Earlier in my career I used to be involved in the management of a concentrated equity fund. Part of which entailed selling it to potential clients.
Luckily my job wasn’t so much the selling, which was a good thing because I’m a terrible salesman. I was there to answer any of the more technical questions the clients had, while the business development managers did the important bit.
One of the selling points of the strategy was that it had what’s known as a “high active share”. A high active share (say, over 90%) means a portfolio is very different to the index it’s being compared against. A low active share (say, below 10%) means it’s closely tracking the index. A concentrated portfolio holding between 25 and 50 stocks will likely have a high active share, because indices tend to hold many hundreds of stocks.
Having worked in the industry, and having read plenty of research on the subject, I’m familiar with the risks of investing in a concentrated portfolio and some of the ways to mitigate those risks. But to more inexperienced investors, or those on the other side of the table from a skilled salesman, it can sometimes be difficult to figure out whether investing in a concentrated portfolio is a good idea.
This post aims to shed some light on active share, by examining the following questions:
- Does active share predict future performance?
- How should I approach investing in concentrated portfolios?
- How useful is measuring active share?
- The evidence
- Why doesn’t active share predict performance?
- Other active share considerations
- The benefits of active share
There’s no doubt that most active managers underperform. The evidence is clear. (A compilation of research is included in the Active vs Passive section of this blog).
What’s left is to decide how best to improve the state of active management. Many have (correctly) claimed that an increasing reason active managers are underperforming is because they’re not being active enough. Thanks to the rise of passives, many active managers are now reluctant to deviate too far from a passive benchmark as they know their performance will be compared to it. They hold positions they don’t find especially attractive, simply to ensure they don’t fall too far behind their peers.
So, is concentration the key?
From a manager perspective, should managers try to improve their performance by being more concentrated? And from an asset allocator perspective, should allocators be using active share as a way to identify managers with higher chances of outperforming?
Let’s dive into the evidence on active share.
The notion of active share being used as a predictor of future returns got started back in 2009.
Academics Martijn Cremers and Antti Petajisto published a controversial paper called ‘How Active Is Your Fund Manager? A New Measure That Predicts Performance’.
The authors state:
“We introduce a new measure of active portfolio management, Active Share, which represents the share of portfolio holdings that differ from the benchmark index holdings… Active Share predicts fund performance: funds with the highest Active Share significantly outperform their benchmarks, both before and after expenses, and they exhibit strong performance persistence.”
The paper, understandably, caused a bit of a storm in investment circles, and threw light onto the measure of active share as a performance predictor. Other researchers, curious to investigate Cremers and Petajiso’s findings, probed the conclusions reached in the paper and conducted their own further research.
Larry Swedroe summarises the main academic evidence following the original paper in his excellent article, ‘Active Impacts Returns and Volatility’, a few of which I’ll highlight here.
Finance has a poor reputation when it comes to replicating past research findings (some have called it a ‘replication crisis’), but researchers from AQR went straight for the jugular when it came to this one, and tested Cremers and Petajisto’s findings using exactly the same data which was used in their original paper. They presented their findings and conclusions in the March 2015 paper Deactivating Active Share:
“Using the same sample as Cremers and Petajisto (2009) and Petajisto (2013), we reevaluated the empirical evidence of active share’s return predictability. We found no statistically significant evidence that high- and low-active-share funds have returns that are different from each other… We conclude that active share does not reliably predict performance and that investors who rely on it to identify skilled managers may reach erroneous conclusions.
Our results should not be too surprising. Active share is a measure of active risk, and simply taking on more risk is unlikely, by itself, to lead to outperformance.”
The authors concluded that, controlling for benchmarks, active share has no predictive power for fund returns.
I like that they also gave a shout-out to our well-known friend the Arithmetic of Active Management, which I seem to end up mentioning in every article I post:
“In general, if the universe of mutual fund managers holds the market portfolio, we know that the market clears: Before fees, every dollar of outperformance must be offset by a dollar of underperformance. Low-active-share investors who simply track the market (Closet Indexers) will match market returns before fees and underperform after fees. As a result, investors who take large bets (and have high-active-share results) will also match market returns before fees and underperform after fees (Sharpe 1991). Among the high-active-share investors will be winners and losers, but as a group, they cannot systematically outperform the Closet Indexers.”
A particularly telling indictment against using active share to predict returns is Cremers’ own update of his original study. He updated his 2009 findings in his October 2016 study ‘Active Share and the Three Pillars of Active Management: Skill, Conviction and Opportunity.’ In it, he recommended that for best results, investors should combine active share with low turnover.
The problem with his finding was that it only held true for the first half of the period he studied (1990 – 2001), which is pretty clear from looking at one of the paper’s charts below:
Source: Financial Analysts Journal
That light blue line is the highest active share funds which have had the longest holding period. Most of the outperformance looks like comes pre-2002.
Larry Swedroe put his detective hat on, and contacted the author for the results for the second half of the period. Here’s the data for the period 2002-2015:
The table shows the manager alphas for each quintile of active share, with all generating negative alpha. That doesn’t look like a particularly ringing endorsement for the efficacy of active share – even when combined with lower turnover.
While the paper’s conclusion was that active share should be combined with low turnover to maximise its effectiveness, the conclusion on high active share being correlated with higher returns is far more circumspect:
“We find no evidence that high Active Share funds have underperformed on average in the long-term, suggesting that investors interested in individual stock pickers could use high Active Share as a starting point for fund selection, but with no ex-ante expectation that the typical high Active Share fund is going to either underperform or outperform.”
The combination of the paper’s conclusion that active share has no bearing on performance and the mixed performance of the high active share/low turnover combination isn’t particularly convincing for using active share as a return predictor.
Piling on to the active share bashing came researchers from Blackrock, who also released a study investigating active share, called ‘Estimating Time-Varying Factor Exposures’. This provided an out-of-sample test (using data after 2009) of the original Cremers and Petajisto findings.
They found that the measure of active share proposed was actually negatively correlated (-0.75) to fund returns after controlling for factor loadings and other fund characteristics. Thus, they concluded:
“We found no evidence that active share is associated with larger active returns; the opposite is true across the whole sample when controlling for such factors as fund size and fees. Managers may have a very high active share, perhaps because they select stocks from only a fraction of the universe yet maintain distinct tilts to factors over time.”
Another strike for active share.
And then in come Morningstar.
They weighed in on the active share debate in several different articles and papers, all of which reach similar conclusions (we’ll get into the main whitepaper in the next section on the reasons behind active share’s lack of predictive power). As an example of one such article, their post ‘Portfolio Concentration Has Little Sway on Returns’ takes their universe of US funds and splits it by concentration quartile. The results are below.
NB: Remember success rates are the percentage of funds which both survived and outperformed their benchmark:
With the three-year holding periods, the most concentrated quartiles had lower success rates than the least concentrated ones in five of the US stock categories and four of the foreign stock categories.
The author concludes:
“These results aren’t any different from what we might expect from chance alone. In about half of the categories, the most concentrated funds did better, and in the other half the least concentrated did better. In all cases, the difference in success rates between the most and least concentrated quartiles was within 10 percentage points. This suggests that there isn’t a significant relationship between portfolio concentration and the odds of beating the market.”
So there’s another study coming to the same conclusion that active share alone has no predictive power.
As if this dead horse hadn’t been flogged enough already, the final piece of evidence against active share being a return predictor comes from Vanguard, in their 2019 paper ‘Urban Legends of Active Share’.
To determine the relationship between active share and performance, they tested the difference in the means between the highest and lowest quintiles of active share, as well as the cost per unit of active share against returns on both a gross and net basis.
We’ll come back to active share and costs in the next section, so let’s focus on the left-hand side of the table for now, which focuses only on the relationship between active share and performance:
We can see from the left half of the table that the relationship between active share and excess return is inconsistent through time.
In the 2009–2013 time frame, for example, there’s a statistically significant positive relationship between high active share and gross excess return. But that relationship completely reverses in the subsequent five years. There was no statistically significant relationship in the preceding five years either, or – probably most importantly – over the entire time frame (2004 – 2018).
Vanguard reach the same conclusion as the other papers I’ve mentioned, noting “our research confirmed prior findings that increasing active share does not lead to outperformance.”
Hopefully that’s enough evidence to make you think twice about using active share as a reliable indictor of outperformance. The evidence seems pretty clear about the lack of relationship between the two.
Why doesn’t active share predict performance?
The evidence is pretty clear that active share doesn’t predict performance.
But perhaps a more interesting question is: why not?
I’ll touch on three reasons behind the lack of relationship between active share and returns here, and have split this section up into (short) segments for clarity:
At the risk of stating the obvious, one reason why active share doesn’t predict returns is there’s no reason it should. Active share only tells you how different your returns are likely to be from the benchmark.
The charts below, from the same Vanguard report we saw above, put this idea into some pretty pictures.
The charts show that across market capitalizations and time periods, higher active share seems to lead to a wider dispersion of outcomes and a relatively symmetrical distribution of excess returns across the range of active share.
Active share does what it says on the tin. It tells you how different a fund is to its benchmark.
Just because your fund is different, it doesn’t mean it’s better.
One thing investors need to bear in mind before allocating to high active share funds is that higher dispersion increases the cost of getting it wrong. Yes, you could you have stellar performance on the upside, but the downside’s are also magnified.
A second reason why high active share doesn’t correlate with high returns is fees.
I wrote about fees being the single best predictor of future returns in this article, and the principle shows up again when looking at active share.
Here we can take a look at Morningstar’s whitepaper ‘Portfolio Concentration’, which shows that while there wasn’t a significant relationship between concentration and returns, there was a positive relationship between concentration and fees.
(NB: A summary/commentary on the paper is here for those who’d rather just get the highlights.)
Morningstar found the most concentrated funds (those in the fourth quartile – “Q4”) tended to charge higher fees than the least concentrated (“Q1”):
The authors state:
“This is consistent with the idea that investors are willing to pay up a bit for bolder active bets. Higher fees make it harder for these funds to outperform.”
Vanguard explored the same idea in their paper, and come to a similar conclusion. They found a positive relationship between active share and costs—the higher the active share, the higher the cost.
Taking it one step further, they then work out the relationship between “cost per active share” – i.e. how much the genuinely active part of the portfolio costs – and after-fee returns.
For those, like me, who prefer looking at pictures to trying to decipher numbers, the charts below plot the relationship between 5 year returns and cost per unit of active share:
The charts show there’s not a significant relationship between cost per unit of active share and excess return on a gross basis, but on a net basis it looks like there’s a slight negative relationship, suggesting that cost isn’t a reasonable proxy for skill.
The summary numbers behind the charts, which are slightly more difficult to get your head round but much more persuasive, are in the furthest right column of the table below:
The numeric results confirm the finding of a reasonably consistent negative and statistically significant relationship between cost per unit of active share and performance on a net return basis across time and market capitalisation.
Comparing this to the left-hand columns we looked at before, which only looked at the relationship between active share and performance, we can see that the key when figuring out the predictive power of active share is to combine active share with the price paid for the level of active share received.
While active share in itself may be inconclusive in terms of its predictive power, the cost per unit of active share looks like it’s more useful in its ability to predict future returns.
Funds which charge high fees for low amounts of active share are more likely to underperform.
This is useful for investors when deciding between funds. If you’re buying an expensive fund, make sure you’re getting some active share for your money. And if you’re buying a fund with high active share – make sure you’re not paying too much for it.
3) Positive skew
The third reason why high active share doesn’t equate to higher performance is the fact that the vast majority of stock market returns come from a tiny minority of stocks.
This was first highlighted by Bessembinder’s famous paper, ‘Do Stocks Outperform Treasury Bills’, the findings of which have been confirmed many times since.
His findings are hugely important for portfolio construction, and a great argument for diversifying as widely as possible:
“When stated in terms of lifetime dollar wealth creation to shareholders in aggregate… slightly more than 4% of the firms account for all of the net stock market gains. The other 96% of firms that issued stock collectively matched one-month Treasury bill returns over their lifetimes.”
This idea that a tiny number (4%) of stocks produce the vast majority of gains is known as positive skew:
Source: Do Global Stocks Outperform US Treasury Bills? (this is from the 2019 update to his earlier 2017 paper)
The fact that markets exhibit such positive skew makes the life of a concentrated fund manager difficult, because those funds holding only a handful of stocks are much less likely to be holding the very few which generate all the gains. They’re fighting against the maths.
To hammer home the point, research from S&P Dow Jones does a great job of putting this into perspective with a simple but interesting thought experiment.
We start by imagining a market with five equally weighted stocks, whose performance over one year is shown in the table below:
Source: S&P Dow Jones Indices LLC
Averaging their returns, we can calculate that the market’s overall return for the year is 18%, but this is mainly driven by the outstanding return on stock E. This is therefore a positively skewed distribution (just like stock returns in the real world).
Using these five stocks, it’s possible to form portfolios of various combinations of each stock. For example, there are five possible portfolios with only one stock in them – one portfolio containing only stock A, one portfolio containing only stock B, one portfolio containing only stock C, etc. Four of these portfolios underperform the market as a whole – the portfolios containing only stocks A-D. They each return 10% vs the market’s 18%.
These are obviously the most concentrated portfolios possible.
If we create more diversified portfolios by holding four stocks rather than one, there are five possible portfolios containing four stocks. Four of these now outperform the market as a whole – all of them except the portfolio which contains only stocks A-D. This portfolio returns 10% vs the market’s 18%.
The expected return of the complete set of one-stock and four-stock portfolios is the same 18%, but the distribution of portfolio returns is different. In this case, holding fewer stocks increases the likelihood of underperformance, because the portfolios are less likely to contain the stock which has generated most of the gains (stock E).
Put simply, only the portfolios which contained stock E didn’t underperform. The more concentrated the portfolio, the more likely it was not to hold E.
Other active share considerations
As a quick recap, we’ve seen that high active share doesn’t predict outperformance because:
- Active share only predicts dispersion of returns, not returns themselves
- High active share funds tend to be more expensive
- Stock market returns are positively skewed
Here I’ll cover off a few further considerations it’s worth investors bearing in mind when examining high active share funds. I’ve again split it into segments for clarity:
- The behaviour gap
- Luck vs skill
- Survivorship bias
1) The behaviour gap
When a concentrated fund is outperforming, it’s the best feeling in the world for investors.
As we saw from the research above, the higher the active share is, the more different the returns are compared to benchmark. So when a fund with high active share outperforms its benchmark, it can do so in spectacular fashion.
On the one hand, this is great – everyone wants a fund with the potential to shoot the lights out.
But the flip side is equally true – the more different to a benchmark a fund is, the more pronounced the periods of underperformance will be.
Everyone’s able to stick with a fund which outperforms by 20% per year thanks to its concentrated high-conviction bets, but when a fund suffers a couple of periods of 20% underperformance, then sticking with your high active share fund suddenly becomes much more difficult.
This not only makes intuitive sense (and is consistent with my own experiences), but is also supported by evidence from Vanguard.
Returning to the Vanguard paper we saw before, they examine the effect of active share on investor behaviour. They use the difference between the time weighted return on a fund (i.e. the return quoted on the factsheet), and the return investors actually receive (which includes their timing of when the bought/sold the fund) to measure what’s known as the “behaviour gap”.
This is supposed to measure the difference between how much an investor would have received if they’d bought a fund at the start of a period and done nothing but hold it for the whole period, versus how much they actually received in returns because of the investor trying to time the market by buying and selling the fund throughout the period. A positive behaviour gap shows the investor outperformed by buying and selling at the right times, and a negative behaviour gap shows the investor received less than if he’d simply bought and held.
The behaviour gap is notoriously difficult to measure (it’s debatable whether IRR is a good measure of investor behaviour), and I’ll be writing a full post all about it in the future. But for now, it’s at least interesting to have a look at Vanguard’s analysis.
The chart below shows the behaviour gap varying by active share. Dots above the line show a positive behaviour gap (investors successful at timing buys/sells), dots below the line show a negative behaviour gap (investors chasing performance and getting their buy/sell timings wrong):
It’s slightly annoying that there’s no numerical data to go along with the table, but from a simple eyeball we can see two things.
Firstly, as active share increases, so does the dispersion in the behaviour gaps. Investors are more prone to market-timing attempts when investing in funds with higher active share.
As to whether this is effective or not, this is where it’d be handy to have the underlying data. Just from looking at the chart though, it looks like there are quite a few more dots below the line than above the line – with the ratio increasing as active share rises.
This would indicate that those investors using higher active share funds are not only more prone to market-timing, but they’re getting it wrong more often then they’re getting it right – and hurting their returns in the process. They’d be better off buying and holding.
The research agrees with what I think’s pretty intuitive anyway. Higher active share funds, because of their more extreme relative performance, tempt investors to invest after periods of strong performance, and make it difficult for investors to stay the course after periods of underperformance.
We know from my previous post why chasing performance leads to bad outcomes, and concentrated portfolios only add fuel to the fire.
Overall, higher active share strategies encourage bad behaviour.
2) Luck vs skill
I’m fascinated with the idea of trying to disentangle luck from skill in investing. It’s why Michael Mauboussin’s book ‘The Success Equation: Untangling Skill and Luck in Business, Sports, and Investing’ was one of my favourite books of last year.
The luck vs skill debate is an important consideration for investors when looking at active share, too. A great illustration of how increased portfolio concentration makes distinguishing skill from luck more difficult comes by the way of another great thought experiment from S&P Dow Jones.
If I was a stock-picking god, descended from Mount Olympus for the sole reason of generating top-decile risk-adjusted returns, then I’d want as many opportunities as possible to display my skill.
In order to demonstrate my godlike stock-picking skills to the mere mortals, I’d need to pick a large number of stocks – all of which must go up. If I only buy a few stocks which go up, then my performance could very well be due to luck. In order to prove my skill beyond any reasonable doubt, I must be correct many times.
S&P illustrate this idea using coin flips.
In a coin-flipping game with a biased coin, one coin has a 53% chance of heads and the other coin has a 55% chance of heads. We win the game if we get more heads than tails.
Source: S&P Dow Jones
As the graph shows, the chance of winning grows as the number of tosses rises and, for any number of tosses, the chance of winning is higher with the more favourable coin – that’s exactly what we’d expect. No surprises there.
But if the number of tosses varies between the two coins, the key idea is that at some point, it’s preferable to have a worse coin and more tosses. For example, I’d rather have the worse coin with 101 tosses than have the better coin with only 1 toss.
Because the game has an element of luck, the more tosses we have the more accurately the results will reflect the true statistical properties of the coin – i.e. the longer we flip for, the more likely we’ll be able to tell which one is biased and by how much.
The analogy to stock selection is straightforward.
Instead of flipping a coin, imagine a manager picks stocks with a given probability of outperforming the market. Let’s now say we have two managers – one who has a 55% chance of picking an outperforming stock, and one less skilled manager who only has a 53% chance.
The more picks the more skilled manager makes, the more likely it is that his skill dominates his luck. As with the coin-flipping game, for a constant number of stocks, a more skilful manager is more likely to outperform than a less skilful manager. But if the number of picks varies, it’s more likely that a less-skilled manager picking more stocks will outperform a more skilful manager who buys fewer stocks.
Concentrated portfolios can make it difficult to tell between skilled managers.
Talking of skill, in this example both managers have an over 50% chance of picking an outperforming stock. But what happens with a manager with below-average skill?
In this analogy, he’s also flipping a biased coin, but his coin has less than a 50% chance of coming up heads. Ironically, this manager has a better chance of winning the game the smaller the number of tosses (just as a skilled manager has a better chance the more he tosses).
The fewer stock picks the below-average manager makes, the more likely it is that luck will dominate, his low skill will remain hidden, and he’ll be seen as preferable to other more-skilled managers.
Concentrating portfolios, in other words, makes it more likely that good managers will look bad, more likely that bad managers will look good, and more likely that asset allocators’ decisions will be informed by luck rather than skill.
To add fuel to this fire, let’s not forget about one of the most important aspects of manager selection – survivorship bias.
3) Survivorship bias
Let’s say I was suddenly put in charge of a fund management company, but the catch was that my only employees capable of selecting stocks were 100 chimpanzees.
Here’s my plan.
I create 100 concentrated portfolios with room for, say, 25 stocks each – then get each of the 100 monkeys to throw darts at pages of Financial Times. The first 25 stocks each of them hit would become their portfolio.
Next, give it a few years for each to build up a track record, then throw the marketing budget behind the ones which outperformed.
I bet there’d be some performance charts in there so good you could make investors forget their fund was being run by literal monkeys.
Of course, we’d know this performance was all due to luck. But the only thing potential investors see is a bunch of graphs showing funds which massively outperformed their benchmarks. Concentration magnifies both outperformance and underperformance – hide the underperformers and you look like a hero.
Given how concentration increases dispersion, survivorship bias becomes even trickier to combat. You won’t see any of the equally concentrated funds which didn’t make it, and the funds you do see will be even more attractive because they’ll have stronger outperformance.
The bottom line is concentrated portfolios make distinguishing luck from skill much trickier for investors.
One final point on why investors should approach high active share funds with caution is because the level of active share is dependent on the benchmark chosen.
I could run a FTSE 100 tracker fund with an active share of 0%, then decide to change my benchmark to the S&P 500 and have an active share of 100%. The portfolio hasn’t changed, but my active share has rocketed because I chose a different (totally inappropriate) benchmark.
When the benchmark for any active share analysis is selected by the fund manager, rather than an independent third party, it’s difficult to tell whether the active share statistic is useful or not without assessing the validity of the benchmark, which makes drawing any conclusions more complicated.
This is less of a problem when looking at conclusions drawn from analysis conducted by independent third parties like Morningstar, as in this case it’s Morningstar defining which benchmarks are suitable for which funds.
But when you’re relying on data compiled by the fund manager, like when reading an in-house whitepaper or being on the receiving end of a marketing pitch, be aware that benchmark flexibility makes drawing conclusions on active share trickier.
The benefits of active share
I’ve painted a pretty bleak picture of concentrated, high active share portfolios for both managers and investors.
We’ve seen it’s difficult for fund managers to outperform using a concentrated portfolio because:
- Concentrated portfolios tend to charge higher fees
- Positively skewed returns means concentrated funds are unlikely to own the mega-performers
We’ve also seen it’s difficult for investors to use active share as a basis for investment decisions because:
- It has no correlation with outperformance.
- Higher active share funds have higher dispersion and a higher cost of getting it wrong.
- Higher active share funds are likely to encourage worse investor behaviour.
- They make skill harder to detect, and worsen the effects of survivorship bias.
- Active share is dependent on the benchmark being used.
But despite all that, active share does still have its uses.
Importantly, it’s helpful for sussing out closet indexers.
There are plenty of incentives for managers, over time, to drift towards benchmark weights. Strategy capacity limitations causing style drift, the rising popularity of passives as benchmarks, the proliferation of intermediaries, and career risk being some of the main culprits. So being aware of a fund’s active share, and how it evolves over time, can be a useful measure for ensuring your active fund is staying active. You don’t want to be paying active management fees for benchmark performance.
It can also be useful for ensuring the fund your manager is comparing themselves against an appropriate benchmark.
This is especially important for those funds charging performance fees, as these managers are rewarded for outperformance, but not penalised for underperformance. This fee structure incentivises choosing a less relevant benchmark, to maximise the magnitude of potential outperformance. A fund with a high active share may signal an inappropriate benchmark is being used (although this is obviously not always the case – many funds choose the most relevant benchmark possible, and just choose to be different).
Using active share as an ingredient in the cost per active share calculation is also useful.
The evidence above suggests cost per active share does seem to have some predictive power. When conducting fund due diligence, this looks like a useful metric to bear in mind. If you’re buying an expensive fund, make sure you’re getting some active share for your money. And if you’re buying a fund with high active share – make sure you’re not paying too much for it.
- The evidence shows there’s no correlation between active share and performance.
Why doesn’t increasing active share increase returns?
- Dispersion – active share only predicts dispersion of returns, not returns themselves.
- Funds with higher active share tend to be more expensive.
- There’s a negative relationship between cost per active share and outperformance.
- Positive skew – the more concentrated a portfolio is, the less likely it is to own the few stocks which generate most of the returns.
Other active share considerations:
- Higher active share funds, because of their more extreme relative performance, are likely to encourage worse behaviour.
- Concentrated portfolios make it more likely that good managers will look bad, more likely that bad managers will look good, and more likely that asset allocators’ decisions will be informed by luck rather than skill.
- Given how concentration increases dispersion, survivorship bias becomes even trickier to combat. You won’t see any of the equally concentrated funds which didn’t make it, and the funds you do see will be even more attractive because they’ll have stronger outperformance.
- When relying on data compiled by the fund manager be aware that benchmark flexibility makes drawing conclusions on active share trickier.
- Identifying closet indexers
- Ensuring benchmark appropriateness
- Calculating cost per unit of active share
We’ve seen how heavily the odds of outperformance are stacked against concentrated high active share funds.
Given this, I was initially surprised to see there wasn’t a stronger correlation between high active share and underperformance. Intuitively I would’ve expected to see a stronger correlation between the two, so the fact most of the evidence points to no correlation at all is (in my mind) a small win for active managers.
It suggests concentrated managers must be having at least some success with their high-conviction ideas. If they weren’t, the evidence would show a clear negative correlation.
But the fact remains that concentrated managers are facing an uphill battle. The evidence shows despite having some success with their high conviction ideas, it’s clearly not enough to improve their odds of outperforming.
This presents a tricky dilemma for investors.
On the one hand, we’ve seen that concentrated portfolios are no more likely to outperform, have higher dispersion, encourage worse behaviour, make skill harder to detect, and increase the cost of being wrong.
On the other hand, we’ve know we don’t want our active funds to be too diversified to point where they’re tracking an index.
So where’s the middle ground? Is there one?
I think this question essentially boils down to “Should I include concentrated funds in my portfolio and, if so, how do I go about choosing them?”
Although I’m not personally a fan of concentrated funds, I think they can work in in portfolios looking for active exposure. For those wanting to allocate towards them, there are a few things to bear in mind before doing so.
First of all, investors are going to want to try and identify managers displaying a high active share, low cost per active share combination. The strongest correlations we saw in the evidence on active was that between high cost per unit of active share and underperformance. That’s a combination to avoid.
Secondly (and this goes for all active fund selection), ensuring the manager has a long track record is especially important here, given how much more difficult it becomes to separate luck from skill when dealing with concentrated position sizes.
Thirdly, as with all things in investing, remaining diversified is key. Diversification means ensuring the concentrated positions of the underlying manager don’t become too large a part of a broader portfolio. Investors should be allocating to high active share funds with a view to creating and maintaining a balanced portfolio – which I think can be achieved in a couple of ways.
One approach could be to use one or two concentrated funds as ‘satellite’ positions, allocating a small weighting to a couple of high-conviction funds alongside a much larger core holding in a broadly diversified portfolio.
Another could be blending a larger number of concentrated funds together in such a way that their combined exposures result in increased diversification (but ensuring you don’t end up with index-like performance.
I think either approach can work.
From a behavioural perspective, the challenge for these investors will be trying to avoid zooming in too heavily on relative performance. Concentrated funds will underperform, but by integrating them into a broader, diversified portfolio, it’s more likely investors will be able to refrain from making emotionally charged decisions (which the evidence shows is damaging for returns).
For me, I avoid using concentrated funds for a few reasons.
Firstly, I don’t like the dispersion. The costs of getting it wrong increase the higher up the active share spectrum you go, and we saw in this post, ‘Smooth is what we aim for’ that minimising dispersion is useful for maximising returns through avoiding the volatility tax.
It also takes a huge amount of time to build up a track record which implies any outperformance is due to skill rather than luck. We saw in this article that it takes at least 20 years to be 90% confident of skill, and given the larger role luck plays in concentrated portfolios, I’d have a difficult time finding funds I’d be confident in.
Of those, there are even fewer which I’d have enough faith in to believe the skill which delivered those returns would remain for the rest of my investing lifetime. Given how quickly fund mangers, companies, markets, and economies change over time, I think it’s unlikely that skill can survive for long enough to not only prove its existence, but to stick around for long enough for investors to benefit from it.
As a result, I’m unlikely to be able to stick with my chosen funds through their inexorable bouts of underperformance.
This is a conundrum faced by all active investors, but it’s especially difficult for those looking to allocate to concentrated funds, given the high return dispersion and the costs of getting it wrong.
Speaking of difficulties with traditional active investment, high active share funds also suffer from some of the difficulties I face when investing in other active funds.
These factors are personal to me, and I’m sure not everyone feels the same. But I tend to avoid active funds for a few reasons, including them taking up more mental bandwidth (I’d rather think about other things than worrying about underperformance), increased levels of decision-making required (I will always make the wrong decision), increased time required for monitoring (I’d rather not spend my time assessing style drift), increased stress during periods of underperformance (I like passives because they can’t underperform), not knowing when to sell (I’ll either end up regretting selling too early or not selling soon enough), and taking on the risk of the fund merging/closuring (they’re not all Woodfords, but mergers due to underperformance are more common than you’d probably think).
All of these become magnified with concentrated funds, so I tend to steer clear.
Overall, I’d exercise caution before relying too heavily active share alone to inform investment decisions. It clearly has no positive correlation with future performance.
But active share can be a useful statistic, both when selecting funds and in their ongoing monitoring. For those looking to allocate to active managers, funds with high active share can be valuable when selected carefully, and when used in a fully diversified portfolio.
They’re just not for me.