 To most of us in the financial sector, the events of 2008 were shocking. We have all struggled with the aftershocks and asked ourselves what happened and what led us there. Some have argued that this is an unprecedented situation, which may be true in terms of the speed and magnitude of the shock, and possibly the long-term impact on the real economy. But others see echoes of the past in terms of the elements that triggered the crisis and the way it has been unfolding.
In fact, history points to many precedents. Rogoff and Reinhart [2008] identified 18 major banking crises since World War II, and a casual count of recent crises shows a pattern of a major market event at least every 10 years. Most of the time, these crises have been linked to excess liquidity in the economy, combined with a general misjudgment on the benefits of a certain type of innovation. This time is no different. Excess savings from emerging markets were reinvested in developed financial and housing markets, while financial innovation in the form of sophisticated securitized instruments contributed to a false sense of security surrounding systemic risk reduction.
The experience of this last market shock highlights several shortcomings in the risk management and asset allocation practices that have been commonly adopted by industry practitioners. It is now clear that the underestimation of the frequency and magnitude of extreme events left many unprepared. In particular, the consequences of contagion on portfolios have been overlooked. By increasing correlation across asset classes, contagion significantly reduces the benefits of diversification in otherwise-well-diversified portfolios. In addition, liquidity dried up in many parts of the capital markets, leading to cash-flow issues for most investors, including many long-term investors such as endowments and pension funds. Finally, the intensity of the crisis exacerbated the negative consequences of nonmarket risks, such as counterparty risk, operational risk and concentration risk. This is vividly captured by Warren Buffett’s now-famous quote: “Only when the tide goes out do you discover who’s been swimming naked.”
In our view, the events of last year will force a number of changes in investment practices in the following areas:
- Management of Extreme Events: Business continuity planning (BCP) is a standard operational risk control practice for organizations, enabling critical processes to function in all conditions. Asset managers and plan sponsors need BCP plans for their portfolios. Achieving this will require methods to measure and model tail risk, rules to trigger BCP plans on predefined market conditions, and a consistent framework for stress-testing portfolios and planning for extreme event scenarios.
- Strategic Asset Allocation: Investors are rethinking asset allocation practices with a view to better managing downside risk and avoiding investment horizon mismatches, in particular for individual retirement products. They may more formally take into consideration cash-flow requirements and eventually move toward more risk-based allocations.
- Integrated Risk Management: There is wider recognition now that investors need an all-encompassing view of portfolio risk. They need tools that help them understand the sources of risk, not just the absolute levels. Breaking asset class silos will allow investors to view the sources of risk across all assets in the portfolio. New tools will be needed to look beyond market risk into other types of portfolio risks, such as counterparty risk, concentration and liquidity risk.
The rest of this paper addresses these three areas in more detail by looking at the main issues of last year’s crisis and possible solutions to mitigate their impact.
I. Managing Extreme Events
Extreme events can be characterized by volatility jumps, increased risk aversion, negative returns for risky assets, and increased correlation across asset classes. Such events actually happen more often than is commonly perceived. In just the last 21 years, we have experienced 10 major market events: Black Monday (1987), Gulf War (1990), European ERM Crisis (1992), Mexican Crisis (1994), Asian Crisis (1997), Long-Term Capital Management (LTCM) (1998), Tech Bubble Crisis (2000), September 11 (2001), Quant Crisis (2007) and Credit Crisis (2008).
Aside from the frequency of occurrence, it is also interesting to note the volatility spikes and clusters through time. Figure 1 shows the one-day losses in the MSCI World Index over a period of 20-plus years and reveals how market volatility can change drastically and unexpectedly in extreme events.
Understanding Risk Regime Shifts And Flight To Quality
A high-volatility regime usually develops with a shock in one asset class, extends to other asset classes as investors worry about the consequences of the shock and, in a herding movement, snowballs into panic. At that point, returns from assets are inversely proportional to their riskiness, which is the opposite of expected behavior in normal times. This is illustrated in Figure 2, which compares the returns of various asset groups in the worst month of five major crises.
While equities were the worst performers in all these crises, bonds were also affected by the widening of credit spreads. In addition, assets that were previously negatively correlated may no longer remain so in a crisis, which significantly alters the diversification picture of a portfolio.
Figure 3 illustrates this effect by displaying the correlations between equities and bond premia in the last 10 years. In general, correlations are unstable and tend to change over time. However, during the periods of crisis as highlighted in orange, correlations generally rose substantially, which caused equities to become more correlated with bonds and hence reduced diversification benefits.
Since correlations can change quickly over time, particularly in crises, a well-diversified portfolio in normal times may still be subject to unexpected volatility in extreme circumstances. It is therefore also important for us to understand tail risk— particularly, how to estimate and manage it appropriately.
Modeling Extreme Events
A common statistic currently used to measure downside risk is value at risk (VaR). This statistic is defined as the maximum loss a portfolio is expected to incur over a specified time period, with a specified probability or confidence level. For example, if the one-week 99 percent VaR of a portfolio is 20 percent, then there is a 99 percent chance that the loss of this portfolio over a week is not more than 20 percent. In other words, there is only a 1 percent chance that the weekly loss would exceed 20 percent.
Traditional VaR methods, however, tend to underestimate the likelihood of extreme events because they usually assume that returns are normally or lognormally distributed. In reality, the empirical distribution of asset returns shows that extreme events, referred to as “fat tails,” are more likely to occur than normal distribution patterns imply. Assuming a normal distribution may underestimate the likelihood and magnitude of extreme losses. Risk estimates acceptable during normal periods are prone to fail during severe downturns. This problem of fat tails is not unique to financial markets and has received much attention in other disciplines, such as hydrology and structural engineering. Researchers in these disciplines approach this problem using extreme value theory, which focuses on the distribution of returns at the tails. This is precisely what is required for analyzing market crises, since these are extreme or tail events. The framework of normal distribution is only adequate as a reflection of average returns, not extreme ones. In extreme value theory, the distribution of the tail values instead follows the generalized extreme value (GEV) distribution. Once the tail distribution is determined, the VaR can then be computed as in the normal distribution.
Figure 4 provides an example, extended from Goldberg et al. [2008], of how extreme value theory can provide a better reflection of the downside risk. We compare the relative robustness of the traditional VaR and the extreme-value VaR for a variety of portfolios composed of U.S. equities. Daily returns are taken from December 1996 to October 2007, a period that covers major crises that include the Asian crisis, LTCM, Tech Bubble, Sept. 11 and the Quant Meltdown in August 2007. VaR figures are generated using two methods: the traditional way in which returns are assumed to be normally distributed and exponentially weighted across time, and through using extreme value theory. We choose a confidence level of 99 percent and a time horizon of one day, so that the resultant VaR figures should represent the maximum daily loss that would be incurred with 99 percent probability.
To compare the two measures over a variety of different portfolios, we evaluated 74 factor-tilted portfolios. It is important to note that the VaR numbers generated here are forecast values. In Figure 4, the horizontal axis is divided into interval ranges that denote the percentage of days in which actual losses are greater than the VaR, while the vertical axis displays the number of portfolios (out of the possible 74 in our sample) within each interval. Ideally, all portfolios should be in the expected range, left of the broken line. This is true for the majority of the portfolios when using the extreme-value VaR, but not in the case of the traditional VaR. While about 80 percent of the portfolios meet this criterion under the extreme-value measure, only 3 percent do so in the case of the traditional measure. Clearly, the estimates of risk generated using VaR are starkly different depending on whether the distribution is normal or not.
In addition to better modeling of fat tails, extreme value theory also introduces a tail-risk measure that provides a more complete reflection of the expected loss in a worst-case scenario. While VaR tells an investor his worst expected loss in 99 percent of the trading days, it does not indicate how severe the loss would be in the remaining 1 percent. Expected shortfall measures the expected loss within that worst 1 percent. Goldberg et al. [2009] have demonstrated how this new concept can be integrated in the standard tool kit used to measure portfolio risk, and show that different portfolios have distinct downside characteristics.
Options To Manage Tail Risk
BCP is a standard risk control practice for organizations, enabling critical processes to stay in operation even if a disaster strikes. This practice is well-established in many industries, including finance, as well as in nonbusiness organizations such as the military. We suggest that an analogous concept, which will be referred to as Portfolio BCP, would be relevant for portfolio management.
Generally, Portfolio BCP would require the following steps to be implemented. Firstly, the definition of an extreme event has to be determined. This could be based on returns, volatility, tracking error, VaR, drawdown or a combination of these measures as captured in a scenario. The probability and severity of the extreme events can then be quantified, as was carried out in Bhansali [2008].
Secondly, thresholds have to be decided upon, so that the conditions that trigger BCP are clear to everyone involved. Thirdly, scenarios should be elaborated to cover the most likely current threats. In order to rehearse these potential extreme events, stress tests should be performed to simulate the performance of portfolios under such situations. Finally, portfolio trades reflecting the mitigating decisions should be prepared and preapproved by the various investment committees for fast and consistent execution, should those extreme events happen.
In this context, the importance of stress-testing under extreme conditions should be emphasized, since this would help determine how much tail risk should be hedged away. In recent years, stress-testing has attracted the attention both of regulators and practitioners as an important measure that complements traditional risk measures such as volatility, tracking error and VaR. Increasingly, regulators such as the Basel Committee and the EU Commission (UCITS III Directive) require that practitioners incorporate stress-testing into their regular risk management practice. Stress-testing is particularly important because it helps to mitigate the overdependence on recent historical data. Multi-asset class risk systems often include dozens of predefined stress-testing scenarios.
II. Rethinking Strategic Asset Allocation
The issues of 2008 also highlighted the mismatch between the investment horizon, the level of risk investors can bear and the characteristics of their portfolio. The most problematic cases were seen in the defined contribution space, but this issue also affects mature pension plans.
A retiring worker who expects to exit the workforce in one year should take on much less risk in his investments than a young worker in his 20s who is looking at another 40 years of employment. Sounds obvious? Yet individuals retiring in 2009 who put their savings in a 2010 target-date fund would typically have seen the value of their savings shrink by 20–30 percent. For someone purchasing an annuity at retirement under these circumstances, it means a permanent loss of revenues in the same proportion.
Understanding Investment Horizon And Downside Risk
Asset allocation decisions have to be made in the context of the risk-return characteristics of various asset classes and the tolerable downside risk. The latter is linked to the length of the investment horizon. The key investment problem is determining the acceptable level of downside risk while at the same time maximizing long-term real returns, subject to protection against inflation or deflation.
The investment horizon is an important element of this problem because it is related to the appetite for downside risk. While the returns of an asset may be positive in the long run, in the short run the possibility of losses cannot be ruled out. Long-horizon investors have greater capacity to withstand short-term losses because they see beyond these short-run fluctuations to the long-term trend that will in time reassert itself. Figure 5 illustrates the various dimensions of the problem.
This chart compares the historical returns of U.S. stocks and bonds, net of inflation (real returns), over different holding periods or investment horizons from one year to 30 years. The dotted lines represent the average returns for stock and bonds for all periods and the solid ones show the bad outcomes (divider between the best 95 percent and the worst 5 percent). When you look at the simple average, stocks offered systematically higher real returns, even after the two major down markets of 2002 and 2008. These results for the U.S. markets are largely consistent with findings in other markets around the world over similarly long periods, even if the U.S. equity market has one of the highest returns, as highlighted by Dimson, Marsh and Staunton [2009].
However, when looking at downside risk, as measured by the worst 5 percent of returns, the picture is completely different. Equities were very volatile over a one-year horizon, while bonds were less so. Only cash provided more protection in extreme events. More generally, the downside risk was higher in the case of stocks for relatively short holding periods of up to five years. For longer investment horizons, the drawdown risk was not as significant, and in fact for horizons of more than 10 years, stocks even enjoyed a slight advantage.
Why is this particularly important for individual investors? Most retirement solutions offered to individuals today are trying to find optimal solutions for average investors, which is different than solving for the constraint of an individual retiring at a unique point in time. This is not surprising given that most research has been conducted in the context of vehicles such as defined benefit plans that pool assets and liabilities. Unfortunately, an individual in a defined contribution world does not benefit from pooling liabilities with others. An individual planning for retirement needs to have sufficient funds for his or her maximum life expectancy, not the average one. On the contrary, an insurance company providing an annuity service would be pooling the liabilities, allowing it to focus on the average life expectancy, since individual differences are canceled out at the aggregate level. Similarly, while the average investor will not retire in a catastrophic year since such years are not regular events, the prudent individual investor cannot rule out the possibility that he or she will retire in a year like 2008.
Segregation of assets and liabilities forces individual investors to put more emphasis on tail risk than pension funds with very long horizons. The optimal solution for an individual is therefore found in the context of the worst case scenarios, not the average ones. The logical consequence of this statement should lead to a dramatic change in the asset allocation of individual retirement products, such as target-date funds, toward less risky assets as retirement dates approach.
Toward Risk-Based Asset Allocations
Another reason why diversified portfolios did not offer as much downside protection as anticipated during the events of 2008 is that diversification strategies were often misapplied. In particular, it has become clear that the categorization of asset classes in a portfolio influences the approaches chosen for diversification. Many pension plans are still using a categorization of asset classes that groups assets into three buckets: equities, fixed income and alternatives. The category of alternatives includes private equity, private real estate, hedge funds, commodities and other real assets.
This segmentation reflects more the structure of asset management practices than the role that the assets are supposed to play in the portfolio, and that has led to some undesirable effects. Firstly, the fixed-income category has evolved to include a mix of low-risk government bonds with higher-yielding assets like corporate high-yield bonds and emerging markets bonds. The rationale for including these higher-risk fixed-income instruments in the fixed-income segment was that these assets were providing higher returns and diversification to the rest of the bond portfolio. That is true, but only if the bond portfolio is viewed in isolation. However, as we have seen in the first section, high-risk assets tend to be highly correlated in times of crisis. At the portfolio level, those risky fixed-income assets are in fact reducing the downside diversification that you expect from your allocation to bonds. Peters [2008] offers a particularly clear and elegant explanation of this phenomenon.
Secondly, the alternative asset class has often been viewed as in a world of its own, where its risk-return profile has no relation with the two other segments. Obviously, that is not correct. One of the characteristics of these alternative asset classes is illiquidity. Illiquidity can create the illusion that assets are uncorrelated if naively compared with publicly listed equivalents. In reality, there are many fundamental reasons to link alternatives to equities and bonds. Many studies have shown that hedge funds’ strategies have a high component of traditional beta. Private equity returns should closely follow public equity returns. Privately held and managed real estate assets are subject to the same real estate cycles as public real estate assets.
This general misperception of asset characteristics and correlation demonstrates how investors, in general, have lost sight of the fundamental aim of portfolio construction. In response, several leading asset owners are moving away from the traditional asset class categorization to a system that more explicitly accounts for the role of each asset in the portfolio. We will refer to this approach as risk-based asset allocation.
Under this approach, as illustrated in Figure 6, a risk-based asset allocation could be structured along four broad segments: equities, real assets, liability hedging bonds and absolute returns strategies. For investors adopting this approach, the equity segment opportunity set could be represented by global public equities covering the broadest investable universe, such as the one captured by the MSCI All Country World Investable Market Index (ACWI IMI). Allocations to private equity and long/short equity hedge funds could complement this core equity allocation, providing a more diversified return stream along with a strong alpha component. The primary purpose of the equity allocation is to provide the highest long-term real returns possible, matching long-term economic growth.
This core equity allocation could be complemented by real assets. The real assets category could cover real estate, timber and farmland, as well as commodities. Infrastructure investments could also be put in this category, although some may argue it belongs to the equity segment. Real assets would principally be included to provide additional and more effective protection against inflation risk. The third component of this asset allocation framework is liability-hedging. Given the high level of downside risk of risky assets, the closer a retirement plan is to the payout phase, the better the liabilities need to be matched with assets of similar nature and duration. This constraint calls for a mix of low-risk government bonds, including some allocation to TIPS.
A framework that included only these assets would still leave a high number of sources of risky returns unexploited. These sources could be found in the various risk premia associated with the fundamental factors driving traditional asset classes, such as the small-cap or value premium in equities or the credit premium for corporate bonds. For example, an investor holding small-caps receives a market return for investing in equities, plus a risk premium to compensate for the risks of small-cap securities.
Similarly, high-yield bonds yield a return or beta that equals the corresponding yield on a government bond of a comparable maturity, plus a risk premium or spread for assuming the higher risk of holding such a bond. Strategies that aim to capture a specific risk premium through the execution of systematic trading rules also qualify for the risk premium approach. For example, arbitrage strategies such as merger arbitrage or convertible arbitrage qualify under that scheme.
These additional sources of return could be captured in the absolute returns strategies segment of this strategic allocation framework. Briand, Nielsen and Stefek [2009] analyze the risk return characteristics of these risk premia and show that it is possible to create portfolios of risk premia that offer similar returns to a traditional portfolio composed of 60 percent equities and 40 percent bonds with significantly lower volatility. The returns of an equal-weighted one are displayed in Figure 7.
III. Developing An Integrated View Of Risk
Finally, the recent crisis in financial markets has raised awareness of the importance of some types of risk that often receive little attention under normal market conditions. These risks include counterparty risk, operational risk, liquidity risk and concentration risk.
Counterparty risk captures the potential loss from derivative contracts (including swaps, CDS, etc.) should a counterparty default. The bankruptcy of Lehman Brothers left many pension plans and asset managers scrambling to measure their counterparty exposure to Lehman, which is not a good sign of preparedness.
Operational risk refers to the risks attached to people, processes, technology or external events such as fraud. Brown, Goetzmann, Liang and Schwarz [2009] provide a detailed description of how to build a scoring model to detect operational risk in hedge funds.
Concentration risk is the risk arising from all investments in securities of a particular issuer, sector, country or other common factor. Concentration risk emerged in this recent crisis when securitized subprime assets were found to be present in many parts of investors’ portfolios without proper management of the aggregate exposure. Many cash management products, money market funds and other commonly used investment vehicles were found to be exposed to subprime assets.
The recent credit crisis has also reminded us of the importance of liquidity risk. Market liquidity may fall very quickly as risk aversion suddenly rises in a crisis situation, effectively shutting down certain segments of the market and increasing the cost of selling in most others. For investors, this raises the danger of being unable to meet cash-flow obligations. In 2008, the endowments that were heavily invested in illiquid asset classes, such as private equity and real estate, found themselves with a sharp liquidity mismatch and had to liquidate part of their holdings at distressed values.
Given the variety of problems, the dynamic nature of portfolios and the links between assets and asset classes, it is hard to imagine efficiently managing risk at the portfolio or enterprise level without a proper framework to measure, monitor and aggregate risk. Multi-asset class risk systems can help address this problem.
Fundamental Factors Driving Portfolio Risk And Return
While VaR and volatility act as barometers to provide an advance warning signal against heightened volatility, these measures are not useful in understanding the sources of risk. Looking at the fundamental factors driving each asset reveals the sources of rising risk. The individual impact of such factors can be derived using multiple-factor models, which not only measure and forecast risk for a portfolio, but also break down that risk according to the contribution from the various factors used in the model.
In the Barra risk models, for example, each stock receives an exposure to various factors such as country, sector or style (value, growth, leverage, liquidity, size, nonlinear size and momentum) that drive risk. These country and industry factors help to determine how much of a portfolio’s risk is specific to a particular country or industry after netting out other effects. For instance, a portfolio of Korean stocks is clearly subject to the country risk of Korea, but also reflects the risk of the technology sector that has a disproportionately high weight in that market. Using country and industry factors helps to disentangle these effects and identify “pure” exposures of a portfolio to country and industry risk. Similarly, a portfolio of bonds has different exposures to different changes in the term structure, whether it is a shift of the whole curve, a steepening or flattening of the curve or a change in its convexity.
At the portfolio level, these fundamental factors will be linked across different asset classes. For example, the value and size factors in equities will be linked to yield curve factors in fixed income, as well as other factors from commodities and other asset groups. Figure 8 illustrates how factors can be integrated into a multi-asset class correlation matrix.
With an integrated factor model, factors from the country level are in fact combined together to arrive at the global factors that apply to all markets. This greater granularity allows for different factors in different markets, thereby facilitating a finer analysis of exposures of various assets worldwide. By combining all factors from different asset groups, a portfolio manager can more effectively analyze portfolio exposure to multiple levels across asset classes.
In the recent crisis, many investors did not have sufficiently strong modeling capabilities to comprehensively cover the many derivatives in their portfolio, thus causing them to miscalculate the true risk of these instruments. It is now clear that there was an overreliance on credit ratings, which in some cases were based upon wrong or flawed models. Given the importance of models in risk measurement, modeling risk cannot be overlooked. The integrity of the models—and that of the modelers—should be assessed. Complex and untested instruments should be subject to additional prudent constraints.
Conclusion
This review has revealed the complexity of dealing with the many facets of portfolio risk. Hopefully, it has also introduced a high-level road map that may help investors find better ways to handle extreme market events.
The 2008 crisis has been an acid test for risk management. It has shone a bright light on institutions that put risk management front and center, and cast in bold relief those that may have neglected the tools and key concepts. While risk management did not fully prevent downside in portfolios, it is clear now that the organizations that invested in the intelligent dissection of risks and acted on their findings fared significantly better than those that did not.
At the end of the day, risk management is a state of mind, not a technique. To be successful at managing risk, one needs a desire to be prudent, the means to understand risk and the discipline to make difficult decisions.
References
Bhansali, Vineer [2008]. “Tail Risk Management,” Journal of Portfolio Management (Summer 2008), pp 68–75.
Briand, Remy, Frank Nielsen and Dan Stefek [2009]. “Portfolio of Risk Premia: A New Approach to Diversification,” MSCI Barra Research Insights (January).
Brown, Stephen, William Goetzmann, Bing Liang and Christopher Schwarz [2009]. “Estimating Operational Risk for Hedge Funds,” Financial Analysts Journal (January/February),
Vol. 65, No. 1, pp 43–53.
Dimson, Elroy, Paul Marsh and Mike Staunton [2009]. “Looking to the Long Term,” Credit Suisse Investment Returns Yearbook 2009, pp 11–18.
Goldberg, Lisa, Michael Hayes, Jose Menchero and Indrajit Mitra [2009]. “Extreme Risk Management,” MSCI Barra Research Insights (February).
Goldberg, Lisa, Guy Miller and Jared Weinstein [2008]. “Beyond Value at Risk: Forecasting Portfolio Loss at Multiple Horizons,” Journal of Investment Management, Vol. 6, No. 2, pp 73–98.
Peters, Ed [2008]. “Does Your Portfolio Have Bad Breadth?” First Quadrant Perspective, Vol. 5, No. 4.
Rogoff, Kenneth and Carmen Reinhart [2008]. “Is the 2007 U.S. Sub-Prime Financial Crisis So Different? An International Historical Comparison,” American Economic Review,
Vol. 98, pp 339–344.
|
