Imperfect information about financial frictions and consequences for the business cycle

Abstract

Reproduction permitted only if source is stated.ISBN 978-3-95729-134-9 (Printversion) Non-technical summary Research QuestionFinancial linkages between savers and borrowers are exposed to agency problems which may arise for several reasons. Imperfect information about an investment project can cause moral hazard behavior of borrowers if lenders do not have sufficient information about them. Our aim is to introduce imperfect information in a contract related to a limited enforcement problem. The way agents process and update new information is key for the optimizing behavior of the individual and hence for aggregate macroeconomic variables. In this paper we ask what the consequences for the business cycle are when agents have to learn about the behavior of banks. ContributionThe idea of our paper is to implement imperfect information into the banking sector of an otherwise standard New-Keynesian model, in which limited enforcement creates an agency problem. In our setting, economic agents learn about the size of changes in the diverting behavior of bankers. Everything else in the economy, i.e. both the structure and the parameter values, is known. Then we contrast the learning approach to full information rational expectations and analyze their respective roles for the business cycle. ResultsFor the period during which agents learn about the economy, the whole economy exhibits higher volatility and different outcome paths for macroeconomic variables compared with rational expectations. In particular, the introduction of imperfect information amplifies the responses of variables as the leverage ratio in the economy is higher for a longer period. Output is also higher due to a boost in investment before it undershoots the rational expectations benchmark outcomes. This goes hand in hand with an increase in uncertainty and higher volatility of all macro-variables. Compared with rational expectations output, investment and the leverage ratio display an increase in their respective volatility of between 1% and 8%. Output becomes even slightly more persistent in the learning case. Nichttechnische Zusammenfassung AbstractIn this paper, we discuss the consequences of imperfect information about financial frictions on the macroeconomy. We rely on a New Keynesian DSGE model with a banking sector in which we introduce imperfect information about a limited enforcement problem. Bank managers divert resources and can increase the share of diversion. This can only be observed imperfectly by depositors. The ensuing imperfect information generates a higher volatility of the business cycle. Spillovers from the financial sector to the real economy are higher and shocks in general are considerably amplified in the transition period until agents' learning is complete. Volatility and second-order moments also display an amplification under the learning setup compared with the rational expectations framework.

Learning, robust monetary policy and the merit of precaution

“…A number of studies introduce financial frictions in New Keynesian models. 18 Such frictions tend to amplify the fluctuations in inflation and the output gap, especially when private agents adopt learning behaviors, implying that monetary policy must be more aggressive in response to inflation shocks than under RE (Caputo, Medina, and Soto 2011;Rychalovska, Slobodyan, and Wouters 2015;Hollmayr and Kühl 2016). These results suggest that the interactions between learning and robust monetary policy in the presence of financial frictions could be quite different from those in the absence of such frictions.…”

Learning about banks’ net worth and the slow recovery after the financial crisis

“…The challenging part is to introduce this pattern as well as possible into the model. We therefore rely on the approach by Cogley, Matthes, and Sbordone (2015) and Hollmayr and Kühl (2016), which is rather intuitive and fits the description of the behavior of agents pretty well. The upside to this implementation is that learning stays very close to full information rational expectations.…”