CREDIT AND ECONOMIC ACTIVITY: CREDIT REGIMES AND NONLINEAR PROPAGATION OF SHOCKS.(Critical Essay)

Publication: Review of Economics and Statistics

Publication Date: 01-MAY-00

Bookmark this article Print this article Link to this article Email this article Digg It! Add to del.icio.us RSS

COPYRIGHT 2000 MIT Press Journals

I. Introduction

In much of the recent literature on financial market frictions, credit acts as a nonlinear propagator of shocks. For example, Bernanke and Gertler (1989) construct a model in which the balance-sheet conditions of firms can amplify fluctuations in output and in which negative shocks are likely to have a greater effect than positive shocks. Azariadis and Smith (1998) develop a model in which it is possible for the economy to switch back and forth between a Walrasian regime and a credit-rationing regime. Blinder (1987) develops a model in which monetary shocks have different effects when the economy is in a credit-rationing regime than at other times. In all of these models, credit conditions need not be an important source of shocks but are, nonetheless, an important propagator of shocks. Interestingly, these models imply nonlinear dynamics such as regime switching and asymmetric responses to shocks.

Empirical evidence of the importance of credit conditions for aggregate economic fluctuations is mixed.(1) Ramey (1994) finds that credit variables such as credit velocity and the loan-to-securities ratio provide little additional predictive content for output above and beyond that contained in money. Alternatively, Stock and Watson (1989), Friedman and Kuttner (1992, 1993), and Kashyap, Stein, and Wilcox (1993) find evidence that other proxies for credit conditions such as the spread between commercial paper and treasury bills--or the fraction of bank loans as a fraction of total short-term external finance--do have predictive content for economic activity.(2) Perhaps one reason for the mixed evidence is that it is based almost entirely on linear regressions or linear vector autoregressions (VAR). Standard linear time series may have difficulty detecting credit's role as a nonlinear propagator of shocks as envisioned in much of the recent theoretical literature on the role of credit.

In this paper, we employ nonlinear time-series analysis to examine credit's role as a nonlinear propagator of shocks. Specifically, we test for and estimate a threshold vector autoregression that changes "structure" if credit market conditions cross a critical threshold. Here, credit regime changes can be endogenous as shocks to other variables, such as the Fed funds rate, can result in a switch in regimes. Using nonlinear impulse-response analysis, we attempt to isolate the relative effects of shocks and the nonlinear structure on the time-series behavior of output. Among the findings, it appears that shocks during a "tight" credit regime have a larger effect on output than do shocks in the "normal" regime. Furthermore, there is evidence that contractionary Fed funds shocks have larger effects than do expansionary shocks. Finally, we calculate nonlinear analogs of historical decompositions to examine the role that tight credit regimes played in the propagation of macroeconomic fluctuations.

The analysis in this paper is related to that in McCallum (1991) in that he estimates a threshold model in which the coefficients on money in an output equation change depending on credit conditions. However, we examine three alternative measures of credit conditions that have been the focus of much of the recent analysis on the role of credit for fluctuations. Second, we adapt the simulation methodology proposed by Hansen (1996) in order to conduct proper inference. Third, by estimating a threshold vector autoregression, we allow switching into and out of the tight credit regime to be endogenous.

II. Empirical Methodology: Testing and Estimating Threshold Models

In this paper, the separate role that credit may play as a nonlinear propagator of shocks is captured by a threshold vector autoregression (TVAR) model. A TVAR is a relatively simple and intuitive way to capture nonlinearity such as regime switching, asymmetry, and multiple equilibria (which, in a time-series context, might be reflected in multimodal stationary distributions) implied by theoretical models of credit and macroeconomic activity. In addition, a TVAR allows credit regimes to switch as a result of shocks to other variables besides credit, so that credit regimes are themselves endogenous.

Consider the following...

Read the full article for free courtesy of your local library.