VoxEU Column Financial Markets International Finance

Ring-fencing is good, but no panacea

The Vickers Commission recommends separating commercial and noncommercial banking activities in order to protect core financial functions from riskier activities. This column warns that such ring-fencing may fail because there are still incentive problems in traditional banking activities. The accompanying risk-weighted capital requirement recommendations will address this only if we do a better job of measuring risks.

The recent report issued by the UK's Independent Commission on Banking, chaired by Sir John Vickers, provided recommendations on capital requirements and contained a proposal to ring‑fence banks – in particular, their retail versus investment activities. I view ring-fencing as potentially useful but argue that the more important question is whether the risk weights in current Basel capital requirements are appropriate.

The backdrop of the Vickers Commission report is that countries such as the UK, Sweden, and some others, where the financial sectors are rather large compared to the size of the countries, are getting increasingly concerned about facing the kind of banking crises that we faced in 2008. The risks of a double-dip recession and a slowdown in global growth have increased given the tentative recovery in the US and the sovereign debt problems in Europe. Hence, some countries are trying for something more substantial in financial sector reforms than what the Basel III reforms are offering.

Sweden has gone for relatively high levels of capital requirements. The UK is unique in considering the ring-fencing solution, which involves trying to separate the riskier parts of banking activities (mainly investment banking and proprietary trading) from what are considered the core or ‘plumbing’ aspects, such as payment and settlement systems, deposits, interbank markets, and bank lending, which in turn are primarily centered in the commercial banking activities. The Vickers report concludes that the risks that the commercial banking system, which is at the centre of the plumbing, faces from noncommercial banking activities are serious enough in the current economic climate that we ought to think about some ring-fencing of this sort.

While ring-fencing seems reasonable when viewed in this manner, there is an important risk that any ring-fencing operation will have to worry about – ring-fencing in and of itself is not a panacea. In particular, banks may be encouraged to take greater risks with activities that are inside the fence, such as mortgages, corporate loans, and personal loans.

Ring-fencing ensures that if risks hit the noncommercial banking aspects or if some mistakes happen there – maybe the current UBS trading loss is an example – then the risks will not directly spill over into the commercial banking aspects. However, if there are risk‑taking incentives inherent within the commercial banking arm as well – and if what we saw happen during 2003-07 through the trading aspects was just a reflection of that deeper problem – then of course we really would not have solved the real problem. We would have likely just transferred it somewhere else.

Therefore some other fixes are crucial. For instance, it is crucial that a resolution authority be in place globally to wind down in an orderly manner a large, complex financial institution (which even some pure commercial banks are) and ensure that their capital requirements are in sync with the kind of systemic risk that such institutions are undertaking.

In fact, a point in favour of focussing more on improving regulation of the traditional banking aspects is that in the end what really brought down banks and complex organisations in the crisis of 2007-08 was not just the quality of their trading activities. A large part of the portfolios of risky mortgages and mortgage‑backed securities were held just as straight commercial banking exposures. That is, these risks appeared in their traditional banking mortgage books themselves. Ring-fencing, by itself, would not necessarily have reduced these exposures.

In this regard, it is useful to consider the capital requirements announced in the Vickers report. Banks will be required to hold equity capital of at least 10% of risk‑weighted assets in the ring‑fenced business, and both parts of the bank will be required to have total loss‑absorbing capital of at least 17% to 20%. This seems substantially higher than what Basel III has proposed. The requirements are in fact more in line with the levels of capital requirements that Switzerland has been implementing. The requirements are also somewhat higher than those tentatively discussed in the US, though we are waiting for further clarity on what will be the systemic capital surcharge for systemically important financial institutions in the US under the implementation of the Dodd‑Frank Act.

The key point is that whatever the capital requirements – 10% and 17% or 18% – it is as a function of risk‑weighted assets. The fundamental question that has not been put on the table is, are the current risk weights – and the overall framework for determining them – right? In particular, we have very low risk weights on residential mortgage‑backed securities. What that did was actually increase the lending to the residential mortgages as an asset class. Endogenously, therefore, ie, as a response to the capital requirement itself, the residential housing became a systemically important asset class. However, all through the crisis, and even post‑crisis, we have continued with a relatively attractive risk weight on this asset class – in spite of the fact that the crisis effectively told us that what banks were holding as capital against these assets was not adequate from a resiliency standpoint as far as bank creditors and investors were concerned.

We are in fact facing a similar problem with respect to the sovereign debt holdings of the troubled countries in Europe. Their bonds are being held by banks all over the world, especially in other Eurozone countries. These bond-holdings have so far, as long as they are in banking books of banks (and hence, not ring-fenced), not received substantial haircuts in regulatory capital assessments. In turn, banks have not been asked to raise substantial capital against them (though we might finally see some pan-European bank recapitalization plan). Investors are, however, treating these bond-holdings as risky so that regulatory bank capital and market values of bank capital are completely out of sync.

Therefore, while increasing the level of capital under some circumstances makes sense, we ought to ask whether the risk weights that go into calculating the required capital are right or not, because otherwise we might be raising capital to 20% of risk-weighted assets but the risk might have been poorly calculated. Worse, banks have incentives precisely to hold those assets whose regulatory risk weights are lowest (or most poorly calculated) relative to the implied market-required weights (which can be implied from market valuations, for example).

The second point is that often the current level of capital held by an institution is not as important as what its level of capital is going to be if it is hit by a substantial crisis. This issue ties in with my first point about risk weights. Today, most regulators who are considering subjecting the banking system to stress scenarios would consider as a stress scenario a substantial haircut on the sovereign debt holdings of some of the peripheral countries in Europe, but treat all others as essentially riskless. Recent moves in the Eurozone countries’ credit risk have shown that bank recapitalization will put sovereign balance sheets under stress, even in the case of relatively stronger sovereign balance sheets. Now, if banks are charged zero risk weight on bond-holdings of these countries, then it is clear that the levels of capitalization required by the stress tests are never going to be adequate for any future stress on expected recoveries and valuations of bonds of these sovereigns.

In contrast, a capital requirement that deals adequately with future systemic risk should be based on bank losses in stress scenarios, where the scenarios consider losses in assets that have not yet experienced any significant risk revisions. Such a requirement has the feature of charging higher risk weights to those assets that are going to lose under future stress scenarios. Such capital requirements can be conceptually formalized as well as empirically implemented, as explained in another piece in this book (“How to Set Capital Requirements in a Systemically Risky World” by Viral V Acharya and Matthew Richardson) which explains how NYU Systemic Risk Rankings are implemented.

In summary, I support the push for higher capital requirements, but I stress that regulators – in Basel, the UK, Switzerland, and the US – all fundamentally need to rethink whether static risk weights, which do not change when the market’s risk assessment of an asset class permanently changes, are really the right way to continue.

If we just keep raising capital to risk-weighted asset ratios but we do not improve our measurement of risks in determining the denominator of these ratios, we have a serious problem on hand. We have moved from a credit bubble in one low risk-weight asset class (housing) to another (sovereign bonds) over the past decade and both have resulted in among the worst crises of our times. It is time to change this state of affairs. 

Author's note: This essay is based on a VoxTalks audio interview between the author and Viv Davies recorded on 16 September 2011.

3,465 Reads