Forecasting the Recent Behavior of U.S. Business Fixed Investment Spending: An Analysis of Competing Models

In this paper, we evaluate the performance of a number of forecasting models of U.S. business fixed investment spending growth over the recent 1995:1-2004:2 out-of-sample period at multiple forecast horizons. The forecasting models are based on the conventional Accelerator, Neoclassical, Average Q, and Cash-Flow models of investment spending, as well as empirical models developed more recently by Barro (1990) using real stock prices and Lettau and Ludvigson (2002) using excess stock return predictors. None of the forecasting models stands out above the others at forecast horizons of 1-2 quarters. However, at forecast horizons beyond 2 quarters, forecast encompassing tests indicate that the Barro (1990) real stock price model contains most of the information useful for forecasting U.S. business fixed investment spending growth for 1995:1-2004:2 relative to the other models. This points to an important role for the stock market in tracking the recent course of U.S. business fixed investment spending. JEL classification codes: C22, C53, E22, E27