Active asset management largely doesn’t work. Most managers do not beat the market and most investors would probably be better off using passive instruments. This finding is practically common knowledge at this point in financial circles.
In this piece, I outline some reasons for the belief, approaching the problem from several different perspectives. I then conclude by talking about some of the implications of these findings for retail investors.
Most arguments levied against active management cite the fact that it appears that most active managers do not beat the index which it is most apt to compare them against. There are various studies looking at this phenomenon. The SPIVA U.S. Scorecard 2019 report, for example, noted the underperformance of all domestic funds in the U.S. This study is particularly good because, unlike many, it takes survivourship bias into account.
Not pretty reading for American asset managers. This trend was pretty much consistent across fund type:
Looking at performance over the whole period, nearly 90% of funds under-performed the market. This is in large part due to the fact that less than half (44.53%) of funds actually survived this 15-year period. Many went bust. Plenty of investor money went down the toilet.
The picture is not much better for fixed income:
Here, it appears that performance is more volatile. We see some ludicrous numbers – over 99% of High Yield Funds were out-performed over a 15-year time horizon – but also some positive signs for asset managers. Overall, and in the long-term, however, the conclusion is the same: most funds underperform the market. And once again, many funds don’t survive:
Note: middle column indicates 15-year survival rate.
This is not just a U.S.-specific problem. Research from Lyxor ETF examined the consistency of performance amongst European funds. Note that they did not take survivorship bias into account, so their results will be skewed in favour of asset managers.
No fund type out-performed its benchmark in aggregate over the examined 10-year period.
Some argue that although asset managers don’t beat the market on average, many managers out-perform and do it consistently. Investing in these funds should deliver superior returns. Although this may be true, it seems that in reality past performance is not a good indicator of future performance. Performance does not typically exhibit consistency.
The SPIVA 2019 Persistence Scorecard examined the ability of managers to consistently outperform the market. Were successful funds following-up success with more success? The answer for domestic equity funds was a resounding no:
With similar conclusions being applicable to fixed income funds:
Despite their apparently-higher levels of culture, it seems that Europeans – like their compatriots across the pond – can’t repeat out-performance either.
Why is this underperformance occurring? Why don’t funds exhibit consistency?
These at first appear to be difficult questions. How can this happen to smart people, who work very hard and are obsessed with what they do? Because of the role of luck.
If we think of returns as having a large element of luck involved with them, some of these conclusions start to make a lot more sense.
Assume that every fund has a 50% chance to beat the benchmark each year that is randomly determined; at the start of each year, God flips a coin to decide whether each fund will out-perform or under-perform. In this scenario, an equal amount of funds will beat the market as will not beat the market each year. It is as if someone’s gain is someone else’s loss. This is roughly how it works in the capital markets: generally for some entity to win, some other entity has to lose (relatively speaking). It’s kind of a 0-sum game. However, active managers have fees. This means that, although half of them are beating the market in our environment, after fees most of them do not beat the market.
This hypothetical world can also explain the lack of consistency in returns. If out-performance is randomly determined, essentially by a coin flip, then successive periods of out-performance are increasingly unlikely (just like how 10 heads in a row is pretty uncommon). The chance of outperformance for year one is 50%. To continue this outperformance in year two the chance is 25%. For year three it’s 12.5% and year four it’s 6.25%. Look back again at the consistency data from SPIVA and Lyxor to see how this is very roughly consistent with what we observe in reality.
This effect is known as regression to the mean. Although in each successive period there is an equal chance of under-performance and over-performance, if we take a period as a whole, a more even split of both types of performance is more likely than all one type or the other.
We can test this conjecture against reality via the use of simulation. Let’s take the performance of the asset managers in the U.K. over the last 10 years (using data from Trustnet):
Note: % returns accurate to 2.d.p. Investment period listed in years.
These are actually pretty good returns considering the returns of the FTSE over the same period. But please remember that these funds should not all be benchmarked against the FTSE 100 and that this table fails to capture the survivourship bias talked about earlier.
For our first test, we take the same number of firms who have 1-year returns listed on Trustnet – 1,686 – and simulate performance over 10 years. We give each fund a 50% chance of outperformance in each period and assume that three consecutive years of outperformance leads to liquidation. After three years of poor returns, the investors want their money back (they have heard of another fund that has had 30%+ returns three years in a row and will invest in them). Simulating this universe 1,000 times we observe that roughly 830 funds survive the entire 10-year period. This is somewhat similar to the figure observed in reality – 762 funds were listed that had track records of at least 10 years.
Secondly, we assign returns from a Normal Distribution with mu of 0 and standard deviation of 15 to each fund for each period. We assume that if the fund has three periods of negative returns in succession, it shuts down and is excluded from the final statistics. Again using 1,000 simulations we can compare the statistics generated from this simulation to those of reality:
Note: % returns accurate to 2.d.p. Investment period listed in years.
Both the mean and median are remarkably similar to reality.
Why is the maximum significantly less?
This is associated with the distribution of returns. You see, we have assumed that returns are taken from the Normal Distribution but, in reality, they have much fatter tails. This is evident by looking at the histogram of the 10-year returns of the funds:
Note: histogram of the 10-year % returns for funds listed on Trustnet.
Ignore the x-axis absolute values here, it isn’t really important – just look at how far some of the observations are from the bulk of the distribution. That tail ain’t skinny…
Our simulations are crude and faaaaar from a reflection of reality. However, they do show how a system in which performance was completely random could have similar outputs to what we observe in the real world. I am not suggesting that all performance is random, just that it’s probably more random than most people fully appreciate, especially naive retail investors (this a point Nassim Taleb stresses in Fooled By Randomness).
The average asset manager doesn’t matter. Identifying top-half performers doesn’t matter. Just invest in all-star fund managers, who have extensive track records and have shown they can beat the market on a consistent basis.
Unfortunately, it’s not that easy.
Someone’s probably gunna win the lottery
Looking again at the U.K. asset managers, it is likely that one will have ten consecutive years of out-performance if we assume a 50% chance of doing so in each year. If we give fund managers a 60% chance of beating the market in each year, one out of those 1,686 has a reasonable chance of out-performance 15 years in a row. Many years of out-performance is possible by pure chance, given a sufficiently large sample size.
I have seen two responses from Warren Buffet to this phenomenon. He stated that although 1,000 monkeys making predictions will likely generate one that is right for ten consecutive years, no-one would invest in a fund operated by a monkey. His point being that lucky managers get found out: you should be able to tell if a manager is bullshitting or if they are genuinely skilled. Unfortunately, this is easier said than done. We are easily convinced by a good story. I’m not sure any manager that has 10 years of good performance would admit that it was luck, this would not attract many new clients. All have a plausible story which is extremely potent when coupled with a consistent track record.
Secondly, he claims that if all/most of the prescient monkeys came from a specific zoo in Omaha, you may want to see what the zookeepers are feeding them. If most investors with attribute X outperform, then X may cause outperformance. This is correct but very hard to determine in practice. Additionally, X may be causal but neither necessary nor sufficient. It could be the case that 50% of investors with attribute X out-perform but this still leaves you with the problem of which manager with X to invest in. Also, some investors who don’t have X may also still be able to out-perform. So even knowing that X is causal still leaves us with a lot of work to do in determining whether a fund will out-perform or not.
Let’s say some managers can consistently out-perform the market due to skill (I believe this to be true). The question now becomes whether it’s possible to identify these out-performers a priori. It’s all well and good admiring the returns of certain funds after the fact. In these cases it may even be possible to determine whether the out-performance was due to skill. However, determining whether a fund will have exceptional performance before it does is significantly more difficult. To do this, you need to be able to distinguish between out-performance due to luck and out-performance due to skill. We have already seen how this can be tricky.
This is also difficult because even if the performance was due to skill, this doesn’t necessarily mean that these skills will be useful in the future, something that Howard Marks alluded to in The Most Important Thing. A manager can employ a successful method that works, but it may not work over the next 10 years. The market changes. Investors adapt. Specifically, the market is very good at destroying profitable strategies because as soon as a profitable strategy is known, it is instantly eroded as more and more investors attempt to exploit it.
Having said all of that, in some cases it’s pretty undeniable that out-performance is occurring due to skill. Take the Medallion Fund of Renaissance Technologies, for example:
$10,000 invested in 1988 in the fund would be worth a shit-load today. The longevity and sheer size of the out-performance makes a random argument almost impossible. This is supported by the fact that no one knows what the fuck they actually do and how they make such large returns. Those strategies that are public knowledge are less likely to be genuinely consistently profitable.
Implications for Retail Investors
Asset management is a remarkable industry.
As previously stated, it has practically become common knowledge that most asset managers do not beat the market. Yet asset management continues to thrive as an industry. This is a testament to:
The sales ability of asset managers. They explain their strategy. It seems clever. It seems like it should work. It’s a good story. A good story that is backed up by 5 years of out-performance. It is very hard not to be convinced by this.
Superstar managers/returns. Some managers can vastly out-perform the market. In some cases, over a long period of time. Whether this is due to luck or skill is largely irrelevant. The possibility of these returns is a very tempting prospect.
Laziness. People are generally reluctant to change their behaviour. You buy a certain brand of cereal (you still eat breakfast?!) because you have always bought that brand of cereal. Humans don’t want to make loads of decisions every day so we fall into routines. And rightly so – this is very helpful. However, these routines can be harmful. Many people still invest in certain funds because that’s what they have always done/that’s what their pension fund is set up to do.
Ignorance. Although the under-performance of active management might seem like common knowledge in financial spheres, this idea has not yet been fully adopted by the general public. Most people are utterly clueless about all things financial and are highly susceptible to good sales pitches.
I don’t think most investors should be looking at active asset management as a viable investment option. There are simply too many traps, too many ways to be conned and too many ways to lose money.
However, the market is not completely efficient. Price does not always reflect intrinsic value and certain investors have the ability to consistently profit from this mis-pricing. Those aware of the pitfalls of investing in active funds may be able to select the right funds, and may be able to beat the market by doing so. This is hard, very hard, to do. But it is possible.