Website CV Paper
Fields
Econometrics, Macroeconomics
Job market paper
Are Professional Forecasters Inattentive? Evidence on Alternative Sources of Information Rigidity
Since expectations about the future shape key macroeconomic decisions, the question of what drives these expectations has motivated a large and growing literature. Despite their importance, a recurring finding is that forecasts are excessively sticky, as economic agents often do not update their expectations. In my model, an unchanged forecast can arise for three reasons: (i) the new information led to a small change in belief that was rounded to zero; (ii) the agent was attentive but deliberately decided to cling to their prior beliefs; and (iii) the agent was purely inattentive. To allow for these three possibilities to coexist in the population in a unified framework, I develop a triple-hurdle zero-inflated forecast revision model and estimate the weights placed by agents on their prior beliefs. A distinguishing feature of this model is that the sample separation information between the inattentive and attentive zero revisions is unobserved, and hence the model is estimated using the marginal likelihood approach with identification coming from cross-sectional, cross-variable, and temporal information. I study Blue Chip forecasts from 1986 to 2024 observed over 24 successive monthly horizons. In addition, I supplement the analysis with individual SPF and the Federal Reserve's Tealbook forecasts. I show that failing to distinguish among the three sources of zero revisions substantially distorts estimates of the anchoring of expectations. In particular, the weights on prior beliefs for GDP growth, CPI inflation, and unemployment are overstated when every zero is treated as a genuine zero revision, and understated when every zero is attributed to inattention.