The belief that risk management means merely minimising or eliminating investment risk has few followers today. It is now widely accepted that investment risk is necessary to drive returns, and that it is the function of risk management to enable the asset manager to maximise the use of risk to get the appropriate return
The concept of risk budgeting is also well understood. Risk should be thought of as a scarce resource, Bob Litterman, head of quantitative resources at Goldman Sachs Asset Management suggests, and must be spent wisely. It should not be hoarded or squandered.
Pension fund managers now scrutinise asset managers’ performance to check whether they are using the risk they have been budgeted to use.
There are standard risk measurement tools to help them do so. They can use tracking error to measure the portfolio risk of traditional asset managers and Value at Risk to measure the trading book risk of hedge fund managers.
In theory, risk management and risk management tools should provide the information about how much risk investment managers are taking, whether intentionally or unintentionally. In practice, this is not always possible.
Kenneth Winston, global head of risk management at Morgan Stanley Investment Management (MSIM), has suggested, provocatively, that there is a case for saying that risk management tools work only when they are not needed.
He points out that under certain circumstances risk management tools might break down. He describes these circumstances as ‘poor risk prediction regimes’.
“We need to realise that no risk analysis tool is perfect. Financial markets are constantly changing and mathematical models have regimes during which they work well, and they have regimes where they don’t. “ he says.
“There is a danger of blindly relying on risk management tools or on a particular set of tools. We need indicators of when the standard risk management tools may begin not to work, and we need to know what other tools we can use in their place.”
Winston and his colleagues at MSIM have tried to identify the indicators of poor risk predictions regimes, the clues that conventional risk management tools may not be able to manage. One area they have considered is the general level of volatility in the markets, measured by the VIX, the Chicago Board Options Exchange (CBOE) Volatility Index.
In March 2003, during the run-up to the Iraq war, the VIX was exceptionally high, says Winston. “You might think that this was a time when, with a turbulent market as defined by overall levels of volatility, your risk processes would break down. As it turned out they didn’t, and the predictive power was reasonably high during that period.”
In other words, there was no apparent correlation between the overall level of market risk and risk predictive power.
One indicator of a possible ‘bad prediction regime’ is lower average correlations between securities, whether grouped by industry, sector or region, says Winston. “As correlations go down, things behave more and more differently. Systematic factors begin to work less and idiosyncratic factors begin to work more. That generally leads to poorer prediction regimes.”
Another indicator of a breakdown in risk predictive power, he suggests, is intra-market or cross-sectional volatility; that is, how much stocks move relative to other stocks or relative to the market at any point in time.
A MSIM study of the cross sectional volatility of stocks, rather than their volatility through time, showed that returns were dispersing. “To quote Yeats, ‘things fall apart, the centre cannot hold’. And that’s a bad sign for the ability to predict risk.”
Cross-sectional volatility itself has also decreased. The significance of this is that tracking error - the conventional measurement of investment risk – has become compressed. As a result, it appears that investment managers are taking less risk than their risk budgets allow.
Augustin Sevilla, chief investment officer at Axa Rosenberg in London has been observing the fall in both types of market volatility – volatility over time and cross-sectional volatility for more than a year. “During this period both measures have come down, but the one that’s been most remarkable is the cross-sectional view of volatilities.”
The probable explanation for this is the growing risk aversion of investors, he suggests. “Market participants are expecting the markets to continue to trend sideways, and hedging their positions accordingly by selling ‘at the money’ volatility to buy ‘out of the money’ protection.”
Lower cross-sectional volatility may also reflect the fact that, over the past 20 years, the source of cross-sectional volatility has changed, Sevilla says.
“Until the mid 1990s the more volatile stocks tended to be the small cap stocks. Because they are young, unproven companies, it is hard for investors to agree how to price them. So there’s a lot of volatility in their prices compared to larger companies, which are carefully followed by the markets.
In the late 1990s two things happened to change this situation, he says. “Large companies suddenly became very volatile with the TMT bubble, and companies went from being small cap companies to large cap companies almost overnight.
“So the area of uncertainty shifted from small to large cap. And because large cap companies comprise a large part of the market, that gave the appearance that the overall market was much more volatile.
“What we have seen happening recently is that the market is reverting to the less volatile levels of the early 1990s.”
The problem with this situation is that risk measurement tools such as tracking error are providing readings that suggest that managers of active equities are not using their full risk budgets.
This is unavoidable, Sevilla says. “Risk measurement tools were developed during the 1980s and 1990s, and the data they contain will average out these different periods of time. So they will tend to give the appearance that tracking error and volatility is much lower than it has been for a very long time. It would follow from this that managers aren’t taking enough risk.”
Some clients and consultants have noted the apparent change in investment strategy by hitherto active managers. “We have had inquiries from our clients and consultants asking why our tracking error has come down, whether it is something we’re concerned about and whether we are planning to do anything about it. Some have asked us to put together a more risky version of our strategy.”
Yet increasing the riskiness of the investment strategy to achieve the agreed tracking error is not the way forward, Sevilla says: “Our view is that what we’re measuring is historical cross-sectional volatility. It doesn’t necessarily follow that because volatility has been low it will continue to be low.
“It would be difficult to argue that we should somehow increase the riskiness of our portfolio, because the variable that we’re trying to affect is the volatility of the market – a variable that we cannot possibly control.”
The decision whether to increase risk levels will depend not only what information a risk model is providing, but how the risk manager interprets this information, “A risk model is just an approximation of reality and the output of any such model has to be interpreted,” says Neil Brown, global head of risk management at Credit Suisse Asset Management.
“In the case of lower risk levels, then we would be asking managers whether anything has changed in the way the money is being managed, or whether the model is actually producing different sets of numbers.
“If the model is something that is trusted explicitly, then the message coming across will be that risk levels have lowered, quite substantially in some cases. If this is due to data change rather than portfolio management change then we will have to decide what to do about it.
“Do we do nothing and say we will continue to manage the money in the same way and the market will come back at some point? Do we have a dialogue with the client to make sure that they’re happy with that stance? Or do we actually increase some of the risks in the portfolio in order to use up the risk budget or pick up the performance potential?”
The decision asset managers come to will depend on their view of the future volatility levels, he says. “If the expectation is that market volatility levels will remain low, then something will have to be done at some stage to try to increase the risk taken and the performance earned.
“On the other hand, increasing risk levels to current market conditions would mean that if market conditions change you could rapidly find yourself above a risk budget.”
Market uncertainty means that there must be an active dialogue about risk levels between the risk management function, the portfolio managers and the clients, he says, particularly if specialist mandates are involved.
“It’s a consequence of moving away from balanced to specialist mandates that we can only see the riskiness of the part of the portfolio we are managing, and it can sometimes be difficult for us to know exactly how to respond.
“If every manager the client is using is experiencing exactly the same events in the market and all of them decide not to do anything about it, there may be a significant drop in the perceived riskiness or aggression within the overall client portfolio.”
Other factors may make it more difficult for managers to identify the risk in their portfolios. Risk management tools operate least effectively in trendless or sideways markets, Brown points out. “Risk tools are most useful when strong market moves occur. In some respects that is exactly what you want. If the market is moving dramatically then you really want to know what the risk in your portfolio is. If it’s moving steadily and slowly then perhaps the need for accuracy and urgency is diminished.
“At the moment it’s slow moving and stock specific, and risk tools don’t do that terribly well. If we get back to thematic or trending, then risk tools are likely to do it better.”
There are other challenges to risk tools in the current market, notably hedge funds. As pension funds and other institutional investors show increasing interest in hedge fund investment, there has been a growing demand for risk transparency
Olivier Le Marois, the chief executive officer of Riskdata, a risk management solutions provider, says “Risk transparency is a critical issue if you want to attract institutional money. Investors are typically locked in to a hedge fund for one month, and they want to know what the risk is going to be at the end of that month. The importance of risk transparency is being able to predict hedge funds exposure to various market events.”
The usual way of looking at this is positional risk. Positional risk provides a snapshot of a portfolio at a particular point in time. This has its uses in portfolios with traditional asset classes. Yet it can conceal rather than reveal the risk in a hedge fund portfolio, says Le Marois.
One reason is the complexity of many hedge fund investments. “Certain hedge funds are dealing with such complex assets, mortgage backed securities for example, that positional transparency brings nothing to the party. In fact it can be completely opaque.”
Another reason is that hedge fund strategies are dynamic rather than static. This means that risk levels can rise and fall. Hedge funds themselves can use the dynamics of their portfolio to reduce or increase risk in a way that would not be captured in a snapshot, Le Marois points out
On one hand they can mitigate the risk with simple stop-loss or stop gain rules. On the other hand they can reinforce the risk, for example by shorting an option without having any options in their portfolio.
“Hedge funds are dealing with very dynamic strategies, so you must incorporate the dynamic of the portfolio in your risk management model,” says Le Marois. “If you just have a photograph of merger arbitrage strategy portfolio you are missing something important if there is a crisis in the equity markets. This is because merger arbitrage strategists, who are normally neutral against the equity market, have to move, since they are supposed to hedge their exposure to the equity market when they are doing an arbitrage within a merger.
“They will see their returns going down simply because, when the equity market is falling, the number of merger opportunities is decreasing. That type of arithmetical pattern is very important for an investor.”
Le Marois says that institutional investors need a risk management tool that enables them to understand the fundamental drivers of the risk in a hedge fund, including the dynamic of the portfolio.
Riskdata’s approach has been to build a proxy, considering the hedge funds as a distinct asset class, and capturing their dynamic and statistical details. It is applying to hedge funds the same factor analysis techniques that will be familiar to equity risk analysts, Le Marois says:
“We are saying the hedge fund behaves in this way against these factors. Based on that we have a proxy, which is a model of the hedge funds. We can then apply this model, which is based on the past, to predict what is going to happen in one month.”
The next step is to compare the predicted return with the real return. “If the gap between the prediction and what really happened is higher than the risk itself it simply means that if you used this system to hedge your risk you are going to increase your risk. If you have return that is 50% lower it means you are capable of reducing your risk by 50%.”
Le Marois says that pension funds investing in hedge funds through funds of funds need this kind of risk transparency: “It is important for institutional investors to be able to predict risk, because at the end of the day the concern of pension funds is asset liability management. They want to be sure the way their liabilities behave is going to be matched by their assets.”
Yet among pension funds there may not be a consensus about how risk should be managed in current markets. There is evidence in the UK, for example that pension fund trustees are more risk averse than the pension fund sponsors.
Kevin Carter, senior investment consultant at Watson Wyatt, says that there has been a polarisation of attitudes to risk. “Usually the sponsoring company is more inclined to take more risk in the pension plan than the trustees in the plan feel is appropriate.
“The sponsoring company feels that the assets performance is the best way to address the deficit rather than relying on more contributions from the sponsor.”
Trustees are warier of market risk, however. “They are highly sensitised to the fact that deficits exist and what might happen if those deficits increase. If you increase return-seeking assets like equities you may, if you’re fortunate, get the result the sponsoring company wants. But you may also get the opposite position where equities fall rapidly and suddenly the deficit grows.
“And trustees are disinclined to have that outcome occurring and so are rather saying to sponsors we’ll take a controlled amount of risk in the pension plan and if that means you have to pay up more contributions in due course so be it.”
The new international accounting rules, in particular IAS 19, has reinforced this polarisation of perspective, Carter says.
“The accounting rules have created a dichotomy between sponsors’ time horizons and those of trustees, which is unfortunate. The accounting rules that sponsors have to live under are increasingly causing them to take a shorter-term view of risk in the pension plan. Whereas the trustees, with obligations that stretch out decades into the future, have a longer-term perspective of risk taking.”
Robin Saunders, head of risk management for Europe, Middle-east Africa and Asia at JP Morgan’s Treasury & Securities Services, suggests that attitudes to risk will depend to some extent on who is bearing it.
“On the sponsors’ side there is far greater focus on short-term volatility because of the accounting standards and the impact on the annual profitability of the company.”
The greatest risk the sponsoring company faces is the stock market’s perception of the company’s short-term earnings volatility, driven by volatility in its pension fund’s deficits, he says. “That is perceived risk, because, whether we wish it or not, there is no doubt that short-term volatility is not something that is attractive to the market.
“So the perceived greatest risk is short-term earnings volatility. That may not be its greatest risk but that is the perception, and it has to be managed.”
Members of defined contribution plans, on the other hand, may take a different view of risk, he says. “The individual member may be able to take a slightly longer term view on his or her pension plan. The evidence is that they’re not taking that view, but are taking a short-term view of investment return.”
The greatest risk a pension fund faces is its inability to hedge longevity risk in its pension plan, Saunders believes. For pension funds, the perils of longevity risk entirely eclipse those of market risk. “Market risk measurement tools are useful and interesting, but in terms of magnitude market risk is materially outweighed by longevity risk, not least as it cannot be measured or controlled in the same way.
“Market risk does not therefore represent the most significant risk that pension funds run in terms of their overall liability profile,”
This is a bleak message for pension funds. For despite all that risk management can achieve in the management of market risk it can do nothing to manage longevity risk.

Topics