Central bankers habitually argue amongst themselves about the appropriate conduct of monetary policy; everyone knows that. The 24/7 coverage of central banks in the media and financial markets fixates on a never-ending intellectual tug of war between 'hawks' and doves' within the central banks. Academics use the dissent that we can see in the votes on monetary policy to shed a more dispassionate light on how the policy process works. There is just one small problem: if you focus on what policymakers vote for, then there is no evidence that they disagree with one another in any meaningful sense. There are no hawks, no doves, only consensus.
The mirage of dissent
When the Bank of England was given operational independence for the conduct of monetary policy in 1997, the architects of the new regime were conscious of the need to encourage a robust internal policy debate, and the process was designed accordingly. The creation of a nine-member Monetary Policy Committee (MPC) to set interest rates reflected the conventional wisdom among economists – but not necessarily psychologists (Stoner 1968) – that committees take better decisions than individuals because committee members can debate and learn from one another (Blinder and Morgan 2005). One key feature of the new regime was that MPC members were individually and publicly accountable for their votes (King 2007):
“Disagreement among the Committee is inevitable; it is also desirable because it represents the individual judgements of members, rather than an attempt to create a false consensus. It is a source of strength.”
The chief architect noted with pride that the process appeared to work: dissent was a more frequent and pervasive phenomenon at Threadneedle Street than in other central banks (King 2007):
“Not only is the number of dissenting votes greater on the MPC, the frequency of more substantive disagreement – where one quarter or more of the voters dissented – is markedly greater. Voting patterns on the MPC reflect the distribution of views about how to interpret the economic data, not a hint about where rates will go. Voting is certainly not used as a signal by the Committee.”
The metric that Governor King focused on – the incidence of dissent – has become the accepted academic metric of disagreement within policy committees. The pattern of dissenting votes within policy committees has been used to investigate a range of questions: to assess whether dissent contains information about the future path of rates (Gerlach-Kristen 2004); to examine the role of private assessments of the state of the economy in driving committee decisions (Hansen et al. 2014); and to explore whether the characteristics of policymakers (their background, their legal status on the policy committee) can help explain their voting record (Besley et al. 2008). These are interesting lines of enquiry but a narrow focus on how often policymakers disagree with one another risks missing the bigger picture.
If we want to quantify the extent of the disagreement within the MPC we need to factor in how much MPC members disagree with each other as well as how often. Now we arrive at a very different conclusion to the one King reaches: we can only find one occasion between late 1998 and the outbreak of the financial crisis in summer 2007 when an MPC member voted for a rate setting that was more than 25 basis points away from the one favoured by the majority. So the debate hinges on whether the difference in interest rates from the majority view in every other instance of dissent (25 basis points) is material or not. An objective review of the literature on the interest rate multiplier – the impact of changes in official interest rates on the variables that the MPC cares about (output and inflation) – suggests that it is not.
According to the Committee’s own estimate of the multiplier, raising interest rates by a full percentage point (i.e. by four times as much as the default level of dissent) and then holding them there for a year, has a peak impact on inflation of just over 0.3 percentage points (MPC 1999). On that basis, the 25 basis point difference between votes implies an underlying disagreement over the inflation outlook of fewer than 10 basis points. The implied difference in view on inflation is indistinguishable to the naked eye when placed in the context of the Committee’s estimate of the uncertainty around the outlook for inflation: the Bank’s fan chart for inflation currently spans an interval of around six percentage points, and the central band (which includes only 30% of the mass) spans an interval of around one percentage point (Bank of England 2016).
To sum up, all we see in the data is token dissent when you analyse the votes through the lens of the MPC's mandate: what does this imply for inflation? The frequency of token dissent is irrelevant. Even if there is a five-to-four split on the MPC over the stance of policy, it is a bit of a stretch to label that a ‘substantive disagreement’ if all concerned can agree on the level of rates to within 25 basis points.
Why don’t policymakers disagree?
If the pattern of votes genuinely reflects the views of individual members then we should conclude that to all intents and purposes MPC members agree all of the time. This would be a remarkable state of affairs.
If one considers the uncertainty about the structure of the economy and the shocks currently working through the system then it is surprising that nine people could independently reach the same conclusion about the appropriate monetary stance at a particular point in time, let alone all the time. Of course, the fact that MPC members do not reach their conclusions independently of one another but are instead exposed to the same information set and analysis in the internal policy process could explain a tendency to herd on a particular view. Nevertheless, reasonable disagreements within the Committee about the conjuncture (e.g. over the current size of the output gap) or strategy (e.g. how to respond to building financial imbalances) should still have routinely translated into much larger differences of opinion over interest rates than we actually observe, given the Committee’s own estimate of the interest rate multiplier (you need to move interest rates a lot to offset non-trivial ‘shocks’). That is a problem because a review of the speeches that MPC members delivered over this period reveals that there were material differences of opinion within the Committee on the key issues of the day (Barwell 2016).
We are left with the conclusion that MPC members did not vote their view. Contrary to Lord King’s claim, it appears that voting is used as a signal by MPC members, who reveal the direction of their dissent in their vote but not the magnitude. The question is, why? MPC members may collectively agree that revealing the true scale of the differences of opinion within the Committee would shake confidence in the process and the institution, and potential amplify the uncertainty around the future path of rates. Alternatively, policymakers may see virtue in small, serially correlated changes in the monetary stance because that ‘gradualist’ strategy might give the central bank greater traction on long rates (Bernanke 2004); if so, dissenting members would likely advocate only modest deviations from the current strategy in the direction of their unconstrained preferred stance. Or perhaps policymakers perceive a threat to their personal reputation should they advocate a position that is far from the one favoured by the majority and subsequent events do not validate their stance. Finally, some policymakers may perceive that they are under an obligation to support the majority view within the Committee – particularly in moments of crisis – even if they don’t believe it to be appropriate.
If the acid test of dissent within a policy committee is how people vote – rather than how they talk – then to paraphrase Lord King, all the MPC ever does is consensus. The hawks and doves are birds of a feather: they agree, all the time. In this column, we have focused on the MPC during the Great Moderation, but the point applies more broadly. For example, it is interesting that there is far more dissent within the FOMC over the near-term path of interest rates (e.g. ‘the 2017 dot’) than over the current level of interest rates. Likewise, if one looks at the pattern of votes during some of the most high profile episodes of dissent within central banks that have fascinated the media and markets alike – such as ‘Svensson versus the Riksbank’ in 2010/11 or ‘Blanchflower versus the Bank of England’ in 2008 – we see the same mismatch: material differences in views but marginal differences in votes. Whether this state of affairs is optimal is debatable. A case can be made that the quality of the internal policy debate and the external communication strategy would benefit from policymakers providing a frank assessment of the appropriate monetary stance alongside their assessment of the economy. What cannot be disputed is that either policymakers essentially agree all the time or they do not vote their view.
Bank of England (2016), August Inflation Report.
Barwell, R (2016), ‘Nine votes, one view’, in Chadha, J, Crystal, A, Pearlman, J, Smith, P and Wright, S (eds.) The UK Economy in the Long Expansion and its Aftermath, CUP.
Bernanke, B (2004), ‘Gradualism’, Speech, 20,May.
Besley, T, Meads, N and Surico, P (2008), ‘Insiders versus outsiders in monetary policymaking’, American Economic Review (Papers and Proceedings), 98(2), 218–223.
Blinder, A and Morgan, J (2005), ‘Are Two Heads Better than One? Monetary Policy by Committee’ Journal of Money, Credit and Banking , 37(5), 789-811.
Gerlach-Kristen, P (2004), ‘Is the MPC’s voting record informative about future UK monetary policy?’, Scandinavian Journal of Economics, 106(2), 299–313.
Hansen, S, McMahon, M and Rivera, C (2014), ‘Preferences or Private Assessments on a Monetary Policy Committee?’, Journal of Monetary Economics, 67, 16-32.
MPC (1999), The transmission mechanism of monetary policy.
Stoner, J (1968), ‘Risky and cautious shifts in group decisions: the influence of widely held values’, Journal of Experimental Social Psychology, 4(4), 442-459.-59, 1968.