Association for Public Policy Analysis & Management

Popular political rhetoric often pits teachers’ unions vested interests in opposition to the interests of students (Moe, 2017). Advocates for reforms to teachers’ unions and their collective bargaining agreements (CBAs) often portray the restrictions found in CBAs, such as seniority rights and tenure policies, as negatively impacting students’ access to high quality teachers. In contrast, proponents of teachers’ unions often argue that the working conditions of teachers are the learning conditions of students, and that improvements to CBAs should enhance students’ opportunities to learn. Of the growing number of studies on the relationship between CBAs and student performance, the results are still mixed, and confined to only a handful of states (i.e., California, Ohio, Michigan, and Washington) (e.g., Strunk et al., 2018).

In this study, I again take up the issue of CBAs and student performance, this time using a novel dataset of CBA policies from Pennsylvania. I ask and answer the following research questions: 1) To what extent do CBA items vary across districts within Pennsylvania? 2) To what extent do political, demographic, financial, and geographic spillover factors predict the strength of Pennsylvania CBAs?3) To what extent is CBA strength related to student achievement and graduation rates?

I assembled a database of 33 high-profile contract items (identified by Strunk et al., 2018) from all 499 Pennsylvania district contracts. Using factor analysis, I created overall and subarea measures of CBA strength (defined as academic items pertaining to class sizes and teacher evaluation, and non-academic items pertaining to bonuses, professional growth plans, preparation time guarantees, hiring, and dismissal). To answer research question 1, I depict variation among items using GIS mapping of school districts in Pennsylvania. To answer research question 2, I run a series of OLS, probit, and spatial autoregressions. Finally, to answer research question 3, I use propensity score weighting with generalized boosted models (GBM) in order to achieve balance among districts across the distribution of contract strength on observable characteristics. For the continuous contract strength constructs, I test several cut-off scores as treatments.

Because of the mandates within the Pennsylvania School Code, I find relatively little variation across many of the items, except among the seniority hiring and dismissal rules. In answer to research question 2, I find a significant negative relationship between the percentage of registered Republicans in a district and the strength of the CBA. I further find that CBAs are stronger in districts with larger enrollment. In answer to research question 3, I find a significant positive effect between the academic provisions and students’ math achievement in grades 3-8, and in high school. In contrast I find a significant negative effect between the non-academic items and graduation rates. My results, though not causal, suggest that subject-area and grade-level differentiation in collective bargaining items related to class sizes and teacher evaluation might benefit students.