Predicting growth in US durables spending using consumer durables-buying attitudes

Abstract

This study employs the Michigan Surveys of Consumers (MSC) data on durables-buying attitudes to predict monthly growth in durables spending in real time. We specify and estimate a univariate moving-average (MA) and five augmented-MA models that utilize past information in consumer durables-buying attitudes (calculated using all survey responses and the responses of participants in three income categories). We show that the augmented model that utilizes durables-buying attitudes of consumers in both bottom and top income categories produces the most informative forecasts for 2008-2019. These forecasts show over-prediction and yet are directionally accurate under asymmetric loss. We further show that the forecast errors are generally orthogonal to changes in both interest rates and income growth, and to changes in important relevant MSC variables. In concluding, we first note that consumer durables-buying attitudes disaggregated by income help detect useful information for predicting growth in durables spending, and then discuss the policy implications.