Corresponding author. Address: Department of Economics, University of California, Berkeley, 530 Evans Hall 3880, Berkeley, CA 94720-3880. Email: firstname.lastname@example.org. We gratefully acknowledge funding from the National Science Foundation, the Laura and John Arnold Foundation, and the Spencer Foundation. We are indebted to SEII research managers Annice Correia and Eryn Heying for invaluable help and support. Thanks also go to Isaiah Andrews, Pat Kline, Guido Imbens, Rick Mansfield, Chris Nielson, Stephen Raudenbush, Jesse Rothstein, Doug Staiger, and seminar participants at the 2014 All California Labor Economics Conference, the APPAM Fall 2014 research conference, the 2014 AEFP meeting, the 2015 ASSA annual meeting, the 2015 SOLE/EALE annual meeting, the 2015 NBER Summer Institute, the Federal Reserve Bank of New York, the 2015 Becker/Friedman Applied Microeconomics Conference, the University of Chicago Committee on Education Workshop, Brown University, and the University of Chicago Workshop on Quantitative Research Methods for suggestions and comments.
Conventional value-added models (VAMs) compare average test scores across schools after regression-adjusting for students’ demographic characteristics and previous scores. This paper tests for VAM bias using a procedure that asks whether VAM estimates accurately predict the achievement consequences of random assignment to specific schools. Test results from admissions lotteries in Boston suggest conventional VAM estimates are biased, which motivates the development of a hierarchical model describing the joint distribution of school value-added, bias, and lottery compliance. We use this model to assess the substantive importance of bias in conventional VAM estimates and to construct hybrid value-added estimates that optimally combine ordinary least squares and lottery-based instrumental variables estimates of VAM parameters. The hybrid estimation strategy provides a general recipe for combining non-experimental and quasi-experimental estimates. While still biased, hybrid school value-added estimates have lower mean squared error than conventional VAM estimates. Simulations calibrated to the Boston data show that, bias notwithstanding, policy decisions based on conventional VAMs that account for lagged achievement are likely to generate substantial achievement gains. Hybrid estimates that incorporate lotteries yield modest further gains. JEL Codes: I20, J24, C52.
The Author(s) 2017. Published by Oxford University Press, on behalf of President and Fellows of Harvard College. All rights reserved. For Permissions, please email: email@example.com