When highlighting the merits of the profession of active manager, stock picking is often presented as the leading factor in the creation of value. After the downturn in the markets in 2000, benchmarked management was vilified in favour of investment management based on ‘convictions’ where small cap stocks, which had been left aside for a long time, were given a more significant weighting in portfolios. The sharp rise in these stocks over the last three years has been a strong influence on investment management, leading to the role of stock picking being emphasised. Some funds1 define themselves as ‘integral stock pickers’ without referring to any reference index whatsoever.
In fact, stock picking is currently a sales and advertising selling point that is overwhelmingly favoured by asset management firms. According to a study carried out by EDHEC on a sample of 100 sales documents issued by 45 French asset management companies and devoted to equity management, 82% gave first place in the number of quotations in the sales proposition, to the manager’s capacity to carry out stock selection, as shown by his capacity to beat his benchmark, outperform the market or obtain absolute performance, compared to only 6% for tactical allocation (market timing). The balance of the quotations represented 12% of the cases.
Among the 82% that highlighted stock picking in investment management, more than a quarter (28%) of the sales claims either did not mention or formally excluded market timing as a source of performance.
Faced with this exceedingly popular marketing approach, how can one genuinely evaluate the impact of stock picking and timing on fund returns? How can one stock picker be compared to another?
To answer these questions, we extracted the best2 funds invested in French equities3 from the EuroPerformance database – Style Analytics, which represents a sample of 50 funds that have a minimum of three years of historical returns.
The goal of active management is to perform better than the market or better than a benchmark that is chosen as a reference. Funds that are managed according to traditional methods are part of the framework of active management. These funds represent a considerable share of the funds available on the market and they are often quite specialised.
They turn to financial analysis, which involves choosing each stock in an individual manner – we then speak of stock picking – by basing their choice on the balance sheets and financial characteristics of the companies. In addition to, or as a result of, those stock picks, active management funds can see their performance affected by a variation in their exposures to market, sector or style risks. They are then faced with a consequence that is either desired (the result of tactical allocation or market timing) or incurred, with a badly controlled choice of stocks leading to beta biases in comparison with the benchmark.
A manager’s capacity to add value by practicing timing between asset classes or management styles can be measured through a generalisation of the Treynor and Mazuy (1966) approach. By applying this approach to the sample of French funds, we have been able to evaluate the importance of timing and the timing capacity of French managers of French equity funds. By comparison, we have also been able to highlight the funds that did not practice timing and for which the alpha was the fruit of a pure stock picking activity.
We observe that only 22 of the best 50 French equity funds, or 44%, are ‘pure stock picking’ funds. Most manage their timing between asset classes and styles.
Moreover, table 1 shows that more often than not managing timing destroys value: managers who carry out market timing or style timing do not do it efficiently. However, for the selected funds with an alpha considerably greater than zero, stock picking more than compensates for the negative effects of timing on average over three years.
The practice of timing mostly involves arbitrage between market caps (small caps vs. large caps). Apart from very rare exceptions, this does not add value.
Arbitrage between the Growth and Value styles is also practiced on numerous funds, but to no greater avail.
So, how does one compare the stock pickers? Decomposing performance with the help of multifactor models uses a principle which involves isolating the performance that comes from the choice of benchmark from the performance that comes from stock picking. The benchmark in that case is made up of the linear combination of the returns’ explanatory factors (in the present case, the style and the size indices).
The residual return term not explained by the factors corresponds as a result to the asset selection term (alpha). The style ratings developed by EuroPerformance and EDHEC are constructed according to this approach.
As such, alpha is a risk-adjusted performance measure that enables a relevant classification of actively managed funds to be established. Indeed, it is the most objective criterion for comparing stock pickers among themselves.
It is only possible to measure alpha if we know a fund’s ‘true’ benchmark. Often, when calculating the benchmark, the investor, or indeed the rating agency, relies on the assertions of the managers who mention a reference index in the prospectus. However, there is no guarantee that this reference index really reflects the risk taken by the manager.
Besides, the regulator and the rating agency focus on the respect of categories attributed to the funds. The definition of these is fairly broad and leaves room for extensive possibilities in the choice of styles, sectors, classes and therefore indices. An EDHEC study has shown for example that a very large proportion of French funds that were considered to be Euro-zone, North American or global large cap in the Morningstar categories were in fact mostly invested in small-cap stocks. This incorrect classification generally leads to the wrong index being chosen to evaluate management outperformance.
Furthermore, numerous managers refuse to communicate a benchmark or use their opportunistic management approach, for which the risks cannot be summarised by a benchmark or market index, as a justification. We note that of the top 50 funds, 18 do not indicate any reference index and limit themselves to a presentation of the management objectives.
A more rigorous approach involves constructing benchmarks that take account of the risks that were really taken by the manager for each of the funds without turning to the assertions of the asset management companies and the biases that they could thereby introduce. It then involves selecting the style indices that represent the strategic allocation and thus the risks taken over the study period. This analysis is carried out in the form of a multi-linear constrained regression such as that developed by the Nobel Prize winner William Sharpe, a method also known as return based style analysis.
Using this selection of style indices which is specific to each fund and the weights that are relative to them (customised benchmark), it is then possible to calculate the excess performance of the fund by taking into account the risks to which the fund is really exposed.
We can evaluate the quality of the two methods by comparing the explanatory power of the fund returns. The ‘true’ benchmark allows the fluctuations in the fund returns over the study period to be reproduced more faithfully.
We have, therefore, compared the coefficients of determination (R²) of the regression of the fund returns onto those of the reference index with the r² of the regression resulting from the style analysis (return based style analysis).
In 54% of the cases where a reference index is mentioned, the benchmark that is reconstituted using the style analysis is better than the reference index. We can also observe a relative asymmetry that is favourable to the regression with the benchmark because the negative differentials are less significant than the positive differentials: when the ‘self-proclaimed’ index is better, it is not better by much.
Furthermore, this approach allows one to attribute a benchmark to funds that have not declared one and to understand that their performance is not only the fruit of active management decisions but also of an average exposure to risks over the whole analysis period.
Another way to measure the approximation that the reference index represents is to calculate the distance between the reference index and the genuine benchmark. In order to be able to compare them we carry out a style analysis on the reference indices. As a second step, we calculate the distance on the three axes: value, growth and small caps. To make the table easier to read, the measurement of the differentials is expressed as a percentage.
For the funds that declare a reference index, the divergence ranges from 13% (MID France from Atlas Gestion) to more than 70% (Pluvalca France from La Financière Arbevel). The average is 35%. These very high figures show that, much more than the stock picking claimed, it is the exposure to styles that are different from those of the index put forward by the asset management firms that explains the difference in performance with the index.
In view of the poor performance of style timing and market timing, we can say that the essential part of the positive performance in comparison with the fund’s benchmark comes from stock picking. Alpha measurement therefore allows French managers to be ranked on their ability to deliver outperformance through stock picking.
The three-year rankings, to 23 December 2005, reveal the dominance of French subsidiaries of foreign groups (UBS, HSBC, etc) and independent firms (Financière Arbevel, Financière de l’Echiquier, etc).
For the top 50, the alphas range from 12.2% for UBS Mid Cap to 0.1% for SogeActions Opportunités. The average is 4.2%.
While pure stock picking only concerns 44% of the best French ‘equity’ funds in active management, the practice of timing between asset classes and styles is nearly always unsuccessful.
Measuring the stock picking capabilities of the funds is made difficult by the poor communication from the funds:
q More than one-third of the funds do not declare a reference index;
q And for the population that does declare an index, the index only imperfectly reflects the risks taken in the portfolio in 54% of the cases.
In total, the information provided to subscribers by more than 70% of funds is incomplete or imprecise.
In this context, the highlighting of stock picking and in general of the value added by management (alpha) is more a communications and marketing operation than evidence of a desire to respond to professional obligations and provide subscribers with information.
Noël Amenc is director of the EDHEC Risk and Asset Management Research Centre and Frédéric Picard is managing director of EuroPerformance
1EDHEC European Asset Management Practices survey, initial results, ongoing survey, 2006
2The best funds in this study are the funds that obtained a positive alpha over the study period of December 27, 2002 – December 23, 2005, and as such hold 4 or 5* within the framework of the EuroPerformance–EDHEC Style Ratings
3These funds correspond to the French Equity Mutual Fund category of the French regulatory authority (the AMF)