Report

Consumer sentiment and household expenditure: reevaluating the forecasting equations

Abstract: This paper reestimates the simple forecasting regressions in Carroll, Fuhrer, and Wilcox (1993) (CFW), which investigate the predictive power of consumer sentiment for consumption growth. Durability in the consumption categories analyzed implies that the error term may be distributed as an MA(1), indicating that ordinary least squares (OLS) estimation is inappropriate when variables lagged one period are used in the forecasting equation. I reestimate the forecasting regressions using nonlinear least squares (NLLS), explicitly accounting for a moving average error structure. In addition, I include two financial indicators as controls in the forecasting regression. These changes produce notable qualitative differences with the results obtained in CFW, and with my own results using OLS. In particular, using NLLS and financial controls, consumer attitudes appear to have little incremental forecasting power for categories of consumption other than motor vehicles.

Keywords: Consumer behavior; Forecasting; Consumption (Economics);

Bibliographic Information

Provider: Federal Reserve Bank of New York

Part of Series: Research Paper

Publication Date: 1996

Number: 9636