Effective corporate governance depends on achieving an appropriate balance between cost and risk and return, both in day-to-day business and in strategic management. For this reason, cost control and risk control strategies should seek to limit the scope for adverse variations in earnings, as well as negative B/S impact. To a considerable extent, adverse variations can be effectively judged through stress scenariosarising from costs and material risks being assumed.3 Simple linear models help only up to a point in cost reductionand risk identifi- cation. Stress tests provide more rigorous responses, particularly in relatively new business lines and in complex or unusual transactions; also, in response to exter- nal events that affect a vital business activity, such as the continuous monitoring of ourportfolio, and evaluation of each transaction’s P&L. Stress tests should be:
● Administered consistently, and
● Cover all risk categories and their detail.
The frequency of their implementation must be increased after a significant mar- ket change, budget overrun, or growth in exposure. But what is really meant by a stress test? The most straight-forward answer comes from engineering: putting a man-made product or system under stress conditions. Usually, though not always, the aim is to make inference about a product’s or system’s behaviour in a compressed timeframe.
Take quality of incandescent lamps for the consumer market as an example. Such lamps are supposed to work for an average of 1000 hours under 240 volts. Curves relate voltage to useful life, and by testing the lamps under 360 volts it is possible to estimate the statistical distribution of a manufactured lot in a much shorter time, since lamps get used faster under higher voltage. This is destructive stress testing.
By contrast, in finance we use statistical stress testing. An example is provided by looking at Figure 16.1, which shows a normal distribution curve with a long leg and a spike. Within two standard deviations, s, each side of the mean, x, lies the 95% of the area under the curve. This rises to 99.99% within four standard deviations either side of the mean. At that 99.99% level, it looks as ifpractically
every event has been covered. But it is not so. The reason is not only that the normal distribution is an idealized case; it is also that statistics depend on past information – while present and future events that we want to study may fall out- side these limits as the nature of the distribution evolves.
The effort of trying to capture nowthe after-effect of likely but not so probable future events is served through a method similar to that of testing incandescent lamps. Such events may lie at the long leg of a distribution, and this is investi- gated through a stress testat 5, 10 or 15 standard deviations, depending on what we are after.
Putting assumed exposure in the trading book and banking book under stress conditions provides a certain assurance that future risks involved in transactions can be studied beforehand in terms of their after-effect. This also helps in the sense that some of their characteristics become transparent and subject to appro- priate risk evaluation. For this reason:
● Stress loss measures should be most extensively implemented for trading activities, including credit risk, market risk, legal risk, country risk, and other exposures.
● Default stress testing, too, is very important not only for the loan portfolio, with particular emphasis on lower-rated borrowers, but also for invest- ments and derivatives trades.
A stress lossframework is a dynamic process which must be continually enhanced and progressively extended to all classes of risk categories. The identification and
x 1s FREQUENCY
2s
3s
VALUES OF VARIABLE UNDER STUDY SOME 90% OF
ASSET CLASSES OCCUR HERE
LONG
LEG SPIKE
Figure 16.1 The normal distribution curve is an approximation: future events whose impact we study may well be outliers
quantification of risk under stress conditions has become a vital and integral part of a properly tuned internal control, as the introduction has explained.
While sensitivity to risks being assumed is an important, never-ending business, equally critical is the ability to effectively communicate to all levels of management the results of risk tests, so that corrective action can be taken without delay. Delays increase the pain and cost of corrective processes, while at the same time they diminish their effect. A similar statement is valid in regard to cost control measures.
In the general case, whether the theme of analysis is cost control or risk man- agement, an effective communication of stress testing results must be made in a comprehensive manner. This should account for company culture and manage- ment style. Key to well-designed management style approaches is the ability to:
● Follow the way executives make decisions, and
● Frame the reported risk in a way that works in synergy with these decisions.
Moreover, valuation models must be used to identify strong and weak positions, accounting for the fact that financial markets are discounting mechanisms. This should make us sceptical in regard to the often-made assumption that ‘the future will replicate the past’. Except (sometimes) in the very long run, this never really happens.
Because the future holds surprises, managerial and professionals must appreci- ate the importance of experimentation and simulation, including the steady development of better methods and sharper tools to avoid the impact of stan- dards fatigue (see section 4). Just as important is a system design that permits the company’s internal control system to work both vertically and horizontally – vertically by digging up greater detail, which reveals the hidden side of expo- sure; horizontally by addressing the risk assumed by:
● Any instrument
● With any counterparty
● By any officer or trader
● In any business units
● In every country of operation, and
● At every time of day or night
The rationale of experimentation, and the way to go about it, has been explained in Chapter 15. Supported through analytics, the enterprise management system must not only reflect on each and every exposure, and on the synergy which exists
between them, but also correlate the level of exposure to market liquidity, market volatility, and our bank’s financial staying power. As an example, Figure 16.2 presents a frame of reference for senior management reporting.
Taking the banking industry as an example, the complexity of this job is increased by the steady innovation characterizing financial instruments. As a result, the entirety of the enterprise risk management problem cannot be analysed, and can- not be accurately modelled, by using only the currently available artifacts. We need methods and metrics able to identify differences between models like VAR required by regulators, and an holistic solution which is future-oriented.
● Determining ourbank’s current exposure is reasonably straightforward.
● Determining ourbank’s potential exposure is much more complex, and it cannot be done without stress tests.
Not only do we need to anticipate which risks might increase beyond prudential limits but we also should consider which new ones enter into the risk and return equation. Once these exposures and their potential magnitude have been identified,
OUR BANK'S LIQUIDITY VERSUS THE MARKET'S
MARKET LIQUIDITY
MARKET LIQUIDITY VERSUS
MARKET VOLATILITY
OUR BANK'S RESERVES
MARKET VOLATILITY
EFFECT OF MARKET VOLATILITY ON OUR RESERVES
Figure 16.2 Liquidity, volatility, and financial staying power
the next move is to determine the type of information necessary to track them, and keep them in control. Steady vigilance, timeliness, and accuracy are at a premium.
In the case of the highly leveraged portfolio of Orange County, for example, prior to the December 1994 meltdown of its treasury most of the repo desks at investment banks which acted as counterparties had not even been calculating daily price change of the collateral. Instead,
● They were marking to market this collateral once per month
● But while doing so they did not know the value of structured notes they sold and neither did they appreciate their counterparty’s ability to with- stand market stress.
Never believe theoreticians who say that hedging creates a risk-free portfolio.
What silly statements like that don’t bring into perspective is that even the more perfect hedge can be invalidated because of market changes. As the price of the hedged commodity fluctuates, the investor must alter the composition of his or her portfolio, or be subject to an inordinate amount of risk.
For this reason, only steady vigilance involving alert risk management policies and procedures, which are supported by high technology, can ensure that our institution has under control the risks which it is assuming. Let me add that while it is evident that this cannot be done through manual methods, mathe- matical models and computers will not automatically provide the needed rigor- ous solution. Snapshots are most often rushes to judgment for which senior management feel guilty later on. Rigorous analytical processes are best supported through long-term stress testing.