Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Coibion and Gorodnichenko (2015) provide a useful framework to test the null hypothesis of full-information rational expectations against two popular classes of information rigidities, sticky information (SI) and noisy information (NI). However, the observational equivalence of SI and NI in average forecast errors gives no power in the test for one against the other. We identify the source of information rigidities by estimating the equations for the average forecast errors and variance of forecasts. The results show the importance of both SI and NI, but favor a type of NI in which agents can quickly learn the underlying state.