Academic journal article Economic Inquiry

Consumption Spending and the Paper-Bill Spread: Theory and Evidence

Academic journal article Economic Inquiry

Consumption Spending and the Paper-Bill Spread: Theory and Evidence

Article excerpt

I. INTRODUCTION

One of the more interesting recent findings in empirical macroeconomics is that the paper-bill spread, the difference between the yields on six month prime commercial paper and six month T-bills (henceforth, the "spread") helps predict a number of important macroeconomic series in post-war U.S. data. For example, Stock and Watson [1989a], after an exhaustive examination of 55 possible variables (themselves selected from a larger initial field of nearly 280 variables), include the spread as one of only seven variables in their new index of leading economic indicators. Friedman and Kuttner [1992, 1993a,b] show that the spread helps forecast real output growth in three and four variable systems including lags of output growth, the spread, inflation, and either credit or any of several measures of the money stock or mid expansion federal expenditures.(1) Finally, Bernanke [1990], Bernanke and Blinder [1992], and Kashyap, Stein, and Wilcox [1993] show that the spread helps forecast a whole range of key series in addition to output growth. These series include, among others, the unemployment rate, the level of employment, personal income growth, retail sales, the inflation rate, and consumption spending.(2)

Aside from general discussions as to why the spread might help predict output growth or the overall state of the economy,(3) these studies have not provided any theoretical insights as to why the spread should help predict such a lengthy list of important variables. In particular, no explanation has been advanced as to why the spread should help forecast consumption growth. In this paper, I present a theoretical model of the relationship between consumption growth and the spread. The purpose of the paper is threefold: First, I develop an intertemporal optimizing model of consumption choice and portfolio allocation which explains why, under certain conditions, lags of the spread should help predict growth in consumption spending. Second, I test the model empirically by estimating both reduced form and structural models of consumption growth which are implied by the theoretical model. The model appears to be consistent with post-war U.S. data. Finally, the model developed below suggests that the rule of thumb model of consumption studied by Campbell and Mankiw [1989, 1990] and Cushing [1992] is misspecified. Thus, the third purpose of the paper is to include the spread in an empirical rule of thumb model. I show that doing so reduces the estimated relative importance of non-optimizing behavior.

The paper is organized as follows: In Section II, I update and extend previous empirical evidence on the predictive power of the spread for the change in consumption. In Section III, I develop an intertemporal optimizing model in which consumption growth depends on the current one period real interest rate and the current spread. This model shows that if the spread is serially correlated, the significance of lags of the spread for predicting consumption growth simply reflects the optimal response of consumption growth to contemporaneous changes in the spread.

Sections IV-VI present the empirical analysis. Section IV reports results from an extension of the model of consumption growth and asset returns pioneered by Hansen and Singleton [1983]. The extension, which is based on the model of Section Ill, permits household utility to depend on private holdings of government debt. The empirical findings generally support the model. In Section V, I use an instrumental variables method with lags of the spread included among the instruments to estimate a model in which consumption growth depends on the private net real interest rate and the spread. Since the spread is strongly autocorrelated and the coefficient on the current spread is strongly significant, the empirical results are consistent with the theoretical model. In Section VI, I reconsider the "rule of thumb" model of Campbell and Mankiw [1989, 1990], Cushing [1992], and others. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.