This article publishes the annex to a letter sent to the Treasury Select Committee from David Aikman on the topic of FPC accountability and specifically how to quantify its financial stability objective.
The goal of financial stability policy is invariably expressed in imprecise, nebulous terms, affording significant discretion to policymakers in how it should be translated into operational targets and ultimately into policy decisions. The Financial Policy Committee (FPC)’s primary objective to “protect and enhance the resilience of the UK’s financial system” is no exception. In this piece, I discuss some of the reasons why this is problematic and offer some practical suggestions for how we might move forward to begin to remedy this issue.
Problems created by having a vague primary objective
The current formulation of the FPC’s financial stability objective creates several problems.
First, it impedes accountability. The FPC cannot be held to account for its actions if its objective cannot be measured. Instead, we have a situation where it effectively marks its own homework. Moreover, it is unclear from the outside how the FPC even takes its decisions. Accountability only happens ex post in the event of a crisis. Left unresolved, this situation arguably calls into question whether a framework of delegated responsibility for macroprudential policy is even appropriate.
Second, it creates the risk of poorer decision-making. One concern is that, over time, the FPC could end up paying excessive attention to the more measurable aspects of its mandate (competition, competitiveness etc.) to the detriment of the fuzzier but core goal of protecting resilience. These measurable aspects are, after all, where the financial industry’s lobbying efforts will be targeted in the years ahead. A related problem is that an imprecise objective will tend to create a bias towards the status quo by making it easier to avoid taking uncomfortable decisions.
Third, the current objective is unclear on how the FPC should weigh certain considerations in its policy decisions. One is what level of resilience to aim for, and hence how to navigate the poorly understood trade-off between long-run growth and resilience. Another is the distributional effects of its policies, which are typically felt to be larger than is the case with monetary policy actions.
We see an example of some of these problems in the FPC’s judgement (from its December 2021 FSR) that risks to the UK’s financial system at present are “standard”. On the face of it, this seems like an odd adjective to describe present circumstances: besides the unpredictable future evolution of the Covid pandemic, risks to inflation and hence interest rates are higher than they have been for decades. There remain well-documented fragilities in the shadow banking sector. And the backdrop is one with very little room for further monetary (and perhaps fiscal) stimulus, so in the event a shock occurs the resulting stress hitting the financial system may be more severe than otherwise.
Some thoughts on ways forward
I offer three suggestions on ways forward for addressing the problems raised above. The first two are aimed at enhancing accountability and understanding of the FPC’s decision-making process taking the current formulation of the primary objective as given; the third suggests an approach to quantifying the FPC’s objective, and would require a research programme be set in train to develop the idea further.
Greater transparency over decision-making process
My first suggestion is that the FPC could substantially increase the transparency it provides over how decisions are taken. Specifically, it could consider:
- publishing (with a suitable lag) the full transcript of its policy meetings, alongside the agendas and essential staff briefing papers that inform its discussions during the policy process; and
- moving away from its consensus-based decision framework to voting on certain specific policy issues.
It is now a little over ten years since the FPC was first established, and this is a good time I think to revisit the transparency rules it has operated under up until now. In particular, I think it would aid FPC accountability if it were to commit to publishing a full transcript of its policy meetings after a suitable lag. This would bring it into line with the MPC, which in 2015 committed to publish its transcripts with an 8-year lag. Accompanying this, I suggest the FPC commit to publishing the agenda for its meetings and all essential staff briefing papers that inform its discussions. Taken together, this would give the public a greater understanding of the full set of arguments considered in FPC discussions and the weight of argument supporting each.
These issues were last considered as part of the December 2014 Warsh Review. While the focus of that report was the transparency of the Monetary Policy Committee (MPC), Warsh did consider the extent to which his recommendations mapped to the FPC case. On balance, he felt this would have been premature at the time given that the statutory FPC has not long been in existence and the Committee was still establishing its operating framework. He recommended revisiting this issue 3-5 years hence – a date that has now passed.
One objection that is sometimes given against publishing transcripts is that the FPC relies on privileged institution-specific information to inform its debate, information that it would not be appropriate to publish. This is a weak argument in my view. While it is true that institution-specific information is presented to FPC, the focus of its discussions is the stability of the UK’s financial system as a whole. The key elements of its deliberations have this system-wide focus and would be valuable to publish even if supervisory information on individual firms was excluded.
A more convincing argument against publishing meeting transcripts is that it could impair the quality of debate at committee meetings, as members could become more reticent in sharing views that challenge the status quo. The evidence from the Federal Open Market Committee (FOMC) of the Federal Reserve, which began publishing transcripts of its meetings in 1993 (with a 5-year delay), is mixed in this regard. But this risk can perhaps be managed by choosing a suitable delay period and releasing a transcript of only the “Policy” meetings, and not the “Issues” meetings which tend to be more deliberative in nature.
While I recognise these innovations would not be a silver bullet, they would I believe enhance accountability by providing outsiders with a far better sense of the deliberations that underlie policy decisions. In my experience, the FPC’s discussions are wide-ranging and very well informed, and the FPC should have nothing to fear from this information eventually entering the public domain.
My second suggestion is that the FPC should move towards voting on certain key policy decisions that lend themselves to such a process. Voting could take place, for instance, over decisions on the countercyclical capital buffer, on adjustments to the FPC’s housing measures, on adjustments to the leverage ratio requirement – in essence, any quantitative policy decision taken by the Committee.
This would have two benefits in my view. First, alongside the transcript, the voting record would provide outsiders with a clearer understanding of the decision-making process, in particular which areas are particularly contested. Second, voting would guard against status quo bias in my view. It would allow dissenting voices to be aired as they are, without pressure to join the consensus. As a side benefit, this, I expect, would also result in a rebalancing of power within the Committee towards external members.
The architects of the UK’s macroprudential framework opted for a consensus decision model because of concerns that it could prove difficult to design a simple voting scheme that would work well when there were multiple policy options available to address any risk faced. A contrast was made with the MPC, for which voting worked well because the policy decision was one-dimensional. (Should Bank Rate change by 25 basis points or not?) In practice, this risk seems to have been overblown. FPC policy decisions have tended to be one-dimensional in the main, focusing on the appropriate calibration of each tool in isolation. Moreover, MPC decisions themselves have become multi-dimensional (deciding the size of asset purchases, which assets to purchase, the level of Bank Rate) and voting has continued to work well.
A more detailed description of its risk assessment framework
To its credit, the FPC has been admirably transparent in recent years in setting out its view of the appropriate resting point for the countercyclical capital buffer (2% in its latest thinking), alongside the appropriate level of capital requirements for the UK banking system as a whole in normal times. The logical next step in this process is to provide outsiders with an understanding of when it might be appropriate to deviate from these normal-times settings. So, my second suggestion is that the FPC could identify in advance the developments that would lead it to change its view of the level of risks facing the UK financial system such that a tightening in macroprudential policy was required.
Developing the framework in this way would have several benefits. From an accountability perspective, it would provide those charged with holding the FPC to account with a much clearer understanding of how it interprets its mandate and how it intends to act. In particular, it would tell us how “cyclical” the FPC expects itself to be with its tools. This would allow outsiders to debate whether these planned responses are too aggressive or not aggressive enough. From a decision-making perspective, it would help guard against the status quo bias in the Committee’s decisions, helping parliamentarians hold the FPC’s feet to the fire if it fails to act.
What I have in mind with this suggestion is something that goes beyond a purely qualitative description. The FPC’s most recent Financial Stability Report contains a risk outlook chapter which offers a useful starting point from which a quantitative framework of “constrained discretion” could be built. This chapter refers to various indicators, which we can assume implicitly play a central role in the FPC’s thinking. One option for the FPC to consider would be to put numbers or ranges on these specific indicators – or conditional forecasts of these indicators – telling us what readings would lead it to be concerned that risks were growing to material levels, requiring a policy response.
For this to be useful, the list of indicators and associated trigger points would need to be relatively parsimonious. The so-called “Core Indicators” which until recently the FPC published at the back of its Financial Stability Report and which have now been relegated to a separate web annex, are not fit for this purpose. These 58 indicators provide useful data for researchers. But an indicator list of this size is of no use in holding the Committee to account for its actions. This, I suggest, could be a useful occasion to also refresh this indicator list, bringing it into line with the quantitative information the FPC finds useful in its discussions.
In making this suggestion, I recognise that the FPC will not be able to specify in advance all the states of the world that may require it to act. The financial system is evolving continuously, and the next crisis may well have its roots in a product or innovation that does not even exist today. A cornerstone of the UK’s framework is that the FPC must have discretion to act if it sees problems emerging. My suggestion is that the ex ante conditions identified by the FPC should be treated as sufficient but not necessary conditions for action, i.e., there should be a presumption of policy action if indicators reach a level identified as dangerous by the FPC, but the Committee need not wait until such a point before acting. This could be usefully accompanied with a qualitative description of the principles that would lead the FPC to act outside of this framework.
Developing models of financial stability risk
The suggestions above are not sufficient to address the problems created by the vague primary objective of the FPC. They leave two gaps. First, thresholds for simple indicators are not a substitute for a fully-fledged model of financial stability risk that weights information from various sources. The analogy would be the MPC relying on a particular threshold for wage growth or unemployment, say, to guide its monetary policy rather than the inflation forecast. Second, these suggestions do not provide a suitable means for the government to specify the FPC’s goal in more precise terms. We should be aiming for a framework in which the FPC has “instrument” but not “goal independence”.
Recognising this, my third suggestion is that the FPC and Bank should accelerate the process of developing economic models that aggregate information from various indicators to inform its judgement of financial stability risks. The state of the art in the economics-finance literature is a concept called “GDP at risk”. In a nutshell, think of this as an estimate of the most severe recession we might reasonably expect to experience over a defined period ahead, e.g., over the next 3 years. The government could task the FPC will ensuring that the odds of a 5% decline in GDP over this period are no larger than 1-in-20 for example (or whatever tolerance level the government deems optimal).
The key challenge for the FPC would then be understanding how GDP-at-risk changes over time in response to developments in the financial system, as well as in response to its macroprudential policies. We might expect, for instance, a housing boom to increase GDP in the short term but increase the probability of a large recession over the medium term if boom turns to bust. The FPC might choose to tighten its loan to income limit in response, by an amount that brings GDP-at-risk 3-years from now back to its desired level.
This is an attractive concept for financial stability policy for several reasons.
First, it puts the focus squarely on “tail risks”, i.e., low probability, high impact events – on what could happen, rather than the most likely outturn for the economy. In this way, it provides a natural complement to the perspective of the MPC, whose deliberations tend to be focused on expected outturns. There is a natural link too to the stress test scenarios the FPC already sets each year. Second, it allows us to express financial stability risks in common units, in terms of what they mean for overall economic growth, a measurement that is well understood. Third, it provides financial stability policymakers with a consistent framework to guide their decisions – helping to ensure members of the FPC approach their task with the same goal in mind.
There is a debate to be had about whether the relevant index should be GDP or an alternative indicator over which FPC policy has more direct bearing. GDP after all is influenced by many factors, some of which (e.g., pandemics) the FPC has little influence over. Recognising this, if the FPC was given a GDP-at-risk target, there might need to be a clause recognising that certain shocks were outside its remit. An alternative approach worth considering would be to focus on risks to measures of credit supply rather than GDP – the focus could be cast in terms of reducing the risk of a large tightening in spreads on bank loans and on corporate bonds for instance. Another challenge is that large shocks to the economy are, by definition, rare events. So, we have fewer data points to calibrate and validate models against than is the case with conventional macroeconomic modelling.
The GDP-at-risk approach has attracted lots of interest from the central banking and regulatory community in recent years. It already features prominently in the IMF’s regular assessment of prospects for global financial stability. And the European Systemic Risk Board recently published a detailed study examining the how it could be used within a systematic macroprudential policy framework.
Developing a fit-for-purpose model or models of GDP-at-risk would be a significant undertaking. There is no off-the-shelf approach that can simply be taken and implemented. Instead, it would require substantial R&D from the Bank of England and from academia. Despite the scale of the challenge, I think this would be a worthwhile investment. As a first step, the Bank’s research department could be asked to produce a paper exploring options for developing models that tie together developments in the financial system with future risks to economic activity.