CMStatistics 2017: Start Registration
View Submission - CMStatistics
B1033
Title: Nonparametric estimation of Lerner indices for measuring bank competition Authors:  Paul Wilson - Clemson University (United States) [presenting]
David Wheelock - Federal Reserve Bank of St Louis (United States)
Abstract: The Lerner index (L) is a well-established measure of market power at the level of individual firm and is determined by the spread between marketprice of a firm's output (P) and its marginal cost (MC) via $L=(P-MC)/P$. In banking studies, firm-specific Lerner indices are typically estimated using the observed ratio of total revenue to total assets to measure output price P, and computing MC from an estimated parametric,translog cost function. Recent studies have called the translog specification for banks' cost function into question. The local likelihood method is used to estimate the cost function for US bank holding companies while allowing for inefficiency. In addition, we add to total assets the credit-equivalent values of off-balance sheet items to obtain a more comprehensive measure of output price. We compare estimates obtained using one or both of these innovations with estimates using traditional approaches to estimate Lerner indices and show that previous studies have mis-stated the level of competition among US banks.