Generating Optimal Policies with the Murphy Model

R.D. Herbert & R.D. Bell

Department of Computing
Macquarie University
Sydney1

This paper is concerned with economic policy evaluation
using the Murphy model and control theoretic principles. The
approach taken here is to build a learning (or observing)
model which aims to behave in the same manner for policy
targets as the Murphy model to a change in policy
instruments. Optimal policies are then developed through the
learning model and applied concurrently to the Murphy
model. The emphasis is on policy generation necessary to
achieve a desired outcome and what are the implications of
such policies.

1 Introduction

This paper investigates the generation of policies for the Australian economy using the Murphy model and optimal control theory techniques. Control theory techniques have frequently been used for policy evaluation both in a single economy and multiple economy setting with models ranging from being small, linear and deterministic to large econometric models. See, for example, Currie & Levine (1993), Petit (1990), Fisher (1992), Holly & Hughes-Hallett (1989), and McKibbin & Sachs (1991). In this paper we apply control theory techniques to a non-linear rational expectations econometric model of the Australian economy.

One of the problems with optimal control has been that of time inconsistency (Kyland & Prescott, 1977) whereby forward looking agents with rational expectations cause ex post optimal policy generated by some policy reaction functions to become ex ante sub-optimal. This has lead to the consideration of precommitment, reneging and the

1 E-mail contact [email protected]

?R.D.Herbert & R.D.Bell 22 December 1994 Page 1