سیاست های پولی و تورم تداوم اختیاری

Rational expectations models of staggered price/wage contracts have failed to replicate the observed persistence in inflation and unemployment during disinflationary periods. The current literature on this persistency puzzle has focused on augmenting the nominal contract model with imperfect credibility and learning. In this paper, I re-examine the persistency puzzle by focusing on the discretionary nature of monetary policy. I show that when the central bank is allowed to re-optimize a quadratic loss function each period, imperfect credibility and learning, even in the absence of staggered contracts, can generate a significant amount of inflation persistence and employment losses during a disinflationary period.

The New Keynesian sticky price/wage framework has been criticized for its empirical failure to generate sufficient inflation and output persistence.1 This so-called persistency puzzle has led researchers to more closely examine the theoretical problems of generating observed output and inflation dynamics. In particular, a group of models has emerged attempting to improve the empirical fit of the New Keynesian framework by augmenting it with imperfect credibility and learning.2 This literature asserts that the persistency puzzle arises because of the empirically questionable assumption of perfect policy transparency, and not because of any intrinsic shortcomings of the contract structure. The idea is that if policy suffers from imperfect transparency and credibility, then the public is forced to learn the true intentions of the monetary authorities by observing real outcomes. It is this learning process that is likely to generate additional persistence in inflation and output dynamics. For example, Erceg and Levin (2001) use a contract model similar to Taylor (1983) and argue that by including imperfect information and learning they can account fairly well for the dynamics of inflation and output following the Volker-disinflaition in 1979. The problem with these learning models is that they typically assume that the monetary authorities do not behave optimally, but instead simply follow an exogenously determined Taylor rule. As a consequence, the explicit relationship between discretionary monetary policy and inflation and output dynamics is disregarded.3 The main reason for sidestepping the central bank's optimization problem is the difficulties of modeling optimal policy in an environment of imperfect information and learning. However, this is not without consequences. For instance, the common specification of the Taylor rule includes lags of the central bank's control instrument. This exogenously assumed policy inertia is not only likely to create inflation persistence in itself but also makes it harder for agents in the economy to learn the true nature of the current regime and thus reinforce the persistence in both output and inflation. The question then is how much of the persistency is really generated endogenously through imperfect credibility and transparency and how much is exogenously assumed by eliminating discretionary monetary policy in favor of an appropriately specified Taylor rule.4 In this paper, I re-examine the persistency puzzle by focusing on the discretionary nature of monetary policy. In particular, I study how credibility and transparency affect inflation and unemployment dynamics through discretionary policy as opposed to nominal rigidities. Using a modified version of Barro and Gordon's (1983) neoclassical model of discretionary policy, I show that when the central bank is allowed to re-optimize a quadratic loss function each period, imperfect credibility and transparency, even in the absence of staggered price/wage contracts, can generate a significant amount of inflation persistence and employment losses during a disinflatonary period. Hence, I identify an additional channel, unrelated to nominal rigidities, through which credibility and transparency affect inflation and unemployment dynamics.5 The relationship between optimal policy and inflation dynamics has been discussed extensively in the past. In their seminal paper, Kydland and Prescott (1977) demonstrate that discretionary policy can be dynamically inconsistent and therefore produce excess inflation. Barro and Gordon (1983) extend the framework and show that dynamically inconsistent monetary policy predicts a positive relationship between inflation and the natural rate of unemployment. Thus, any persistence in inflation should be directly related to persistence in the natural rate. Ireland (1999) set out to test this hypothesis using quarterly U.S. data. Indeed, he finds a strong positive long-run relationship between inflation and unemployment. However, his results also indicate that the Barro–Gordon model (henceforth BG model) fails to explain the short-run relationship between the two time series. In this paper I suggest that the latter result most likely arises for two reasons. Firstly, the preferences of the monetary authorities changed over the sample period. For example, most observers would agree that the Volker-disinflation reflected a change in the Federal Reserve's objective function. Secondly, even if regime shifts were taken into account, the BG model predicts that a disinflation induced by a change in the preferences of the monetary authorities is immediate and completely costless in terms of employment losses. This, of course, is due to the assumption that the private sector has perfect information regarding the preferences of the monetary authorities. Taking these two concerns into account, this paper analyzes a disinflationary regime shift in an environment where monetary policy is discretionary, but where the public has imperfect information regarding the true preferences of the monetary authorities. I show that in this setting the BG model does predict persistence in unemployment and inflation following a regime shift, and that the degree of this endogenous persistence is strongly influenced by the level of credibility and transparency. The intuition for this result is straightforward. Assume that the central bank announces that it is going to be tough on inflation. If the central bank lacks credibility, the announcement will have little effect on inflation expectations. Since the monetary authorities do not believe that they can credibly manipulate inflation expectations, they perceive a quick disinflation to be highly costly. Consequently, in order to limit short-run employment losses the monetary authorities find it optimal to reduce inflation gradually.6 The problem is that since policy is not perfectly transparent the public assesses the likelihood of a regime shift from observing real outcomes. Hence, a small reduction in inflation provides little information in this regard and is likely to only affect inflation expectations marginally. This, of course, forces the monetary authorities to continue its gradual disinflation. Thus, the failure of discretionary policy to incorporate the impact of credibility and transparency on the dynamics of inflation expectations leads to a dynamically inconsistent disinflation characterized by excessive employment losses and persistent inflation. The paper is structured as follows. Section 2 presents the original BG model. Section 3 introduces incomplete information and learning and analyzes, through simulations, the nature of the relationship between inflation and unemployment persistence and credibility and transparency. Section 4 reviews the work of Ireland in light of the Volcker disinflation, and shows how the model presented in Section 3 can potentially explain the empirical failure of the simple BG model. Section 5 concludes.

The neoclassical proposition that monetary policy does not matter rests on three assumptions: (1) fully flexible prices and wages, (2) rational expectations, and (3) perfect information. The New Keynesian literature relaxes the first assumption by introducing friction in the price/wage setting process, which allows monetary policy to affect the real economy in the short-run. However, several researchers have pointed out that this framework fails to account for the observed persistency in inflation and real activity during disinflationary periods. Much of the current literature on this so-called persistency puzzle has focused on augmenting the nominal contracts models with learning and imperfect credibility (i.e., relaxing the assumption of perfect information). However, one undesirable feature of these learning models is that they assume that the monetary authorities do not behave optimally but instead simply follow an exogenously determined Taylor rule. Hence, the explicit relationship between discretionary policy and inflation and unemployment persistence is disregarded. This paper presents a modified version of Barro and Gordon's (1983) neoclassical model of discretionary monetary policy. By incorporating imperfect information regarding the true preferences of the monetary authorities and a consequent learning process on the part of the public, I show that imperfect credibility and transparency can, even in the absence of nominal rigidities, generate a substantial amount of persistence in inflation and unemployment following a disinflationary regime shift. This result is due to the failure of discretionary policy to incorporate the impact of credibility and transparency on the dynamics of inflation expectations, which leads to a dynamically inconsistent disinflation characterized by excessive employment losses and persistent inflation. Hence, in its simplest form the model suggests, in sharp contrast to the New Keynesian framework, that long-run persistence in inflation is governed by the persistence in the natural rate of unemployment, but short-run persistence is due to dynamically inconsistent monetary policy coupled with incomplete information.28 However, simulations indicate that although the model can easily generate a high degree of inflation persistence, it has more difficulty generating large sacrifice ratios. Interestingly, Taylor (1998) argues that the inclusion of learning or lack of credibility in a rational expectations model of staggered contracts can easily explain the observed disinflations costs, but that such a framework is less likely to generate observed inflation persistence. The answer to this puzzle may be that while nominal rigidities are primarily responsible for the observed employment losses, inflation persistence is largely due to the discretionary nature of monetary policy.