سیاست های پولی با ترجیحات نامشخص بانک مرکزی

This paper considers monetary policy when policy makers’ preferences are private information. I show that in the first period of a two-period term, all policy makers but the least inflation averse inflate less – but respond more to shocks – than if there were no private information. Moderately inflation-averse policy makers may reduce their inflation most. A tendency toward increased conservatism in their second period increases inflation in the first. With T<∞ period terms, inflation depends solely on the policy maker's time left in office. With unchanging preferences and no discounting, inflation is lower the longer he has left.

An important factor in monetary policy is how averse policy makers are to inflation relative to their dislike of output loss. Their preferences are their private information and this has implications for their behavior. The intent of this paper is to analyze the effect of unobservable preferences on policy makers’ incentives to inflate. In the basic model, policy makers serve two periods and each period they minimize a loss function which is decreasing in output and increasing in squared inflation. Output is increasing in unexpected inflation. Stochastic shocks, realized after the public's expectations are formed but before monetary policy is made, provide a role for activist monetary policy. Policy makers vary by the weight they put on output loss, relative to inflation. This weight is a policy maker's private information and there is a continuum of policy maker types.1 The policy maker takes current expected inflation as given when he chooses current inflation. The public has rational expectations. If it knew the policy maker's preferences, then on average it would predict inflation correctly in equilibrium. If it knew the policy maker's preferences, then on average there could be no unexpected inflation in equilibrium. Thus, he does not take into account the rise in equilibrium current expected inflation associated with a higher current inflation choice. As a result, inflation is too high. This is the familiar time-inconsistency problem. If, however, the public believed the policy maker to be more inflation averse than he actually is, then its expectation of inflation would be too low and on average unexpected inflation would be strictly positive. Thus, output would be higher than it would be if preferences were guessed correctly. Thus, policy makers have an incentive to increase the public's perception of their inflation aversion. The public realizes this and knows the inflation the policy maker would pick, given its type. I show there exists a unique separating perfect Bayesian equilibrium, where the policy maker's type is revealed by his inflation choice.2 I show that all but the least conservative policy maker (that is, the one who puts the most weight on output) inflate less during their first period in office than they would with known preferences. This is because at their within-period optimal inflation, current welfare is insensitive to small changes in inflation. But, a small decrease in inflation raises the public's perception of their inflation aversion and increases welfare in their second period in office. The least conservative policy maker does not inflate less. His type is revealed in equilibrium and inflationary expectations are as bad as they can be. Thus, he has no incentive to signal and chooses his within-period optimal inflation. This result is similar to Vicker's (1986). Two novel results are derived in the basic model. First, policy makers with intermediate preferences may lower their inflation the most as a result of their private information. The very conservative place little weight on output and, thus, put little weight on future expected inflation and have relatively little incentive to signal. The less conservative care more about future expected inflation. However, because the least conservative policy maker does not lower his inflation, relative to what he would do with known preferences, slightly less conservative policy makers do not have to reduce their inflation by much to distinguish themselves. Second, unknown preferences make policy makers less inflationary in their first period in office, but more responsive to shocks than they otherwise would be. A shock increases within-period optimal inflation and the increase is larger the less conservative the policy maker. Thus, for all but the least conservative policy maker, inflation increases by more than within-period optimal inflation does because more conservative policy makers do not have to lower their inflation, relative to that of less conservative policy makers, by as much to distinguish themselves. The tradeoff between central bank conservatism and activism may be less severe than Rogoff (1985) suggests. In addition to the separating equilibrium just described, where the policy maker's type is revealed, the model can have a continuum of pooling equilibria. These equilibria are supported by beliefs that seem implausible and a further equilibrium refinement is used to rule them out. The model is extended to consider changing preferences. I find the novel result that if policy makers tend to become more conservative while in office, this causes lower inflation in their second period and higher inflation in the first. Inflation rises in the first period because a lower weight is expected to be placed on future output, relative to inflation, thus lowering the gain from future unexpected inflation and the incentive to signal. The model is also extended to T<∞ periods and two new results are derived. Inflation depends solely on the time left in office. If the probability of a preference change is zero and the discount factor is one, inflation is lower the longer the policy maker has left in office. Thus, longer terms lead to lower inflation on average. This paper is related to the Backus and Driffill (1985) paper on uncertain central bank preferences. There, two types of policy makers inhabit a finite-horizon Barro–Gordon (1983) world. One type is strategic and maximizes social welfare; the other is mechanistic, always choosing zero inflation. A policy maker's type is his private information. The strategic type may attempt to masquerade as a mechanistic type early in his term in office to increase the public's belief that he might be mechanistic. This lowers future inflationary expectations, providing a more favorable output-inflation tradeoff later in his term. Another related paper is Cukierman and Meltzer (1986). (Faust and Svensson (2001) is a recent extension.) In their infinite-horizon model, the central bank has private information about its time-varying preferences and chooses money growth. Because the central bank chooses money growth, rather than an interest rate, Cukierman and Meltzer argue that the central bank's action (planned money growth) is imperfectly observable. This turns what would otherwise be a signaling problem, similar to the one here, into a statistical inference problem. This results in striking differences between their model and the one here. Unlike the model here, there are no pooling equilibria; the separating equilibrium may be unique, but is not revealing; the outcome depends upon the public's beliefs about the distribution of preferences.3 The policy maker in this model unilaterally chooses inflation. This setup is relevant for countries where policy is made by a monetary policy committee with both instrument and target independence and which is dominated by its chairman.4 The United States may be an example. It is also relevant for countries where the government delegates instrument, but not target, independence to the central bank. In this case, the policy maker should be interpreted as the government or government official who chooses the target, not the central bank. An example might be the United Kingdom, where the Chancellor selects the target. In Section 2, the basic model is presented. Section 3 discusses the model's implications for the tradeoff between policy maker conservatism and activism. Section 4 allows for uncertainty, in the form of random changes in the policy maker's preferences. Section 5 extends the basic model to T<∞ periods. Section 6 concludes.

This paper shows that if the weight monetary policy makers put on output loss, relative to inflation, is private information, all policy makers but the least inflation averse inflate less and react more strongly to stochastic shocks than they would without private information. Intermediate types may reduce their inflation by most. Increasing in inflation and decreasing in output and where output is increasing in unanticipated inflation. The existence of a shock, realized after expectations are formed, provides an activist role for monetary policy. If the weight the policy maker puts on output, relative to inflation, is known then equilibrium inflation is too high to minimize the equilibrium value of the policy maker's loss function. If his preferences are his private information and there are a continuum of potential policy maker types. If policy makers tend to become more conservative while in office, it is shown that they will inflate less in their final term and more in their first term than if preferences were constant. If policy makers serve T<∞ periods terms it is shown that inflation depends solely on their time left in office, and not on how long they have served or what they have done previously. If preferences do not change and policy makers do not discount the future while in office, then they inflate less the longer they have left in office.