The Revolution in Risk

Spurred by powerful computing technology, risk budgeting challenges many basic assumptions about money management. Beyond theorizing, plan sponsors are actually rolling the dice.

In April the Ontario Teachers Pension Plan Board invested $405 million in Canadian Occidental Petroleum. That gave the fund a nearly 20 percent stake in the oil and chemical products maker and a position more than six times larger than the company’s normal market weighting. Such a huge bet on a single stock, regardless of its fundamentals, would make most pension fund managers queasy. Not Leo de Bever, who oversees the pension plan’s investments.

“We regard the CanOxy investment as risk-reducing,” says de Bever. Of course, he likes CanOxy’s financial prospects and is confident that the Ontario fund knows the company well. More important, the investment allows Ontario to protect itself from the vagaries of the Canadian stock market, where it already has a large exposure via the passive portion of its roughly $12 billion Canadian equities portfolio. By holding a large position in a stock that makes up less than half a percentage point of the Toronto Stock Exchange’s total market capitalization, de Bever’s overall portfolio is less correlated to the equity market’s performance, thus cutting his risk. And because de Bever’s main concern is creating an excess return over his liabilities without violating his risk parameters, the CanOxy investment is prudent.

Guiding the 52-year-old former central bank economist is a practice known as risk budgeting, the latest wrinkle in a growing arsenal of portfolio management tools. Enabled by more powerful risk measurement technology that ties together once disparate strands of data, risk budgeting marks a new way of allocating the risk that pension funds are willing to take as they try to achieve the returns they need over the long run to meet their liabilities.

For a growing number of investors like de Bever, who believes that funds like Ontario’s shouldn’t be run solely on the basis of their returns against a benchmark, this new way of viewing investments is, in some respects, a return to basics. “What has happened is that the pensions industry has changed the true problem, which is to manage the surplus a fund has above its liabilities, to managing return relative to benchmarks,” de Bever says. “That is not a philosophy we buy at all.”

Risk budgeting is the process of measuring the risk that pension funds are actually taking, assessing their appetite for risk, quantifying how much and in what areas they are willing to take it and then deciding how to allocate it, either internally or with outside managers. This decision-making process took on new urgency following the Orange County, California, bankruptcy in 1994 and the Long-Term Capital Management debacle of 1998.

It sounds simple, but assessing risk across a portfolio is in fact very complicated. Measuring the risk in a multibilliondollar portfolio made up of thousands of individual investments in a number of different markets is a huge task that couldn’t be cackled without the exponential gains in computer technology of the past few years. Tying together a series of theoretical advances like Nobel Prize winner William Sharpe’s information ratio, which attempts to quantify a manager’s risk-adjusted returns against a benchmark, and then making them all easily accessible on a manager’s desktop requires additional computing firepower.

This process of allocating risk is what has come to be known as budgeting. In short, funds decide how much risk they want to allow a manager. Generally, this is expressed relative to the most appropriate benchmark. In the case of a large-cap equity manager, say, the target might be a 200-basis-point differential—or tracking error—from the Standard & Poor’s 500 index.

The implications of this process are potentially revolutionary. Why? Because outperformance—bearing benchmarks and peers—is what managers live and die for. Pension funds don’t want to discourage high returns, but they do want to achieve them within a tighter set of risk parameters and controls. Using risk budgeting, a fund would review its managers based on the consistency of their returns relative to their benchmarks, rather than on their ability to bear those benchmarks. If the fund decides it wants a certain manager to take more risk, it simply allocates a bigger variation in tracking error.

As de Bever describes Ontario’s experience, using a risk budgeting approach gradually erodes the influence of benchmarks or return on assets measures on a portfolio manager’s or plan sponsor’s thought processes. “What happens when you adopt a risk budgeting methodology is that because the [risk measurement] numbers are in your face every day, people start to think in terms of producing a return on risk rather than a return on assets,” says de Bever. With the benchmark removed, an active management strategy becomes an absolute return strategy within a predefined risk framework.

Risk budgeting is winning over a number of influential proponents in the academic, pension fund consulting and asset management worlds. Bob Litterman, co-author with the late Fischer Black of an influential 1991 paper on quantitative approach to asset allocation, describes himself as a “big fan” of risk budgeting. Says Litterman, now head of quantitative resources at Goldman Sachs Asset Management: “Risk is a scarce resource that needs to be allocated optimally. That’s where risk budgeting comes into its own. It says that the way to approach the problem is to maximize the return at the margin for each risk allocation within the overall budget.”

Leading plan sponsors in Europe and North America have started to experiment with the concept, albeit in small ways. In addition to the $45 billion Ontario pension fund, two Dutch plans—PGGM, the $50 billion fund for the country’s health workers, and ABP Investments, Europe’s largest pension fund, with $150 billion in assets—have begun to use it. The $108 billion Florida State Board of Administration is applying risk budgeting to several asset classes and hopes to have a planwide risk budget in place by the end of the year. Other large U.S. public funds, including the $120 billion New York State Common Retirement Fund and the $175 billion California Public Employees’ Retirement System, have said they are evaluating the merits of risk budgeting.

Many plan sponsors focus on the most basic use, known as the implementation level. Here the sponsor sets its budget after the fund has already made its strategic asset allocation decisions. This form of risk budgeting does not ask too much of plan sponsors. It takes as a given the current portfolio mix. If, say, the fund has decided to invest 40 percent in equities and 60 percent in bonds, a risk budget might then determine how far the fund is permitted to stray from those figures. At ABP, for example, the risk budget has permitted a 2 percent tracking error to the overall strategic allocation benchmark (currently 53 percent fixed income, 40 percent equity, 7 percent real estate) over a one-year time horizon. ABP will maintain those allocations within a couple of percentage points and rebalance its mix to keep them relatively constant. Beyond that, ABP has also set a tracking error limit for the regional allocation within each of the asset classes using a fund-of-funds structure.

PGGM, at a similar stage right now, hopes to expand its use of risk budgeting. “Next year we will begin to look at the next level down, applying risk budgets to the individual asset classes,” says risk manager Wouter Peters. “We are taking a gradual approach that is informed by the data we generate.” By applying a risk budget to asset classes, PGGM will set a target tracking error for an investment class, such as global fixed income, and measure that against a standard investment bogey, such as the J.P. Morgan government bond index.

The Florida Stare Board of Administration is building from the bottom up. It has set risk budgets for two individual asset classes, non-U.S. equities and international bonds. It is in the process of setting one for domestic equities as well. Only then will it begin to put a risk budget in place for the whole plan. “The end destination for us is the overall fund,” says Kevin SigRist, the plan’s assistant chief economist.

The process of establishing a risk budget for both the overall fund and the individual asset classes subtly affects the way pension fund managers view their assets. Instead of aiming to meet a return target within set asset allocation parameters, they look to take a defined amount of risk and maximize the return from the asset allocation process. Having a limit on the total amount of risk, they must decide where they want to spend it.

With two years of risk budgeting under its belt, ABP is preparing to venture into this realm. “What we are now examining is where ABP truly adds value,” says Thijs Coenen, the pension fund’s head of risk management. In international fixed-income markets, for example, managers typically add value through their duration and currency decisions. If Coenen were to discover that ABP is a weak interest rate forecaster but a strong currency manager, then it would make sense to create a bigger risk budget, allowing more tracking error, for currencies than for duration.

Ontario Teachers takes risk budgeting to its logical next step. In addition to using risk budgeting to monitor portfolio performance, Ontario uses it to decide what its strategic asset allocation mix should be. “Active manager selection to emphasize certain types of risk is significant, but it’s not the most important part,” says de Bever. “Finding the best match between assets and liabilities and the mix that gives the optimum excess returns above liabilities is the key.”

One might assume that a fund with a strategic asset allocation mix of 70 percent fixed income and 30 percent equities is less risky than one in which those positions are reversed. But that is not necessarily the case. If, for instance, the plan sponsor is a start-up dot-com in which the average employee will not be a beneficiary for 30 years, investing 70 percent in fixed income could be perilous. The fund might not grow enough to meet its future liabilities.

Ontario is also exploring other dimensions and deeper levels of complexity in risk budgeting. Such issues as contribution policy—at what level of overall solvency the sponsor will take money from the fund or make higher contributions—are considered. Plus, the fund begins to bump up against some of the limits of actuarial forecasting, which attempts to project an uncertain future based on historical experience. This can lead to frequent revisions of expected liabilities as both the markets’ and the company’s fortunes change.

Gus Boender, a professor of asset-liability management at Amsterdam’s Free University and one of the founders of Ortec, a Rotterdam-based econometric consulting firm that specializes in asset-liability modeling, believes that risk budgeting can ultimately be made compatible with actuarial science, but that the convergence will first require additional technological gains and analytical tools. As he notes, bringing a plan’s liabilities into the mix makes risk budgeting that much more complicated.

Ontario’s de Bever argues that the effort to unravel these complexities will be time well spent. “If you bring in an external consultant once a year who looks at the asset allocation and an external actuary that looks at the liabilities, of course it is difficult,” he says. That’s because the fund isn’t tying the two together. To promote a unified view of assets and liabilities, Ontario Teachers centralizes its asset allocation and liability assessment functions.

Even for funds not venturing nearly as far as Ontario Teachers, risk budgeting may change asset management arrangements by reminding investors of the value of adding uncorrelated investments. In a risk budgeting world, products are assigned “hurdle” rates, which quantify their correlation to the overall portfolio. The higher the hurdle rate, the higher the product’s risk relative to the overall portfolio.

For example, most plan sponsors would regard a market-neutral hedge fund as very risky. But if the hedge fund has little or no correlation to equity risk, by far the largest risk in most funds’ portfolios, then it could have a very low hurdle rate, making the hedge fund much more attractive to plan sponsors.

“That is an interesting and not necessarily obvious process,” says GSAM’s Litterman. Low hurdle rates, he notes, can lead sponsors to make supposedly risky investments because their risk isn’t directly related to equity risk. As he points out: “Commodities, which are generally negatively correlated to equities, might have a lower hurdle rate than ‘risk-free’ T-bills. Another example of a product that would have a very low hurdle rate is a currency fund.”

For sponsors the idea that pork bellies might be less risky than short-term Treasuries or that a foreign currency play might be more conservative for an overall portfolio than largecap stocks may take some getting used to. Broadly deployed and taken to its logical conclusion, risk budgeting suggests widespread changes in investment strategy.

De Bever, who notes that he would not have considered an investment like CanOxy without risk budgeting, says: “We used to avoid some investments because we could not calculate target levels of return. But when you set a risk budget and produce a return on that, you do not care what the reference rate of return on assets is. There may not even be any underlying notional assets.”

Risk budgeting also promises to reinvigorate the debate about active versus passive management. Most investors perceive passive management as low risk. Risk budgeting, which is applied only to active management, offers a means to quantify and limit risk. With risk under careful control, some proponents believe that active management should be expanded.

“If I am an active manager, what I am typically doing is overweighting or underweighting certain stocks relative to my benchmark,” explains GSAM’s Litterman. “So long as all my managers aren’t doing the same thing, that active risk is relatively uncorrelated and therefore should have a lower hurdle rate than equities. In a risk budgeting context, if an active manager generates positive returns relative to the index, even if the return per unit of risk is much less than equities’, then that is beneficial [because it is unrelated to the core equity holdings]. Our institutional clients find this conceptually fascinating. In short, I would not rush to do too much indexing.”

Don Ezra, director of strategic advice at Frank Russell Co. and co-author with Kevin Ambachtsheer of Pension Fund Excellence: Creating Value for Shareholders, believes that individual selection of stocks and bonds can still produce incremental gains without undue risk. “We have found [that] by carefully combining specialist active managers, we can eliminate the need for a passive core,” he says. Like de Bever, he views the strategic asset allocation mix as by far the most important decision a sponsor makes in meeting its liabilities. Once that’s done, the fund should take active risk to implement its plan. On this point he is unequivocal: “100 percent active management.”

Given the enormous popularity of indexing, whether these arguments prove persuasive remains to be seen. Indeed, critics say that some risk-based management systems actually promote closer indexing.

THE RISE OF RISK BUDGETING IS JUST ONE manifestation of the new pervasiveness of risk measurement, management and awareness.Risk management owes less to a desire to maximize returns than to a determination to avoid losses. By all accounts, the initial spur was the embarrassing 1994 bankruptcy of wealthy, Republican Orange County after its treasurer ran up paper losses of $1.6 billion in a government investment pool.

Following the ensuing wave of negative press coverage, representatives from 11 large public, corporate, foundation and endowment funds formed the Risk Standards Working Group in 1995. The group’s report, published the next year, included 20 specific recommendations relating to auditing and compliance issues, which amounted to a best practices checklist for pension funds. The impact of the report was widespread, prompting many funds to start to incorporate risk management functions into their daily operations.

“The decision to create a risk management group grew directly from the work of the Risk Standards Working Group, which we were a part of,” says Harris Lirtzman, director of risk management at the $110 billion New York City Retirement Systems. “We really wanted to make a clear statement that we thought risk management was important in the asset management business.”

That message was delivered. Henrik Neuhaus, a senior risk management consultant at SSB Citi Asset Management, reports that the involvement of large pension funds sparked the creation of a risk management function at the firm this past January. “Some of the larger pension funds are adopting the role of activists in pushing the risk management agenda forward,” he says. “They are producing guidelines and we have to be in a position, commercially, to respond to what they want.”

Aside from keeping pace with clients, asset managers have their own well-documented reasons to pay heed. Deutsche Bank was forced to inject £180 million ($272 million) into its Morgan Grenfell subsidiary in 1996 after discovering that the manager of two European equity funds had taken large unauthorized positions in unlisted securities. The matter is still under investigation. Rather than risk massive redemptions, Deutsche stepped in with cash to restore investor confidence.

The legal action by the trustees of Unilever Superannuation Fund against Merrill Lynch Mercury Asset Management predictably focuses even more attention on the huge risks asset management firms run. So far this one case has cost Mercury £1 billion in assets, a bout £15 million in annual fees and an incalculable amount of prestige. Unilever is seeking damages that are estimated to exceed £100 million.

The stormy dispute revolves around the parameters of risk that are appropriate for an asset management firm to take. As Richard Greenhalgh, chairman of the trustees of Unilever Superannuation Fund, said in a statement last October: “The advice that the USF trustees have been given is that the U.K. equity portfolio selected by MAM was far too risky for the agreed investment mandate. It should have been clear to MAM that if the investment judgments they made turned out to be wrong, there would be underperformance far beyond the agreed downside tolerance.”

The MAM-Unilever case also illustrates a new risk for asset managers. “If you represent yourself as a sophisticated provider of a service, you can expect to be held accountable to a standard above and beyond what might be in the letter of a contractual agreement,” says Peter Davies, CEO of risk management software vendor Askari.

Even a small gray area in these contracts is not enough for some funds. Bruce Hawver, senior investment officer at the New York State Common Retirement Fund, says that partly in response to the Unilever case, he has just completed a review of exception reporting. “When a manager does something that is outside their mandate,” he explains, “we wanted to ensure that the contracts were clear about how this is reported and what the process is.”

This pressure from clients to deliver what is expected and to keep surprises to a minimum plays to the strengths of such risk management tools as risk budgeting, where variances are closely tracked and quickly quantified.

Asset managers must provide “truth in labeling” to customers and that concept must infuse their whole culture, according to Erwin Martens, the former head of global market risk for Lehman Brothers, who now leads one of the industry’s largest and most sophisticated risk management efforts at Putnam Investments. “We have spent a lot of time agreeing on a common risk language and methodology across the firm. One central tenet of that is the risk ribbon. We define the risk ribbon as a spectrum of risk and return characteristics,” says Martens. To ensure that they deliver what is promised, he adds, “products will find themselves within relatively tight ranges on this risk ribbon.”

At GSAM, another firm that has made a significant commitment to risk management, Litterman shares Martens’s vision. “If l had to summarize what the purpose of the risk department is, I would say it is quality control of the investment process,” he says. “It is concerned with making sure we deliver a consistent product and that consistency is defined by the tracking error. Quality is defined by the information ratio.”

Robust risk management is fast becoming an economic necessity for asset managers, both for winning business and avoiding potentially expensive problems. Says Nicholas Watts, head of the European investment practice at consulting firm Watson Wyatt Worldwide: “What we want to see is a culture of risk management at an asset management company. We want to see that risk management is part of the investment process and at the same time independent from it.”

Instilling a risk management culture in an asset management firm is not easy—and it’s never cheap. Obviously, it demands substantial support from senior management. At a conference in Paris in May, Jacques Longerstaey, co-head of risk management at GSAM, told the audience that the firm spends 0.5 basis points of total assets, or approximately $10 million annualized, on risk management. That number surprised many of his listeners.

Risk managers must prove that this is money well spent, that it enhances the core business of the firm: delivering good investment performance. Says Putnam’s Martens: “It’s not just about reporting. It’s a better way to manage money.”

Risk budgeting and its kin would not be possible without recent technological gains that are increasingly able to draw together many different forms of portfolio analytics and put them on the desktop of a plan sponsor or portfolio manager. In addition, the Internet has “democratized” the flow of information from all corners of the asset management world, allowing plan sponsors to break down and quantify aspects of their portfolios that once defied easy analysis.

Plan sponsors and money managers are eagerly reshaping and using this information in new ways. A group of ten pension funds and fund managers, including Virginia Retirement System, Wellington Management Co. and SSB Citi Asset Management, banded together in 1998 with Askari to form the GlobalView Partnership Progam. The group spent a year helping to design new risk management software, known as GlobalView, that integrates portfolio valuations, a number of forward-and backward-looking portfolio risk measures and compliance limits. It became available in June.

At New York City Retirement Systems, which will begin using Barta’s Total Risk for Asset Management software in the third quarter, Lirtzman says such systems will alter the whole nature of discussions with his money managers. “At the moment, the conversations we have with our money managers tend to be wide-ranging and general, about the direction of markets and issues like that,” he says. “This technology will enable us to focus discussion on portfolio dynamics.”

These data-rich, technology-driven discussions are likely to become ever more empirical, according to Ronald Layard-Liesching, a partner at Pareto Partners, the world’s largest currency overlay manager. “If a global bond manager tells you that they add value by currency management,” he notes, “you will be able to test it. The eloquence of the smoothest salesman can’t compete against real data.” And once sponsors have all this information at their fingertips, says Lirtzman, “all the players in the business will have their roles redefined. Everyone will have to earn a place at the table by the value-added they bring to the investment policy process.”

Related