The business models of large interconnected banks and the lessons of the financial crisis.

Blundell-Wignall, Adrian ; Atkinson, Paul E. ; Roulet, Caroline 等

This paper looks at the urgent and ongoing need to change the business models of global systemically important banks particularly those that dominate the OTC derivatives markets which carry massive counterparty risk via collateralisation practices. It explores the three main lessons of the financial crisis: too big to fail, excess leverage and conflicts of interest. While regulatory reforms have been plentiful, none have adequately addressed the main source of the problems which lie in the very nature of the business models of large interconnected banks.

Keywords: Financial crises; systemic risk; derivatives; financial reform; non-operating holding company

JEL Classifications: G01; G18; G21; G24; G28

I. Introduction

At the broad generic level, the three most important lessons of the global crisis are: the presence of 'too big to fail' (TBTF) interconnected financial institutions, which causes risk to be underpriced; excessive leverage that permits financial institutions essentially to use other people's money too often to carry strategies, something akin to 'bad private equity deals', (1) with taxpayers bearing all of the downside risk; and conflicts of interest in the way those institutions operate and interact with clients and the regulatory environment. These same crisis mechanisms remain in place today, and policymakers are only just beginning to understand that it is the business models of global systemically important financial institutions (GSIFIs) that need to be changed. The world is only at the beginning of what will in the end have to be a period of massive change in the banking industry.

This short paper cannot hope to deal with all of the relevant issues, which have been developed at length by the current authors from the very beginning of the crisis. (2) Instead it focuses on GSIFI institutions and derivatives to illustrate some of the key issues for policy, and expressly focuses on OECD research and views published by the OECD since 2007. This is not to say that traditional banking didn't also play an important role in the crisis. It is clear that house price and mortgage boom-bust mechanisms were also present--in the USA, Spain, Ireland and the United Kingdom, for example. But even here it is difficult to separate this from the role of financial innovation, including securitisation of mortgages and derivatives to structure these products for distribution to investors. The role of GSIFIs at the top of the food chain in the interbank market, their domination of derivatives trading and their key roles in market making, underwriting, prime broking and product innovation for distribution throughout the financial system makes the structure of their business models a prime concern for policy.

2. Derivatives and too big to fail

GSIFIs are simply too big to fail. While derivatives have an outstanding notional value (relevant for exposure and fees), they are contracts between two counterparties that trade on margins collateralised in the most part with cash (and sometimes government securities). These collateralising margins are referred to as 'independent' amounts for bilateral contracts or 'initial margins' for cleared contracts. As volatility and prices of the reference securities move, so too does the value of derivatives, and losing (out-of-the-money) counterparties transfer 'variation margins' in favour of winning counterparties. ISDA surveys suggest that roughly three quarters of trades comingle these margins in non-segregated accounts, and over 90 per cent of the cash is used in re hypothecation. (3) The latter refers to the practice whereby as title to the collateral is transferred contractually in the trade, this can be re-used by the recipient as collateral for new derivatives trades, resulting in multiplier effects for gearing and exposure (sometimes referred to as the 'velocity of collateral'). The amount of leverage in this process is enormous and this leverage of collateral is by its very nature one that gives rise to interconnectedness between financial firms.

A simple illustration is provided in figure 1 for the CDS contract. In this example notional protection of 100m [euro] is bought, and a 50 per cent recovery rate in the event of an actual default is assumed (so the maximum final value of the contract payout in a default would be 50m [euro]). A four-period model is used. In the first period, four successive revaluations of the survival in each of the subsequent periods are considered: 95 per cent, 90 per cent, 70 per cent and 30 per cent. The bottom rung shows the value of the contract where the probability of the reference entity surviving in each of the four periods is 95 per cent. So the probability of default over the life of the contract is only 19 per cent, shown on the lefthand side, and the value of the contract is 4.6m [euro]. (4) The second rung shows a rise in the value to 11.7m [euro] as the survival probabilities have fallen, resulting in a 34 per cent probability of default over the life of the contract. This rises to 33.3m [euro] for a 76 per cent chance of default and 45.2m [euro] for a 99 per cent chance. There has been no default event, but at each stage in this process the position has to be collateralised, and the writer of this contract has to provide the cash (frequently borrowing it).

If a bank's counterparty fails to post collateral in such cases, the hole in the balance sheet raises the risk of transacting with that institution, and all banks and intermediaries will begin to take defensive action. A dealer bank, A, at risk to the insolvency of the writer bank, B, can try to cover by borrowing from B, or by entering into further offsetting new OTC derivative contracts with B (that can be netted). However, all of these actions exacerbate B's weak cash position. The most likely defensive response of exposed counterparties would be to request novation away from the bank concerned. This creates huge pressure for the bank under attack, as it has to transfer cash collateral to the new bank. This means selling assets and unwinding trades at possibly fire-sale prices.

[FIGURE 1 OMITTED]

The moment a bank does not have a sufficient cash buffer, short-term securities of sufficient quality, or the ability to borrow to meet collateral calls, it is essentially, in the absence of direct official support, going to go rapidly into a failure situation. This most recently occurred with Dexia in October 2011, when it failed to meet margin calls worth around $22bn. For GSIFIs such public support is always there, and this support itself becomes a part of a much bigger problem. Asset managers, hedge funds and the like prefer to deal with GSIFI banks precisely because this public support can be relied upon. Were it not there, the cost of capital would be much higher. First, counterparties to the bank would demand segregated accounts and no re-hypothecation--collateral re-use plays a fundamental role in reducing collateral costs. Second, any securities used for collateral would attract bigger haircuts. Third, risk premia for lending collateral would rise. This implicit TBTF cross-subsidisation from the taxpayer results in risk being underpriced, resulting in excessive risk-taking, and bouts of unexpected volatility can lead to failure. The best example of this is the bailing out of AIG, which had written CDS contracts for most of the GSIFI banks for the purposes of regulatory arbitrage. (5) The cost of public support, when close out netting began, is shown in table 1. In the case of some of the GSIFI banks this single counterparty exposure amounted to more than a third of their equity less goodwill. It would have been higher had the rescue not been mounted when it was. Some GSIFI banks claim that they received the payments but were not exposed to the risk of AIG failure, on the grounds that they had bought CDS against AIG. (6) However, only other GSIFI banks could have provided this 'insurance', and they too were all heavily exposed to AIG--had AIG not been saved this insurance would have been a mirage.

3. Derivatives and leverage

Derivatives are the ultimate form of leverage. First, by making a small cash down-payment (the independent amount or the initial margin), broker/dealers can take large speculative positions in the market, and transform the riskiness of their assets and income flows, while booking revenue from fees and OTC derivative spreads. These activities enable the structuring of tax-effective products that are attractive to clients. Through re-hypothecation, this leverage can be multiplied throughout the financial system. Second, derivatives play a critical role in regulatory arbitrage under the Basel system, which essentially permits banks to have a wide discretion in risk-weighting their assets for regulatory capital charges.

Under the Basel system, lending between banks and appropriately regulated financial firms was treated for risk-weighting in the same way (at 20 per cent). Glass-Steagall might have separated securities activities and insurance from banking, but this was not reflected in the cost of capital for lending between institutions. Securities businesses in general became too large as regulators contributed to the underpricing of risk, particularly after 2004. Leverage accelerated as a direct consequence of perverse regulatory changes. First, once Glass-Steagall was removed at the end of the 1990s, banks previously excluded from securities businesses entered the market through acquisition and organic growth in order to improve their share of the lucrative structured product sector built around derivatives and leverage. Second, Basel II did not alter the 20 per cent risk weight for lending between financial firms and, in addition, actually gave GSIFIs the right to run internal models to assess the riskiness of their portfolios for regulatory purposes. This possibility opened the way for banks to use model-based risk-weight optimisation and the structuring of risk via derivatives to determine the size of risk-weighted assets (RWA) to which the new capital rules would apply. Third, in 2004 the SEC agreed to rules permitting stand-alone US investment banks to be regulated on a 'consolidated entities basis' (to enable them to operate according to the minimum Basel standards required in Europe--effectively removing the leverage constraints on the broker arms of securities firms that had hitherto applied). From 2000 until the advent of the crisis, GSIFIs reduced the ratio of RWA versus total assets (TA) to minimise their cost of capital (see figure 7 below). (7) Derivatives were integral to this process and to the explosion of bank leverage over this period.

Figure 2 shows the notional value of derivatives as a share of world GDP and the same measure for primary securities (consisting of all issued domestic and international securities, bank intermediated credit and equity market capitalisation). From the late 1990s, the notional value of derivatives rose from around 3.5 times world GDP to a staggering twelve times world GDP on the eve of the crisis. Primary securities, on the other hand, have been relatively stable at around three times world GDP. The strong growth in OTC derivatives has been a crucial aspect of the recovery of securities market profitability, despite the increase in the number of players. Specialist securities firms like Morgan Stanley (MS) and Goldman Sachs (GS) were making a 30 per cent return on equity just prior to those profits collapsing during the crisis, and the new players like Citi and Bank of America (BAC) were not far behind (see figure 3).

[FIGURE 2 OMITTED]

It is often argued that the leverage implied by notional exposures is irrelevant from a risk viewpoint, for two reasons. First, because the settlement exposure--i.e. Gross Market Value (GMV) that measures what it would cost to replace all the trades at current market prices--is markedly smaller. The global derivatives GMV is shown in figure 4. While the notional value was around $586tn on the eve of the crisis (December 2007), the GMV at the same time was only $16tn--not unlike the net settlement amount on which the trades are based (at about 2.7 per cent of the notional). Second, financial firms have offsetting positions that can be netted and banks expressly hedge most of their positions. Figure 4 also shows these netting amounts--if all positions were to be closed at a point in timeout, then only the net amounts would be settled. The GMV less netting is the Gross Credit Exposure (GCE).

[FIGURE 3 OMITTED]

It is against the GCE that collateral is held. Figure 5 shows the global value of GCE and estimated collateral. So from $585tn notional on the eve of the crisis, and $16tn GMV, the GCE was a mere $3.3tn and, against this, $2.1tn collateral was held. The net global open exposure of only $1.2tn in December 2007 seems barely measurable versus the notional--yet a massive global crisis involving derivatives was about to ensue. The idea that derivatives are not a source of systemic risk because the open position may seem small is one of the great misconceptions about derivatives.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

Derivatives fund nothing, but serve to shift exposures from one party to another and work through margins (collateral), yet they carry all the bankruptcy characteristics of debt for the out-of-the-money party. A sudden move in volatility can shift the GMV quickly, and netting provides no protection for this. Netting is about settlement amounts using prices at the point of close out. Netting does not protect any financial firm from market risk. On the eve of the crisis, the GCE net of collateral was $1.2m, and the 22 GSIFI banks considered in this paper which dominate the derivatives business only had around $1m in capital. (8) As the crisis hit in 2008, the GMV rose by $19.5m (from $15.8 to $35.3), the GCE rose $1.7m (from $3.3m to $5m), and estimated collateral had to rise by a whopping $1.9m (from $2.1tn to $4.0m). Counterparties had to make these margin calls in a highly risky environment--the question of who would lend to them would go to the very centre of the financial storm that was unfolding.

Figure 6 shows the size of the balance sheets of the US Federal Reserve, the ECB, and the Bank of England, in USD, on the right-hand scale of the chart. It also shows the ratio of estimated total collateral outstanding (including collateral re-used via re-hypothecation). The sudden surge in demand for collateral led to an extreme shortage of cash, as financial firms engaged in the defensive actions noted earlier, such as refusing to lend to each other to settle margin calls, requesting novations, and the like. A rough measure of this stress is provided by the ratio of collateral to the balance sheets of the central banks. At the worst point in 2008, this ratio reached 97 per cent. Subsequent policies by all three central banks relieved that pressure. The central banks became the interbank market, and this lender-of-last resort function helped to settle market volatility. (9) The GCE less collateral shown in the chart fell significantly as the quantitative easing policies took effect. It is difficult to imagine what might have happened if a series of financial institutions in addition to AIG and Lehman Brothers had failed to meet margin calls. It is also difficult to imagine how all this additional central bank money can be safely removed without changing the business model of GSIFI banks.

[FIGURE 6 OMITTED]

4. Derivatives and conflicts of interest

The structuring of products via securitisation, swaps, use of seniority tranches and CDS insurance was an integral part of the growth of derivatives discussed above. The search for higher yields in a low interest rate environment and tax arbitrage presented a profitable business model, provided credit ratings could be found to make them seem attractive to investors. (10)

Consider a high (H) and a low (L) coupon bond, H at a 10 per cent coupon and L at an 8 per cent coupon. One investor is tax exempt while the other investor is subject to a 50 per cent tax rate on bond H and a 25 per cent tax rate on bond L. The non-taxable investor can buy bond H with the proceeds of shorting bond L and capture 2 per cent per annum of the face value traded with no initial investment (other than initial margin). The taxable investor can buy bond L with the proceeds of shorting bond H and capture a 1 per cent spread after tax with no investment. Both traders gain as long as the taxable investor can utilise the tax deductions. The combined profits realised by both trading partners, after tax, comes at the expense of a reduced government tax liability. With performance-based remuneration, these sorts of transactions give strong incentives for GSIFI banks to create structured notes that are very interesting to investors, giving rise to returns and risk profiles that they might not otherwise achieve. Since the tax system is essentially a parameter of the trade, and there are plenty of high-yield securities in the pipeline, this activity can go on for a long time prior to a crisis, provided credit ratings can be achieved to make them saleable to pension funds, mutual funds, insurance companies and private clients. The issue>pays model for credit ratings is a massive conflict of interest that led to ratings inflation, and hence to the successful sale of such products. Some GSIFI banks, once the crisis set in, were able successfully to construct products, sell them to clients and then short the incorrectly rated products for their own proprietary account. (11)

A different form of conflict of interest concerns regulatory arbitrage. As noted earlier, the Basel system allows GSIFI banks to run their own models to determine risk weights and this, together with the use of CDS and swaps, allows banks to optimise their risk weights to reduce the amount of capital they are required to hold (calculated as a percentage of RWA). A simple example is illustrative: bank A lends to company XYZ (100 per cent risk-weighted) and then buys CDS insurance from a bank B counterparty which is only 20 per cent risk-weighted. By shifting the promise to pay from company XYZ to bank B, the capital charge to bank A is all but removed. Bank B, in turn, then shifts the promise again by underwriting the CDS contract with a re-insurer outside of the banking system and possibly in another jurisdiction. The charge for this is very small. In this example, the bank reduces capital charges and leverage is expanded. Figure 7 shows the average ratio of RWA to TA of the GSIFI US, UK and European banks--where US banks are put on an equivalent IFRS basis. (12) The ratios have been systematically reduced, and in some individual banks are as low as 20 per cent. (13) Only a leverage ratio on IFRS assets can provide effective capital rules that do not encourage banks to engage in regulatory arbitrage. (14)

[FIGURE 7 OMITTED]

Both of these examples could in principle be avoided if corporate governance refused to take advantage of the enormous opportunities to arbitrage the tax and regulatory systems and to collude commercially with credit rating agencies in the process. But this is like leaving free money on the pavement and expecting passers-by not to pick it up. Corporate governance did not and does not work to take the place of prudential policymakers. Indeed conflicts of interest within the governance structure itself are rife, and the need for reform--not just of compensation rules that are easily avoided--is essential if business models are to work properly. These conflicts include, inter alia:

* The process of selection of board members, where the CEO in practice makes the appointments, making a 'nonsense' of 'independence'.

* The fit and proper person test is assessed only in terms of the absence of a history of fraud or bankruptcy, as opposed to any genuine need for expertise in banking and the technicalities of risks in derivatives and related products.

* The Chairman of the Board and the CEO that he or she oversees is the same person in many cases.

* A risk officer reporting line only to the CEO who can hire and fire them.

* The practice of large payouts even for CEOs who fail.

* Granting options to directors and management, both of whom have inside information about the company.

* The ability of management to determine their own compensation packages, with non-independent directors and the ability to win voting battles at general meetings (as a consequence of non-active institutional investors, non-voting share structures, use of borrowed shares, influence on proxy advisory companies, etc.).

* The use of proxy advisory companies that recommend on how to vote, but which also have commercial advisory relationships with the companies.

* The use of borrowed securities in voting.

5. Changing the business model of GSIFI banks

The OECD has long argued that the best way to deal with the three lessons of the financial crisis--TBTF, excess leverage and conflicts of interest--is far reaching changes to bank business models and a rethinking of the Basel approach to leverage.

TBTF and changes to the GSIFI bank business model

The distinction between universal banks and Glass-Steagall separated banks, insurance companies and investment banks was less important in the 1980s and 1990s, because traditional banking products and securities were segmented by 'incomplete markets', and the leverage of broker/dealer arms of investment banks was tightly controlled by the SEC. However, the process of innovation in financial markets began increasingly to transform the nature of traditional bank products, for example via securitisation and the advent and greater use of swaps and CDS, at a time when financial deregulation was taking place. The financial sector moved from a world of 'incomplete markets' in bank credit to a state of 'complete markets', where products could be created and traded long or short like any other security, and leverage rules were relaxed.

The OECD favours the separation of GSIFI banks into business segments which deal in products that have quite different prudential qualities- and not merely separating retail and investment banks as such--it is what banks do that matters. (15) This paper has focused on the example of the prudential risks of derivatives and structured products and their funding through credit channels. These products should be created and distributed via separate affiliates or entities within the banking group. But this achieves nothing if: (i) the entities are not ring-fenced legal affiliates that cannot pass assets, liabilities and capital between each other or via the parent (without regulatory approval); and (ii) a resolution regime is not in place for those affiliates that will explicitly be allowed to fail. How does this proposal address the TBTF cross-subsidisation of risk?

A natural minimum separation is between: (i) the traditional bank, with retail deposits from unsophisticated investors and assets such as loans that are amortised cost accounting products; and (ii) securities businesses, particularly those dealing with OTC derivatives. With respect to derivatives businesses, the presence of implicit and explicit guarantees, including access to the lender-of-last resort, cross-subsidises the activity. How so? When a fund manager, insurance company or hedge fund is setting up a counterparty relationship, it is very important to have access both to as much capital of the group as possible, in the event of a problem, and the strong likelihood that the group will be saved via the lender of last resort or other forms of aid in the midst of a crisis. Consider the following two choices:

* A counterparty relationship with a TBTF GSIFI bank with $40bn in capital, and access to the lender of last resort; or

* A counterparty relationship with a Cayman Islands armslength affiliate of the TBTF GSIFI, with only $5m in capital.

The answer is obvious enough. Now imagine that the regulations ensure that capital and assets cannot be passed between the affiliates and the parent in the event of a crisis, and the securities affiliate will be allowed to fail under a resolution regime. This removal of implicit guarantees will cause the cost of capital for the securities firm to rise. With less capital to chase in the event of failure, and the absence of official support, credit ratings would be lower and clients would demand segregated margin accounts, greater independent amounts/initial margins and 100 per cent cleared variation margins. The securities business would become smaller and less levered and the bar would be raised for dealing in high-risk products. Securities businesses with separation would be less likely to fail and, if they did, they could be more easily resolved without the disruptions seen in 2008/9, 2010 and 2011.

The CEO of the separated securities business would have to earn profits with clients aware that it would have access to less capital in the event of failure, and that failure would not result in the socialisation of losses. The experience of hedge funds during the crisis provides some guide. They did not benefit from implicit cross-subsidisation from the official sector, and for the most part did not lever the positions to the same extent as GSIFI banks. Hedge funds failed, but they did not create systemic crises.

Alternative proposal: 'bail-in bonds'

Other proposals have been made to deal with the TBTF problem and GSIFI resolution without separation. For example, bail-in bonds have been proposed to deal with these issues. (16) Essentially, just before the point of failure, supervisors intervene and equity holders are wiped out first, then bail-bonds are converted to equity. This approach has many problems.

* Supervisors are notoriously poor at following what is happening at a bank and therefore declaring a point just before failure. With derivatives--the focus of this paper--JP Morgan's recent loss of an amount that began at $2bn and has since risen substantially occurred despite a substantial number of supervisors in residence. Dexia failed through the margin call mechanism. The 'point of failure' would be declared most likely after the real point of failure. In this case the bond holders will take a loss anyway, and the institution would not be a going concern unless, like Dexia and similar cases, the government intervened directly and paid out its creditors.

* If a bond bail-in is declared well before the point of failure, on the other hand, it could well accelerate selling of all of the securities of the bank.

* It is unclear why in-the-money holders of the derivative contracts should be settled at the expense of other security holders--unless the bonds were sold in the first place with this possibility transparently specified in the contract. But if this is the case, it is unclear why anyone would buy them. Selling covered calls on the bank would provide a similar risk return trade-off for investors.

Even if all of these issues could somehow be addressed by very clever supervisors, and bail-in bonds could be sold in sufficient quantity to fundamental investors like pension funds and insurance companies, the entire approach remains one that does not provide the incentives to avoid the underpricing of risk in the first place. In other words, in terms of cross-subsidising risk-taking through derivatives, it does not matter whether the implicit guarantee comes from the government or a pension fund. The fact is that counterparties to the bank will be paid regardless of the source, and risk will remain underpriced. It does not address one of the key causes of the crisis. This is not the case with the separation mechanism proposed by the OECD--an ex-ante approach that works by shifting the cost of capital to reflect risks without the TBTF distortion.

Alternative proposal: liquidity ratios

Reflecting the concerns outlined earlier, the Basel framework has proposed the liquidity coverage ratio (LCR), which requires banks to hold unencumbered, mainly zero-weighted, assets (such as cash or government debt) to withstand 30 days of continuous liquidity stress. Aside from biasing bank balance sheets against lending, this approach does not focus on changing the ex-ante cost of capital differentially for prudential segments within a banking group. (17) Instead, it focuses on the ability to survive once a liquidity crisis hits. It does not seem sensible to tie up vast pools of liquid assets that will affect traditional bank lending to the private sector, when most of the liquidity pressure in a crisis is likely to be due to counterparty pressures in securities businesses. It is the business model of banking groups and the appropriate ex-ante pricing of risk that are the main issues.

The NOHC proposal

The OECD has long proposed a non-operating holding company (NOHC) structure. The US Dodd-Frank Act ('Volcker rule') and the UK Vickers Report have come to similar broad conclusions on the need for business models to be changed by separating traditional banking from securities businesses to a greater or lesser extent. The three regimes are compared in table 2.

The Vickers report has more in common with the NOHC model. (18) The main differences between Dodd-Frank and Vickers/NOHC are:

* Dodd-Frank focuses on prohibiting insured depository institutions from engaging in certain activities involving proprietary trading (see table 2), i.e. "engaging in trading activity in which it acts as a principal in order to profit from near-term price movements'. (19) But there are important exceptions (see table 2) and this requires qualitative rules to be written separating activities in terms of the intent of the trade.

* The Vickers/NOHC approach, on the other hand, focuses on different prudential 'buckets' within a banking group that can be ring-fenced, so that the adjustments fall on the pricing of risk in the different activities, thereby helping to avoid (inappropriate) cross-subsidisation of risk-taking via TBTF channels.

* The Dodd-Frank Act also has extra-territoriality issues that are more pervasive than Vickers or the NOHC. (20)

Under the Volcker Rule the guidance for rule-making in key securities market activities is that they be: "designed not to exceed the reasonably expected near-term demands of clients, customers, or counterparties.'" There is a 'backstop' prohibition for the permitted activities: "Under the Volcker Rule, permitted activities are subject to a "backstop" that prohibits these permitted activities if they result in a material conflict of interest, result in material exposure to high-risk assets or high-risk trading strategies, pose a threat to the safety and soundness of the banking entity, or pose a threat to financial stability". (21) This rule-based approach to separation may be difficult to interpret, implement and monitor in practice, and banks have been very adept at finding ways around such complex rules in the past. (22) Three permitted activities, underwriting, market making and prime broking, serve as an illustration:

* Underwriting: is exempted from Dodd-Frank provided it is done on behalf of clients as a service, but not for profit driven by market prices. Yet underwriting a securities issue puts the bank's capital at risk and it would be unreasonable to expect this activity to be undertaken without a profit incentive. If the book build on behalf of a client proves to be inadequate, so the issue of (say) a bond is under-subscribed, the bonds would have to be taken into the banks' own inventory and laid off later. The issue may have been mispriced, or fundamentals might change in the future, resulting in prices moving for or against the bank. This will result in profits or losses. The bank will need to have some profit motive in mind in running an underwriting business over time, or it will not be able to continue with this activity. Underwriting could simply migrate to the shadow-banking sector or to foreign banks not subject to the Volcker rule, which brings about separation. But, depending on the final writing of the rules, such proprietary risk may well be allowed to rest with the bank.

* Market making: this is also exempted if it is not done on a for-profit basis. Yet market making is inherently a form of proprietary trading. (23) The bank provides immediacy to a client by buying at a quoted price, taking the security into inventory, and laying it off at an average price over subsequent periods. As before, the bank won't undertake the activity if it can't quote with a profit motive in mind to compensate the risk of adverse price movements in the uncertain future. This may lead to the migration of the business to elsewhere as a consequence of the Volcker rule. But if the interpretation is that the rules will allow it, then considerable proprietary risk is likely to stay in the bank.

* Prime broking with a covered fund in which the banking entity has an ownership interest: these transactions are exempted provided they meet the de minimis rules (see table 6), the bank provides no guarantees to the fund, and sections 23A and 23B of the Federal Reserve Act (dealing with armslength transactions) are complied with. Prime broking, however, requires banks to provide immediacy to hedge funds in terms of the borrowing stocks and the execution of sales. These functions require inventory--which drive profit and loss via short-term price movements that have to be managed with profit motives in mind.

Controlling leverage

Figure 8 shows the equity positions of major GSIFI banks. The 22 banks were selected with a view to containing all of the large holding companies and investment banks that own banking licenses (allowing them to benefit from lender-of-last-resort functions, deposit insurance and other such implicit subsidisation of their business models) while including all the banks that dominate the derivatives business. Sufficient geographical coverage of countries that have been prominent in the global financial crisis was also a consideration. The banks covered are:

* United States: Bank of America (BAC), Citi (C), JP Morgan (JPM), Morgan Stanley (MS), Goldman Sachs (GS) and Wells Fargo (WFC).

United Kingdom: Barclays (BCS), Royal Bank of Scotland (RBS), Hong Kong and Shanghai Banking group (HSBA), and Lloyds (LLOY).

* Germany: Deutsche Bank (DBK) and Commerz Bank (CBK).

* France: BNP Paribas (BNP), Societe Generale (SGE) and Credit Agricole (ACA).

* Switzerland: Union Bank of Switzerland (UBS) and Credit Suisse (CS).

* Italy: Unicredit (UCG) and Intesa San Paolo (IS-NPY).

* Spain: BBVA (BBV), Santander (SAN), Banco Popular Espanol (BPOP), Banco de Sabodell (SAB).

[FIGURE 8 OMITTED]

To make the US banks as comparable as possible to their European counterparts, the GMV of derivatives, not including hedging derivatives, is added back into the US bank balance sheet figures. (24) The leverage ratio is total assets (so defined) versus tangible equity, i.e. equity less goodwill and other intangibles. The chart shows the ratio of equity less intangibles to total IFRS assets. The problem of lack of capital to absorb losses is most acute in Europe.

[FIGURE 9 OMITTED]

The average equity ratio of GSIF1 banks in Europe is just above 3 per cent, compared to around 4.5 per cent in the USA and the UK. 'Good deleveraging', involving the injection of equity into banks has been carried out more forcefully in the USA and the UK, whereas Europe has seen more 'bad deleveraging', involving less new equity and more balance sheet contraction. (25)

The detailed bank positions for the individual banks are shown in figure 9. The positions at the end of 2011 are shown in the black columns and the positions at the end of 2008 are shown with the red columns. The OECD 5 per cent Rule is also shown with the horizontal line. Only a few of the GSIFIs (which have high Tier 1 ratios) meet the OECD view that well-capitalised banks should meet a leverage ratio of 5 per cent. Risk-weight optimisation is the main cause of this difference, with the Basel system having no view on the appropriate ratio of RWA to TA.

The most recent GSIFI failure via the margin call mechanism was Dexia: in a recent press release explaining events they reported a total equity position consisting of -320m [euro] in December 2011, i.e. negative net equity failure, while their Tier 1 and core Tier 1 ratios were a relatively healthy 7.6 per cent and 6.4 per cent, respectively. (26)

The leverage ratio proposal for twenty times (the 5 per cent rule) would apply at the NOHC group level. For the bank, the capital standard would most likely be more conservative and not conducive to creating housing/lending bubbles. The securities business would require capital too, and the leverage ratio there would be a residual depending on the difference between the bank and group level ratios. A further great advantage of this structure (in the current environment of 'bad deleveraging') is that the separated traditional bank could go on lending to households and SMEs regardless of liquidity issues and possible failures in securities market affiliates.

Conflicts of interest

The main way to deal with conflicts of interest is to reduce the opportunity set of conflicts and to improve corporate governance. Amongst things that should be taken off the table are:

* The Basel III risk-weighting approach based on internal models which gives rise to risk-weight optimisation. Banks have been brazen enough to write to clients at crucial policy change periods to say this is exactly what they will do--change the model rather than raise more capital. (27) A binding leverage ratio based on IFRS assets is required.

* Different regulatory and tax ratios within and between jurisdictions that provide the arbitrage opportunities for derivatives trading. (28)

* The user-pays model for credit rating agencies. Getting elements of investor pays is important, and this can best be achieved in a platform-pays model favoured by the authors. (29)

* The possibility of the parent shifting assets, liabilities and capital between affiliates for insolvent affiliates without full regulatory approval.

* TBTF cross-subsidisation, which is a massive temptation for bonus-hungry businesses to make easy profits in 'normal' periods prior to crises--using other people's (cheap) money while claiming there is somehow some skill in this. (30)

* Corporate governance anomalies that are conducive to taking advantage of conflicts of interest vis a vis clients and short-term profits versus long-term prudential goals.

The OECD put forward a comprehensive compendium of necessary basic reforms very early in the crisis. (31) These include, inter alia:

* A transparent accounting of the real situation for losses and bank balance sheet positions, separating bad assets and proactively recapitalising banks. A set of rules for dealing with the disposal of 'bad' assets in the exit strategy.

* The implementation of the NOHC structure with ring-fencing for bank business models in the world of counterparty risk.

* A leverage ratio based on IFRS assets, to get away from low capital and regulatory arbitrage.

* Separating the roles of Chairman and CEO. The CEO should have no role in board nominations.

* Technical expertise in banking, accounting and risk being a formal requirement in the 'fit and proper person test'.

* Term limits on all board members.

* Direct access of the chief risk officer to the board, and his/her terms of employment having some independence from the CEO.

* Transparent compensation arrangements and payout terms for management, based on long-term performance and subject to shareholder approval.

* Clearly-defined fiduciary duties for board members closely tied to the bank, without multiple- and cross-directorships.

* Strong firewalls in proxy advisory companies.

6. Concluding comment

The global financial crisis resulted in extreme losses, the full ex-ante extent of which will never be fully known. To avoid failure many major banks employed less than transparent ways of dealing with mark-to-market losses, and regulators and policymakers had to make decisions in the trade-offs between transparency and the costs of bank resolutions and government guarantees. The unprecedented use of central bank balance sheets in response to the urgent demands for liquidity of interconnected financial firms provided a breathing space. The regulatory response has been extensive--including the move towards Basel III, revised trading book rules, Dodd-Frank, including the separation of proprietary trading (Volcker rule) and separation of retail from wholesale banking under the Vickers Report in the United Kingdom. Nevertheless, and despite all of this, GSIFIs continue to resist vigorously the need for change in their business models, even as large losses and failures continue long after 2008 and the emphasis on reform and the need for improved governance: the most recent examples being JP Morgan Chase, Dexia, MUFG, UBS and others.

The world is only at the beginning of thinking about how to address the problems with GSIFI business models.

REFERENCES

Blundell-Wignall, A. (2007a), 'innovation, growth and equity', speech at the OECD Forum, May.

--(2007b), 'Structured products: implications for financial markets', Financial Market Trends, issue 2, OECD.

--(2012), 'Ratings agencies issues', presentation to the French Senate, 27 March.

Blundeli-Wignall, A. and Atkinson, P.E. (2008), 'The subprime crisis: causal distortions and regulatory reform', in Bloxham, P. and Kent, C. (eds), Lessons from the Financial Turmoil of 2007 and 2008, Reserve Bank of Australia, July.

--(2010), 'Thinking beyond Basel III: necessary solutions for capital and liquidity', Financial Market Trends, issue I, OECD.

--(2011), 'Global SIFIs, derivatives and financial stability', Financial Market Trends, issue I, OECD.

--(2012), 'Deleveraging, traditional versus capital markets banking and the urgent need to separate GSIFI banks', Financial Market Trends, issue I, OECD.

BlundelI-Wignall, A., Wehinger, G. and Slovik, P. (2009), 'The elephant in the room: the need to focus on what banks do', Financial Market Trends, issue 2, OECD.

Duffle, D. (2012), 'Market making under the proposed Volcker rule', Rock Center for Corporate Governance at Stanford University Working Paper no. 106.

Financial Stability Board (2011), Effective Resolution of Systemically Important Financial Institutions: Recommendations and Timelines, FSB Consultative Document.

Financial Stability Oversight Council (2012), Study and Recommendations on Prohibitions on Proprietary Trading and Certain Relationships with Hedge Funds and Private Equity Funds, January.

General Accounting Office (2009), Troubled Asset Relief Program, Report to Congressional Committees, 'Status of Government Assistance Provided to AIG', September.

ISDA (2012), ISDA Margin Survey 2012, international Swaps and Derivatives Association, May.

OECD (2009), The Financial Crisis: Reform and Exit Strategies, Paris, OECD.

--(2010), Competition and Credit Rating Agencies, Hearings of the OECD Competition Committee.

UBS (2008), Shareholder Report on UBS Write-Downs, UBS, AG, 18 April.

Vaughan, L. (2011), 'Financial alchemy foils capital rules as EU banks redefine risk', Bloomberg News, 9 September.

Zhou, J., Rutledge, V., Bossu, W., Dobler, M., Jassaud, N. and More, M. (2012), 'From bail-out to bail-in: mandatory debt restructuring of systemic financial institutions', IMF Staff Discussion Note, 12/03.

NOTES

(1) That is, gearing up the world economy, extracting the carried interest for management and leaving others (the taxpayers) to deal with the debt when it was pushed too far and the business failed.

(2) See: BlundelI-Wignall (2007a; b); BlundelI-Wignall and Atkinson (2008; 2010; 2011; 2012); and BlundelI-Wignall, Wehinger and Slovik (2009).

(3) See table 2.1, ISDA (2012)

(4) This is based on a recovery rate of 50 per cent, a discount rate of 6 per cent, a premium of 4 per cent, and a notional of 100m.

(5) See: General Accounting Office (2009), Figure 9.

(6) This claim was made in a response to a question by the CFO of Goldman Sachs, at a conference on the crisis at Stanford University in June 2011.

(7) See BlundelI-Wignall and Atkinson (2008; 2010; 2011).

(8) Herfindhal indexes suggest for interest rate derivatives (easily the bulk of the total OTC derivatives business) that around the equivalent of eleven or twelve equal-sized banks dominate the global derivatives market. See BlundelI-Wignall and Atkinson (2011).

(9) Derivatives are priced off volatility.

(10) See BlundelI-Wignall (2007).

(11) Relatedly, Goldman Sachs has been criticised by both houses of government in the USA, and fined by the SEC for misleading clients.

(12) See below for banks covered here. The GMV of derivatives to put US bank assets on an IFRS comparable basis is not available for all banks prior to 2007.

(13) This point was first raised in BlundelI-Wignall and Atkinson (2008), and again in BlundelI-Wignall and Atkinson (2011).

(14) In a sense the Basel system is based on many flawed concepts, including inter alia: that there is such a thing as ex-ante risk in a world of derivatives and risk modelling; and that two risk concepts (leverage and the intrinsic riskiness of financial products) can be controlled simultaneously with one ratio.

(15) For example, if mark-to-market securitised mortgages guaranteed by CDS are held on a traditional bank's balance sheet, as with Northern Rock, and these products give rise to different prudential considerations than would apply to amortised cost-accounting loans, then they may need to be separated into an armslength SPV (regardless of the fact that Northern Rock didn't have an investment bank). See Blundell-Wignall, Wehinger and Slovik (2009).

(16) See Financial Stability Board (2011) and Zhou et al. (2012).

(17) See BlundelI-Wignall and Atkinson (2012) for a lengthy discussion of liquidity rules and bias issues.

(18) The Australian legislation including the way restructuring instruments and transfer certificates may be used to create an NOHC structure is certainly worth examining--particularly when so many banks claim it is too difficult to achieve. Australia's Macquarie Group restructured to the NOHC with a bank and a securities structure in a short space of time once supplementary legislation to the Banking Act was passed in 2007, permitting practical restructuring instruments and transfer certificates. The structure is not, however, mandatory. One simple reason for this is that Australia is not an investment banking centre, and the same systemic risks do not apply on the scale of banks in the USA, the UK, France, Germany and Switzerland.

(19) See Financial Stability Oversight Council (2012).

(20) A foreign parent that owned a substantive share of a securities business abroad would not be allowed to own a depository institution in the USA unless all of the securities business activities are executed wholly outside the US, do not involve any US residents as counterparties and no personnel involved in the transaction are physically located in the USA. Furthermore, the Volcker exemption for dealing in US government debt does not apply to dealing in foreign government debt.

(21) Financial Stability Oversight Council (2012), p. 17.

(22) Just as they were with the Basel II rules, where regulatory arbitrage was rampant.

(23) See Duffle (2012).

(24) US GAAP accounting permits derivatives subject to netting agreements to be reported on the balance sheet on a fully net basis to measure TA. IFRS includes fair value derivatives exposure in TA with very limited netting--there must be the specific intent to settle the contract on a net basis, or a specific intent to realise the asset and settle the liability simultaneously.

(25) See BlundelI-Wignall and Atkinson (2012).

(26) See Dexia, press release 23 February 2012, "Deterioration of the environment in the second half of 2011 leads the Group to announce radical restructuring measures", Financial Appendix 1 and 2. it is believed that the authorities had to settle their counterparty exposures as also occurred with the AIG rescue.

(27) See Vaughan (2011).

(28) The authors fully realise the forlorn nature of this hope, and believe that the impossibility of this underlines why strict control via separation and a leverage ratio are essential for safeguarding the financial system in the world of counterparty risk.

(29) See OECD (2010) and BlundelI-Wignall (2012).

(30) During the crisis the most classic ploy in a universal bank in the good times in the run-up to the crisis, was to take even a low-risk spread and gear it up using capital that the group treasury could access at a low LIBOR rate for initial and variation margin. A I per cent spread geared up 50 times, is an easy return of 50 per cent, simply for having the access to the name of the parent plus leverage is the source of the return--there is no skill in it. See UBS (2008) for a wonderful description of these practices.

(31) See OECD (2009).

Adrian Blundell-Wignall, * Paul E. Atkinson ** and Caroline Roulet *

* OECD; e-mail: Adrian.BLUNDELL-WIGNALL@oecd.org and Caroline.ROULET@oecd.org. ** Groupe d'Economie Mondiale de Sciences-Po; e-mail: patkinson@noos.fr. The views in this article are those of the authors and do not purport to represent those of the OECD or its member countries.

Table 1. US government payouts to AIG CDS
counterparty losses

 Collateral Payments to As a
 postings securities share of
 for credit lending capital (c) at
 default counter- end-2008
 swaps (a) parties (b) Total (%)

Goldman Sachs 8.1 4.8 12.9 29.1
Soci6t6 Generale 11.0 0.9 11.9 28.9
Deutsche Bank 5.4 6.4 11.9 37.4
Barclays 1.5 7.0 8.5 20.0
Merrill Lynch 4.9 1.9 6.8 77.4
Bank of America 0.7 4.5 5.2 9.1
UBS 3.3 1.7 5.0 25.2
BNP Paribas - 4.9 4.9 8.3
HSBC 0.2 3.3 3.5 5.3
[memo: Bank of
 America after
its merger with 12.0 [18.1]
 Merrill Lynch]

Notes: (a) Direct payments from AIG through end-2008 plus
payments by Maiden Lane III, a financing entity established by AIG
and the New York Federal Reserve Bank to purchase underlying
securities. (b) September 18 to December 12, 2008. (c) Common
equity net of goodwill; net of all intangible assets for Merrill
Lynch and HSBC.

Table 2. Volcker, Vickers and the NOHC: comparisons and contrasts

 Regulatory approaches to bank separation

 Vickers Volcker rule

AIM Insulate retail banking & Reduce speculative losses
 SME lending from to banks and focus on
 international shocks in intermediary functions
 capital markets. Improve serving customers.
 resolvability within
 groups.

 Better prudential safety
 of core functions of
 banking. Reduce taxpayer
 risk.

Features Ring-fence retail from w/ Prohibits proprietary
 sale and investment bank trading by any insured
 (IB) divisions within a depository institution in
 holding co. Tougher commodities, financial
 capital standards for securities and
 ring-fenced bank, but not derivatives that do not
 for IB. Promoting benefit customers. (Banks
 competition in retail and BHCs included).
 banks and IBs separately
 was the main driver. Prohibits deposit
 institutions from
 Foreign branches are investing in private
 welcome to join the equity firms and hedge
 retail space. funds (no more than 3% of
 any one firm's capital
 and no more than 3% of
 the bank's capital in all
 such investments) to
 ensure bank capital isn't
 standing behind them.

Functions Maintains all business Permitted activities:
 units in the group. traditional banking,
 trading in US government
 Bank mandated to take debt; underwriting and
 deposits and provide making markets (broking
 overdrafts/loans. only); risk mitigating
 hedging: trading for
 Bank prohibited from: customers; liquidity
 structuring OTC management; investments
 derivatives; equity and in 'Small Business Inv.
 ETF functions; security Companies'; limited
 market making; investments in covered
 underwriting; services funds in asset management
 outside the EEA. and advisory businesses;
 foreign trading by non-
 Trade finance and aspects US banks; foreign covered
 of w/sale are not fund activities of non-
 excluded. US banks.

Administrative UK banks and lobbyists A disadvantage to US and
and claim that it is foreign deposit-taking
industry difficult to implement institutions in the USA.
concerns and have convinced the Extra-territoriality as
 authorities to give them it applies to foreign
 until 2019 to sort it banks operating in the
 out. They would do well US. Foreign parents can
 to examine the NOHC do all the prop. trading
 legislation in Australia. etc, as long as their
 Applies to foreign branches or subsidiaries
 subsidiaries operating in in the US do not. Banks
 the UK. have to separate
 functions according to
 whether they are
 intermediating for a
 client or trading on
 their own account--this
 is difficult to monitor.
 Making a market for
 example requires
 principle risk.

 Some foreign governments
 claim it will reduce the
 liquidity on trading in
 their own bonds.

Effective 2019--linked to Basell 2012 (July)
date III timetable.

 Regulatory approaches to bank separation

 Non-operating holding company (NOHC)

AIM Better allocation of risk and improved
 resolvability. Continuous provision of
 services by some affiliates in the group in
 the face of distress in some others.

Features The parent is non-operating, raises equity
 and invests in ring-fenced affiliates with
 their own governance. Ring-fenced entities
 (eg the bank from securities businesses)
 are essential in order to protect against
 creditors of distressed affiliates seeking
 redress against the others in the group.
 This allocates risk more efficiently
 between the prudentially regulated bank
 and less regulated entities.

Functions Maintains all business units in the group.
 Decides on the regulatory treatment of
 each--with banks being more heavily
 regulated for capital and with dep.
 insurance for retail investors. Banking
 vs securities functions are the broad
 divide. Risk is priced differently in the
 different segments with the price of risk
 rising in the less regulated. The volume
 of functions is determined by supply and
 demand at prices reflecting the true risks,
 not by detailed rules on what can and
 can't be done.

Administrative It is not mandatory (Macquarrie Bank
and only).
industry
concerns Easy to implement--entities and
 functions defined by prudential regulator
 and balance other laws. Restructure
 instruments grant relief to regulatory
 impediments to separation arising from
 corporate income tax laws to the entities
 specified by the prudential regulator.
 'Transfer certificates' issued provide
 for the transfer of assets and liabilities
 between the entities. This was carried out
 by MQG in the same year as the enabling
 act was passed. Does not apply to foreign
 banks, or any bank choosing not to
 implement (at this stage).

Effective 2007 (July)
date

Source: OECD and various Treasury, regulatory agency, legislative
documents, and inquiry reports.