The first description of dynamic inconsistency remains the most poignant: you must bind me hard and fast, so that I cannot stir from the spot where you will stand me … and if I beg you to release me, you must tighten and add to my bonds. — The Odyssey
The introduction of dynamic inconsistency into formal economic analysis came much later, in the seminal paper by Strotz (1955-1956). Strotz modeled the problem of an individual choosing a plan of consumption for a future period of time so as to maximize utility at the present moment. He posed the following question: If this individual is free to reconsider his consumption plan at later dates, will she abide by it or disobey it even though her original expectations of future desires and means of consumption are verified? Strotz’s model provides the same answer found in Homer’s epic: the plan that is optimal at the present moment is generally one that will not be obeyed, or that the individual’s future behavior will be inconsistent with the optimal plan. For example, if this inconsistency is not recognized, the individual will behave as a "spendthrift," over-consuming in the future relative to the original plan. If the inconsistency is recognized, the rational individual will do one of two things. She may "precommit" her future behavior by precluding future options to ensure that it conforms to the original plan. Or, alternatively she may modify the chosen plan to take account of future disobedience, realizing that the possibility of disobedience imposes a further constraint on the set of plans that are attainable.
Kydland and Prescott (1977) moved the application of dynamic inconsistency squarely into the arena of public policymaking. Their work had a substantial impact on the field of political economy in part because it revealed a gaping flaw in traditional analyses of public policy. Importantly, it demonstrated policy failures that have nothing to do with the underlying motivation of a policymaker, who Kydland and Prescott assume to be interested only in maximizing social welfare. As a source of policy failures, the dynamic inconsistency framework differs fundamentally from other political economy models that rest on the idea that policymakers are driven by myopia, corruption, or some other electoral incentives that cause them to deviate from socially desired goals.
The Kydland and Prescott analysis showed (most famously) that when policymakers have discretion to select the monetary policy that is best in each period, the result is excessive rates of inflation without any reduction in unemployment. The reason is that forward looking economic agents form expectations about future policy choices and, future policymakers are not likely to remain committed to choices made by the present policymakers. Suboptimal policies thus arise in the dynamic inconsistency perspective because there is no mechanism to force future policy makers to take into consideration the effect of their policy on current policymakers. Unlike the dilemma facing Odysseus, the public policy sphere offers no failsafe mechanism to bind future policy makers "hard and fast." Yet Kydland and Prescott still conclude that monetary policy should be governed by rules rather than discretion. They recommend an institutional arrangement that legislates monetary rules that become effective only after a two-year delay. In their view, such an institutional arrangement would be costly to change and thereby accommodate price stability.
Fischer (1980) illustrates the concept of dynamic inconsistency in the realm of tax policy with a useful example. Consider a government policymaker who seeks to promote long-run economic activity; this objective requires tax revenues to finance a public good. The policymaker sets taxes in two sequential fiscal cycles, period 1 and period 2, and two tax instruments are available: taxes on labor income or taxes capital income. Knowing that a tax on capital will discourage investments in productive private capital, the policy maker in period 1 levies a tax on labor income, the lesser of two evils regarding distortions in private economic activity. Firms then make irreversible investments in capital and begin to produce in period 1. In period 2, the policy maker’s assessment regarding optimal tax policy changes. She now decides to tax capital income rather than labor income and thereby minimize the distortionary effects of taxation. In response to this revised policy in period 2, the labor supply increases and capital remains fixed. As this simple example illustrates, the choice of the optimal tax instrument is time-inconsistent. The best policy at one point in time is not the best policy at a future point in time, despite the fact that the government’s objective remains unchanged (in this case, to maximize economic activity over the two periods).
Of course, potential capital investors might anticipate the government’s future policy shift, which would temper their investments in period 1, despite the government’s announced policy not to tax capital. And obviously, economic and policy decisions are made over more than two periods. Governments would not be able to make this type of policy shift more than a few times before economic agents catch on: once burned, twice shy.
This Fischer pedagogical example and the caveats again reveal that the dynamic inconsistency problem derives from two fundamental elements: (i) desired policy choices have a temporal dimension, and (ii) government policies are unenforceable contracts. In political transactions third-party enforcement is not possible simply because the parties to the agreed-upon transaction can subsequently change the rules or renege without legal sanctions. In the tax policy example, suppose the government declares that its policy to tax labor and not capital is permanent. Capital investors, one party to this agreement has no legal recourse if the government were to renege on its promise. Even if the tax policy laws were enacted, legislators (current and future) would not be bound by past agreements; they could enact new laws revoking the old.
It is important to note that the pure theory of dynamic inconsistency abstracts from potential institutional sources of durability. The very absence of legal or institutional mechanisms to maintain long-term policy commitments stands behind the suboptimal policy choice. One line of research emphasizes that political institutions and rules emerge that make currently enacted policies difficult to alter. This was the novel insight exposited by Landes and Posner (1975). In their perspective, a host of elements of the political process can be understood as durability-enhancing mechanisms (Crain and Oakley 1995).
A second approach found in a variety of theoretical models follows more closely in the mold of Kydland and Prescott. These models, sometimes labeled "strategic fiscal policy," have in common the basic theme that a current political regime might use fiscal policy variables as a means of controlling policy choices by future regimes. Once again, the inability of present period voters (or their policy making representatives in the present political regime) to make binding contracts with voters in the next period (the future regime) creates the basic dilemma. The novelty in these models, however, is that policymakers respond by making fiscal choices designed lock-in, bind, or otherwise constrain the choices available to future political decision makers. In other words, a political majority today might use fiscal policy to lock-in a current policy that a future majority would predictably oppose.
It is worth reiterating that an important wrinkle in strategic fiscal policy analysis is that inefficient government policies are driven by the representative voter and not by pressure group demands for wealth redistribution. As long as the current government can affect some policy variable that enters into its successor’s decision calculus, it can influence to some degree the policy carried out by the successor government. In the process of binding future fiscal outcomes, however, the current government selects a different (and suboptimal) policy relative to what it would have preferred if it expected to remain in power. This occurs when the current and future regimes have different, or time-inconsistent, fiscal policy preferences. Three models illustrate this tradition, and surveys are provided in Perrson (1988) and Alesina (1988).
Perrson and Svensson (1989) develop a model in which a current government uses the level of the public debt as an instrument to control the level of spending by a future government. They construct a principal-agent model in which government (or the decisive voter) today is the principal and government in the next period is the agent. The intuitive example in their model posits an incumbent conservative regime that expects to be replaced in the next election by a liberal regime. The current regime will put in place a fiscal policy that features lower taxes and higher deficits than it would otherwise prefer in order to control the ability of the future liberal government to embark on large spending programs. As long as public debt enters negatively into the policy preferences of the future liberal regime, it responds to the conservative regime’s legacy of deficit financing by spending less than it otherwise would prefer.
Alesina and Tabellini (1990) develop a related model, the key difference being that succeeding regimes champion different spending priorities. For example, the current regime favors large defense budgets and minimal welfare budgets, and the future regime favors the opposite policy mix. Alesina and Tabellini (1990) also assume that public debt enters negatively into the preference functions of both regimes. In the case of time-inconsistent spending preferences, the current regime moves to constrain future spending (on the welfare programs it detests) by running a larger deficit than it would if it were assured of remaining in power.
Glazer (1989) develops a strategic model in which voters have a bias toward capital-intensive projects in the absence of durability enhancing institutions. Rational voters show a consistent bias in favor of capital projects, which they would oppose were the decision theirs to make individually in a private market environment. Glazer’s formal derivation is not repeated here; an intuitive understanding is straightforward and sufficient to illustrate the durability-motivated strategic fiscal choice.
Because current period voters cannot make contracts with next period’s voters, one possible strategy is to limit future policy options by constructing a long-lived capital project. This maneuver eliminates from the next period the option to renew or reject the services from the capital project. An inefficiently large public capital stock is predicted under majoritarian rules, irrespective of the cost efficiency of the capital project.
Glazer’s conclusion does not require any assumptions about the cost structure of the projects. Suppose that the benefits of two short-lived projects are equivalent to the benefits of one durable project, yet the costs of constructing two short-lived projects are less than the cost of building the long-lived (durable) project. Suppose further that the decisive voter in period 1 would like the services of the project in both periods. However, if the decisive voter in period 1 expects the short-term project to be rejected in period 2, he prefers the more expensive durable project in period 1. This would be the case if the benefits derived from the shortlived project over the two periods exceed the costs of the relatively more expensive durable project. In other words, the decisive voter selects a second-best outcome to prevent the worst-case outcome: no project in period 2. Glazer labels this source of capital bias a "commitment effect." Alternatively, suppose the decisive voter in period 1 has no strict preference for either the durable or the single-term project and that he expects the decisive voter in period 2 to choose the short-term project. If building the durable project is cheaper than building two successive short-term projects (i.e., there are economies of scale), the decisive voter in period 1 may select the durable project, even though the benefits are less than the cost, because it is less costly than the two short-term projects. This is what Glazer calls the "efficiency effect," which motivates a capital bias under collective choice as long as the difference in the benefits and costs of the single short-lived project exceed those of the durable project.
As this brief summary indicates, strategic fiscal models are based on the idea that choices in a given electoral period take into consideration expectations about preferences of decision-makers in succeeding periods. Fiscal variables such as spending, taxing and borrowing are used strategically as devices to control future choices if current policy makers expect the preferences of future policy makers to differ from their own. The process of binding future fiscal outcomes causes the current government to select second-best (and suboptimal) policies relative to what it would have preferred if it expected to remain in power.
Because these policy choices are second-best from the standpoint of the current regime, political conditions and the presence of institutions that enhance policy longevity should reduce the motivation to use fiscal variables strategically. Political conditions and institutions that facilitate policy durability predictably lower the incentive for strategic fiscal policy choices. In effect, strategic fiscal choices substitute for institutional sources of policy durability.
Crain and Tollison (1993) examine the tradeoff between strategic fiscal choices and institutional sources of durability using American state data. They find that such factors as term limits and the stability of the majority party controlling the state legislature reduce strategic behavior of the type described in the Perrson and Svensson (1989) and Alesina and Tabellini (1990) models. Crain and Oakley (1985) specify an empirical model to investigate implications of the Glazer model. Specifically, political conditions and institutions that facilitate policy durability predictably lower the capital intensity of government spending. Also using American state data, the findings indicate that institutions such as term limits, citizen initiative, and budgeting procedures significantly affect infrastructure spending across states. The results further indicate that political conditions such as majority party stability and voter volatility are systematically related to infrastructure differences across states. These two empirical studies indicate a fruitful common ground between models of dynamic inconsistency and the institutional models in the tradition of Landes-Posner. This ground remains largely unexplored, particularly in formal theoretical models, and represents a promising area for future research.
