ارتباطات در بازی های تکرار شده سیاست های پولی

A central bank and the public are engaged in an infinitely repeated monetary policy game with communication. For reasonable discount factors, there exists an equilibrium in which the central bank fully reveals its private information. The fully revealing equilibrium is superior to the uninformative equilibrium. The welfare gain of transparency increases with the slope of the Phillips curve, the natural rate of unemployment, and the degree of heterogeneity in the population. Transparency results in lower inflation but a higher variability of inflation.

Central bankers around the world seem to spend a considerable amount of effort on communicating their policies. Examples include speeches given by central bankers, policy announcements, press releases, testimonies, and minutes of policy meetings. In recent years, some central banks have been increasing the amount of information available to the public and making it more transparent. As Paul Jenkins of the Bank of Canada says, “while central banks used to say little and let actions speak for themselves, today it would be accurate to say that words can, and often do, speak louder than actions” (see Jenkins, 2004). In this regard, the examples of the Bank of Canada and the Federal Reserve in the U.S. are instructive. In the early 1990s, the Bank of Canada (1) did not disclose its decisions on its policy instrument and thus did not explain them; (2) did not change the short-term interest rate in fixed increments; (3) did not hold policy decision meetings at fixed dates; and (4) did not make sufficiently detailed commentaries on the Canadian economy and monetary policy (see Macklem, 2005). In the mid-1990s, the Bank took several important steps towards greater transparency. In 1995, it started publishing its semi-annual Monetary Policy Report that details the Bank's overview of the economy and its outlook for growth and inflation; in 1996, it started issuing a press release after each policy decision; since 2000, it has had eight fixed policy decision dates per year. In the U.S., the past bias towards secrecy of the Federal Reserve is well documented (see Goodfriend, 1986). Over the last decade this has changed. In February 1994, the Federal Open Market Committee (FOMC) decided to issue a press release after each meeting at which a policy action was taken; prior to this date the market had to infer the federal funds rate target. Starting May 1999, a press release had been issued after each meeting at which there was a major change in views about future developments. A “policy bias”, part of such press release, was widely interpreted by the market as a hint at future policy actions. Since March 2002, the roll call of the vote on the federal funds rate target has been released after each meeting; this would include the preferred policy choice of the dissenters. In December 2004, the FOMC decided to release the minutes of its meetings three weeks after each meeting instead of six weeks. Each of these events is a step forward towards greater transparency and an increasing number of occasions to talk to the public. Sandra Pianalto, the President of the Federal Reserve Bank of Cleveland, says that the FOMC “has learned to talk” when referring to the evolution of the FOMC from being viewed as secretive to being viewed as more transparent. “Do markets understand our behavior because we are more predictable or because we communicate better? I suspect that both forces are at work. As they say, talk is cheap, and I am not suggesting that better communications are of any value without actions that back up the words” (see Pianalto, 2005). This statement underscores one of the main questions this paper addresses: Can a central bank talk informatively, Does the public listen to the central bank, and What does it take to be listened? In this paper we consider only a special kind of communication by central banks, in which sending a message does not change the payoff structure for the parties involved.1 This means of communication contrasts with costly signaling such as the education game in Spence (1974). We build a model with many private agents and a central bank and show how this model can be viewed as a game of two players. We use a version of the Sender–Receiver cheap-talk game introduced by Crawford and Sobel (1982). In cheap talk games — for an overview, see Farrell and Rabin (1996) — players' payoffs are not affected by message sending. Talk is “cheap”. We model the relationship between a central bank and the private economy as a repeated monetary policy game with cheap talk. There are two players, a central bank (CB) and the public (P). The CB can be of two possible types, “low-inflation” or “high-inflation”. After learning its type, the CB announces it (not necessarily truthfully). Next, the P chooses its action (which results in the formation of inflation expectation), and then the CB chooses inflation. The actions of both players are observable. First, we show that when the CB is reasonably patient, there is a fully revealing equilibrium in which the CB truthfully announces its type. The intuition behind the first result is that when the CB is sufficiently patient, reputational considerations are important. Second, we demonstrate that social welfare is higher in the best fully revealing equilibrium than in the best babbling equilibrium. The reason behind this result is that when the type is truthfully announced, the P and the CB can attain the best possible outcome under rational expectations.2 It is also found that the welfare gain of transparency increases with the slope of the Phillips curve, the natural rate of unemployment, and the degree of heterogeneity in the population. We study several different environments. We allow the CB's type to be either permanent or change every period following an i.i.d. process. In the latter case, the CB makes announcements every period. We also consider the case when the CB does not have perfect control over inflation, and its inflation target is not observable. In these new environments, as before, the best fully revealing equilibrium is superior to the best uninformative equilibrium. The paper is organized as follows. Section 2 describes the environment. Section 3 contains the main results. Section 4 presents two extensions of the basic model. Section 5 reviews the related literature and concludes.

In this paper we study central bank communication. We analyze a repeated monetary policy game of cheap talk. We have shown that when the central bank is reasonably patient, there is a fully revealing equilibrium, in which it can talk truthfully and credibly. We demonstrate that social welfare is higher in the best fully revealing equilibrium than in the best babbling equilibrium, in which the central bank talks uninformatively. Discount factors larger than 0.5 can support the best fully revealing equilibrium. In the macroeconomic literature, it is common to use discount factors between 0.95 and 0.99. From this we conclude that the transparent outcome is possible as long as the CB has an opportunity to reveal its private information, i.e., when the game has the announcement stage. This, in our view, highlights the benefits of having such channels of communication of central banks' private information as required testimonies, speeches, releases of internal forecasts etc. The welfare gain of the transparent outcome is shown to increase with the slope of the Phillips curve, the natural rate of unemployment, and the degree of heterogeneity in the population with regard to preferences over inflation. It has also been shown that the transparent outcome enjoys a lower level of inflation. However, the variability of inflation is higher under transparency than in the uninformative outcome. Even though nowadays the openness of central banks seems to be the prevalent phenomenon, it was not the case some 15 years ago. Both pros and cons of secrecy have been offered. Goodfriend (1986) first considered the Federal Reserve's arguments for secrecy, and by informal arguments rejected validity of all of them. Other contributions include Walsh (1999), Thornton (2000), Faust and Svensson (2001), and Geraats (2008). Spanjers (2008) considers cases of an optimistically and pessimistically inclined public (and a central bank); he finds that a lack of confidence in the central bank is harmful in an environment of strategic ambiguity. Arguments for secrecy have also been provided. Bhattacharya and Weller (1997) find that secrecy about the scale of an intervention operation in the foreign exchange market is desirable. In Amato, Morris and Shin (2002), the presence of both a public and private signals and the preference of private agents to choose actions close to the actions of others results in the public signal crowding out the private signals: the agents would ignore the latter and use primarily the former. As a result, the market overreacts to the public signal. Svensson (2006) argues, however, that for reasonable parameter values in the Amato, Morris and Shin model, social welfare actually increases in transparency. We now comment on two strands of literature on which our work is built. The first was started by Stein (1989) who applies the Crawford and Sobel (1982) cheap talk model to rationalize ambiguity of the Fed's announcements. In this two-period model, the Fed has private information regarding its target for the exchange rate. Because of quadratic preferences the Fed has an incentive to make the public believe that its target is larger than it actually is. This makes precise announcements on the part of the Fed impossible. Nevertheless, there is an equilibrium with imprecise announcements: the Fed can credibly announce the interval within which its actual target lies. Other applications of cheap talk in monetary policy games include Garfinkel and Oh (1995) and Moscarini (2007). Cheap talk is not the only way to convey private information. The second strand of literature uses costly signaling. Cukierman and Meltzer (1986) provide a rationalization for central bank ambiguity in a game of both incomplete and imperfect information. The main objective of their paper is to derive conditions under which central banks prefer ambiguity to full disclosure of objectives. This is achieved by allowing the central bank to choose the level of noise in money control. The higher the variance of the noise the higher is ambiguity; that is, ambiguity is measured by this variance. The key result is that under certain assumptions the optimal level of the variance is positive. Other important contributions in this strand include Canzoneri (1985) and Faust and Svensson (2001). The first approach addresses strategic issues concerning central bank announcements by using cheap talk games; but these models are static. The models in the second approach are dynamic but do not explicitly model central bank announcements. Both Stein (1989) from the first group and Cukierman and Meltzer (1986) from the second one justify ambiguity of central bank announcements. Our approach combines the strengths of these two papers: we study repeated cheap talk games. A key result in our paper is that the reputation issue arising in the repeated game may drastically change their results. Central banks can talk more precisely, and this improves welfare. Using an index of central bank transparency developed by Eijffnger and Geraats (2006) for nine major central banks, it has been shown by some researchers that transparency leads to better inflation performance. For example, Demertzis and Hughes Hallett (2007) show that inflation variability becomes lower as a result of transparency. And Geraats (2008) demonstrates that greater transparency has been followed by lower inflation. Our results are consistent with the latter finding. We show that in the transparent equilibrium, inflation is lower than in the babbling (opaque) equilibrium. However, unlike in Demertzis and Hughes Hallett (2007), the transparent equilibrium exhibits higher inflation variability than the opaque one. The aspect of transparency addressed in this paper has to do with speeches, testimonies, press releases, or forecasts. The recent move towards transparency may be viewed as a result of a shift to a new regime with more opportunities for central banks to reveal their private information such as their forecasts, outlooks for the economy etc. The old regime corresponds to a game with no communication stage in our model, which, in turn, is equivalent to the babbling outcome. The paper does not address the reasons why the new regime with more opportunities for central banks to communicate has recently arisen. This is an interesting political economic issue that may be addressed in future research. The results of this paper may imply the following policy recommendation: more channels of communication for central banks are welcome; releases of central banks' private information such as their outlook for the economy as well as inflation and output forecasts are welcome.