“…6 In fact, we require only a small and quite natural deviation from rational expectations, based on the econometric learning approach increasingly utilized in macroeconomics. Recent applications include the design of monetary policy (Bullard and Mitra (2002), Evans and Honkapohja (2003), and Orphanides and Williams (2005a)), recurrent hyperinflations in Latin America (Marcet and Nicolini (2003)), US inflation and disinflation (Sargent (1999), Orphanides and Williams (2005b), Bullard and Eusepi (2005)), asset prices (Timmermann (1993), Brock and Hommes (1998), Bullard and Duffy (2001), 3 Fama (1984) demonstrates that, for this to happen, the variance of the risk premium must be greater than the variance of expected depreciation, and their covariance must be negative.These requirements do not appear to be supported empirically.…”