The answers to these questions are contained in the Fundamental Law of Active Management. It tells us that the three drivers of consistent outperformance are having accurate forecasts (information coefficient), having forecasts on as many different assets as possible (breadth) and taking advantage of these forecasts in as clean or direct a manner as possible (transfer coefficient).
In this article, we argue that it is this third driver of performance that separates most asset allocation overlays from traditional investment strategies, thereby leading to their superior performance. Therefore plan sponsors aiming to improve the performance of their portfolios might be better off relaxing the constraints placed on their asset allocation mandates rather than relying entirely on finding a “skilled” manager who may be operating under over-restrictive constraints.
So, the Fundamental Law of Active Management says that while manager skill is essential to delivering good performance, the more often he can apply his skill, or the more efficiently he can exploit his insights, the more consistent the expected outperformance (or information ratio) is. Stated differently, even if a manager’s skill is high, if the efficiency in exploiting his insights is low, then one should not expect a high level of outperformance. You might have great forecasts, but if you can’t implement them then you can’t expect strong investment performance.
Portfolio efficiency
Investment managers sometimes ignore the importance of portfolio efficiency when they design their strategies. The relationship between forecasts and the invested portfolio is as important a driver of the information ratio as the relationship between forecasts and realised returns. In many cases, it is far easier to improve efficiency than skill.
Transfer coefficient is the proportion of the overall insight theoretically available that is captured in a live portfolio once practical considerations such as transaction costs and constraints are taken into account. To illustrate the importance of the transfer coefficient, we simulated the performance of a typical global tactical asset allocation (GTAA) mandate. It was a constrained strategy that invested in 30 global stock indexes, bond indexes and currencies.
The constraints were to be long-only and unhedged, and have a maximum active position of 5 per cent in any asset except US equities and US bonds, which were at 20 per cent. The maximum US equity versus international equity active position was plus or minus 10 per cent. The benchmark was split between in stocks, international stocks and bonds, and it was run at an active risk level of 2 per cent. Chart 1 shows the cumulative active performance from a backtest of this strategy. Its information ratio is –0.4.
Simulated past performance numbers are shown for four years and three months. They should be treated as examples and are not guaranteed. They are not maximum or minimum amounts.
The natural first reaction is that the forecasts on which this backtest is based are poor - this must be an unskilled manager. However, dropping the long-only, unhedged, and maximum position constraints reverses the negative performance and gives an information ratio of +1.3:
Simulated past performance numbers are shown for four years and three months. They should be treated as examples and are not guaranteed. They are not maximum or minimum amounts.
So a low skill level did not cause the poor performance in the traditional GTAA strategy. Both of these simulations used identical forecasts, meaning that they had the identical skill levels. And both invested in the same universe of assets, meaning they had the same breadth.
The only difference was the constraints, which decrease the portfolio efficiency, or transfer coefficient in the GTAA simulation. The lower transfer coefficient drove the realised information ratio from 1.3 down to -0.4.
Figure 3 overleaf shows the transfer coefficient for each of the strategies. The average transfer coefficient for the constrained GTAA strategy is 0.30. This means that only about 30 per cent of the insights are being reflected in the portfolio, with the remaining 70 per cent being eaten up by the constraints.
In contrast, the average transfer coefficient for the unconstrained strategy is 0.95, which means that a full 95 per cent of the insights make it into the portfolio. The fundamental law of active management tells us that we should therefore expect the unconstrained strategy to outperform the constrained strategy by a factor of about three.
Further research reveals that two sets of restrictions drive most of the reduction in the transfer coefficient. The long-only constraint explains about two-thirds of the reduction, while the “unhedged” restriction explains most of the rest.
Let’s use the May 2002 portfolios to illustrate the impact of the long-only constraint. During that month, the transfer coefficient was 0.26. In chart 4 (see download file), the blue bars are the (risk-adjusted) forecasts that month, while the gold bars are the portfolio holdings. For simplicity, only the equity positions are shown.
Two impacts of the long-only constraint are apparent, both of which drive down the transfer coefficient. First, the inability to express negative views, for example, in Spain, Sweden and UK. In these markets, we had a very negative view, but the long-only constraint prevents us from expressing these views.
Second, the inability to express positive views. We liked the US market, but were unable to express that view. The inability to express negative views in the smaller markets causes the larger markets (like the US) to become the only source of funding for long positions elsewhere in the portfolio.
The second driver of the low transfer coefficient is the inability to treat currencies as a separate asset class. To illustrate, in December 2002 we liked the euro, but disliked European equities. In the unconstrained strategy, we were long on the euro by 12 per cent and short on European equities by 4.4 per cent.
We were right on both counts, as the euro appreciated by 5.6 per cent and European equities fell by 10 per cent. However, in the traditional (constrained) GTAA mandate, we cannot take separate currency views, so we must get our currency exposure through the equity and bond markets. Our fondness for the euro exceeded our dislike for equities, so we were overweight unhedged European equities by 7.6 per cent. But unhedged equities depreciated by 4.5 per cent, so the position lost money even though we were right on both pieces.
These are two examples that demonstrate how restrictions imposed on an investment strategy can reduce the transfer coefficient and therefore hurt investment performance. It is the imposition of these kinds of constraints (especially the no-shorting constraint) that causes traditional asset allocation strategies to underperform asset allocation overlays, which usually permit short positions.
Andrew Dales is director of currency research at Barclays Global Investors
