Most analyses in macroeconomics take the rational expectations (RE) approach which assumes that economic agents have a great deal of knowledge about the economy. However, in applied work economists, who postulate RE, do not themselves know the parameter values and must estimate them econometrically. It, therefore, appears more natural to assume that agents in the economy face the same limitations on knowledge as do statisticians or econometricians. A more plausible approach to rationality is that the agents act like statisticians when making forecasts relevant to their economic decisions. They may do this by running regressions on actual time series data on economic variables. Such a process might lead them to make forecasting errors which they correct over time as new data becomes available. This type of analysis would determine whether policy prescriptions advocated under the RE approach continue to be desirable under this alternative approach of "adaptive learning".
The adoption of inflation targeting monetary policy by a number of major central banks in the past decade (including the UK) has led to widespread interest and research in the type of monetary policies which would be desirable for banks to adopt. This led to a development of micro-founded monetary models incorporating nominal rigidities which analysed optimal monetary policy as well as simple monetary policies for adoption by central banks.
However, initially the research conducted in this area made the assumption of RE. This assumption, especially in the context of monetary policy, is overly strong and unrealistic: it is well known that central banks face substantial uncertainty about the future and the decision lags inherent in the consequences of monetary policy make policy formulation especially difficult. With economists advocating ”good” monetary policies to central banks, at the very minimum, it seems sensible to check that these policies do not lead to disastrous consequences when central banks (and households or firms in the economy) have much less information about the economy than is typically done in the analyses which assume RE.
The adaptive learning approach has subsequently been applied to the field of monetary policy in macroeconomic models. The learning viewpoint has been very popular in analysing monetary policy design: the seminal papers of Bullard and Mitra (2002) and of Evans and Honkapohja (2003) examined the performance of Taylor type rules and optimal monetary policies respectively. Bullard (2006) and Evans and Honkapohja (2003b, 2009) provide surveys of this huge literature. Indeed researchers in both academia and central banks have come to accept the importance of learning in assessing policy design. Chairman of the Federal Reserve, Ben Bernanke (2007) argues “In sum, many of the most interesting issues in contemporary monetary theory require an analytical framework that involves learning by private agents and possibly the central bank as well.”
Researchers: George Evans and Kaushik Mitra
Related CDMA Working Papers:
0802 George Evans (Oregon and St Andrews) and Seppo Honkapohja (Bank of Finland and Cambridge) 'Expectations, Learning and Monetary Policy: An Overview of Recent Research'.
George Evans (Oregon and St Andrews) and Seppo Honkapohja (Bank of Finland and Cambridge)
'Robust Learning Stability with Operational Monetary Policy Rules'