Staff Reports
Auto Credit and the 2005 Bankruptcy Reform: The Impact of Eliminating Cramdowns
October 2016 Number 797
JEL classification: D12, G21, K35

Authors:  Rajashri Chakrabarti and Nathaniel Pattison

Auto lenders were perhaps the biggest winners of the 2005 Bankruptcy Reform. Cars depreciate quickly, so borrowers often owe more than their car is worth. Prior to the Reform, these borrowers could reduce the principal on their auto loan to the market value of the car through a “cramdown” in Chapter 13 bankruptcy. The Reform prohibited cramdowns during the first two and a half years of an auto loan. This paper is the first to estimate the causal effect of this anticramdown provision on the price and quantity of auto credit. The authors use a novel empirical strategy that relies on the fact that eliminating cramdowns affected only one of the two types of consumer bankruptcy: Chapter 13. They exploit persistent historical variation in states’ usage of Chapter 13 generated by differences in local legal culture. Using difference-in-differences regressions, their empirical strategy compares pre-Reform and post-Reform auto loans across states with persistent historical differences in the share of bankruptcies filed under Chapter 13. They find that eliminating cramdowns decreased interest rates on auto loans in the average state by 15 basis points, with a larger decline in states where Chapter 13 is more common. The decline in interest rates occurs in the exact month that the Reform went into effect, and the authors rule out other aspects of the Reform as possible causes. Next, the authors estimate the effect on the quantity of auto credit. Using quarterly data from the FRBNY Consumer Credit Panel based on Equifax credit reports, they examine the effect of eliminating cramdowns on the number and size of new auto loans. The estimates show a small, negative, and insignificant impact on the number of new auto loans. The authors do find some evidence, however, that loan sizes increased among subprime borrowers.

Available only in PDF pdf 44 pages / 1,366 kb
Author disclosure statement(s)
Tools
E-mail Alerts