I am still in my research on manager selection, so apologies to everyone who doesn’t find that too interesting. We already touched the question on what to do with underperforming managers in AgPa #59 and #60. This week’s AGNOSTIC Paper, however, examines this problem somewhat more generally and delivers some really simple (but psychologically hard-to-execute) common-sense conclusions.
Everything that follows is only my summary of the original paper. So unless indicated otherwise, all tables and charts belong to the authors of the paper and I am just quoting them. The authors deserve full credit for creating this material, so please always cite the original source.
Setup and Idea
There are certainly exceptions, but most investors probably don’t think about firing their asset manager in periods of strong outperformance. In fact, this week’s authors argue that many of them simply chase the latest high-performing funds and replace their losers with some recent winners. Although the empirical evidence (see below) clearly suggests that this is not optimal, the authors explain that such behavior is not too surprising. Investing in a recent winner is much more comfortable and intuitive than going against the consensus and doubling-down on a loser. Or to quote the Darwinian wit from the paper, “Our ancestors on the African veld did not survive by running toward a lion, so it should not be surprising that we, today, still instinctively avoid what has caused us pain and losses while seeking more of what has given us joy and profits.“
Despite the enlightening quote about our ancestors, the issue of performance-chasing is actually more complicated. There is strong empirical evidence for both Momentum and contrarian-strategies like Value. So it is not necessarily stupid to invest in a recent winner with the thesis that it will go-on somewhat further. Over longer time-horizons, however, most studies find that the uncomfortable contrarian approach dominates.
While the results of this week’s paper are fully in-line with this literature, the key message is even more general. Forgetting about all the technical stuff for a moment, the authors basically just make the point that investors shouldn’t select funds solely based on past performance. In particular, investors should examine past performance together with changes in the fundamental valuation of the underlying assets. More often than not, winning funds just won because their underlying assets got more expensive. This is of course nice for those who invested early enough, but going forward, paying more for the same dollar of earnings or cash flow reduces the expected return of any investment. In contrast, losing funds that got fundamentally cheaper should offer higher expected returns.1Of course, the last two statements are only true if nothing else changes…
I think it is very hard to disagree with this reasoning and this is what I meant with really simple common-sense conclusions in the introduction. We all know that past performance doesn’t guarantee future returns and that there is much more to successful manager selection. From my own experience, however, I can confirm that such simple common-sense frequently leaves the room when the missed-out performance of recent winners just looks too good…
Data and Methodology
For their empirical analysis, the authors collect a sample of 3,331 US funds within the period between January 1990 and December 2016 from Morningstar. Morningstar is a standard professional database for funds, so the data quality should be very good. Most importantly, Morningstar handles survivorship-bias which is particularly important for funds as many of them simply disappear after some periods of underperformance. The authors also use the Morningstar Style Box to classify funds into nine categories to obtain a reasonable peer group. In addition to that, they compare the performance of funds to a passive benchmark of the US market and also calculate alphas versus the Fama-French-5-Factor-Model augmented by Momentum and Low Beta (7-Factor-Alpha).
The heart of the analyses is the relation between past and future performance to test if there is any rational base for performance-chasing. For this purpose, the authors regress rolling 1, 3, and 5-year returns and factor alphas on their corresponding lags and the funds’ trailing expense ratios. There are some statistical issues with such overlapping data, but the authors use some more advanced methods to ensure that their results are robust. I think a deeper discussion about clustered standard errors is beyond this post and doesn’t really help at this point.
Important Results and Takeaways
Current winners tend to be future losers
Starting with the simple return of funds without any adjustment for the market or factors provides a very clear result. Over 3 and 5-year horizons, the regression coefficients are significantly negative and therefore suggest that funds with strong past performance tend to do worse in the future. Over 1 year, the coefficient is not significant and close to zero, and thus doesn’t indicate a short-term momentum-effect. Looking at those results, the idea of chasing the winning funds of the last 1-5 years goes against all empirical evidence. The relation between past and future returns is at best insignificant, but over longer horizons most likely negative.
This pattern is less severe but generally similar for relative returns against the overall market. Over the short 1-year period the relation between past and future outperformance is insignificant. Over the longer 3 and 5-year periods, however, it is again significantly negative, although of less magnitude than for the simple returns. So the message remains unchanged: past outperformance doesn’t indicate future outperformance. If anything, investors should bet that the current outperformers tend to underperform in the future.
For the relative returns of funds with respect to their Morningstar peer-group, there are no statistically significant results at all. Throughout all horizons, the regression coefficients are basically zero. Given that Morningstar classifies funds according to three styles (Value, Blend, Growth) and market capitalization (Small, Mid, Large), these results already suggest that the characteristics of funds matter and that investors should look beyond past performance.
No matter how we spin it, the results are pretty clear. There is no empirical evidence that funds with strong recent performance continue to outperform in the future. The evidence actually speaks for the opposite and suggests a mean-reversion of fund performance over 1-5 year cycles. The authors use those results to demonstrate the stupidity of performance-chasing with sufficient scientific politeness and rigor. And they are not alone with this view. For example, the author from AgPa #60 shows that even large institutional investors are quite bad at hiring and firing asset managers. The average fired manager tends to outperform the average hired manager after the firing-decision…
High fees are the most reliable way to underperform
Before coming to their main advice for fund investors, the authors briefly touch on the very important issue of fees. As I mentioned above, they include the trailing expense ratio as control variable and also report the corresponding regression coefficients. Quite unsurprisingly, the relation between fees and future performance (after fees) is generally negative. Fees apparently doesn’t matter too much over the short 1-year period as the regression coefficients are insignificant. Over the longer 3 and 5-year horizons, however, the relation is massively negative. For the relative return of funds to their peer group and the 7-Factor-Alpha, the results are even more extreme and also significant over the shortest 1-year horizon.
What is more interesting though, is the magnitude of the coefficients. For most specifications, they are somewhere between -1 and -2. Following the simple logic of linear regression, this means that a fund which increases its fees by 0.1% tends to lose more than 0.1% in future performance. This is quite striking and the authors explain this phenomenon by hidden fees like unnecessary transaction costs from sloppy implementation.
So the idea of buying a high-price product in the hope of good service (future performance) doesn’t apply to the world of asset management. If anything, investors should be very careful about fees as they directly reduce performance and are one of the few parameters we know with certainty at the time of making the investment.2These results are also not new and the authors from AgPa #17 find a similar relation between fees and future returns for a sample of European funds. Jack Bogle already said back in 2005 that “In Investing, You Get What You Don’t Pay For”. The evidence still supports this statement and high fees remain one of the safest ways to underperform.
Investors should use factor exposures and valuations to evaluate fund managers
Coming back to the analysis of past and future returns, the results are drastically different for the 7-Factor-Alpha. Instead of the mean-reversion they find for simple and relative returns, the authors here find a small but highly significant positive relation. In English: fund managers who generated alpha in the past, tend to also generate alpha over the next 1, 3, and 5-years. Despite those promising results, the authors highlight a few important practical limitations.
First, calculating the 7-Factor-Alpha is a nice intellectual exercise to determine the skill of a fund manager in theory. Most practical investors, however, focus on beating their benchmark or making money in absolute terms. Both is not necessarily the same as generating alpha versus a 7-factor-model. Second, the seven factors in this analyses were not as popular as they are today at the start of the sample in 1990. The analysis thus suffers from look-ahead-bias. Third and related to the first points, past outperformance in terms of alpha only correlates with future alpha. The relation to simple fund returns and relative returns against the market – which ultimately matter for most investors – is not very stable.
Despite those weaknesses, factors are still useful to examine funds’ characteristics and to form more informed expectations about their future returns. In the first step, the authors show that factor portfolios, just like all other assets, also become fundamentally cheap or expensive over time. For example, the Value factor underperformed massively from 2018 to 2021 and became historically cheap against fundamentals. In 2022, this pattern finally reversed and Value strongly outperformed. The authors document such long-term relations between fundamental valuations and future returns for all of the seven factors. The logic is always the same: the cheaper the factor, the more attractive its future expected return. The cheaper you buy something, the better. It is just uncomfortable for our brains to live by that…
Since we can observe the current valuation of factors, we can use the historical data to make our best estimate about the expected return for each factor going forward. The authors further show that most funds have relatively stable factor exposures over time. Putting those two things together, they recommend to evaluate funds by their Factor Implied Return. This is just the product of the previous ideas. You first check the fundamental valuation of each factor and form your expectation about its future return. In the next step, you identify the exposure of your fund to each factor and calculate its Factor Implied Return as product of the factor exposures and the corresponding expected factor-returns.
The authors find that this approach delivers significant predictive power for fund investors. In both their original US sample and within an international sample used as robustness test, the Factor Implied Return predicts the subsequent one-year return of the fund relative to the overall market with high statistical significance. The authors argue that this approach is much better than just looking at historical performance because it incorporates more information about the fund, its underlying assets, and their fundamental valuation. As I mentioned before, I think it is very hard to disagree with that and argue that just looking at past performance is the better way to go…
Conclusions and Further Ideas
While this post got somewhat longer than I originally planned, there are actually just a few simple key-messages from the paper. First, several investors probably suffer from the psychological bias of performance-chasing because our brains are wired to prefer comfort over pain. Second, there is no empirical evidence that performance-chasing leads to better outcomes. In fact, the data actually suggests the opposite. Investors with a sufficiently long horizon should therefore follow a contrarian-approach and focus on the recent underperformers that are fundamentally cheap. Apart from that, the paper is sometimes also named as a major reference for the idea of timing factor strategies based on their fundamental valuation.
Condensing the information even further, the most important points of the authors are even more general. Selecting funds solely based on their recent (out-)performance is an appealing mental shortcut, but most likely wrong. There is no rational reason to ignore fundamental valuation ratios or other relevant fund characteristics like fees or factor exposures. The empirical evidence clearly suggests that more comprehensive analyses lead to better results.
Factors are just a (maybe somewhat complicated) toolkit to execute such analyses, but the general idea is (in my opinion) just common sense. Unfortunately, common-sense is not always easy to execute, especially when it is uncomfortable. Closing the circle to the challenge of manager selection, I think the takeaway from this paper is very clear. Create a comprehensive analyses-process and execute it without getting lost in appealing psychological shortcuts. And of course, never replace losers by recent winners just because they are winners…
Given that what is comfortable is rarely profitable, having the discipline to follow a much less orthodox and quite uncomfortable approach to investment may translate into far better performance. […] The capital markets do not reward comfort. In investing, we generally find our best rewards in our discomfort zone.Arnott et al. (2018, p.12-13)
Finally, I think there is one more thing to mention. Rob Arnott and his company research affiliates are known for contrarian value investing. Given their stellar reputation and long track-record, I don’t think this impacts the intellectual honesty of the results, but it still makes sense to keep that in mind. With the benefit of hindsight, however, the advice of being a contrarian in this 2018-paper was massively wrong. Several fundamentally attractive factor strategies suffered from massive drawdowns between 2018 and 2021 and became cheaper and cheaper. This just shows out-of-sample that being a contrarian is easier said than done. It is uncomfortable and our brain is not designed to like that. But that is probably also the reason why it should be profitable at some point…
- AgPa #66: Machine-Learned Manager Selection (2/4)
- AgPa #65: Machine-Learned Manager Selection (1/4)
- AgPa #64: Fund Manager Multitasking
- AgPa #63: Fire the Winners and Hire the Losers
This content is for educational and informational purposes only and no substitute for professional or financial advice. The use of any information on this website is solely on your own risk and I do not take responsibility or liability for any damages that may occur. The views expressed on this website are solely my own and do not necessarily reflect the views of any organisation I am associated with. Income- or benefit-generating links are marked with a star (*). Please also read the Disclaimer.