Systematic errors in growth expectations over the business cycle

Abstract

We document the fact that the growth forecasts made by professional forecasters in advanced economies exhibit systematic errors, and analyze how these errors depend on the business cycle state. On average over our full sample, forecasters overestimate GDP growth. However, this result masks considerable differences across business cycle states. Growth forecasts for recessions are subject to large negative systematic errors, while forecasts for recoveries are subject to small positive systematic errors. In contrast, forecasts for expansions do not exhibit systematic errors. Thus, there is evidence that forecasters try to issue forecasts which are unbiased conditional on being in an expansion, rather than forecasts which are unbiased overall. We also show that forecasters adjust their forecasts slowly around business cycle turning points. Furthermore, we show that cross-country differences in systematic forecast errors during expansions cannot be explained by changes in trend growth rates.