A1543
Title: Noisy past and business cycles
Authors: Naoko Hara - Seikei University (Japan) [presenting]
Abstract: While standard macroeconomic models assume perfect information about the current and past states of the economy in real-time, accurate information often becomes available only after a long delay in practice. To assess the impacts of mismeasurement on economic activities and expectations, a structural vector autoregression model is estimated using different data vintages of the U.S. productivity over 1968Q1-2008Q1. The estimation results show that the measurement errors in early releases of productivity data significantly affect the underlying economic fundamentals and expectations in the short run. Furthermore, professional forecasts from the survey of professional forecasters positively respond to these errors, particularly to long-lived measurement errors that persist even after multiple data revisions. The heavily revised data may not yet fully reflect recent shifts in industrial structures and other key information that describe recent structural changes because the information is only incorporated into the data through comprehensive revisions generally conducted every five years. The long absence of such information in the data could explain why the long-lived errors continue to impact expectations. The findings suggest that this absence of information can influence the economy through expectations.
