Unionization and Productivity: Evidence from Charter Schools

• Published date: 01 July 2015
• Date: 01 July 2015
• DOI: http://doi.org/10.1111/irel.12097

Unionization and Productivity: Evidence from

Charter Schools*

CASSANDRA M.D. HART and AARON J. SOJOURNER

This paper studies the relationship between teacher unionization and student achieve-

ment. Generally stable patterns of teacher unionization since the 1970s have histori-

cally presented challenges in measuring the effects of unionization on educational

production. However, the blossoming of the charter school sector in recent decades

provides fertile ground for study because while most charters are nonunion, teachers

at some charters have unionized. Using a generalized difference-in-difference

approach combining California union certificationdata with student achievement data

from 2003–2012, we find that, aside from a one-year dip in achievement associated

with the unionization processitself, unionization does not affect student achievement.

Introduction

Student learning matters greatly to our society’s health and depends cru-

cially on teachers’productivity, which varies greatly. Based on conservative

estimates from recent work (Chetty et al. 2011; Hanushek 2010), a one stan-

dard deviation improvement in teacher productivity per teacher per year—

equivalent to producing a 0.2 standard deviation increase in student achieve-

ment—generates about $200,000 in net present value social returns by raising

students’future private earnings and reducing their demand for public expendi-

tures. Given the importance of variation in teacher productivity for long-term

economic potential, policymakers have a keen interest in exploring the condi-

tions that foster or undermine teacher productivity. In recent years, unioniza-

tion has attracted great attention as a candidate factor that may affect teacher

productivity.

1

*The authors’affiliations are, respectively, University of California, Davis, Davis, California. Email:

cmdhart@ucdavis.edu; University of Minnesota, Minneapolis, Minnesota. Email: asojourn@umn.edu.

The authors would like to thank the California Public Employee Relations Board, especially Les Chis-

holm, and the National Association for Public Charter Schools for sharing unionization data with them. They

are grateful to Brigham Frandsen, Ezra Golberstein, Colleen Manchester, Elton Mykerezi, Heather Rose,

Rob Vellela, and two anonymous reviewers for comments, and thank Charles Niemeyer for research assis-

tance. Any errors are their own.

1

State laws governing teachers’unionization rights vary widely, as do states’teacher unionization rates,

which range between 14 and 77 percent (Sojourner 2013). In 2012, half of U.S. elementary and secondary

teachers belonged to a union and half did not (Hirsch and Macpherson 2013).

INDUSTRIAL RELATIONS, Vol. 54, No. 3 (July 2015). ©2015 Regents of the University of California

Published by Wiley Periodicals, Inc., 350 Main Street, Malden, MA 02148, USA, and 9600 Garsington

Road, Oxford, OX4 2DQ, UK.

422

The effect of unions on average labor productivity in organizations generally

(Freeman and Medoff 1984) and in schools specifically (Eberts 1987; Hoxby

1996) is theoretically ambiguous. First, consider how changes in effort and job

design might affect productivity, holding the set of workers and managers fixed.

On the positive side, increased commitment between the worker and firm and

greater training opportunities may raise productivity (Acemoglu and Pischke

1998). On the negative side, unions may advocate for restrictive work rules (such

as strict limits on on-site workday length or division of tasks between job catego-

ries) and blunt incentives (such as seniority-based personnel rules and resistance to

performance pay), which may lower effort and productivity (Mue 2011). Second,

consider changes in resource allocation. Unions might shift the mix of inputs and

raise school budgets in productivity-enhancing ways by increasing teachers’ability

to bring their superior expertise in educational production to bearon policymaking,

or they might produce shifts in the allocation of resources in ways that enhance

teacher satisfaction at the expense of students (Rose and Sonstelie 2010). Third,

unionization may affect who works at an organization in complex ways, leading

unionized schools to have different sets of teachers and managers than nonunion

schools (Card 1996; Freeman and Kleiner 1990; Pettengill 1979). Which of these

channels dominate remains an important, unsettled empirical question in organiza-

tions generally and in schools particularly (Eberts 2007; Lovenheim 2009).

Credibly identifying union effects on student achievement is difficult.

Unionized schools may differ systematically from nonunion schools in many

ways other than unionization. For instance, if urban schools are more likely to

unionize and to serve more-disadvantaged, lower-achieving students than non-

urban schools, unionized schools will have lower average student achievement.

However, we would not want to interpret this negative association as causal.

To help guard against such problems, we condition on a rich set of observable

student and school characteristics at the school-grade-year level. This leaves

the possibility of systematic differences between unionized and nonunion

schools in unobservable determinants of achievement.

To address this possibility, we avoid making comparisons across schools.

Instead, we use panel data to study only how changes over time in unioniza-

tion status are associated with changes in student achievement within school-

grade-subject clusters. This generalized difference-in-difference design

conditions out any stable, additive differences in unobservable achievement

determinants between school-grade-subject clusters. The remaining threat to

identification is the possibility of changes in time-varying unobservables

correlated with changes in unionization status. We use the multiple years of pre-

unionization data to address this concern. If soon-to-unionize schools were

experiencing differential achievement trends compared to other schools, this

test would detect a spurious preunionization “effect”of future unionization,

Unionization &Productivity in Charter Schools / 423