Stability of Sunspot Equilibria under Adaptive Learning with Imperfect Information
This paper investigates whether sunspot equilibria are stable under agents’ adaptive learning with imperfect information sets of exogenous variables. Each exogenous variable is observable for a part of agents and unobservable from others so that agents’ forecasting models are heterogeneously misspecified. The paper finds that stability conditions of sunspot equilibria are relaxed or unchanged by imperfect information. In a basic New Keynesian model with highly imperfect information, sunspot equilibria are stable if and only if nominal interest rate rules violate the Taylor principle. This result is contrast to the literature in which sunspot equilibria are stable only if policy rules follow the principle, and is consistent with the observations during past business cycles fluctuations.
Sunspot-driven business cycle models are popular tools to account for the features of macroeconomic fluctuations that are not explained by fundamental shocks. US business cycles in the pre-Volcker period are considered to be driven by self-fulfilling expectations, so-called ”sunspots” (see Benhabib and Farmer, 1994; Farmer and Guo, 1994). Those non-fundamental expectations are considered to stem from the Fed’s passive stance to inflation (see Clarida, Gali, and Gertler, 2000; Lubik and Schorfheide, 2004). Even recently, global financial turmoils in the last decade had historic magnitudes that could not be explained by fundamental reasons, and hence it is analyzed in models with sunspot expectations (see Benhabib and Wang, 2013; Gertler and Kiyotaki, 2015)