Active Share: a misleading measure

By Charles Younes FE Research and Opinions

Updated on Friday, 18 December, 2015

Active Share has been causing quite a stir in the industry over the past year and while some managers have quite readily employed it as a marketing tool on their factsheets, the increasingly wide use of Active Share as a marker for a manager’s ‘activeness’ has been contentious. As the measure’s popularity continues to grow, FE Research looks at just how useful it is.

FE Research have been surprised recently by the increasing use of ‘Active Share' (AS), a measure of the percentage of stock holdings in a manager's portfolio that differ from the benchmark index, in the wider industry.

Early this year, Neptune Asset Management and Baillie Gifford, among others, took the decision to publish Active Share figures for the first time. However, the measure's usefulness continues to be hotly debated and as such, we take a look at just why it is so attractive as well as to explain why this measure may be hindering investors rather than helping

An introduction to Active Share: definition and popularity.

The concept of Active Share was introduced to the investment industry in the 2000s. Researchers Martijn Cremers and Antti Petajisto released their first paper on Active Share in 2009, where they define and use Active Share in conjunction with Tracking Error, to categorise domestic equity mutual funds by their degree of active management.

Cremers and Petajisto interpret Active Share as the “fraction of the portfolio that is different from the benchmark index”. It is calculated as the sum of the absolute value of the differences between the weights of the securities of the portfolio and the weights of the securities of the fund’s benchmark index, divided by two:

Active Share formular_active_share.png

The role of Active Share in fund selection has become increasingly important as active managers are having to defend and quantify how ‘truly active’ they really are, especially against passive fund strategies and beta strategies – the popularity of the latter two having grown drastically in recent years.

The ‘closet indexers’, or ‘closet trackers’, are being named and shamed as investors rightfully demand clarity. If you want and are going to pay for active management, you have every right to receive a truly active, benchmark unconstrained proposition.

In addition to their definition of Active Share, Cremers and Petajisto observed a positive historical correlation between high Active Share and higher Excess Return. In conclusion to their work, some investors and advisors have begun to use Active Share as an explicit proxy for a manager’s potential to generate Excess Return in the future. Therefore, in order to benefit from the rising popularity of this measure, several asset managers have decided to calculate and include Active Share on their fund factsheets.

A misleading measure, if used in isolation.

We introduced the concept of Active Share, as a measure of how different an active manager is from an appropriate passive benchmark. We now introduce the concept of R Squared, a standard ratio available on the FE Analytics tool.

The R-Squared measure is an indication of how closely correlated a fund is to an index or a benchmark. It can be treated as a percentage, showing the proportion of a fund’s movements that can be attributed to those of the benchmark. Values upward of 0.7 suggest that the fund’s behaviour is increasingly linked to its benchmark.

Since Active Share shows how different an active manager is from an appropriate passive benchmark, there should be a negative correlation between Active Share and the R-Squared measure. In theory, the higher the Active Share, the lower the R-Squared. In order to understand this relationship, we ran a regression analysis on those funds disclosing the Active Share information on their factsheets (35 funds). Please find below the results of this study (using a five year performance period):

 

Misleading_measure.png

We can observe a negative relationship between the two ratios, but the significance of the test is so low that one cannot conclude or confirm the statement. We cannot therefore say that there is actually no relationship between the Active Share and R-Squared metrics.

Returning to the Cremers and Petajisto paper, we remind you that they introduced the concept of Active Share in conjunction with Tracking Error. Logic would therefore suggest we associate a higher Active Share with a higher Tracking Error. Similarly to the R-Squared measure, we ran a regression analysis on Active Share and Tracking Error over a five year performance period. Please find below the results:

 

active_share_vs_tracking_error.png

 

Once again we can see a positive relationship but the significance is far too low, so one cannot infer anything from the positive correlation between Active Share and Tracking Error. In fact, when used together - both ratios may complement each other to understand how “active” the manager really is.

Cremers and Petajisto state that an active manager can position a portfolio to be different from its underlying index benchmark index via security selection – picking individual stocks that the manager expects to outperform the benchmark, while holding a similar exposure to factors such as sector, industry and market capitalisation. This is a very similar approach to the stratified sampling method used by some passive providers, as the latter try to replicate index performance at a cheaper cost.

Cremers and Petajisto added that alternatively, the fund manager could engage in factor timing, which changes the exposure to these systematic factors over time; or the manager could do both.

A portfolio can be 80% unique at stock level with only a 2% Tracking Error, as there are similarities between the factors. To do so, a portfolio manager can divide the stocks within the fund’s benchmark index in a number of categories (e.g., market capitalization, industry, value, and growth). Each index stock is placed into the different categories that best describes it. The manager would then build a portfolio by selecting one or two stocks from each category and ensuring that the sum of the weight of the stocks purchased from each category corresponds to the category’s weight in the benchmark index.

The combination of Active Share and Tracking Error allows us to distinguish between these two approaches to active management. Active Share is one appropriate metric to measure stock selection and Tracking Error is one appropriate metric to measure factor timing. They help a fund selector in identifying how active a fund manager is. Cremers and Petajisto introduce four groups of active fund managers, using these two variables:

 

Active_share_tracking_error.png

 

Portfolios with a high level of stock selection and factor timing, or “concentrated stock picks”, tend to concentrate on a limited number of securities and factors. If the benchmark is appropriate we will describe the fund as having a “bottom up and top down approach”.  In this category, managers take systematic and non-systematic risks (such as sector or factor risks) in order to beat their benchmarks. Stock picking is not sufficient to understand the fund performance relative to the benchmark. It should be complemented with a sector and style attribution analysis.

“Diversified stock picks” have a high degree of stock selection but express little difference to the benchmark index, with respect to issues such as factor exposure and market capitalisation. We will label this a “pure bottom up” manager; despite a high level of Active Share, the fund performs roughly in line with its benchmark. In this category, managers are trying to align the risks of the portfolios with the benchmark. Only stock picking can explain the fund’s outperformance relative to its benchmark.

Conversely, a fund that is exclusively timing broad factor portfolios but not attempting to choose stocks within such portfolios would have high Tracking Error and a (relatively) low Active Share. We label this a “top-down” managers. The manager will just attempt to outperform the benchmark by playing on one or two factors such as style, industry or market capitalisation.

If an investor does find that one of her investments is a “closet tracker”, it then makes sense to sell this fund and opt for a cheaper “passive” fund. “Closet trackers” funds simply don’t justify their “active” label and importantly their active management costs.

Case Study: Baillie Gifford, Liontrust and Neptune

Let’s now apply this methodology. Baillie Gifford, Liontrust and Neptune were some of the first groups to disclose the Active Share measure onto their factsheets. We will have to rely on this information as FE does not have the relevant details on security weightings for a benchmark index.

We take a look at the 33 equity funds from these three fund groups, which all contain at least a five year track record. We decided to use 4.0 as a triggering value for Tracking Error against their benchmark, and 70% as a triggering value for Active Share. Please find below the graph highlighting the results:

 

graph_active_share_and_tracking_error.png

The graph highlights that the majority of the funds could be described as Concentrated Stock Picks or as “Top down and Bottom Up” investors. Investors will be pleased to see that the majority of the funds sampled justify their “costs” as active managers. Investors will also be pleased to notice that only one fund does not justify its higher “active costs”.

The main purpose of this methodology is to differentiate funds. Two funds display Active Share higher than 70% but with an Annualised Tracking Error lower or equal than 4%. These managers are “bottom-up” stock pickers. The performance of these funds are only linked to the manager’s capacity to invest in the right stocks. Performance can’t be explained by style or sector bets.

On the contrary, we have four funds with an Active Share below or under 70%. Investors may have rejected these funds because they don’t look “active” enough, according to this single measure. Nevertheless, their Annualised Tracking Error is actually high, suggesting they can’t be dismissed.

These fund managers are just playing factor risks, such as style or sectors. Stock picking is secondary.

FE Research believes that Active Share is a useful tool, only if investors combine it with other measures of activeness like Tracking Error or R Squared. On its own, it can only really mislead investors as it does not give any information on how active the portfolio manager is. We believe that investors should actually combine Active Share and Tracking Error in order to classify managers. FE believes that asset managers may try to justify higher “active” costs by playing on the common and rising belief that higher Active Management should generate higher Excess Return – a belief which remains very much still in debate.