Policy preferences of central bankers and the design of a monetary-policy committee

Sylvester Eijffinger, Ronald Mahieu, Louis Raes 07 May 2013



One important practical aspect of monetary-policy concerns decision power and its accompanying responsibility. Most central banks have moved towards a structure in which decisions are taken by a committee. Pollard (2004) provides an early survey in which 79 of 88 surveyed central banks are governed by a committee.

However, even with a committee that is responsible for decisions on monetary policy, many options are possible. Blinder (2007) discusses some aspects that drive what is chosen from these options. One aspect which has received a lot of attention in the academic literature concerns whether a committee should be composed of people who have worked at the central bank or not, or a mix of both. Related to this is the question of whether it is desirable to appoint central bankers who have relevant experience in sectors like the financial sector, the NGO sector or working as a civil servant. The idea is that bringing some particular career experience to the table provides additional knowledge to the committee.

So far, the academic literature has been inconclusive on these matters. We report here on recent research which sheds some light on these issues. Two papers estimated policy preferences of central bankers within a spatial voting model, Hix et al. (2010) and Eijffinger et al. (2013). This spatial modelling approach was previously developed by scholars investigating judicial votes or votes in the US congress and has substantial advantages over the traditional methods economists have used to tackle these questions. The approach is both easy to implement and flexible. To our knowledge only the above two papers have tried to implement this method in the context of voting at central banks. The former paper focuses on the political appointment channel. The latter paper focuses on the questions stated earlier and fits in the literature investigating the composition of monetary-policy committees, see Besley et al. (2008) for a recent review.

New research

Figure 1 shows the estimated policy preferences or ideal points as presented in our recent paper (Eijffinger et al. 2013). The graph presents the estimation results where only voting data on the policy rate was used. The paper also reports results when votes on asset purchases are incorporated. The conclusions do not alter materially. The dot indicates how central bankers rank on a dove-hawk scale. The line indicates the uncertainty surrounding the estimates. The policy preferences are estimated from the votes the central banks have cast. No outside information was used and as a result we can interpret these as truly revealed preferences. The graph presents a relative scale where positive ideal points correspond to hawkish policy preferences, and negative ideal points refer to dovish policy preferences. To interpret these results compare the ideal points of Adam Posen and Andrew Sentence. The results indicate that Sentence is clearly more hawkish than Posen. At any given meeting we would expect that Sentence is more likely to vote for the hawkish policy choice than Posen.

The Bayesian approach provides the joint probability over all parameters. What this means is that one can work with any derived quantity which is a function of these preferences only. Consequently a researcher is able to be very flexible in developing hypotheses and investigating whether there is support for them in the data.

Figure 1. Estimated policy preferences at the Bank of England

Source: Eijffinger et al. (2013).

The analysis leads us to several conclusions:

  • It is not the case that internals vote more dovishly or hawkishly than externals.

However, internals tend to more often occupy the middle ground. When we look at the extreme policy preferences then we find that these are held by external monetary-policy committee members. For example, in the figure above Wadhwani, Blanchflower, Sentence and Besley were external monetary-policy committee members.

  • Background characteristics do not simply correlate with policy preferences.

It is not the case that having worked in a certain industry ensures that someone votes more hawkishly or dovishly. One exception is central bankers with industry experience who tend to be more hawkish than others, but this result is built on a smaller sample so we should be cautious with this result.

  • If one considers the variation in preferences within different groups, then remarkable differences emerge;

External members tend to have much more differing policy preferences than internal members. When considering the career backgrounds we find that the wide variation in policy preferences is mainly due to central bankers with an academic background or a background in industry. The difference with some other career background is remarkable. There is compelling evidence that the variation is much smaller among central bankers who had experience at the Bank of England compared to central bankers who had worked in academia.


What can we make of this?

  • First of all, these results confirm some findings in the literature, in particular that it is not the case that externals are simply more hawkish or dovish than internals.
  • Also, career backgrounds do not simply correlate with preferences on a dove-hawk dimension.

Both these results resonate with the findings from Besley et al. (2008). Because we can identify the policy preferences of individual central bankers it is easier to understand how certain results come about.

  • A new finding is that the variation in preferences changes with career backgrounds;

This is intuitive. If you have worked for years at the Bank of England it is likely that you share some organisational consensus. An academic on the other hand has worked in an environment where having an opinion (and especially a pronounced opinion) is valued. But the implications hereof are important. There may be different reasons to have a committee in place. If one of these reasons is that a wide range of policy preferences should be included in a monetary-policy committee then having sufficient academics in the committee might be a reasonable strategy.

The empirical research discussed in this column builds on a latent variable framework in which monetary policymakers are ranked on a dove-hawk dimension. Some may feel that this is too restrictive and that a multidimensional framework is more suitable to capture voting behaviour. From a theoretical point of view we certainly agree. However the results presented in our paper (Eijffinger et al. 2013) as well as in Hix et al. (2010) suggest that a single latent dimension fits the observed votes quite well. In other words one does not need to complicate the modelling approach by introducing multiple latent dimensions to predict observed votes. In practice the single dove-hawk dimension seems adequate.


Besley, T, N Meads, and P Surico (2008), ”Insiders versus Outsiders in Monetary Policymaking", The American Economic Review, 98(2), 218-23.

Blinder, A (2007), “Monetary policy by committee: Why and how?", European Journal of Political Economy, 23(1), 106-123.

Eijffinger, S, Mahieu, R, Raes, L (2013), “Inferring hawks and doves from voting records”, CEPR Discussion Paper, DP9418.

Hix, S, B Hoyland, and N Vivyan (2010), “From doves to hawks: A spatial analysis of voting in the Monetary Policy Committee of the Bank of England", European Journal of Political Research.

Pollard, P (2004), “Monetary policy-making around the world", presentation, Federal Reserve Bank of St. Louis.



Topics:  Monetary policy

Tags:  central-bank committee

Professor of Financial Economics at Tilburg University, President of the Tilburg University Society and CEPR Research Fellow

Professor of Finance and Innovation, Tilburg University

Assistant Professor, Tilburg University

Vox A/V