ارتباطات در یک کمیته سیاست پولی

We model monetary policy decisions as being taken by a group of heterogeneous policy makers, organized in a committee. Intuitively, when MPC members disclose and discuss the arguments behind their view on the interest rate, the quality of the collective decision should be higher compared to merely taking a simultaneous vote. We show that in some cases this intuition need not be correct. We also find that communication is a relatively effective way to implement the ‘knowledge pooling’ argument in favor of collective decision-making, compared to expanding the size of a committee. Moreover, decision-making with internal communication appears generally more robust in situations when heterogeneity of members is not adequately captured by decision-making rules.

Judging from a survey held in the late 1990s (Fry et al., 2000), a large majority of central banks nowadays delegates policy decision-making responsibilities to monetary policy committees (henceforth: MPC). When members convene for the MPC meeting, they communicate with each other. This process of internal communication is an important characteristic of real-life committee decision-making such as by the FOMC in the US or the ECB Governing Council in the euro area (De Nederlandsche Bank, 2000 and Goodfriend, 1999). Interaction among MPC members involves an exchange of views regarding the current and future state of the economy, the transmission mechanism and the appropriate interest rate decision. Communication thus implies an exchange of information that increases the total knowledge available to the MPC (Berger et al., 2008). However, communication also implies augmenting one's initial views with those heard from others. The latter might be qualitatively less than the former, so the impact of communication on the collective decision is not clear a priori.1 The paper aims to investigate the conditions under which communication within a committee2 improves the quality of the collective decision, which we assume is taken by simple majority voting.3 By doing so, we are able to provide a theoretical rationale for some of the results found in the recent empirical literature on MPCs, such as Chappell et al., 2005, Gerlach-Kristen, 2003a, Gerlach-Kristen, 2003b, Gerlach-Kristen, 2009 and Meade and Sheets, 2005. Our findings indicate that an exchange of views within monetary policy committees will, in general, be beneficial for the quality of decision-making even though it increases correlation in members' voting behavior. Hence, we contradict the results of Ladha (1992), who — on statistical grounds — argued that a high positive correlation of votes above a certain threshold can substantially reduce the accuracy of a simple majority decision (in the limit, the majority will even do worse than an average voter). We show that an exchange leading to a single view on the correct decision yields a collective outcome inferior to the simultaneous simple majority voting only when decisional skills are very unevenly distributed among committee members. In other cases, communication increases the quality of monetary policy and is more effective doing so, the lower the average of MPC members' skills. Finally, we compare two ways of enlarging the pool of knowledge available to a committee of a given size: adding extra members (and having the committee vote without communicating internally) or allowing the members to share their views before voting. The latter approach appears to be more efficient. The structure of the paper is as follows. We start, in Section 2, by briefly describing our analytical framework. In Section 3, we formalize the effects of communication on the quality of the collective decision taken by an MPC consisting of members with identical skills. An MPC consisting of members with heterogeneous skills is considered in Section 4. Section 5 concludes.

5. Conclusions In this paper, we have highlighted a number of results with important implications for actual policy-making, when conducted in a committee. First, communication internal to the MPC is most likely to be beneficial for the quality of interest rate decisions. Only when decisional skills are very unevenly distributed among members could internal communication be detrimental to the quality of policy decisions. Through communication MPC members improve their common knowledge about future economic developments, i.e. the premises for the interest rate decision, which is beneficial for the quality of policy decisions. Secondly, communication can be a more efficient way of improving the quality of policy decisions than increasing the MPC size. Finally, the quality of collective decisions made with communication within the MPC and aggregation of views is less likely to suffer from a lack of appropriate weighing of different views with varying accuracy than the quality of collective decisions made through simple voting. The latter illustrates the fact that deliberations make collective outcomes more robust to the voting rule used. Our set-up is the simplest one that still allowed us to illustrate, analytically, various effects that communication between MPC members might have on (the quality of) monetary policy decisions that are collectively taken. Nevertheless, there are caveats to be kept in mind when interpreting our results. These include the specific simplifying assumptions about the single source of heterogeneity between the members (different ‘mind-sets’ or heuristics), about the form of communication (members only share their information on future inflation shocks Ei, tet + 1) and its impact on the decisions (the information on future inflation shocks is averaged), absence of strategic behavior among the members (see e.g. Tillmann (2011) for the analysis of strategic motives in forecasting) and the single-shot nature of our simultaneous voting model. As to the latter, it would be interesting to see whether the results presented in the previous section also hold in a more realistic repeated game setting. Once the game is repeated, the scope of communication and decision-making could be extended to account for the implications of differences in performance between members as observed in previous meetings. We leave this to future research.