RDP 8801: Time-Consistent Policy: A Survey of the Issues 1. Introduction
January 1988
- Download the Paper 640KB
The impact on the economy of a change in macroeconomic policy depends partly on the expectations surrounding the implementation of the policy. This point has been recognized for many years, but it has become more relevant in the economics literature with the widespread use of models containing forward-looking agents.
There are two main reasons why a policy announcement may not be credible[1]: the policy may be seen to be politically or economically infeasible; or the policy may be feasible but it may be believed that the policymaker has no incentive to carry through with the policy.
A policy of the second type is referred to in the literature as a “time-inconsistent” policy and is the focus of this paper. Specifically, a “time-inconsistent” policy is a policy which future governments (or the current government in future periods) will have an incentive to change in the future. If agents understand the incentives of the policymaker, then a time-inconsistent policy is not credible. A time-consistent policy is credible because the private sector understands that the policymaker has no incentive to change the policy once it is announced. In an important paper, Kydland and Prescott (1977) showed that the optimal policy (found by optimal control methods) in a model with forward looking agents would generally be time-inconsistent and therefore would not be credible to the private sector. If the private sector does not believe the policy, the desired effect of announcing the policy will not be realized. This is a stunning result and has led to a large and growing literature on the problem of policy formulation when the government is seen as playing a “game”[2] against forward looking agents.
A standard illustration of the problem of time-inconsistent policy is the question of patents. To stimulate research into discovering new products, it is beneficial to promise patent rights over any discoveries. Once a discovery is made it is then optimal to tax away the monopoly profit that the patent generates[3]. A similar argument is often made over the provision of incentives for resource exploration. Once a discovery is made, the policymaker has an incentive to tax away the economic surplus resulting from the discovery. This insight also applies to the formulation of macroeconomic policy.
The purpose of this paper is to draw out the main ideas in the recent theoretical literature on macroeconomic policy. One of the important directions of research has been the application of game theory to the problem of setting macroeconomic policy. Once the setting of policy is regarded as a repeated strategic interaction between policymakers and the private sector, understanding the issue of “reputation” in policymaking becomes important. The approach taken in this paper leans heavily on the game theoretic developments.
In Section 2, the time-consistency problem is illustrated in a simple closed economy model. It is shown why, in a deterministic world with forward looking agents and complete information on the objectives of economic actors and the structure of the economy, the optimal policy rule for a government may not be feasible because it will not be believed by the private sector. Section 3 is addressed to the question: Can the optimal policy be made time-consistent and therefore credible under some circumstances? It is shown how credible threats and repeated interaction between the policymaker and the private sector can lead to the optimal policy being believable. It is also illustrated that the introduction of various types of uncertainty can help to make the optimal policy believable. This section illustrates that in a rational expectations model it may be desirable for a policymaker to introduce noise into the policy implementation process. Section 4 shifts the focus of the paper away from the problem of a single government interacting with a private sector, to a model of competing political parties. The allowance for elections adds an interesting twist to the problem of time-consistency. A conclusion and suggestions for future research are presented in Section 5.