Kydland–Prescott Time Inconsistency of Optimal Policy

The Kydland–Prescott time inconsistency of optimal policy is the result that, in dynamic economies with rational private agents, the best plan for a benevolent policymaker is not dynamically consistent: when later periods arrive, re-optimizing yields a different action from the one originally planned. Finn Kydland and Edward Prescott demonstrated it in their 1977 Journal of Political Economy paper "Rules Rather than Discretion: The Inconsistency of Optimal Plans". The contribution, cited in the 2004 Nobel Memorial Prize in Economic Sciences awarded to both authors, transformed macroeconomics by showing that policy must be analyzed as a strategic interaction between the government and forward-looking agents, an idea close to game theory and to the commitment problems studied in mechanism design.

Before the 1970s, policy was analyzed with optimal control theory: a planner chooses a sequence of instruments to maximize a social objective subject to the equations of the economy. This method treats the economy's behavioral relationships as fixed. Lucas (1976) had already criticized the assumption of stable relationships, since private decisions depend on expected policy. Kydland and Prescott pushed the logic further. If agents form expectations about future policy, then today's choices by firms and households depend on the policy sequence they expect, and the planner is not choosing among exogenous outcomes. Control theory does not apply because the "system" being controlled anticipates the controller.

The authors show that in this setting, an optimal plan chosen at date 0 is generally not optimal when re-evaluated at date t > 0. At t, past decisions are sunk, and the policymaker can exploit the fact that private agents have already committed capital, wages, or contracts based on their expectations. The same policymaker, at date 0, would like agents to believe that the exploitation will not happen, because anticipated exploitation distorts earlier decisions. A promise not to deviate is not credible, since at the later date the optimal move is to deviate.

The result can be seen in a two-period model. In period 1 private agents choose an action x (for example, investment) knowing the policy that they expect in period 2. In period 2 the policymaker chooses a policy p (for example, a tax rate) after x is fixed. Let the planner's welfare be W(x, p). The commitment solution chooses p anticipating that x responds to p. The discretionary solution chooses p to maximize W given x already fixed, and agents, with rational expectations, anticipate this. The two solutions differ because under discretion the policymaker disregards the effect of p on x. In general welfare under discretion is lower than under commitment. Discretion yields a Nash equilibrium of the game between the government and the private sector, while commitment gives the government the role of a Stackelberg leader.

This distinction is a defining one: the discretionary equilibrium is time-consistent, because every decision is optimal at its own date, but is inefficient. The commitment plan is optimal but not time-consistent, which is why it needs an institution that makes it binding.

Capital taxation. Once capital has been invested, it is a fixed factor, so a tax on it distorts nothing at that date. A benevolent government that needs revenue is therefore tempted to levy a heavy tax on existing capital. Investors anticipate this, invest less, and the economy ends up worse off. The theory explains why constitutions and treaties restrict retroactive taxation and expropriation, and it is related to the capital levy debate in fiscal policy.

Flood control and disaster relief. Kydland and Prescott's original example is a government that would like to discourage building on a flood plain but, once houses exist, finds it optimal to build dams or provide relief. Residents rationally build in the flood plain, anticipating the relief.

Patents and public goods. Ex post, a government might prefer to grant free access to an invention, but ex ante this destroys the incentive to invent, so a patent system acts as a commitment device.

Monetary policy. The most influential application concerns inflation. Once wages have been set on the basis of expected inflation, a central bank can raise output by surprising workers with higher inflation. Because workers expect this, the equilibrium has higher inflation without higher output. The model is developed fully in the Barro–Gordon monetary policy model and underlies the study of inflation, the Phillips curve, and monetary policy institutions. Time inconsistency has also shaped the study of sovereign borrowing, where a government may be tempted to repudiate debt, as in the Eaton–Gersovitz sovereign debt model.

Kydland and Prescott concluded that policy should follow rules rather than discretion. A rule, such as a constitutional limit on taxes or a fixed money-growth target, commits the policymaker. Subsequent research identified further solutions:

1. Reputation. In repeated interactions, a policymaker may keep promises to protect a reputation. The punishment for deviation is a loss of credibility, supported by trigger strategies in the sense used in game theory.
2. Delegation. Society can appoint agents with preferences different from its own. Rogoff (1985) showed that a conservative central banker who places more weight on inflation than society does can reduce inflation bias.
3. Contracts and institutions. Walsh (1995) proposed a contract for central bankers that penalizes inflation. Independent central banks, inflation targets, and fiscal rules are institutional responses.
4. Timeless perspective. Woodford and others proposed that the policymaker adopt a policy that would have been chosen at some earlier date, which achieves a form of commitment without a binding rule.

The result reshaped central banking, leading to a global shift toward independent central banks and explicit targets from the 1990s. It is central to the theory of credibility and is applied beyond macroeconomics: to regulation, to the hold-up problem in contracts, and to international agreements. Critics observe that rules are rigid and cannot anticipate every contingency, which creates a tradeoff between flexibility and commitment explored by Rogoff and by Lohmann. Empirical evidence for inflation bias is mixed, and some economists argue that the inflation of the 1970s resulted from mistaken beliefs rather than from the incentives described. Nonetheless, the concept stands as one of the most influential ideas in modern policy analysis. A related literature in moral hazard examines how unobservable actions undermine contracts, while time inconsistency emphasizes how observable but unenforceable promises fail.

1. Kydland, F. E., & Prescott, E. C. (1977). Rules Rather than Discretion: The Inconsistency of Optimal Plans. Journal of Political Economy, 85(3), 473–491. https://doi.org/10.1086/260580
2. Barro, R. J., & Gordon, D. B. (1983). A Positive Theory of Monetary Policy in a Natural Rate Model. Journal of Political Economy, 91(4), 589–610. https://doi.org/10.1086/261167
3. Rogoff, K. (1985). The Optimal Degree of Commitment to an Intermediate Monetary Target. The Quarterly Journal of Economics, 100(4), 1169–1189. https://doi.org/10.2307/1885679
4. Lucas, R. E. (1976). Econometric Policy Evaluation: A Critique. Carnegie-Rochester Conference Series on Public Policy, 1, 19–46. https://doi.org/10.1016/S0167-2231(76)80003-680003-6)