Results 1 
6 of
6
From doves to hawks: a spatial analysis of voting
 in the Monetary Policy Committee of the Bank of England, 19972007. LSE PSPEWorking Paper No. 8, London School of Economics
, 2007
"... Abstract. This article examines the making of monetary policy in the United Kingdom between 1997 and 2008 by analysing voting behaviour in the Bank of England’s Monetary Policy Committee (MPC). It provides a new set of measures for the monetary policy preferences of individual MPC members by estimat ..."
Abstract

Cited by 5 (0 self)
 Add to MetaCart
Abstract. This article examines the making of monetary policy in the United Kingdom between 1997 and 2008 by analysing voting behaviour in the Bank of England’s Monetary Policy Committee (MPC). It provides a new set of measures for the monetary policy preferences of individual MPC members by estimating a Bayesian item response model.The article demonstrates the usefulness of these measures by comparing the ideal points of outgoing MPC members with their successors and by looking at changes over time in the median ideal point on the MPC.The analysis indicates that the British Government has been able to move the position of the median voter on the MPC through its appointments to the Committee. This highlights the importance of central bank appointments for monetary policy.
An econometric model of monetary policy decisionmaking with applications to the United Kingdom and Sweden", working paper
, 2009
"... * We acknowledge the research assistance of Supratim Dasgupta and Kathryn Mulligan. Helpful ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
* We acknowledge the research assistance of Supratim Dasgupta and Kathryn Mulligan. Helpful
PowerSharing in Monetary Policy Committees: Evidence from the United Kingdom and Sweden * PowerSharing in Monetary Policy Committees: Evidence from the United Kingdom and Sweden
"... Abstract Committees may make better monetary policy decisions than individuals; however, the benefits of group decisionmaking could be lost if committee members cede power to a chairman. We develop an econometric model to describe intracommittee powersharing across members. Estimation of the mod ..."
Abstract
 Add to MetaCart
Abstract Committees may make better monetary policy decisions than individuals; however, the benefits of group decisionmaking could be lost if committee members cede power to a chairman. We develop an econometric model to describe intracommittee powersharing across members. Estimation of the model permits us to classify monetary policy committees into the typology developed by 1 Monetary policy decisions are typically made by committees rather than by single individuals. Recent theoretical and empirical work on the monetary policy decision process supports the argument that committees may make better policy decisions than individuals. 1 To put the matter succinctly, when compared to individual decisions, committee decisions seem to offer the wellknown benefits of diversification In this paper, we develop an empirical model that can classify MPCs in the Blinder typology on the basis of econometric estimation. Our model describes the monetary policy preferences of individual MPC members in the form of Taylor rules 1 Blinder (2004,(38)(39) offers four reasons why committees may make better policy decisions than individuals. First, when members have different preferences, diversification avoids extreme outcomes. Second, even if members have similar preferences, they might use different models to reach their conclusions. Third, members might use different forecasts or forecasting techniques. Finally, individuals might use different modes of reasoning to approach problems. For additional discussion of the benefits of committee decisionmaking, see the references given in footnote (7) below. for the Riksbank. In each case, the start of the sample followed important reforms that established current institutional arrangements. We chose to end our samples in the fall of 2008, when financial panic and a deepening recession led to radically changing policymaking practices. These changes made it impractical or inadvisable to apply the models that were suitable for the preceding period. 3 Our work is related to several strands of empirical research on monetary policy decisionmaking. There is an extensive literature on central bank reaction functions, most recently focusing on those in the form of Taylor rules. These studies describe how central bank policy instrument choices are empirically related to prevailing or forecasted 2 As we explain in Section III, it would at present be impractical to apply our model to the U.S. Federal Open Market Committee (FOMC). 3 Prior to fall 2008, large interest rate moves were rare. Properly modeling the large moves that followed the financial panic would require an increase in the number of discrete choices possible for each voter decision. This complication alone could make estimation impractical. In addition, the proximity of a lower bound on rates also became an issue, and properly modeling that boundary in the econometric model would have been problematic. Finally, the use of nontraditional policy options (e.g., quantitative easing) added an extra dimension to policymaking that would be difficult to model. 6 5 I. Monetary Policymaking at the Bank of England and the Riksbank I.A. Monetary Policymaking at the Bank of England Current monetary policy institutions in the UK are prescribed by the 1998 Bank of England Act. The Bank of England has independent authority to manage monetary policy through the setting of the official bank rate, a rate at which the Bank lends to financial institutions. The Bank is directed by the government to target inflation, with a current target of 2.0% per year as measured by the consumer price index. Interest rate choices are made by a ninemember monetary policy committee, including five "internal" appointees and four "external" appointees. 8 By statute, the committee meets every month and makes interest rate decisions by majority vote. Since 1999, the interest rate targets of each committee member in each meeting have been publicly reported. Because decisionmaking is majoritarian, the selected interest rate is always the median of the individually reported desired rates. Individuals' reported rates and adopted targets have normally varied in 25 basis point increments, and rate movements larger than 25 basis points in a single meeting have been rare. 9 The BOE's website provides this characterization of the committee's decisionmaking process: Each member of the MPC has expertise in the field of economics and monetary policy. Members do not represent individual groups or areas. They are independent. Each member of the Committee has a vote to set 8 The MPC's five internal members are the Bank's Governor, its two Deputy Governors, the Executive Director responsible for monetary policy analysis, and the Executive Director responsible for monetary policy operations. The four external members are those appointed to the MPC by the Chancellor of the Exchequer. These external members are to have expertise in economics and monetary policy, but there are otherwise no specific requirements for them. We are not aware of previous research that characterizes the monetary policy preferences of individual members of the Riksbank's MPC. Jansson and Vredin I.C. Previous Studies of the Bank of England and the Riksbank II. The Econometric Model Our econometric model has the following elements: As the list above suggests, our proposed model describes how individuals' interest rate preferences are mapped into targets adopted by the committee; it permits individuals' preferred targets to be influenced by the Governor; and it permits formally recorded rate preferences to differ from the preferences that prevailed when collective choices were made. In the rest of this section, we provide details on our modeling of each of these features. The sequence of actions in our model is assumed to mirror that of the following discussion. To facilitate our exposition, II.A. True Interest Rate Preferences As a meeting begins, individual MPC members are presumed to have "true" interest rate preferences, * it R , governed by the following reaction function specification: 13 Although we have noted that the data for both the BOE and the Riksbank include instances where actual or desired movements in interest rates exceeded 25 basis points, our model collapses the preference data into three discrete categories. This simplification is necessary to keep the econometric model computationally tractable, especially for the BOE. We have estimated a model variant permitting five discrete categories for the Riksbank; this has no substantive effect on our results. II.B. Modified Interest Rate Positions Our model permits committee members to modify their positions in deference to the chairman. 17 We assume that members' modified interest rate positions, it R , are 14 The Bank of England's MPC has nine members, and the Riksbank's committee has six. Over time, the identities of the individuals occupying these positions change. For simplicity, our notation indexes only the number of voting positions, but our later econometric analysis will also distinguish the individuals occupying the available slots. 15 Inclusion of the shared error term component permits the model to account for the observed tendencies of members to agree on rate movements, without necessarily attributing that observed agreement to influence from the chairman or pressures for consensus. 16 Riboni and RugeMurcia (2010) derive reaction functions using an explicit optimization framework. Assuming a specific but plausible form for the utility function, they find that reaction function parameters other than intercepts are the same for all members. This finding supports our admittedly restrictive characterization of differences across MPC members. Moreover, our estimation procedure becomes computationally impractical with large numbers of additional parameters. 17 Disagreements inside an MPC may arise because of differences across members in educational and career backgrounds, theoretical perspectives, reputational considerations, or regional interests. The chairman may play key roles in MPC deliberations as a consensus builder and as an agenda setter. He may also gain leverage over the committee if he is responsible for allocating the central bank's resources to rankandfile members and if he serves as the spokesman for the MPC in political and public arenas. In this expression, is the weight members assign to the chairman's interest rate preference. Second, we map that weighted average into a discrete interest rate choice, where the available options include the status quo interest rate, Member i supports member i supports 0.25 and member i supports 0.25 For the chairman, * * 1t i t R R , so these conditions imply that 1t R is the discrete option that is closest to the chairman's true preferred rate. II.C. The Committee's Selected Rate The committee selects a rate, t R , by majority rule, so we specify that the adopted interest rate target is the median of members' modified supported rates: disproportionate influence on FOMC decisions. For additional discussion of the power of the chairman and the role of consensus at the Federal Reserve, see the references given in footnote (6) above. 14 1 2 , ,..., In the event of a tie vote, the committee chairman serves as a tiebreaker. 18 II.D. Recorded Votes After the median is determined, formal votes are recorded. At this stage, the model permits committee members to report formal votes that differ from the policies they advocated when the collective choice was made. Specifically, members may choose to support the committee choice in order to produce a vote that reflects a consensual outcome. In equation We assume that final recorded votes, the it R , are obtained by making this weighted average discrete in a manner that is analogous to the operation previously described in conditions (3). Because the chairman may value consensus differently than rankandfile members, we permit a different weighting parameter for the chairman, c , and replace condition (5a) with condition (5b): 18 At both the Riksbank and the Bank of England, the Governor has formal tiebreaking power. At the Bank of England, the monetary policy committee normally has nine members, so ties can occur only when a member is absent or when a vacancy is unfilled. In our sample period, the BOE's Governor never had to vote to break a tie. However, modeling tiebreaking must be a feature of our model, even for the BOE. Later, when we calculate the likelihood that a particular voting outcome will occur in a meeting, we must correctly account for the possibility that nonobserved outcomes might have occurred instead. The model we have described is complex and nonlinear, and it includes a number of unobserved variables. However, it is feasible to calculate the likelihood function using simulationbased methods. We describe our maximum simulated likelihood (MSL) estimator in the remainder of this section. 15 II.E. Special Cases of the Model 19 Consider a set of trial values for the parameters of the model. To simulate a single observation (a meeting) in our sample, we first draw values of the random error terms, t u and it v . Given these simulated error terms, and given the reaction function parameters, we calculate a simulated "true" interest rate preference for each member using equation (1). Then, using conditions (2) and (3), we calculate members' modified rates after accounting for deference to the chairman and the discreteness of policy options. At this point, we use equation (4) to determine the committee median, the majority voting winner over the three discrete alternatives. Finally, using conditions (5) and the requirement that reported rates be discrete, members' formal votes are determined. This simulation sequence produces a vector of discrete interest rate preferences, one for each 19 The MSL method is described by 17 member in the meeting. The simulated data are analogous to historical data, which also record discrete rate preferences for each committee member. We can use this method to calculate the likelihood for each observation in the sample and, in turn, the likelihood for the complete sample. In order to estimate the model, we must find parameter values that maximize the likelihood function for the sample. The ML PROC routine in TSP is used for this purpose. Because of the complexity of the model and because outcomes are often low probability events, estimation of the model is computationally burdensome. For example, for a ninemember committee voting on a discrete outcome with three options, there are 19,683 possible configurations of reported preferences that might occur in any given meeting. 20 A single run of the model for the United Kingdom can require several months of computing time. For the 12member U.S. Federal Open Market Committee (FOMC), however, the number 20 This number of possible configurations implies that some observed outcomes will be low probability events. Our estimation procedure varies the number of simulated observations over meetings, using more simulations (up to one million) when estimated probabilities for the observed outcome are lower. 18 of possible outcomes in any given meeting would be 531,441, which makes estimation of our model impractical given current computing capabilities. See Appendix A for further discussion of computational matters.
Discussion Paper No.21
, 2008
"... individual voting patterns by Charlotta Groth and Tracy WheelerExternal MPC Unit Discussion Paper No. 21 * The behaviour of the MPC: Gradualism, inaction and individual voting patterns ..."
Abstract
 Add to MetaCart
individual voting patterns by Charlotta Groth and Tracy WheelerExternal MPC Unit Discussion Paper No. 21 * The behaviour of the MPC: Gradualism, inaction and individual voting patterns