This content is from: Home

The cost of constraint

The restrictions that pension funds impose on fund managers to reduce risk can backfire and -- a new study shows -- they take a big bite out of return.

The restrictions that pension funds impose on fund managers to reduce risk can backfire and -- a new study shows -- they take a big bite out of return.

By Andrew Capon
October 2002
Institutional Investor Magazine

Most people could use a little constraint. But how much is too much? When do rules meant to keep us safe become, like a too-tight seat belt, so constricting that they cause us harm?

The question is of immediate interest to money managers, who must contend with restrictions imposed by their pension fund clients -- from not shorting stocks to limiting portfolio turnover. Now a new study concludes that such constraints drag down investment performance far more than anyone suspected. The obvious conclusion: Investment constraints are worth reexamining.

To be published in the November- December Financial Analysts Journal, the study -- "Portfolio Constraints and the Fundamental Law of Active Management" -- seeks to demonstrate that common constraints can subtract several percentage points from returns. The paper has already generated a prerelease buzz in the investment community.

"This is a brilliant piece of work," says Alan Brown, group chief investment officer and vice chairman of State Street Global Advisors, the second-biggest fund manager in the U.S., behind Fidelity Investments. "It might well be the catalyst for change in the money management business."

In the paper three finance Ph.D.s -- Roger Clarke and Harindra de Silva, chairman and president, respectively, of Los Angeles quantitative investment shop Analytic Investors, and Steven Thorley, a Brigham Young University professor -- build on Richard Grinold's fundamental law of active management, a key component of which is the concept of an information coefficient. Now head of advanced active strategies research at Barclays Global Investors, Grinold in 1989 published a mathematical equation for measuring the strength of a fund manager's forecasting "signal" -- how well the manager predicts stock returns.

Clarke and his colleagues have taken Grinold's approach a step further by creating a "transfer coefficient" -- a measure of the extent to which a portfolio manager's forecasting signal gets translated into portfolio weights. The transfer coefficient is, specifically, the correlation between portfolio weights and the manager's best estimates of stocks' future returns. Absent portfolio constraints, the coefficient should be one; adding constraints causes it to decline sharply.

The three researchers calculated transfer coefficients for a series of familiar investment constraints, alone and in combination. For instance, confining portfolio turnover to 50 percent a year results in a coefficient of 0.73. But if turnover is restricted to 50 percent, shorting is forbidden and the portfolio must remain market-capitalization-neutral, the transfer coefficient drops to 0.3.

"People have known intuitively that constraints limit a portfolio manager's ability to add value," points out co-author Clarke, who was one of the pioneers of tactical asset allocation. "The contribution of this paper is to measure that. What we have found is that common constraints can mean that only 30 percent of the potential value of information is transferred into the portfolio."

At that low level, Clarke adds, less than 10 percent of a portfolio's performance can be attributed to forecasting. "That might lead portfolio managers to wonder why they are even bothering to forecast stock returns," he says.

The study's findings are consistent with portfolio managers' experience. Says State Street's Brown, "We have started to measure our own portfolios using the transfer coefficient, and our work shows that constraints do bite and in the magnitude suggested by the paper."

One of the strongest selling points of hedge funds, of course, is that they are free of most portfolio constraints. Long-short stock strategies -- out of bounds for basically all but hedge funds -- achieve high transfer coefficients, in the 0.7­0.8 range. This would seem to justify including hedge funds in mainstream pension fund portfolios. The caveat is that if a hedge fund manager's, or any other asset manager's, forecasting of returns is faulty, having a high transfer coefficient hurts more than it helps.

"Constraints are one reason why a lot of hedge funds are ambivalent about institutional money," notes Maarten Nederlof, global head of the pension strategy group at Deutsche Asset Management in New York. "They are not interested in business that constrains them."

Constraints don't just affect individual portfolio managers; they can have an impact on the whole market. For instance, many pension funds order their bond managers to hold only investment-grade credits. So when triple-B-rated bonds, on the cusp between investment and noninvestment grade, get downgraded to junk, bond managers unload them en masse, causing their prices to plunge. When U.K. telecommunications equipment manufacturer Marconi was downgraded on September 7, 2001, its 2005 bonds plummeted in one day from 94 cents on the dollar to 36 cents.

"The only word for these price movements is 'frightening,'" says Jon Jonsson, director of European credit strategy at Merrill Lynch International in London. But such steep discounts also represent an opportunity: a fixed-income fire sale. "Where a fund manager sees these effects in markets, the smart manager should be asking how to exploit them," counsels Lee Thomas, senior international portfolio manager at Pacific Investment Management Co.

The hitch for managers wanting to hold triple-B bonds for their rich yields but restricted to long-only investing is that the bonds' risk asymmetry is enormous: An upgrade might mean a positive price move of 5 percent; a downgrade could spell huge losses.

Shorting triple-B bonds offers a much better risk-reward trade-off, but that option is open chiefly to hedge funds because of their lack of constraints. Indeed, Thomas's colleague William Gross, manager of Pimco's Total Return Fund, suggested in a recent commentary that bond hedge funds are exploiting the effect of forced institutional sales by deliberately driving down the prices of triple-B bonds. (See "J.P. Morgan and Character -- the Third Edition" at ca/bonds_commentary_index.htm.)

Investment constraints can have other untoward consequences -- such as, paradoxically, adding to a portfolio's risk. "The problem with constraints is that they are binary: You either can't do something or you can," says Pimco's Thomas. "If you tell someone that they can't take risk in one area, then they are sure to take more risk in another. If you constrain the ability to buy sub-investment-grade bonds but give free rein in foreign exchange markets, the inevitable tendency is to take big bets on FX."

The U.S. bond market debacle of 1994 -- the worst year for bonds in two decades -- illustrates how constraints can sometimes backfire. Throughout the winter of 1992 and the whole of 1993, the U.S. federal funds rate had been stuck at barely 3 percent. Desperate for return to retain restless clients but unable to buy high-yield bonds, conventional bond managers avidly snapped up Wall Street's newest concoction: triple-A-rated structured notes, such as inverse and range floaters, linked to permutations on the yield curve. The notes grew ever more esoteric and bizarre.

Then in February of 1994, the Federal Reserve Board hiked interest rates. "Funds invested in structured notes blew sky-high," recalls Bluford Putnam, former chief investment officer for equities at Bankers Trust Co. and now head of New York risk management firm Bayesian Edge Solutions. "These funds complied with their constraints but took on huge risks in these notes, which were little more than toxic waste in a changed rate environment."

Despite the collateral damage wrought by ill-applied portfolio constraints, pension funds are not likely to unfetter their managers altogether. A compelling case can still be made for constraints, apart from the obvious one that they may bar risky, exotic investments and prevent rogue behavior. Pension funds like to put asset managers into style boxes so they can have an overall view of their portfolios. Thus they don't want managers to stray too far from their brief. Further, constraints allow pension funds to measure managers against their style benchmarks and against other managers pursuing the same style. If portfolio managers could do whatever they wanted, asks one pension manager, how could they be objectively measured?

Co-author Clarke says: "The point of the paper isn't to suggest that constraints are unreasonable in all circumstances. What we have aimed to do is show the cost of the constraints."

In a bear market like this, those costs may be steep. "Constraints that limit a manager's transfer coefficient are disproportionately damaging now," contends Deutsche's Nederlof, because "the game for asset managers and pension funds is to capture alpha," or excess return, not to scrupulously avoid tracking error -- and follow the market down.

State Street's Brown has a timely suggestion: "Fund managers and pension funds should spend a lot more time considering whether constraints that are commonly applied are appropriate."