on geographic diffusion of educational reform policies in neighboring states as instruments for teacher contributions. Using an unbalanced panel of 31 distinct state-administered teacher pension plans from 2001 to 2015, our results show robust empirical evidence that stronger teachers' unions have a positive and significant causal impact on the generosity of their pensions. For example, other factors constant, we find that the difference in the present value of benefits earned in a single year that are paid by the plan sponsor varies by $2,400 per active pension member for teachers in states with strong unions relative to teachers in states with weak unions (75th versus 25th percentile). We also find evidence that the marginal effect of unions on generosity was strongest in the pre-Great Recession period before weakening some between 2010 and 2013 when nearly all states enacted benefit reductions.