There may be a shift to relaxing constraints on investment managers, through rising use of absolute returns funds and global tactical asset allocation, but for the bulk of assets under management constraints remain. A recent paper by Charles Jacklin, president of Mellon Capital, shows how constraints on managers lead to inefficient investment outcomes.

Jacklin says constraints are pervasive – varying from limiting a universe of stocks, not allowing short-selling to prohibitions on currency hedging. He argues that not only do such restrictions inhibit alpha, they can also lead to style biases, which can lead to higher correlation in alphas across managers and a subsequent reduction in the diversification benefit of hiring multiple managers. Long-only equities managers still occupy the vast middle ground of institutional investment portfolios. But long-only managers face diminishing returns for any increases in active risk (active portfolio weights being a portfolio’s actual weighting to stocks and the benchmark weighting). Jacklin says that, at very low levels of active risk – say, a constraint that a manager should not deviate from the index by more than 0.25 per cent – a long-only manager will still be able to sell off unattractive stocks and buy more of what it believes are attractive stocks. However, if the level of active risk is increased 20-fold, to a more common constraint or target tracking error of 5 per cent, the long-only manager hits the no-short-selling constraint. It must substitute less attractive bets if the same level of active risk is to be maintained. Jacklin argues that, given the same level of skill, an enhanced index manager should always have a higher expected information ratio than a long-only active manager who takes much higher levels of risk. “The unintended consequences of long-only mandates do not stop with a loss of portfolio efficiency,” Jacklin says. “Unless carefully controlled, long-only mandates can also lead to style biases.” As a long-only manager looks to scale up its active bets, the only stocks that can be significantly under-weighted are the stocks with significant benchmark weights. “It is quite natural for active managers to develop a bias in favour of small capitalisation stocks … Some quantitative managers control for capitalisation and manage to avoid such bias. Nonetheless, even quantitative managers have to substitute away from their ideal portfolio …” An examination of the performance history of 434 US equities managers for the 20 years to 2004 showed that, on average 16 per cent of long-only manager total returns is exposed to a small-cap style. Limiting a manager to stocks within an index (or a smaller component of the broad market index) will reduce the style bias, but it is not without cost. “If a manager has skill, then any limitation of the manager’s investment universe is going to reduce the manager’s ability to add alpha,” Jacklin says. An examination of long/short managers, however, showed that on average only 1.7 per cent of returns is through exposure to small caps (the universe for this was only 27 managers, between 1992 and 2004). Long-only manager returns are highly correlated because of their market risk – beta – in every portfolio. If you separate their alpha from beta, there is no reason for the alphas to be correlated. But, Jacklin says, even if the investment processes of each manager are independent, their investment constraints are not. “To the extent investment restrictions induce biases in individual managers, the managers’ alpha generation will no longer be independent… What follows is that if alphas are dependent then diversification across managers is reduced.” Jacklin calculated the correlations between the 434 long-only managers mentioned above (there were 93,961 ‘pair-wise’ correlations) and found an average correlation of 0.241. The same calculations on the 27 long-short managers showed an average correlation of only 0.112. Jacklin says that if you have three managers with equal weightings and equal risk, the active risk of the total portfolio would be about 9 per cent lower if their alphas had a correlation of 0.112 against 0.241. He also explores the effect of constraints on asset allocation, through the use of global tactical asset allocation mandates, with similar results. “It is incumbent on both the manager and client to understand and appreciate the impact of investment guidelines and constraints,” Jacklin concludes. “Experience suggests that both managers and clients are often unaware of the impact that constraints have on performance. Ideally, constraints act to ensure adequate risk control. Nevertheless constraints should be viewed both on the basis of their ability to control risk as well as their cost in terms of reducing portfolio efficiency.”

Leave a comment