Euromoney 30th anniversary: Heroes and villains
The past 30 years have witnessed the birth, infancy and adolescence of risk management as a profession. Although actuarial science has been practised for over 200 years, it deals primarily with assessing the risks of liability claims; its major contribution to the measurement of the risks of financial assets was the quantification of the time value of money.
In recent decades, the market prices of financial assets have been scrutinized with statistical tools, yielding an array of neo-Greek symbols - alpha, beta, gamma, delta, theta, vega - to complement the basic metrics of probability - mu, sigma, rho. The Nobel Prize work of Miller, Markowitz, Sharpe, Black, Scholes and Merton spawned analytic methods which today can decompose market risk into its fundamental elements.
How will risk management evolve in the next 30 years? The evolution of value-at-risk (VAR) in the mid 1990s and the current emphasis on stress-testing indicate that the pendulum is swinging toward risk-aggregation, in contrast to the previous dominance of risk-decomposition. As long as the supervisors hold out the possibility of reduced regulatory capital, financial institutions will continue to devote substantial resources to mastering risk aggregation. The direction which that development will take is best seen by examining the financial consequences of demographic trends, the basic paradigms of risk intermediation and the fundamental unresolved questions in risk measurement.
The trends which will characterize finance in the first decades of the next century are rooted in demographics: statistics indicate that by 2050, there will be only two working-age Americans for every American over retirement age. Europe leads the US a decade in this regard, and Japan is 10 years ahead of Europe.
There is simply no way that a pay-as-you-go scheme of social security and pensions can succeed in face of such figures. America was the first country to attend to these issues, by forcing companies to fully fund defined-benefit pension schemes (ERISA reform) and by encouraging individuals to create defined-contribution pension accounts (IRA, Keogh and 401K). This prompted a huge shift to mutual funds (from under $100 billion in 1979 to nearly $6 trillion today), since the professional management of the segregated monies was essential to ensure that a defined-contribution scheme achieved sufficient return to fund its liabilities. There are now almost twice as many mutual funds as NYSE-listed companies.
Today, this huge sump of professionally managed but dormant money in mutual funds motivates financial engineers to create and securitize structures containing multifarious risks. Recent popular structures include the securitization of portions of banks' loan portfolios or the reinsurance tail of natural catastrophes like earthquakes and hurricanes. The fact that most mutual funds do not have large short-term liquidity needs makes them ideal for housing longer-term risks, and their ability to invest in many individual holdings makes them uniquely suited for managing risks whose principal mitigation technique is diversification. This rise in the influence of mutual funds shows no signs of abating, despite the fact that Americans are perennially among the world's worst savers (achieving a negative savings rate in the first quarter of 1999). Imagine the situation if Americans were to begin saving significant portions of their incomes, or if the federal government began to privatize the Social Security System.
Europe and Japan will experience the same explosion of mutual-fund activity, because the demographics demand it. Britain, with its PEP initiative in the late 1980s, is further down the road of pension privatization than the US. The arrival of the euro means that the major continental governments can no longer run substantial deficits, so their own fiscal health depends on reducing or eliminating long-term, unfunded liabilities. Corporate Europe, seeking to survive in the newly-competitive EU, will also try to offload pension liabilities in favour of defined-contribution programmes.
Indeed, the key to the success of the EU is labour mobility, one corollary of which is that employees will insist on the right to take their defined-contribution nest eggs with them across the corporate landscape, as Americans now do. Mutual funds are the ideal tool to allow such pension mobility. In Japan, the poor returns of domestic bonds and shares, as well as the weak condition of the financial sector, will lead to participation in foreign investments through mutual funds. Finally, America's big mutual funds such as Vanguard and Fidelity face a saturated market in the US. Their continued growth must involve the creation of new markets overseas, and the natural starting place is in those countries with high savings rates and ageing demographic profiles, particularly those with little domestic competition, like Japan.
Trio of paradigms
The central task of financial intermediaries is to deploy capital from those who have it to those who need it, ensuring that the rights of the former are preserved over time. There are three fundamental paradigms for managing the risks of intermediation (see figure 1). Each of these risk intermediation paradigms will be affected by the major financial trends analyzed earlier.
|Merton's option-based model for valuing debt and equity|
Over the coming decades, governments will exit the pension insurance business, keeping only a safety net for those who cannot successfully provision their own nests. (Governments will always remain last-resort providers of unemployment insurance.)
Life insurance will not change significantly; this industry systematically benefits from the increasing life spans of the populace sufficiently to offset periodic span-shorteners such as AIDS. The biggest controversy in life insurance will be medical privacy, as the mapping of the human genome and better understanding of genetically conditioned diseases will afford insurance companies the opportunity to red-line customers (charge them according to their individual risks) based simply on a DNA sample.
Casualty insurance will see mutual funds compete for reinsurance business with the global private companies that now provide the lion's share; the ability to measure the magnitude of natural catastrophes will make it ever easier to securitize those risks.
The great unknown remains medical insurance. Continual increases in cost and demand make this a difficult business proposition. Look to government to stay involved, particularly where implicit rationing of health care services is necessary.
In the brokerage/underwriting paradigm, those needing capital are distinct from those with capital, but the risk preferences of each group are equivalent (or rather, mirror images of each other). The intermediary matches the two groups and determines a unified clearing price (for example in bond and share issuance). Pure brokers take on very little risk; underwriters absorb large, but momentary, inventory risk and dealers must maintain inventories of securities to match their market-making ambitions.
Risk mitigation techniques include using index derivatives to hedge the general risk of securities; specific risk is managed mainly by astute supply/demand determination and by diversification. This paradigm also requires an entire settlement/custody apparatus for managing the cashflows and legal rights associated with securities.
Within the brokerage/underwriting paradigm, expect the continuing "democratization" of finance with the growth of mutual funds. The spectrum of investments acceptable to these funds will expand until it equals the total investable menu. These funds will compete with re-insurance firms in providing the protection against exogenous tail events that can be clearly defined and measured.
On the other hand, funds will also relieve banks of an increasing proportion of their balance sheet risk, as financial engineers learn to assemble and package corporate loans with the same flair that they brought to mortgage and credit-card debt. Look for an increase in bond funds dealing in non-investment-grade instruments. Also, funds will seek authority to increase their leverage from the current ratio of 1:1 to between 2:1 and 3:1, but this depends on the creation of a consensus metric for leverage, including market, credit and liquidity risks. Of the firms that service the buy-side community, underwriters will see plentiful opportunities; in contrast, brokers involved in secondary trading of securities will see their franchise fatally eroded by the internet, which will then bring into question the large securities inventories which dealers routinely hold.
The banking paradigm is the most complex. Those needing capital and those with capital are distinct groups whose requirements diverge as well. A standard example involves a bank soliciting demand deposits from individuals and offering mortgages to families. Banks have traditionally been expected to intermediate both term-interest-rate risk (borrow short and lend long) and credit risk (borrow from mixed-quality credits but represent a rock-solid credit risk themselves). They now provide a huge menu of derivative products for managing a variety of financial risks.
Principal risk-mitigation techniques include portfolio offsets and dynamic replication. The banking paradigm will see the most profound changes in the next three decades. The explosive growth in the interest-rate swaps market has already eliminated the need for banks to hold term-rate risk on their balance sheets.
Expect to see a similar sequence played out in credit risk: More accurate quantification of risk allows supervisors to construct capital rules based on true risk; banks measure their results on a risk-adjusted basis; banks begin to shed their management of credit portfolios to specialized funds, often staffed by the very professionals who previously performed the task inside the bank; banks effectively de-lever. With the elimination of their second traditional raison d'être (credit risk), banks will be left with two major responsibilities. They will need to risk-manage their vast derivative portfolios, and they will continue to hold the junior, residual tranches (the so-called toxic waste) which inevitably accompany complex securitizations.
Progress in risk-aggregation depends on major enhancements in three areas of risk measurement: credit risk measurement, matching of assets and liabilities, and liquidity risk measurement.
The enthusiasm for modelling, which culminated in the Basle Committee's 1996 amendment to the capital accord covering market risk, augured well for application of similar techniques to credit risk measurement. Indeed, in February 1998, the New York Fed hosted a conference to outline the agenda for achieving model-based capital for credit risk as quickly as possible. But that was before the Long-Term Capital Management bailout and a sober reassessment of the nature of credit risk.
Credit risk does not seem to arrive in neat little packets like market risk; rather it arrives through embedded options in products like revolving loan obligations, letters-of-credit, commercial-paper standby facilities, or changes in the net replacement value of derivatives contracts. Further, credit risk, once assumed, is not easily traded away. Most loan covenants do not allow the assignment or trading of the loan; short-selling of corporate bonds is difficult because of their thin float and non-existent repo market, and financial institutions with appetite for a given counterparty would rather use it to sell their own products to that counterparty than to re-insure products originated by another financial institution.
Finally, because default is an inherently dis-continuous event, there is no tool in the armoury of the credit risk manager akin to dynamic replication in the market-risk setting. The only absolutely reliable risk-mitigation technique in credit is diversification, and it takes many more holdings to achieve diversification in credit (a minimum of 1,000 names) than it does in market risk (comfortable at 100 names). This is because a single default in a portfolio can quickly lose more money than the combined upside of all of the non-defaulting instruments (represented by their cumulative yield above the risk-free-rate).
The outline of the solution is clear, but arduous. Credit risk will need to ultimately reside in portfolios of long implicit duration, which are large enough to contain thousands of individual names and well-managed enough to avoid any significant concentrations or correlated risks. Those managing the portfolio must have sufficient historical data to grade the incoming instruments and to model the performance of the portfolio over the life of the engagements, which would include at least one full business cycle (if the business cycle is indeed still alive!). Supervisors must be prepared to rectify distortions - such as allowing banks unlimited leverage and blunt capital requirements - so that the predatory pricing promulgated by banks ceases to distort the true price of credit risk.
At some point, the creation of an exchange where credit can be traded with sufficient liquidity, independent of the interest-rate risk inherent in bond or loan trading, will be the catalyst that truly vaults this project forward. The difficulty lies in defining a standard product that is meaningful to all end-users. This is compounded by the multiplicity of products containing credit risk, as well as the varying bankruptcy statutes and protocols throughout the industrialized world. Even when such an exchange is started, there will remain the question of who guarantees the contracts. Purchasing credit insurance essentially replaces credit risk to the original obligor with credit risk to the seller of the credit insurance (or credit derivative). The risk to the seller is second-order risk (the obligor must first default before the seller is responsible), but if the fortunes of the seller are highly correlated with the fortunes of the obligor, the purchaser of credit insurance may discover that the transaction provides no real additional protection.
Recent crises demonstrate that the financial industry has much to learn about the mathematical behaviour of correlations, particularly during market shocks. All of this points to the need for a central clearing agent (like the London Clearing House or the Options Clearing Corporation) to act as guarantor for traded credit contracts, which in turn requires a rigorous margining procedure. Best guess at the timeframe for these developments: between five and 10 years.
One fascinating modelling issue, which should be resolved on the road to generic trading of credit, is the relation between the value of a company's equity and value of its debt. The difficulty of establishing this relation is reflected in the current lack of consensus on how to value convertible bonds, which require a unified view of equity and debt for a single firm. In his seminal option paper, Robert Merton indicated that equity and debt could be considered options on the value of a firm's assets, where owning shares was equivalent to being long a call option and owning debt was equivalent to being short a put option (see figure 2). Of course, the "assets of a firm" are a non-observable construct (assets = equity + debt), once book values are replaced with market values. However, some researchers are finding the insight useful, and one risk- technology firm, KMV Corporation, now publicly offers tools for credit evaluation based on the Merton idea. In its first 30 years, the Black, Scholes and Merton paradigm conquered market risk; it is entirely possible that in the next 30 years, it will be a key that unlocks the door to credit risk management.
The second dimension of risk measurement, which will undergo tremendous change in the next decades, is the matching of assets and liabilities for financial institutions. What is now required is the marriage of two specialities - actuary and risk manager - because each understands only half of the picture.
Insurance firms view their assets as relatively stable and focus on the distribution of liability outcomes; banks view their liabilities as stationary and concern themselves with the variation in asset valuation. The truth is that risk cannot be measured by any absolute metric, whether applied to a group of liabilities or a portfolio of assets. Rather, risk is relative, arising from the interplay between assets and liabilities.
For example, a 30-year treasury bond might be a perfect hedge for an insurance company, which must pay out a schedule of cash flows identical with the coupon. But the same bond might be a very risky holding for a securities firm which is funding the position with overnight borrowings. Similarly, a long equity position in an illiquid security might be extremely risky for a fund which allows daily redemptions on the closing price, but it might also be a very good hedge for a pension fund whose defined-benefit scheme is linked to the salary levels in that particular industry. Without knowledge of the liabilities which the asset is intended to service, or, conversely, without knowledge of the asset which backs a particular liability, it is impossible to make a comprehensive assessment of risk. But banks don't usually assign particular assets to particular funding vehicles, they apply a blended cost of funds encompassing all liabilities.
Matching of assets and liabilities leads naturally to valuation issues. Mark-to-market (and mark-to-model) valuation has been the standard among securities firms for years; banks have accepted the regime for their trading books, but are reluctant to extend it to their investment (that is, loan) books. They argue that securities firms, whose portfolios are often funded with short-term money, should continue the mark-to-market discipline, because they may actually need to liquidate quickly if funding sources dry up. However, banks with much longer maturities in their liabilities, and much better access to liquidity, gain little by chasing noisy, daily market prices, as long as there is no systematic deterioration in their assets. Indeed, say the sceptics, forcing everyone to mark all of their assets to market simply makes the entire financial system vulnerable to very public losses in a crisis, further eroding confidence at exactly the time when cooler heads must prevail.
The way out of this impasse is to link the valuation rules for an asset with the basic features of its paired liability. Thus, positions funded with short-term money must be marked-to-market daily; positions supported by long-term funding do not require so volatile a valuation scheme. Since this project will necessarily involve the accounting standards boards (who must first resolve their own battle for global primacy), expect it to take significantly longer than credit risk measurement: 10 to 20 years.
The third and final class of innovation in the next century concerns liquidity risk measurement. Liquidity risk is the danger that a financial institution cannot renew its short-term sources of funding, whether in the interbank or repo markets. Most banking supervisors require a portion of assets to be in instruments which can be easily liquidated, implying that the first level of responsibility in a liquidity crisis lies with the financial institution itself. If a firm has been carrying those assets linked to short-term funding at liquidation values, the initial phase of a crisis can be weathered. However, if the loss of confidence persists, then it is up to the central bank to provide what amounts to re-insurance. Regulatory capital can be thought of as the deductible on that re-insurance coverage. From this perspective, any financial firm whose failure would generate systemic risk (too big to fail) must carry proper capital for all of its activities. (US bulge-bracket securities firms have capital requirements only for their broker-dealer and banking subsidiaries, but not for other operations.) Similarly, no financial firm should encourage firms outside of the capital requirement regime (hedge funds, for example) to assume excessive leverage ratios.
Once all financial institutions are on equivalent capital standards, the emphasis will shift to the question of providing liquidity to central banks in emerging-market countries. If global financial institutions are properly capitalized, the danger of a run on one of those players is far less than the ever-present danger of a run on an emerging-market currency. The global nature of financial markets has rendered emerging central banks impotent in their ability to provide liquidity to their own financial systems. Since much of the bank, corporate and sovereign debt in these countries is denoted in reserve currencies (dollar, sterling, euro and yen), the ability to supply liquidity in the home currency is virtually useless in crisis management. Even a peg to a reserve currency is ineffective, in the absence of a true currency board, and there is some question as to whether a currency board is sufficient (in Hong Kong and Argentina, for example).
There is serious discussion of emerging market countries adopting a reserve currency as their national currency. Setting aside nationalistic emotions, if an emerging-market adopted a reserve currency (Argentina adopting the dollar, for example), what responsibility would that place on the central bank which managed that reserve currency to offer liquidity in the event of a banking crisis in that country? And why would the central bank accept that responsibility if it had no supervisory authority over the banking system in the emerging-market country?
Looking out 20 to 30 years from now, it seems probable that the central banks in those countries which truly want their markets to emerge will need to seek the protection/sponsorship of a reserve-currency central bank. It may not require the abandonment of the national currency, but it will require intervention in the supervisory process of the emerging market banking system as the price of defending a currency peg. The alternative, which is to capitalize the IMF sufficiently to allow it to fund intermittent currency defences against the global financial system, seems entirely too Orwellian to prevail.
Robert Gumerlock, former managing director of risk control for UBS, is a risk consultant based in Menlo Park, California (firstname.lastname@example.org)