Thank you very much for this opportunity to speak today. With the Independent Commission on Banking approaching the halfway point in its work, now is a good time to take stock of some issues concerning how to regulate the capital and corporate structures of banks.
The broad question that I want to raise today is whether, and if so how, structural reforms might relate to other reform initiatives, especially those to enhance banks' capital structures and the credibility of their recovery and resolution plans to cope with crises.
The scale of the problem
One of the roles of financial institutions and markets is efficiently to manage risks. Their failure to do so and indeed to amplify rather than absorb shocks from the economy at large has been spectacular. The shock from the fall in property prices, even from their inflated levels of a few years ago, should not have caused havoc on anything like the scale experienced. Rather than suffering a 'perfect storm', we had severe weather that exposed a damagingly rickety structure.
Hence the magnitude of the questions being addressed internationally of how to regulate the capital and corporate structures of banks (and other institutions) so that the system is properly resilient in the future. Of course the ICB's remit is limited to UK banking, but our work is fully in line with, for example, international work on systemically-important financial institutions. Besides its UK focus, the ICB's remit has two main differences of emphasis compared with most other initiatives.
The first is the focus on competition as well as stability. We have received a lot of evidence on competition, particularly in retail banking, and are also drawing on the work of the Office of Fair Trading and the Treasury Select Committee. My focus today will however be on financial stability.
The second is that we have been asked for views, one way or the other, on possible structural reforms, including forms of separation between retail and investment banking.
Retail and investment banking
By 'retail' I mean not just payment services and deposit facilities, but also activities such as mortgages5 and SME lending. By 'investment' I mean wholesale and investment banking services including lending and operational services to large corporates as well as trading and other capital markets-related activities. Both sorts of banking are risky6, both are important, and in places the boundaries between them are fuzzy. But the policy challenges for retail and investment banking are somewhat different.
For the most part, retail customers have no effective alternatives to their banks for vital financial services, and hence there is an overriding economic, social and political imperative to avert any disruption to the continuous provision of those services. The task is to find better ways of ensuring this, if possible while allowing unsuccessful individual institutions to fail (safely).
Customers of investment banking services, on the other hand, generally have greater choice and capacity to look after themselves. But it is vital to find ways for the providers of these services to fail safely – a point underlined by the mess of the Lehman bankruptcy. Markets for investment banking services are also more international, as must be policy towards them. By contrast national policies, including competition policies, can bear more directly on retail markets, which tend to be national in scope.
Banking in wider context
Banking, including retail banking, is always going to have risks, and there will be 'maturity transformation' so that savers can get access to funds at shorter notice than it is lent to borrowers. It must also be kept in mind that attempts to minimise risks in one part of the system can shift them elsewhere perhaps to a safer place, perhaps not. Ultimately, financial risks have to be borne, and in a market system they should not be borne by the taxpayer providing a generous safety net.
Before turning to the risks posed by banks and potential measures to contain them, it is worth noting that other aspects of public policy can have the unintended consequence of increasing the risks facing, and maturity transformation required of, the banking system.
First, and notwithstanding incentives in the new UK bank levy, the general tax system encourages both corporate and personal leverage. The corporate tax deductibility of debt interest increases the private cost of equity relative to debt. In the case of banks, therefore, the public desirability of more equity capital meets an opposing force from the tax system. For households, the tax system encourages home ownership and therefore mortgage debt, although mortgage interest tax relief was ended ten years ago.
Second, insurance companies and pension funds, which have long-term liabilities, would seem natural holders of a range of long-term assets. But changes in regulation have limited their ability to accept asset price volatility in the short term. Where regulation discourages non-banks from holding long-term assets which are a good hedge for their long-term liabilities, the result is a kind of 'reverse maturity transformation', which adds to the maturity transformation and associated risks for the banking system to handle.
A wider and longer-term review than the ICB's might usefully explore such points further. Their implications, notably for the cost of bank capital, will be part of our analysis.
Risk externalities and policy credibility
As to the risks posed by banks, an economic perspective in terms of externalities may be a useful starting point. Not for the first time in history, banks and borrowers in the last decade took risks that went bad systemically and imposed huge costs on the rest of the economy and society. The consequences for credit flows and possibly even the payments system and ordinary bank deposits would have been unimaginably worse but for government rescues. It would therefore not be credible for governments to say that they would not intervene to save retail banking services (in the broad sense indicated earlier) in future crises.
The result is two kinds of distortion to banks' risk-taking incentives. First, whether or not they perceived themselves as having enjoyed it pre-crisis, and subject to the soundness of public finances, systemically-important institutions now have an implicit state guarantee for risk-taking activities, particularly those related to and/or inseparable from retail banking. This distortion, which is also a distortion to competition with other institutions, should be neutralized or contained.
Second, banks should be discouraged from taking risks that increase the probability and severity of systemic crisis externalities in the first place. Those externalities would seem to be most acute in relation to retail banking services because of the imperative of continuous provision, but apply also to other services unless there are safe resolution mechanisms for them.
Textbook approaches to negative externality problems are taxes and quotas. Capital and liquidity requirements on banks have elements of both. Like taxes, they have (private) costs, and like quotas they constrain the amount of risk-taking (relative to the capital base). Moreover, their purpose is not only to curb incentives to undertake the externality-generating activity, but also to stop the negative spillovers by absorbing them. Alas the loss-absorptive capacity of bank balance sheets proved to be poor in the crisis.
Brittle bank balance sheets
The growth in bank leverage in the run-up to the crisis was explosive. From the 1960s until a decade ago, UK bank leverage was around 20 times on average. But from 2000 it rose sharply, to 30 times and beyond on average, and in some case much more.
This increased the fragility of bank balance sheets and, when the crisis hit, they turned out to be weaker still. For example, a number of supposedly 'off balance sheet' entities leapt onto bank balance sheets once they got into trouble. In the good times the banks enjoyed the regulatory arbitrage benefits of the formal structures they had created; in the bad times a mix of financial, legal and reputational linkages often dissolved formal distinctions as banks felt they had to absorb 'off balance sheet' losses if they could.
But the capacity of their liability structures, beyond equity, to absorb losses was poor. Balance sheets often proved brittle. While some subordinated debt-holders took losses, senior debt-holders generally came out whole as taxpayers rode, or were ridden, to the rescue. In theory, losses are borne sequentially by equity-holders and then debt-holders by inverse order of seniority, with retail depositors compensated to the extent provided for by deposit insurance schemes. In practice, and despite the state of public finances, fear of the consequences of senior debt-holders coming out less-than-whole forced taxpayers to jump the queue of loss-absorbency.
Whether or not taxpayers end up losing money in the process, they were compelled to shield senior debt-holders for fear of what would happen if they did not, which is not how things should be. Since mid-2007, then, we have gone from a position where insured retail depositors in UK banks were expected to take a 10% haircut to one where all senior debt-holders are de facto fully covered, at least in systemic crises if the public finances can bear it.
Can the potential loss-absorbency of bank debt credibly be restored? It might be a useful first step to make insured retail deposits senior to, rather than on a par with, other senior debt-holders, in the creditor pecking order, but that would not itself be enough.
Contingent capital
A case can be made, in the light of what has happened, that equity is the only really credible loss-absorbing buffer between banks and the state. But, in good part for the tax reason mentioned earlier, equity is expensive (for private interests but much less so socially) relative to debt. Moreover, equity holders may be especially reluctant to issue fresh equity when, as in times of stress, much of the benefit accrues to bondholders the debt overhang problem. Hence the attraction, at least in theory, of contingent capital that converts debt into equity if and when signs of stress appear. The conversion trigger might be automatic for example dependent on some observed capital ratio or subject to the constrained discretion of the regulatory authorities. Neither is straightforward, however.
Potential problems with automatic conversion are that the triggering conditions are lagging, or otherwise somewhat arbitrary, indicators of the need for more equity, or else that they are manipulable by speculative market traders. On the other hand, discretionary triggers may run into problems of price uncertainty, regulatory capture by vested interests, and/or time inconsistency (e.g. reluctance to activate the trigger when the time comes). The system-wide dynamic effects of conversions at times of emerging systemic stress might also be unpredictable. There appears to be a wide variety of views on whether such problems can be solved, and hence on the potential role of contingent capital on banks' balance sheets.
So how to increase loss-absorbing capacity?
If, then, one takes the view that the loss-absorbing capacity of banks needs to be massively enhanced and beyond the prospective requirements of Basel III in the case of systemically-important institutions there are dilemmas about how best to achieve that. Senior debt has generally been poor at absorbing losses (but junior debt less so). Equity is good at it, but (privately) somewhat more expensive, in part because of unintended consequences of the general corporate tax system. And while contingent capital may in theory combine the advantages of debt and equity, there are serious questions about how well it would work in practice. We would welcome further views and analysis on all these points.
Asset risk
The ability of banks to absorb losses needs to be assessed relative to the riskiness of their assets. Capital regulation seeks to address this by way of risk weights. Some versions of 'narrow banking' would eliminate most risk by requiring deposits to be fully backed by government bonds. But that would shift lending – including to personal and SME customers – elsewhere, and it is very doubtful that the implicit government guarantee would be confined to narrow banking. What sort of institutions would have loans on their books under narrow banking? Maybe mutual funds – a kind of mass securitisation. But to ban the funding of ordinary credit by deposits could have considerable economic cost.
While the ICB is unlikely to favour radical forms of narrow or limited purpose banking, their aims deserve recognition. In particular, they seek by structural reform and much higher capital and/or liquidity requirements to limit (or make redundant) the government guarantee of risky activities, so that they can be left to the market. Second, they note the useful risk-bearing role that non-banks can play. As we have seen, however, depending on their linkages with banks both directly and through pro-cyclical market effects non- (or shadow) banks can amplify, as well as absorb, risks to the core banking system.
How corporate structure might matter
A theme in the discussion so far has been that the failure of retail banking services credit provision as well as payments system and deposit-taking services would be especially damaging in terms of wider economic and social costs. Retail (including SME) customers are more dependent on their banks than larger corporate customers, and perceived jeopardy to the continuous provision of those services, particularly by major retail banks, would not only have serious effects on economic growth but could also cause widespread panic amongst the public, notwithstanding deposit insurance schemes.
One response to this concern could be somehow to ring-fence the retail banking activities of systemically-important institutions and require them to be capitalized on a stand-alone basis. A variant of this idea would be to require the ring-fenced retail banking activities to be relatively strongly capitalized, while adopting a lighter regulatory policy towards the other activities (if any) of banks, thereby focussing (and limiting) the need for heightened capital requirements on the key retail services. Such ideas meet objections, however, both about practicability (especially if adopted without international agreement) and desirability. Questions about practicability are for another day. What about desirability?
Is retail banking safer with universal banks?
One of the arguments made in favour of universal banking is that it allows diversification of risks with the result that the probability of bank failure is lower than if retail and investment banking are in some way separated.
To examine the logic of this claim, consider a stylised example in which a universal bank U has two operations retail bank R and investment bank I and compare its failure probability with those of the two banks if they are separated. Assume to begin with that the same business strategies are pursued in each case and that there are no synergies. Then a necessary condition for U to fail is that R or I fails, but this is not a sufficient condition unless R and I both fail if either does. So, in this simple setting, the probability of U failing is lower than that of a failure in the separated regime.
It does not follow, however, that retail bank failure is less likely with universal banking. In this respect universal banking has the advantage that a sufficiently profitable or well-capitalized investment banking operation may be able to cover losses in retail banking. But it has the disadvantage that unsuccessful investment banking may bring down the universal bank including the retail bank. In shorthand, in this simple setting, retail banking is safer with universal banking than with separated banking if and only if the probability that I saves R exceeds the probability that I sinks R.
An important question about forms of combination of retail and investment banking is how this balance of probabilities can be shifted favourably, both within institutions and systemically. A challenge for separation proposals is to guard against risks that problems with investment (and shadow) banking could anyway threaten the continuity of retail banking through financial, legal and/or reputational (including 'confidence') connections. Developing safe ways for providers of investment banking services to fail therefore matters not only because of the importance of those services but also because of the risk of collateral damage to retail banking.
Of course the probabilities of failure are not all that matter. Impact of failure is also critical. Depending on the surrounding circumstances, the negative impact of U failing could be greater than the sum of the impacts of R and I failing individually, especially if they are less likely to fail at the same time.
Moral hazard and the efficient use of capital
As to the simplifying but false assumption of constant business strategies and no synergies, several points are worth making. First, there is the moral hazard concern that the investment banking arm of U effectively has a state guarantee that standalone I may not, so that without countervailing measures (e.g. extra capital requirements) it will have incentives to take on more risk.
Second, in assessing potential synergies, it is essential to distinguish real ('social') efficiencies from private gains that come directly or indirectly at others' expense. For example, access to taxpayer-subsidized capital is a private gain but not a real efficiency; indeed it is a distortion.
Third, the largest synergies claimed for universal banking relate to the efficient use of capital and funding, although there may also be synergies on the operational side. But in this regard it seems quite hard to identify and quantify real efficiencies as distinct from purely private gains. There may be some social benefits of diversification that cannot otherwise be achieved, but to the extent that the 'more efficient' use of capital involves risks being run with less capital, there are implications for default probabilities.
Another point in relation to capital efficiency is that with separated capital pots, sub-optimal capital allocation across types of lending may result. However, market economies appear well able to allocate capital and other resources other than by corporate integration.
We would especially welcome further analysis of questions such as these concerning the efficient use of capital, and how it might be affected by alternative ways of regulating the capital and corporate structures of banks.
Credible resolution plans
Complementary to moves towards stronger capital and liquidity to reduce the probability of failure are policy initiatives to contain the fall-out if failure occurs (or approaches). Because normal bankruptcy procedures work so badly for large, complex and interconnected banks, it is imperative to develop credible recovery plans and resolution tools. Much work is under way in the UK and internationally to tackle this problem. The resolvability of global investment banking operations is a particular challenge, and of heightened importance to the UK given the scale of bank balance sheets relative to GDP.
An important question is whether, and if so how, structural reforms could help this reform programme. Credible resolution would seem to require at least some form of separability, and arguably there is a case for some form of ex ante separation so that bank operations whose continuous provision is truly critical to the functioning of the economy can clearly be easily and rapidly carved out in the event of calamity. But perhaps the credibility of resolution plans can be ensured otherwise than by forms of separation, and the benefits of creating such options would of course need to be weighed carefully against costs they imposed.
Conclusion
This talk has contained more questions than answers, as is appropriate at this stage of the ICB's work. Let me conclude by way of two general questions and one observation about relationships between structural reform possibilities and other banking reform initiatives.
It cannot be disputed that banks of systemic importance need much more loss-absorbing capacity than they had a few years ago, and to be much more easily resolvable. There is a wide range of views on how much more loss-absorbing capacity is appropriate for different kinds of institution, and on how to achieve it by equity, contingent capital, etc.
The first general question is whether, and if so how, structural reform of systemically-important institutions might affect appropriate levels of loss-absorbing capacity. If the probability and/or impact of bank failure particularly of retail service provision can be reduced by forms of separation between banking activities, then so too might capital requirements. If so, the case for structural reform might be greater the higher is the cost of bank capital.
The second general question is whether, and if so how, forms of structural separation might enhance the credibility and effectiveness of resolution schemes.
The observation is that, if forms of separation were thought desirable in terms of public policy, there would be the further question of whether they should be required of the institutions concerned, or incentivized, for example by appropriately different capital requirements for different business models. Riskier structures need deeper foundations.
These and other issues will be central to the ICB's agenda in the coming months.
