Alternative Investment Management Association Representing the global hedge fund industry
Clarus Risk Limited
Five years on from the global financial crisis, the focus on risk management is set to increase with the impending arrival of the Alternative Investment Fund Managers Directive (AIFMD). AIFMD requires risk management to be independent, both functionally and hierarchically, from the investment management operating unit. In addition to the requirement to separate these functions, where the AIFM is offshore it will be necessary to demonstrate that the function is carried out rather than delegated. Therefore, for many investment firms this presents a timely opportunity to review the roles, requirements and challenges of both risk measurement and risk management and how they are tested in practice.
While financial re-regulation is still being shaped, investor demands following the financial crisis were more immediate. Diminished trust in the investment industry has led to investor demands for improved transparency and reporting along with greater scrutiny of risk management resources. Institutional investors face their own increased regulatory requirements concerning their alternative investment allocations. A low return environment, particularly for hedge funds, has added further pressure for more frequent and thorough evaluations of the rationale behind alternative fund investments. An investment manager is normally chosen and appraised for their ability to add value by applying a particular strategy often to a specific asset class. While an investment fund will normally describe its process and methodology in a prospectus, it has proven hard to enforce investment mandate criteria legally.
This point has been illustrated once more in a recent case in which a UK judge issued a preliminary ruling against a group of UK pension schemes suing an infrastructure fund over breach of mandate. Cases such as this highlight the benefit to investors in having clear and demonstrable risk criteria and parameters to assist with reducing information asymmetries as well as more generally assisting with corporate governance issues.
Effective risk measurement requires specialisation and a dedicated risk measurement department is particularly common within larger fund management groups. A suitable risk framework will facilitate market and liquidity risk aggregation across asset-classes and trading books and requires expertise along with suitable computational tools. Such a framework not only enables scenario and stress-testing at the fund level but also allows for detailed reconciliation of risk and exposure at the portfolio and security level. The merits in separating risk management from investment management are more contentious, particularly from the perspective of investment managers. The 2012 Deloitte ‘Alternative Investment Fund Managers Directive Survey’ showed that 78% of managers had concerns over the delegation and substance requirements of AIFMD, a driver behind the requirement for independent risk management. A major concern is over the costs involved in changing an existing business model. Also, from the investor perspective, there is a danger that portfolio management teams may perceive themselves as less responsible for risk-taking when an appreciation for risk management should ideally be intrinsic throughout an investment firm. In practice, firms that are more quantitatively-oriented tend to have less of a distinction between risk measurement, risk management and investment management. Whether integrated with investment management or standalone, risk management departments have to improve upon a stereotype associated with an over-reliance on mathematical modelling, a poor appreciation of trading practicalities and most importantly a questionable track record through market crises. A greater degree of independence should also bring increased accountability.
Arguably the most maligned metric in risk measurement is Value at Risk (VaR). Its role within banking and consequently the financial crisis has been well documented. VaR is intended to quantify expected losses with a degree of statistical confidence. Traditionally VaR has been calculated by applying historical data and, as such, is highly sensitive to the conditions of that period, in particular correlation and covariance which are often assumed as constant going forward. It is accepted that the conditions in a crisis are markedly different and, in particular, that diversification benefits disappear in a crisis. The expected losses in normal market conditions are much less important than the expected losses in crisis conditions.
The purpose of stress-testing is, by definition, to model stressed conditions. By applying a measure of financial turbulence, it is possible to distinguish between normal and stressed environments and to condition risk metrics such as correlation and covariance, and by extension VaR, accordingly. While this is one of several ways to improve risk measurement, it is important to test the framework and ensure that it does not exacerbate procyclicality. This has been another criticism levelled at VaR as a risk criterion, as well as being of general concern for financial regulators. Encouraging additional risk-taking in benign environments and de-leveraging and de-risking in periods of heightened volatility (and correlation) serve to amplify risk and in particular systemic risk. Employing risk metrics, which are dynamic, by taking into account the market environment, is more likely to lead to prudent allocations based on more austere and realistic assumptions. Hence, de-risking should be less pronounced. Moreover, the more specific the framework is to the management objectives and style of the fund, the more likely it is to have a positive impact on decision making and performance.
VaR and Expected Tail Loss (ETL) are two of the main risk criteria likely to be regularly measured and tested as part of a risk management process. Market risk parameters often include various exposure-based criteria and limits on the gross leverage employed by an alternative investment fund. Risk concentration metrics, whether by economic sector, asset class, currency or security, can prove helpful in further controlling aggregate risk and also in meeting investors’ ‘style risk’ concerns by ensuring that the fund is committed and provides references to its mandate and prospectus. Another highly important category of market risk normally addressed in risk parameters is liquidity.
While common for each trading or portfolio team to have its own risk tools and criteria, an independent and specialist function is often required to provide a consistent approach to enable aggregation. Also, a risk measurement team tends to focus more on the possibility of extreme losses over a short horizon while an investment manager may be more concerned over the risk of a trade or strategy failing over a longer horizon. The following case study illustrates how risk management processes are often tested by market crises.
Monitoring a hypothetical long / short USD alternative fund during the eurozone debt crisis from 20 July 2011 to 12 August 2011 illustrates the timeline and interaction between market turbulence, estimates of risk, risk parameters and actual losses.
Figure 1 plots a measure of financial turbulence versus the S&P 500. The methodology employed in constructing this daily turbulence index is based on an analysis of the behaviour of a set of global indices over the most recent month against the preceding five-year period. The goal is to have a daily metric which can be used to distinguish between normal and stressed periods. Research has shown that turbulence is persistent and also associated with periods of lower returns. Whether a period is normal or stressed is driven by the level of the turbulence index.
The threshold between normal and stressed periods is breached on 28 July, six days after eurozone leaders announced a second bailout package for Greece. The change in regime is marked in the chart by a change from dark red to light red.
At this point, although the S&P 500 had dropped 4.6% from the start of the period, the transition occurred during the week before the most extreme losses, increasing the market’s volatility amidst the crisis.
Source: Bloomberg, Clarus Risk
Figure 2 shows the P&L realised by the portfolio versus the 1% and 99% $VaR over the period of review.
The portfolio suffered some large losses at the beginning of the period. Importantly, rather than reducing risk, the portfolio became riskier. This is shown by the $VaR boundaries at the 1% and 99% level widening in the top chart of Figure 2. Adjusting for market conditions, it can be seen that the 1% $VaR went from an expected loss of just over $1MM at the beginning of the period to a financial turbulence adjusted 1% VaR of over $2MM by 3 August.
Within this case study, the portfolio breaches a number of risk parameters. Measures of net exposure and two asset concentration limits are breached over this same period, and breaches are shown in the lower chart within Figure 2.
This portfolio suffers an extreme drawdown over the period. The most serious daily loss occurred on 5 August, 11 trading days after the first risk breach and seven days from the turbulence metric indicating a stressed environment. Episodes such as this can clearly unfold over short periods of time but can cause long lasting damage to reputations, investor confidence and in extreme cases the viability of the fund itself.
Risk measurement is part of the process, implementing the results through risk management is another, while the ability of firms to implement policy in an actual crisis is much harder to stress-test.
Source: Bloomberg, Clarus Risk
Figure 3 shows the distribution of 1% $VaR by each underlying position on 5 August. While there are hundreds of positions within the portfolio, there are a small number of positions that account for a disproportionate level of expected loss. These instruments are the drivers behind the concentration risk breaches recorded and the large losses experienced through this period.
This sample portfolio contains a large number of equity and equity index derivatives which require aggregation, not only to the portfolio level but also by underlying instrument for concentration limit purposes.
Source: Bloomberg, Clarus Risk
This case study illustrates some of the challenges already documented and the benefits of a methodology or process which differentiates a normal period from stressed. Risk criteria and parameters should provide a balance between the goals of capital preservation with a suitable investment return opportunity set.
Fall-out from the global financial crisis has brought about a number of improved risk and reporting practices. Investors in alternative funds are in a stronger position to demand more detailed and tailored risk parameters which mitigate information asymmetries and improve corporate governance. Risk measurement techniques have also improved, moving on from the assumptions of Modern Portfolio Theory to differentiate between normal and stressed regimes to provide more robust risk estimates. AIFMD will add further structural requirements including the independence of risk management from investment management and in doing so increases the oversight role and accountability of boards of alternative investment funds. To provide benefit to investors, risk management functions will need to test their ability to implement policy in an efficient and timely manner during periods of stress. Our case study illustrates that the risk management challenges that arise during a market crisis can escalate quickly and while there are conceptual benefits in requiring independence they will introduce additional financial and implementation hurdles. With AIFMD implementation imminent this is a critical time for investment firms to review their risk management framework and functions.
This is an updated version of an article that appeared in Professional Investor Magazine, Winter 2012
Back to Listing