Why can’t professional macroeconomic forecasters predict recessions?

The professional forecasters’ inability to anticipate macroeconomic recessions is well documented. The literature has found that aggregate or consensus forecasts are too optimistic before downturns and too pessimistic before recoveries. This paper explores whether this result also holds at the individual level or is the result of an aggregation bias. Using a Spanish survey of professional forecasters conducted by Funcas, I find that individual forecasters are indeed too optimistic before recessions. The reason is not that they become inattentive when the economy is in good shape. Instead, they put too much weight on the most recent developments when producing their forecasts. The analysis of their forecast revisions reveals that better-than-expected data makes some forecasters to revise their forecasts upwards too much. These revisions raise the consensus forecast and trigger a herd behaviour by other forecasters, who also revise up their forecasts. Both factors lead to subsequent negative forecast errors, especially when a recession occurs. Consequently, professional forecasters could improve their forecasting performance by reacting less aggressively to the latest data releases and by avoiding inefficient herding.I am grateful to Funcas, and in particular to María Jesús Fernández Sánchez, for providing the dataset and for very helpful technical assistance. I thank David Gabauer, Ayaz Zeynalov and participants to the 9th International Conference in Economics (Econworld 2019, Seville, Spain) for interesting comments and discussions. I am also grateful to Susana Tena Nebot for excellent research assistance