土耳其中央银行研究报告.pptVIP

* 28 October 2010 Discussion on “The Financial Accelerator under Learning and the Role of Monetary Policy” by Caputo, Medina and Soto CEPR/ESI 14th Annual Conference Harun ALP Central Bank of Turkey * The Paper Financial accelerator on the demand side of credit market is modelled by using BGG (1999) framework. Borrow at a premium over the risk free rate, which is a function of the leverage of the borrower. Departure from Rational Expectation assumption. Adaptive Learning: Agents in economy act like econometricians. Past data is discounted when agents update their expectations. Combination of the two elements generates a sizable drop in output and asset prices in response to a negative shock. * Consider negative productivity shock under alternative monetary policy rules responding asset prices. The expectations formation mechanism endogenously generates a significant deviation of asset prices from their fundamental values. Responding aggressively only to inflationary pressures is still efficient in this environment. The Paper * Calibration Learning itself introduces a lot of inertia. Milani (2005, 2007): In a New Keynesian model with learning, the estimated degrees of habits and indexations are close to zero. Parameter values for indexations may be too high. Estimate the model to see whether financial accelerator and learning are accepted by data. * Instrument Rules React to the level of asset prices or change in the asset prices. Some fluctuations in asset prices are efficient in the presence of technology shocks. React to the deviations of asset price from its fundamental value (flexible price with no financial frictions) Technology shocks lead to changes in natural rate of interest. React to the natural rate of interest. * Instrument Rules Optimized simple instrument rules under an ad-hoc loss function * More formal welfare analysis Consider optimal policy as a benchmark and look at whether simple instrument rules close to opti