Teacher Collective Bargaining and Student Learning
This week's guest blogs feature Katharine Strunk and Josh Cowen of Michigan State's Education Policy Innovation Collaboration (EPIC). They'll be joined by University of Washington's Dan Goldhaber, Brown's Matt Kraft, USC's Brad Marianno, and Cornell's Michael Lovenheim throughout the week as they discuss teachers' unions and teacher labor market reforms.
Teachers' unions remain one of the most contentious features of the education landscape in the United States. One of the core elements of this debate is how teachers' unions affect student learning. This post provides an overview of what we know about how teacher collective bargaining affects student achievement and why obtaining answers to this question is so difficult.
Proponents and opponents of teachers' unions each have a theoretical basis for their beliefs. Those in favor of teacher unionization argue that teachers should have a voice in the allocation of education inputs. They say teachers have a better understanding of education production than do policymakers, and will negotiate for non-pecuniary benefits that improve teachers' working conditions and thus increase the quality of the teacher workforce.
In contrast, Caroline Hoxby first posed the "rent-seeking" hypothesis of teachers' unions in her seminal 1996 paper on teachers' unions. She posited that collective bargaining allows teachers to soak up more of the resources for themselves without generating any change in the quality of the teacher workforce. This resource distortion is likely to reduce teacher productivity and student learning.
This theoretical ambiguity underscores the importance of empirically estimating the link between teacher collective bargaining and student learning. Doing so has proved difficult for two reasons.
First, districts in which teachers' unions engage in collective bargaining tend to be different from those that don't, usually in ways that correlate strongly with student achievement. We cannot just compare student outcomes across schools with different collective bargaining behavior.
The predominant way researchers have sought to overcome this difficulty is by using the passage of state "duty-to-bargain (DTB)" laws over time, which mandate that school districts negotiate with elected union representatives in good faith. They facilitated a large increase in the prevalence of teachers' unions in the US. Provided that passage of DTB laws is uncorrelated with changes in student characteristics, examining how outcomes change across students who were differentially exposed to these laws provides a way to estimate the causal effect of collective bargaining on student outcomes.
Second, good student outcome data only became available after the teacher collective bargaining movement ran its course. Hence, when we have good outcome measures we do not have much usable variation in collective bargaining, and when we have useful DTB law variation we lack information on student achievement.
The only outcome data from the time period when DTB laws were being passed is high school completion rates from the US Census. Hoxby examined how changes in unionization rates among teachers driven by DTB law passage affects high school dropout rates. She found that increased unionization increases high school dropout. However, I argue in a prior research paper that this finding changes when one uses an arguably more accurate union measure, but I am only able to show these results for three Midwestern states.
A more recent research paper I have undertaken with Alexander Willén re-examines this question by linking long-run outcomes from the American Community Survey (ACS) to how long individuals were exposed to DTB laws during their schooling years. We estimate how labor market outcomes and educational attainment change across cohorts who were differentially exposed to DTB laws because of when and where they were born.
Our results accord closely with the predictions of the rent-seeking model for men. Being exposed to a DTB law for all 12 years of schooling reduces male earnings by almost $1,500 per year, or about a 2.75-percent reduction in annual earnings. Multiplying by all men in the 33 DTB states leads to a yearly reduction in earnings of $150 billion dollars. Furthermore, we find that male hours worked and employment are reduced by 1.3 percent and that men sort into lower-skilled occupations. What's more, the adverse effects in our study are significantly larger for Black and Hispanic men. Thus, teacher collective bargaining negatively affects the long-run labor market outcomes of men and exacerbates racial/ethnic disparities due to the disproportionate impact on non-whites.
Interestingly, we do not find any effects among women. We are unable to tell why there are such strong gender differences in the effects we estimate, though we suspect it is related to emerging evidence that boys are more sensitive to adverse shocks that occur when they are young than are girls.
Our results align with other emerging research on teachers' unions that examines specific teacher contract provisions in collective bargaining agreements. Marianno and Strunk painstakingly code up these provisions for the bulk of California school districts for three iterations of negotiated CBAs. They then estimate how changes in the restrictiveness of union contracts relate to changes in student test scores. While it is not totally clear why union contracts become more or less restrictive, they find that when these contracts are more restrictive student test scores go down (or at least do not go up).
A coherent story emerges from this body of evidence: Teacher collective bargaining leads to worse student outcomes that are reflected in long-run labor market success, and these deleterious effects are driven to some degree by the bargaining process. These findings accord closely with the rent-seeking model of teachers' unions.
From a policy perspective, these studies highlight the importance of developing a more complete understanding of what aspects of teacher collective bargaining are responsible for the worse learning outcomes. It might be possible to alter the scope of collective bargaining or the process itself to protect aspects that teachers value while reducing or eliminating the deleterious effects on students. Recent changes in collective bargaining rights in states such as Wisconsin, Indiana and Michigan may offer new insights into how changing specific aspects of collective bargaining affects students in order to inform optimal teacher bargaining policy.