Authored by Andrew Haldane (Executive Director, Bank of England); originally posted at VoxEU,
Have We Solved ‘Too Big To Fail’?
That is not my pessimistic verdict; it is the market’s. Prior to the crisis, the 29 largest global banks benefitted from just over one notch of uplift from the ratings agencies due to expectations of state support. Today, those same global leviathans benefit from around three notches of implied support. Expectations of state support have risen threefold since the crisis began.
This translates into a large implicit subsidy to the world’s biggest banks in the form of lower funding costs and higher profits. Prior to the crisis, this amounted to tens of billions of dollars each year. Today, it is hundreds of billions (Haldane 2012). In other words, if the market’s expectations are to be believed, the regulatory response to the crisis has not plugged the ‘too-big-to-fail’ sink.
On the face of it, that sounds perplexing. Rarely a day passes without a warning from the financial industry about overbearing regulation of, in particular, the world’s biggest banks. What is certainly true is that, in the light of the crisis, regulation to quell the too-big-to-fail problem has come thick and (at least in regulatory terms) fast. This reform effort falls into roughly three categories:
(a) Systemic surcharges: of additional capital levied on the world’s largest banks according to their size and connectivity. This Pigouvian tax on systemic risk externalities is built on conceptually sound foundations (for example, Brunnermeier 2009). And, encouragingly, good economics has found its way into good public policy. Last year, the Financial Stability Board (FSB) agreed a sliding scale of systemic surcharges for the world’s largest banks. The highest surcharge was set at 2.5% of capital.
Yet therein lies the problem. Based on my estimates (Haldane 2012), a charge levied at this rate would leave the majority of the systemic externalities associated with the world’s biggest banks untouched. The reduction in default probabilities associated with lowering leverage by a percentage point or two would not offset the higher system-wide loss-given-default associated with the world’s largest banks. The systemic tax is being levied at rates which are too low to meet Pigouvian ends.
(b) Resolution regimes: In principle, orderly resolution regimes for banks could lower the collateral costs of a big bank defaulting, thereby tackling at source these systemic externalities. And significant public policy progress has been made on this front, with the FSB publishing (and the G20 approving) some Key Attributes for Effective Resolution Regimes during the course of the past 18 months. A key component of these plans is the ability to impose losses on private creditors – so-called ‘bail-in’ – rather than have those losses borne by taxpayers.
As with systemic surcharges, the issue here is not to so much the bail-in principle, but its application in practice. Bail-in, whether of big banks, sovereigns or companies, faces an acute time-consistency problem. Policymakers face a trade-off between placing losses on a narrow set of tax-payers today (bail-in) or spreading that risk across a wider set of tax-payers today and tomorrow (bail-out).
A risk-averse, tax-smoothing government may tend towards the latter path – and historically has almost always done so, most notably in response to the present financial crisis. Next time may of course be different. But the market is sceptical. For example, in the US the Dodd-Frank Act on paper rules in future bail-in and rules out future bail-out. Yet market expectations of state support for US banks are higher today than before the crisis struck and are unchanged since Dodd-Frank became law. The time-consistency dilemma, at least in the eyes of markets, is as acute as ever.
(c) Structural reform: One way of lessening that dilemma may be to act on the scale and structure of banking directly. Several recent regulatory reform initiatives have sought to do just that, notably the “Volcker rule” in the US, the ‘Vickers proposals’ in the UK and, most recently, the ‘Liikanen plans’ in Europe. While different in detail, each of these proposals shares a common objective: a degree of separation or segregation between investment and commercial banking activities.
In principle, these ringfencing initiatives confer both ex-post (improved resolution) and ex-ante (improved risk management) benefits. Because they act on banking structure, they have a greater chance of proving time-consistent. While this is a clear step forward, those benefits are only as credible as the ringfence itself. The issue raised by some is whether, in practice, the ring-fence could prove permeable. Without care, today’s ring-fence could become tomorrow’s string vest.
If each of these initiatives is necessary but none is individually or collectively sufficient to tackle too-big-to-fail, what is to be done? One solution might lie in strengthening these proposals. For example, re-sizing the capital surcharge, perhaps in line with quantitative estimates of the ‘optimal’ capital ratio (Miles et al (2012), Admati et al (2011)), would be one option for bearing down further on systemic externalities.
Another more radical option, mooted recently by a number of commentators and policymakers, would be to place size limits on banks, either in relation to the financial system as a whole or, more coherently, relative to GDP (Hoenig (2012), Tarullo (2012)). Proposals of this type typically face two sets of criticism.
The first, practical issue is how to calibrate an appropriate limit. Recent research on the link between and financial depth and growth provides a way into this question. This research has suggested that there is a threshold at which the private-credit-to-GDP ratio may begin to have a negative impact on GDP and, in particular, productivity growth (Arcand et al (2012), Cechetti and Kharroubi (2012)). By taking a view on this aggregate threshold, and on an appropriate degree of concentration within the financial system, an institution-specific threshold could be derived.
The second, empirical issue is whether size limits would erode the economies of scale and scope which might otherwise be associated with big banks. The empirical literature on these economies has, until recently, suggested they may be exhausted at relatively low balance sheet thresholds. But a number of recent papers have painted a more optimistic picture, with economies of scale found for banks with balance sheets in excess of $1 trillion (Wheelock and Wilson (2012), Feng and Serilitis (2009), Hughes and Mester (2011)).
Yet this evidence needs to be interpreted cautiously, not least because it fails to recognise the implicit subsidies associated with too-big-to-fail. These would tend to lower funding costs and boost measured valued-added for the big banks. In other words, the implicit subsidy would show up as economies of scale. Bank of England research has recently shown that, once those subsidies are accounted for, evidence of scale economies for banks with assets in excess of $100 billion tends to disappear (Davies and Tracey (2012)). Indeed, if anything, there may even be evidence of scale diseconomies, perhaps consistent with big banks being ‘too big to manage’.
Too-big-to-fail is far from gone. It is even more important it is not forgotten. Further analytical work, weighing the costs and benefits of different structural reform proposals, would help keep memories fresh and policies on the right track.