Title: Estimating financial frictions under learning
Authors: Jacek Suda - FAME|GRAPE (Poland) [presenting]
Patrick Pintus - Aix-Marseille University and CNRS - InSHS (France)
Burak Turgut - CASE - Center for Social and Economic Research (Poland)
Abstract: The collapse of the housing market in the second half of the 2000s triggered the credit crisis and impelled the U.S. economy into the Great Recession. The crisis underscored the importance of both expectations and the role that financial markets (and the housing market in particular) played in the economy and brought back the interest in understanding the links between these markets and the macroeconomy. We study the quantitative implications of the departure from the rational expectations for the financial crisis. We introduce constant-gain adaptive learning into a medium-scale DSGE model with credit-constrained agents. Using data on leverage and usual macroeconomic variables for the period 1975-2008, we estimate the model both under rational expectations and adaptive learning using Bayesian techniques. We find that the learning model has a better fit than its rational expectations version. We find that in the model with learning large negative collateral shock observed in the 2008Q3 had a significant effect, which is compounded by the imperfect information about this process. We assess whether alternative monetary policy rules of the central bank could prevent the housing market bubble from arising and in the aftermath of the crisis. The preliminary results suggest that neither average inflation targeting nor price level targeting would not have been effective in avoiding the crisis, but they do affect the recovery period.