Senior risk advisor,
As the dust of the financial crisis settles patterns of a new order have begun to emerge. Recent market volatility, has led to increased risk awareness at board level, fuelled by additional internal and external regulation driven-demands. The board therefore needs better risk management tools based on stress-testing techniques at the expense of Value-at-risk (VaR) techniques. This implies radical underlying architectural changes, a fact most board members are not sufficiently aware of.
What are the necessary changes? Let us start by imagining a new risk boardroom. At first glance it might not look too different from what already exists. Board meetings will usually still start with the actual position of the firm expressed in terms of market, fair, and book values. It is further informed by additional factors including sensitivity gap and liquidity gap numbers plus other sensitivities, such as duration and option Greeks. There might also be a VaR report. All these numbers reflect a current environment which is, in most cases, the best basis for making decisions. After all, 80 per cent of the time can be classed as “normal” and “quiet” times (which also should be treated in this way). Constantly working as if a black swan is around the corner will not lead to an optimal outcome over the long term.
Having said this, risk management should not be considered paid for normal time, but time spent steering ‘the ship’ through stormy waters; these tools have to be ready to battle the next black swan. Black swans are new situations which cannot be dealt with by history-based statistical techniques. They call for a combination of human intuition and simulation capabilities as found in stress testing tools. Anybody in the room should be able to ask any question relating to a potential disaster. For example what would happen to the value of the firm over the next five years if inflation went up by 20 per cent, short-term rates by 40 per cent and long-term rates by 20 per cent? Or, what would happen if ratings fell and the probability of default rose by X amount for a different sector? The risk manager should then ask additional questions, such as whether book value or fair value is meant and then produce results in a few minutes. In addition to value, he or she should also assess the impact on income or liquidity visible.
How do you build a system that is ready for the next black swan without falling into the same errors as generals who – haunted by memories of past victories and defeats – are accused of “fighting always the last war”? In the following we will outline some of the important characteristics needed to prevent this backward-looking bias. It should also offer an insight into the potential of such a system.
Obviously the first essential for such a stress testing system is the capability to answer questions hitherto unasked. A second important property is consistency. The points mentioned above regarding rate shocks relate to the whole bank running through all positions starting from saving accounts to exotic options. Such questions can only be answered in reasonable time if there is a common underlying structure throughout. Thirdly, the system must be in a generalised simulation environment in order to respond to the diversity of questions that will be asked.
No doubt the biggest challenge is the black swan issue, which demands a system for answering questions not previously asked. This might look like an unsolvable problem but there are examples of a few systems in human history which have created a satisfying solution. For example, the alphabet invented about 1,500 BC - successfully mastered the writing of anything man can express in any language with only minor adaptations. Is a similar system also possible in finance?
The basic architecture of such a system is shown in figure 1. The first level distinction must be between the pink input - and the green and blue analysis elements. The input elements are the factors that make up the wealth and value of a financial institution. In the centre we have the financial contracts surrounded by market conditions (yield curves, FX rates etc.), counterparty conditions (rating, probability of default etc.) and behaviour (contractual properties which can only be expressed statistically like prepayment, saving account withdrawal etc.). Financial contracts produce cash flows dependent on these external conditions from where the green and blue analysis-elements can be derived. The analysis elements represent the information the management finally wants to see. Value, income, sensitivity, liquidity and risk are shown in different ways as income forecast, balance sheets, liquidity gap numbers, VaR even regulatory reports or – our main target – stress test results. There are many more types of analysis, but they are a combination of these basic elements. Besides being an excellent stress testing system, this system is a general framework for financial analysis.
A necessary feature of a general stress testing system is the distinction between fact and fiction. Now, what is hard fact in finance?
The only factual and immovable object in finance is the financial contract itself. A financial contract is a promise between two counterparties to exchange cash flows according to fixed rules. Such fixed rules can be represented mechanically and be hard-wired, making it easy to handle in an IT environment. This puts the notion of the financial contract into a pivotal position.
But what about broken contracts? The breaking of a contract does not devalue the original pledge, which remains fixed. However, when a contract is broken, counterparties are an additional factor that need to be taken into account. Also market conditions have to be added to the system, because many cash flows in today’s financial world are contingent - such as interest payments of swaps, option and future payments. Finally we need behavioural assumptions.
The difference between the central contract and its orbiting parts is the fact that the former remains fixed and the latter can change. These potential changes are the sources of risk that need to be stress-tested. Counterparty, market and behavioural assumptions make up all potential risk factors. Change in probability of default, market conditions or saving account holders are typical cases but many more exist. They can occur individually or in combination, in static or dynamic (forecasted) basis. Having a fixed part in the centre is crucial for the system. When fixed they can be programmed and – once well implemented – be forgotten by the risk manager. The risk manager should never have to worry about the contracts - only about the moving parts which affect them (which can be different in every crisis).
Being in this central position, the system can only work if contracts are standardized homogenously throughout the whole enterprise for analytical reasons. This is a condition hardly met by any bank today not least, because of a failure to understand the concept at board level. Although an imperative for a system needed at the board level, it will only get there once the top management becomes conscious of its importance.
The question might arise at this point, how it is possible to guarantee that such a system will really fight the next and not the previous war? There is no way of proving this mathematically, but a heuristic and an auxiliary condition might help to explain it. Heuristically, the system has been devised more than twenty years ago claiming to be capable of fighting forthcoming wars. These twenty years have been a constant war with changing fronts and conditions and the system got corroborated by it in practice. There was never a need to change its underlying structure. However, it needed some extensions. These extensions are necessary and form the auxiliary argument. Just as the alphabet may need additional letters or signs when transcribed as a new language, the system also needs to evolve over time without changing the base structure. The system has to be built in a way that new contract types, new market risk factors and behavioural elements can be added without pain.
Such a system is not only necessary to fight the black swans of tomorrow but also to avoid a risk management disaster. Without such a strong underlying architecture, the risk management and regulatory reporting task could stifle management both mentally and financially. Risk management would become itself a main risk source for the industry which is to be avoided at any cost.