According to a report in the Financial Times, "European nations are to draw up radical proposals to improve transparency in financial markets and to change the way credit rating agencies operate in an attempt to prevent any recurrence of the financial turmoil arising from the credit squeeze."
Are transparency in financial markets and better rating agencies indeed key to preventing the recurrence of the kind of mess we have been experiencing in the world’s most developed financial systems for these past three months? I intend to take a multi-part romp through the crisis to see what lessons it holds for policymakers and market participants.
The problems we have recently seen across the industrialised world (but not, as yet, in the emerging markets), was caused fundamentally by wanton securitisation, fundamental flaws in the rating agencies business model, privately rational but socially inefficient disintermediation, and competitive international de-regulation. Proximate drivers of the specific way in which these problems manifested themselves were regulatory and supervisory failure in the US home loan market and excessive global liquidity creation by key central banks.
In the UK, the problems were aggravated by:
- a flawed Tripartite arrangement between the Treasury, the Bank of England and the Financial Services Authority (FSA) for dealing with financial crises;
- supervisory failure by the FSA;
- flaws in the Bank of England’s liquidity-oriented open market policies (too restrictive a definition of eligible collateral and an unwillingness to try to influence market rates at maturities longer than overnight, even during periods of serious lack of market liquidity;
- flaws in the Bank of England’s discount window operations (too restrictive a definition of eligible collateral; only overnight lending; too restrictive a definition of eligible discount window counterparties).
Both shortcomings in the Bank of England’s operating arrangements and procedures were due to a flawed understanding of the nature and determinants of market (ill)liquidity, of the Bank of England’s unique role in the provision of market liquidity because of its ability to create unquestioned liquidity instantaneously and costlessly, of the conditions under which there is a trade-off between moral hazard (bad incentives for future bank behaviour) and the ex-post provision of liquidity to (a) markets and (b) specific individual institutions, to prevent collateral damage to the financial system and the real economy.
Traditionally, banks borrowed short and liquid and lent long and illiquid. On the liability side of the banks’ balance sheet, deposits withdrawable on demand and subject to a sequential service (first come, first served) constraint figured prominently. On the asset side, loans, secured or unsecured to businesses and households were the major entry. These loans were typically held to maturity by the banks (the ‘originate and hold’ model). Banks therefore transformed maturity and created liquidity. Such a combination of assets and liabilities is inherently vulnerable to bank runs by deposit holders.
Banks were deemed to be systemically important, because their deposits were a key part of the payment mechanism for households and non-financial corporations, because they played a central role in the clearing and settlement of large-scale transactions and of securities. To avoid systemically costly failures by banks that were solvent but had become illiquid, the authorities implemented a number of measures to protect and assist banks. Deposit insurance was commonly introduced, paid for either by the banking industry collectively or by the state. In addition, central banks provided lender of last resort (LoLR) facilities to individual deposit-taking institutions that had trouble financing themselves.
In return for this assistance and protection, banks accepted regulation and supervision. This took the form of minimum capital requirements, minimum liquidity requirements and other restrictions on what the banks could hold on both sides of their balance sheets.
In the 1970s, Fannie Mae (Federal National Mortgage Association), Ginnie Mae (Government National Mortgage Association) and Freddie Mac (Federal Home Loan Mortgage Corporation) began the process of securitisation of residential mortgages. Asset securitisation involves the sale of income generating financial assets (such as mortgages, car loans, trade receivables (including credit card receivables) and leases) by a company (the originator of the financial assets) to a special purpose vehicle (SPV). The SPV, which might be a trust or a company, finances the purchase of these assets by the issue of bonds, which are secured by those assets. The SPV is supposed to be bankruptcy-remote from the originator, that is, it has to be an off-balance sheet entity vis-à-vis the originator. Cash-flow securitisation works in a similar way, as when the UK government agreed to create the International Finance Facility which was supposed to securitise future development aid commitments.
Private institutions, especially banks, immediately took advantage of these securitisation techniques to liquefy their illiquid loans. The resulting ‘originate and distribute’ model had major attractions for the banks and also permitted a potential improvement in the efficiency of the economy-wide mechanisms for intermediation and risk sharing. It made marketable the non-marketable; it made liquid the illiquid. There was greater scope for trading risk, for diversification and for hedging risk.
There are three problems associated with securitisation (and the generally associated creation of off-balance sheet vehicles).
- The greater opportunities for risk trading created by securitisation not only made it possible to hedge risk better (that is, to cover open positions); it also permitted investors to seek out and take on additional risk, to ‘unhedge’ risk and to create open positions not achievable before. When risk-trading opportunities are enhanced through the creation of new instruments or new institutions, and when new populations of potential investors enter the risk-trading markets, we can only be sure that the risk will end up with those most willing to bear it. There can be no guarantee that risk will end up being borne by those most able to bear it.
- The ‘originate and distribute’ model destroys information compared to the ‘originate and hold’ model. The information destruction occurs at the level of the originator of the assets to be securities. Under the ‘originate and distribute’ model the loan officer collecting the information on the creditworthiness of the would-be borrower is working for the Principal in the investing relationship (the originating bank). Under the ‘originate and distribute’ model, the loan officer of the originating banks works for an institution (the originating bank) that is an Agent for the new Principal in the investing relationship (the SPV that purchases the loans from the bank and issues securities against them). With asymmetric information and costly monitoring, the agency relationship dilutes the incentive for information gathering at the origination stage. Reputation considerations will mitigate this problem, but will not eliminate it.
- Securitisation also puts information in the wrong place. Whatever information is collected by the loan originator about the underlying assets remains with the originator and is not effectively transmitted to the SPV, let alone to the subsequent buyers of the securities issued by the SPV that are backed by these assets. By the time a hedge fund owned by a French commercial bank sells ABS (asset backed securities) backed by US subprime residential mortgages to a conduit owned by a small German Bank specialising in lending to small and medium-sized German firms, neither the buyer nor the seller of the ABS has any idea as to what is really backing the securities that are being traded.
The problems created by securitisation can be mitigated in a number of ways.
1. Simpler structures. The financial engineering that went into some of the complex securitised structures that were issued in the last few years before the ABS markets blew up on August 9, 2007 at times became ludicrously complex. Simple securitisation involved the pooling of reasonably homogeneous assets, say residential mortgages issued during a given period with a given risk profile (subprime, alt-A or prime, say). These were pooled and securities issued against them were tranched, with the higher tranche having priority (seniority) over the lower tranches. This permitted the highest tranche secured against a pool of high-risk mortgages, say, to achieve a much better credit rating than the average of the assets backing all the tranches together (the lower tranches, of course, had a correspondingly lower credit rating).
However, second-tier and higher-tier-securitisation then took place, with tranches of securitised mortgages being pooled with securitised credit-card receivables, car loan receivables etc. and tranched securities being issued against this new, heterogeneous pool of securitised assets. Myriad credit enhancements were added. In the end, it is doubtful that even the designers and sellers of these compounded, multi-tiered securitised assets knew what they were selling, knew its risk properties or knew how to price them. Certainly the sellers did not.
There is a simple solution: simpler structures. This will in part be market-driven, but regulators too may put bounds on the complexity of instruments that can be issued or held by various entities.
2. ‘Unpicking’ securitisation. This ‘solution’ is the ultimate admission of defeat in the securitisation process. A number of American banks with (residential mortgage-backed securities on their balance sheet have been scouring the entrails of the asset pools backing these securities and have sending staff to specific addresses to assess and value the individual residential properties. This inversion of the securitisation matrix is, of course, very costly and means that the benefits from risk pooling will tend to be ignored. It is an ignominious end for the securitisations involved. RMBS
3. Retention of equity tranche by originator. When the originator of the loans is far removed from the ultimate investor in the securities backed by these loans, the incentive for careful origination is weakened. One way to mitigate this problem is for the originator to retain the ‘equity tranche’ of securitised and tranched issues. The equity tranche or ‘first-loss tranche’ is the highest-risk tranche – the first port of call when the servicing of the loans is impaired. It could be made a regulatory requirement for the originator of residential mortgages, car loans etc. to retain the equity tranche of the securitised loans. Alternatively, the ownership of the equity tranche could be required to be made public information, permitting the market to draw its own conclusions.
4. External ratings. The information gap could be closed or at least reduced by using external rating agencies to provide an assessment of the creditworthiness of the securitised assets. This has been used widely in the area of RMBS and of ABS. This ‘solution’ to the information problem, however, brought with it a whole slew of new problems.
2. Rating agencies
A small number of internationally recognised rating agencies (really no more than three: Standard & Poor’s, Moody’s and Fitch) account for most of the rating of complex financial instruments, including ABS. They got into this business after for many years focusing mainly on the rating of sovereign debt instruments and of large private corporates. They have been given a formal regulatory role, (which will be greatly enhanced under the about-to-be-introduced Basel 2 Capital Adequacy regime) because their ratings determine the risk weighting of a whole range of assets bank hold on their balance sheets.
Their role raises a number of important issues because it creates a number of problems.
1. What do they know? This is a basic but important question. One can imagine that, after many years, perhaps decades, of experience, a rating agency would become expert at rating a limited number of sovereign debtors and large private corporates. How would the rating agency familiarise itself with information available only to the originators of the underlying loans or other assets and to the ultimate borrowers? How would the rating agency, even if they knew as much about the underlying assets as the originators/ultimate borrowers, rate the complex structures created by pooling heterogeneous underlying asset classes, slicing and dicing the pool, tranching and enhancing the payment streams and making the ultimate pay-offs complex, non-linear functions of the underlying income streams? These ratings were overwhelmingly model-based. The models used tended to be the models of the designers and sellers of the complex structures, who work for the issuers of the instruments. Models tend to be useless during periods of disorderly markets, because we have too few observations on disorderly markets to construct reasonable empirical estimates of the risks involved.
2. They only rate default risk. Rating agencies provide estimates of default risk (the probability of default and the expected loss conditional on a default occurring). Even if default risk is absent, market risk or price risk can be abundant. Liquidity risk is one source of price risk. As long as the liquidity risk does not mutate into insolvency risk, the liquidity risk is not reflected in the ratings provided by the rating agencies. The fact that many ‘consumers’ of credit ratings misunderstood the narrow scope of theses ratings is not the fault of the rating agencies, but it does point to a problem that needs to be addressed. First, there has to be an education campaign to make investors aware of what the ratings mean and don’t mean. Second, the merits of offering (and requiring) a separate rating for, say, liquidity risk should be evaluated.
3. They are conflicted. Rating agencies are subject to multiple potential conflicts of interest.
a. They are the only example of an industry where the appraiser is paid by the seller rather than the buyer.
b. They are multi-product firms that sell advisory and consulting services to the same clients to whom they sell ratings. This can include selling advice to a client on how to structure a security so as to obtain the best rating and subsequently rating the security designed according to these specifications.
c. The complexity of some of the structured finance products they are asked to evaluate makes it inevitable that the rating agencies will have to work closely with the designers of the structured products. The models used to evaluate default risk will tend to be the models designed by the clients. It’s not just the problem that marking to model can become marking to myth. There is the further problem that the myth will tend to be slanted towards the interest of the seller of the securities to be rated.
There is no obvious solution other than ‘try harder and don’t pretend to know more than you know’ for the first problem. The second problem requires better education of the investing public. The third problem can be mitigated in a number of ways.
1. Reputational concerns. Reputation is a key asset of rating agencies. That, plus the fear of law suits will mitigate the conflict of interest problem. The fundamental agency problem cannot be eliminated this way, however. Even if the rating agencies expect to be around for a long time (a necessary condition for reputation to act as a constraint on opportunistic and inappropriate behaviour), individual employees of rating agencies can be here today, gone tomorrow. A person’s reputation follows him/her but imperfectly. Reputational considerations are therefore not a fully effective shield against conflict of interest materialising.
2. Remove the quasi-regulatory role of the rating agencies in Basel II and elsewhere. Just as the public provision of private goods tends to be bad news, so the private provision of public goods leaves much to be desired (‘the best judges money can buy etc.’). The official regulatory function of private credit risk ratings in Basel I and II should be de-emphasized and preferably ended altogether.
I may get my wish here, because Basel II appears fatally holed below the waterline. It was long recognised to have unfavourable macroeconomic stabilisation features, because the capital adequacy requirements are likely to be pro-cyclical (see Borio, Furfine and Lowe (2001), Gordy and Howells (2004) and Kashyap and Stein (2004)). On top of this, the recent financial turmoil should that the two key inputs into Pillar 1, the ratings provided by the rating agencies and the internal risk models of the banks are deeply flawed.
As regards internal risk models, there are two problems. The first is the unavoidable ‘garbage in – garbage out’ problem that makes any quantitative model using parameters estimated or calibrated using past observations useless during times of crisis, when every crisis is different. We have really only had one instance of a global freeze-up of ABS markets, impairment of wholesale markets and seizure of leading interbank markets simultaneously in the US, the Eurozone and the UK. Estimates based on a size 1 sample are unlikely to be useful. Second, the use of internal models is inherently conflicted. The builders, maintainers and users of these models are perceived by the operational departments of the bank as a constraint on doing profitable business. They will be under relentless pressure to massage the model to produce the results desired by the bank’s profit centres. They cannot be shielded effectively from such pressures. Chinese walls inside financial corporations are about as effective in preventing the movement of purposeful messages across them, as the original Great Wall of China was in keeping the barbarians out and the Han Chinese in – that is, utterly ineffective.
3. Make rating agencies one-product firms. The potential for conflict of interest when a rating agency sells consultancy and advisory services is inescapable and unacceptable. Even the sale of other products and services that are not inherently conflicted with the rating process is undesirable, because there is an incentive to bias ratings in exchange for more business in functionally unrelated areas. The obvious solution is to require any firm offering rating services to provide just that. Having single-product rating agencies should also lower the barriers to entry.
4. End payment by the issuer. Payment by the buyer (the investors) is desirable but subject to a ‘collective action’ or ‘free rider’ problem. One solution would be to have the ratings paid for by a representative body for the (corporate) investor side of the market. This could be financed through a levy on the individual firms in the industry. Paying the levy could be made mandatory for all firms in a regulated industry. Conceivably, the security issuers could also be asked to contribute. Conflict of interest is avoided as long as no individual issuer pays for his own ratings. This would leave some free rider problems, but should get the rating process off the ground. I don’t think it would be necessary (or even make sense) to socialise the rating process, say by creating a state-financed (or even industry-financed) body with official powers to provide the ratings.
5. Increase competition in the rating industry. Competition in the rating process is desirable. The current triopoly is unlikely to be optimal. Entry should be easier when rating agencies become single-product firms, although establishing a reputation will inevitably take time.
3. Excessive disintermediation
There are no doubt solid economic efficiency reasons for taking certain financial activities out of commercial banks and out of investment banks, and putting them in special purpose vehicles (SPVs), Structured Investment Vehicles (SIVs, that is, SPVs investing in long-term, often illiquid complex securitised financial instruments and funding themselves in the short-term wholesale markets, including the ABCP markets), Conduits (SIVs closely tied to a particular bank) and a host of other off-balance-sheet and off-budget vehicles. Incentives for efficient performance, including appropriate risk management can, in principle, be aligned better in a suitably designed SPV than in a general-purpose bank. The problem is that it is very difficult to come up with any real-world examples of off-balance sheet vehicles that actually appear to make sense on efficiency grounds.
Most of the off-balance sheet vehicles (OBSVs) I am familiar with are motivated by regulatory arbitrage, that is, by the desire to avoid the regulatory requirements imposed on banks and other deposit-taking institutions. These include minimal capital requirements, liquidity requirements, other constraints on permissible liabilities and assets, reporting requirements and governance requirements. Others are created for tax efficiency (i.e. tax avoidance) reasons or to address the needs of governments and other public authorities for off-budget and off-balance sheet finance, generally to get around public deficit or debt limits.
OBSVs tend to have little or no capital, little or no transparency and opaque governance. When opaque institutions then invest in opaque financial instruments like the ABS discussed earlier, systemic risk is increased. This is reinforced by the fact that much de-jure or de-facto exposure remains of the sponsoring banks to these off-balance-sheet vehicles. The de jure exposure exists when the bank is a shareholder or creditor of the OBSV, when the OBSV has an undrawn credit line with the bank or when the bank guarantees some of the OBSV’s liabilities. De-facto exposure exists when, for reputational reasons, it is problematic for the bank to let an OBSV that is closely identified with the bank go under.
Banks in many cases appear not to have been fully aware of the nature and extent of their continued exposure to the OBSVs and the ABS they carried on their balance sheets. Indeed the explosion of new instruments and new financial institutions so expanded the populations of issuers, investors and securities that many market participants believed that risk could not only be traded and shared more widely and in new ways, but that risk had actually been eliminated from the system altogether. Unfortunately, the world of risk is not a doughnut: it does not have a hole in it. All risk sold by someone is bought by someone. If the system works well, the risk ends up being born by those both willing and most able to bear it. Regrettably, it often ends up with those most willing but not most able to bear it.Partial solutions.
Mitigation of the problems created by excessive disintermediation will be part market-driven and partly regulatory.
1. Re-intermediation. Either Conduits, SIVS and other OBSVs are taken back on balance sheet by their sponsoring banks, or the ABS and other illiquid securities on their balance sheets are sold to the banks. The OBSVs then either wither away or vegetate at a low level of activity.
2. Regulation. We can anticipate a regulatory response to the problem of opaque instruments held by opaque OBSVs in the form of reporting requirements, and consolidation of accounts requirements that are driven by broad principles (‘duck tests’) with constant adaptation of specific rules addressing particular institutions and instruments. For instance, if the Single Master Liquidity Enhancement Conduit (M-LEC) proposed by JPMorgan Chase, Bank of America and Citigroup ever gets off the ground, it is questionable whether the US regulators will permit the participating banks to keep it off-balance-sheet for reporting purposes, including earnings reports.
4. Competitive Global Deregulation
Regulators of financial markets and institutions are organised on a national basis and are, in part, cheerleaders and representatives of the interests of their national financial sectors. While regulation is national, finance is global. The location of financial enterprises and markets is endogenous. A thriving financial sector creates jobs and wealth, and is generally environmentally friendly. So regulators try to retain and attract business for their jurisdictions in part by offering more liberal, less onerous regulations. This competition through regulatory standards has led to less stringent regulation.
There have been occasional reversals in this process. The Sarbanes-Oxley Act of 2002 was a response to the corporate governance, accounting and reporting scandals associated with Enron, Tyco International, Peregrine Systems and WorldCom. It undoubtedly contributed to a loss of business for New York City as a global financial centre. Because Sarbanes-Oxley compliance is mainly a matter of box-ticking (like most real-world compliance, especially compliance originating in the USA), it has not materially improved the informational value of accounting or the protection offered to investors.
Is this global competitive deregulation process a welcome antidote to a tendency to excessive and heavy-handed regulation or a race to the bottom in which everyone loses in the end? I believe the jury is still out on this one, although I am inclined, if pushed, to suggest that the following are likely to be true
- Principles-based regulation (allegedly what we have in the UK) vs. rules-based regulation is an unhelpful distinction. You need both. You need principles that spell out the fundamental ‘duck test’: (a) Does the institution lend long and borrow short? (b) Does it lend in illiquid form and borrow in markets that are liquid in normal times although they may turn illiquid during period of market turbulence? Do banks have substantial exposure to it? If so, it should be either consolidated for reporting purposes with the bank or treated as a bank it its own right. Then you need rules that are constantly adapted to keep up with developments in instruments and institutions.
- Self-regulation is no regulation unless backed up credibly with the threat that, unless effective self-regulation is implemented, external regulation will be imposed.
- Voluntary codes of conduct are without significance unless they can be and are used by the regulator (through ‘comply or explain’ rules) to impose and enforce standards. That means that if the explanation is not to the regulator’s satisfaction, compulsion can be used.
- The UK’s ‘light-touch’ regulation has become ‘soft-touch’ regulation and needs to be tightened up in a large number of areas.
1. Greater international cooperation between regulators. This is a no-brainer, but very hard to achieve.
2. A single EU-wide regulatory regime for banks, other financial institutions and financial markets. National financial regulators in the EU should go the way of the dodo. An EU-level FSA would be a good idea, although the central banks (the ECB and for the time being still 14 national central banks) should collect more information about individual banks than the Bank of England has done since it lost banking supervision and regulation in 1997 when the Bank became operationally independent for monetary policy.
3. A crackdown on “regulators of convenience”. This requires tough measures towards ‘regulation havens’, some found in the Caribbean, others closer to the UK. One effective approach would be the non-recognition and non-enforceability of contracts, court judgements and other legal and administrative rules from non-compliant jurisdictions.
Borio, C., C. Furfine and P. Lowe (2001): “Procyclicality of the financial system and financial stability: issues and policy options”, in Marrying the macro- and micro-prudential dimensions of financial stability, BIS Papers, no 1, March, pp 1-57.
Gordy, Michael B. and Bradley Howells (2004), “Procyclicality in Basel II: Can We Treat the Disease Without Killing the Patient?”, Board of Governors of the Federal Reserve System; First draft:
April 25, 2004. This draft: May 12, 2004.
Kashyap, A K and J C Stein (2004): “Cyclical implications of the Basel II capital standards”, Economic Perspectives, Federal Reserve Bank of Chicago, First Quarter, pp 18-31.