What do chain store sales tell us about consumer spending?

In recent years, the sales reports of major retail chains have received increasing attention as timely indicators of consumer spending. Despite this attention, a close review of the literature on chain store indexes as macroeconomic indicators reveals that there is no literature! This paper fills this gap, showing how chain store data fit into the broader issues of how we measure and forecast consumer spending. We describe the linkages between chain store data and official measures of consumer spending, highlighting the key seasonality and pricing issues. We then present a battery of in-sample and out-off-sample tests to determine what are the best models for monthly forecasts of retail sales and personal consumption expenditures. While our results question the way chain store data are sometimes used in forecasting, we find strong evidence that the chain store data can significantly improve forecast accuracy, and we show that the best models combine several consumer-related variables with both of the leading chain store indexes. We conclude with some practical advice on the \\"do's\\" and \\"don'ts\\" of consumption forecasting.