Running through all scenarios
It is important that investors do not underestimate tail risk, argues Charlies Prideaux, as he makes the case for the extensive use of scenario analyses.
In a world where many long held assumptions no longer apply, answering the 'What if …' question has become a crucial tool for institutional investors. Based partly on the classic methods used by military intelligence, it offers investors an effective method to ‘road-test' possible investment strategies. In most instances, this involves investigating three-to-five possible outcomes for a given strategy with a view of understanding the interaction of the risks driving each outcome. But beyond the principles, what are the practical considerations institutional investors should bear in mind?
Why use scenarios?
The techniques generally used to estimate the likely risk-return profile of potential investment strategies often come with an excessive focus on a 'central expectation', meaning that tail risks can be underemphasised or ignored completely. Moreover, the results of these techniques can be opaque and complex. For example:Mean variance analysis generates a risk number expressed as the standard deviation of the return, which many investors find difficult to assess in practice. Value at Risk (VaR) frames risk in terms of a specific time horizon and contains hidden assumptions about tail risk.
A stochastic asset liability model, which can generate a huge range of simulations can make it hard to ‘see the wood for the trees' and draw meaningful conclusions. In comparison, scenario analysis, is relatively transparent and can be applied to an investor's individual circumstances to answer key questions such as 'How much might I lose?' and 'What about my funding ratio?' Furthermore, it can act as a reality check on other models used.
Scenario analysis also has a significant advantage of enabling insight into complex issues where risk factors can interact in unexpected ways. This kind of deep-drilling into risk exposures is particularly useful in times of fundamental economic uncertainty such as we currently experience.
How to undertake scenario analysis?
It may seem intuitive, but great care is needed to ensure that the analysis really focuses on the key issues. For instance, the assumed correlation between equities and interest rates is a critical assumption for a typical pension fund and the possible scenarios for this relationship should be investigated carefully.
Once the key issues for the portfolio are determined, the next step usually involves identifying a suitable library of investment crises and analysing how different asset classes reacted at the time. One limitation of this approach is that history is frequently messy, with the 'wrong' result often occurring at the wrong time because of extraneous factors.
For example, following S&P's recent downgrading of US government debt, most investors would have expected long US rates to rise. In the event, rates fell because of a flight to quality amid wider global macroeconomic concerns. A more flexible alternative therefore is to identify the key 'principles' beneath historic occurrences and assess whether they have changed in the interim period. On the downside, this can lead to very complex statistical economic models. We would argue that most investors are better served with estimates that are relatively 'quick and dirty' following the principle that it is better to be approximately right than exactly wrong.
Another important choice facing practitioners is whether to model scenarios simply as an immediate shock or as a developing situation over time. Both approaches have their pros and cons, but it is clear that predicting timing and direction is more difficult than predicting direction alone. In practice, it seems that most of the benefit of scenario planning can be captured by simply analysing one-off shocks without having to get bogged down in complex predictions of timing and phased development of interactions between risk factors.
This pragmatic approach should also apply to the presentation of results to ensure that the key findings are not lost in a raft of numbers. Care must be also taken in labelling scenarios as 'optimistic', 'pessimistic' or 'best case' as this can focus undue attention on quantifying possibilities rather than understanding the nature and interaction of risk factors.
A continuing process
Clearly no scenario analysis will deliver a perfect projection, so it is essential to avoid the trap of becoming committed to predicted outcomes or spurious precision. Furthermore, it is important to see scenario analysis as a dynamic process, particularly given the uncertain world we now live in, and replace old scenarios with new and more pertinent 'What if …' questions as soon as circumstances change. In doing so, scenario analysis allows investors to create a real world framework in which to take their actual investment decisions with a robust understanding of the potential outcomes.
Charles Prideaux, is head of institutional business EMEA at BlackRock.