Layoff costs and dynamic labor demand: Evidence from the U.S. Auto Industrys job security agreement

This paper studies an observable employment adjustment cost introduced in the domestic motor vehicle industry in 1990. The provision, known as the JOBS program, required …rms to pay full salary to a worker for each week spent on layo¤ beyond an allotment speci…ed in the union contract. The paper presents a model in which JOBS generates an incentive to defer layo¤s: a …rm wants to safeguard its allotment of layo¤ weeks for that contract in case future vehicle demand deteriorates. The paper tests these implications on plant-level data. It …rst uses the model to motivate a reducedform regression for the number of weeks per month a plant’s workers spend on layo¤. The reduced form suggests that JOBS deters, on average, at least one-quarter of a week of layo¤per month. The e¤ect is somewhat larger at plants where the allotment appears more likely to bind. The paper next uses the reduced form as a bridge to structural estimation. It carries out indirect inference to estimate the structural parameters and calculate the model’s non-linear optimal layo¤ policy. The structural estimates suggest more modest e¤ects of the JOBS agreement on weekly layo¤s. I am particularly grateful to Charles Brown, Michael Elsby and Matthew Shapiro for their guidance and patience. A conversation with Martin Zimmerman triggered my interest in the JOBS program, and throughout the process, Martin and Dan Luria have served as very valuable guides through the U.S. auto industry. Part of this project was done while on a summer internship at the Federal Reserve Board. I am grateful to seminar participants at the Board and in particular to Alessandro Barbarino, Bruce Fallick, Andrew Figura, Charles Fleischman, Claudia Sahm, John Stevens, and Daniel Vine for their input. Thanks are also extended to Ruediger Bachmann, John Bound, John DiNardo, Chris House, John Laitner, Mel Stephens, and Dmitriy Stolyarov for helpful conversations. All errors are my own. E–mail address for correspondence: [email protected].