Harold Zurcher and Paul Volcker—The Minutes of a Meeting that Never Was

Abstract

We develop a method to robustly estimate dynamic models even if their discount factor is "poorly identified," or not identified at all: We use homotopy continuation to trace the solution to the first-order conditions of the MPEC formulation of the estimation problem (Su and Judd, 2012)—and thus the estimates of the model parameters—as a function of the discount factor. We apply this method to the bus engine replacement model of Rust (1987), a.k.a. the "Harold Zurcher model," whose discount factor is believed to be poorly identified. Instead, we find that it is well identified, with estimates that are greater than one in various model specifications. We demonstrate that this does not contradict economic theory given the prolonged periods of negative real interest rates in the US during the sampling period, 1974 to 1985. Moreover, we implement a natural experiment: In August 1979, Paul Volcker became chairman of the US Federal Reserve, immediately taking drastic steps to curb inflation, which led to a significant rise in nominal and real interest rates. We introduce an unanticipated structural break in discounting, finding that the estimate of Harold Zurcher’s discount factor decreases significantly one month after Volcker took office.

Keywords: Structural estimation, homotopy continuation, discount factor, dynamic programming, structural break, great inflation.