Economic analysis is typically conducted within a single policy regime. However, it is easy to imagine situations where policy changes are implemented by policy makers. Fiscal authorities consider changes to tax or spending policies and monetary authorities act more aggressively in recessions. A number of policy changes are announced in advance and there can be long lags involved in a change of (especially) fiscal policy for eg. the process of changing taxes is very cumbersome.
When these changes are discussed in the media, they will be anticipated by agents which will influence their behaviour before the implementation of the proposed policy change. Such behavioural changes are usually analysed in economic models by assuming a model of expectations formation for agents. The hypothesis of rational expectations (RE) has been the main paradigm for expectation formation in macroeconomics for more than three decades.
RE does serve as a useful benchmark but it is often recognized that the assumption is very demanding since it assumes complete knowledge of the systematic aspects of the economy. The RE assumption is particularly strained when considering the effect of policy changes (generally large structural shifts). Agents need complete understanding of the underlying economic structure including structural parameters, both before and after the policy change, and be able to incorporate this rationally in their decision making.
Since the 1990s, an alternative approach has developed which assumes that economic agents (consumers, firms, and policy makers alike), in situations of incomplete understanding, try improving their knowledge by using methods of scientific inference. This assumption that agents engage in learning behaviour has been incorporated into macroeconomic theory (eg. Sargent (1993) and Evans and Honkapohja (2001)).
The learning viewpoint has recently met with tremendous success in analysing monetary policy design; see Desirability of Monetary Policy Rules under the criterion of Determinacy and Adaptive Learning.
While adaptive learning has proven to very popular in analyzing desirability of monetary policies, the existing frameworks are not suited for the study of announced policy changes since they assume a given economic structure.
The aim in the project is to rectify this weakness of the adaptive learning approach and propose a new methodology whereby announced (anticipated) policy changes lead to immediate changes in the behavior of agents even before the implementation of the proposed change (in the same spirit as they do in the RE framework).
This methodology is potentially applicable to various frameworks. A start has been made by Evans, Honkapohja, and Mitra in Journal of Monetary Economics (2009).
An ESRC grant “Macroeconomic Policy Changes and Adaptive Learning” (Ref RES-062-23-2617 with an award value of £259,000 at Full Economic Costing starting October 2010 for a period of three years) has been obtained by Kaushik Mitra as the Principal Investigator to study this topic in detail in collaboration with George Evans and Seppo Honkapohja.
Researchers: George Evans and Kaushik Mitra
Related CDMA Working Papers:
0717: George Evans, Seppo Honkapohja and Kaushik Mitra, (2009), 'Anticipated Fiscal Policy and Adaptive Learning', in Journal of Monetary Economics, vol.56 (7), pp.930-53.
