The predictive power of fees

When Julius Caesar was 25 years old, he was kidnapped by pirates as he was sailing on his way to Rhodes.

The pirates, having realised he was of some value to Rome, set his ransom at 20 talents of silver. But Caesar wasn’t having any of it. Where most people would have been keen to reduce the ransom as much as possible to increase the likelihood of it being paid, Caesar demanded that the pirates increase the ransom to 50 talents instead. 20 was simply not enough.

Although still a young man, Caesar was well on his way to becoming a powerful person in Rome. He had lofty ambitions, and 20 talents didn’t sound like a worthy ransom for someone as vital to Rome as he considered himself to be. By telling his captors to increase his ransom, he was sending a message to Rome. Caesar was an expensive person, so he must be important.

And if something is that expensive, it must be worth paying for.

This heuristic is still just as relevant today as it was in 75 BC. An expensive pair of shoes will be better than a cheap pair. An expensive television will be better than a cheap one. An expensive hotel will be better than a cheap one.

And that’s why I’m so persistent in writing about fees.

Because the default mode of thinking for most of us is that price can be used as a reasonable proxy for quality, it’s hard to resist the allure of something expensive which promises to be better than the cheap option – especially when the expensive option isn’t available to the riff-raff.

But when it comes to investing, this behavioural bias leads us astray. As John Bogle was fond of saying:

“In mutual funds you don’t get what you pay for. You get what you don’t pay for.”

And there’s plenty of theory to back up that statement.

For example, the chart below (from this post) assumes a portfolio grows at 7% a year for 40 years, and shows the percentage of the portfolio consumed by fees. A 1% fee consumes 31% of the final portfolio value, a 2% fee consumes 53%, and a 3% fee consumes 68%. And these percentages are similar no matter what the assumed returns are (+/- 2% for anywhere between 0% – 10% returns).

Another theoretical example of the power of minimising fees is Sharpe’s Arithmetic of Active Management. It states that before fees, both the portfolio representing the aggregate of all active managers and the portfolio of index-trackers are the same capitalisation weighted ‘market’ portfolio. Therefore, the returns to active investors, on average, and the returns to passive investors must be the same, before costs.

In reality, the higher fees charged for active investment management means that, on average, those paying for active management must underperform those paying lower fees for passive management. After all, they’re both receiving exactly the same return before fees, on average.

As a result, the less you pay for your investments, the less you’re likely to underperform the market by. In short, lower fees predict better performance.

These are just two of my favourites, but there are plenty of theoretical arguments for minimising investing costs.

But what does the evidence say? Do lower fees really predict higher returns?

Contents

• Predicting future returns
• The pervasive power of fees
• Is diversification worth higher fees?
• What about cheap active funds?
• Are fees all that matter?
• What about active share?
• Conclusion

Predicting future returns

In this section, I’ll go through a few pieces of research from Morningstar and Vanguard on the predictive nature of fees. For those who are after a deeper dive, there’s a list of papers at the end of this post, all of which focus on the same topic.

To start with Morningstar, they have, as usual, done some excellent research on the topic of fees and future returns. I tend to like Morningstar research for the same reason I started this blog – there’s a lack of independent research out there. Given Morningstar aren’t tied to any fund manager or promoting any product (aside from their own research and fund ratings), their research is as close to independent as it comes.

As a good starting point for diving into the evidence, the whitepaper linked to in this Morningstar article examines why fund fees are so important.

The article itself is a commentary on the whitepaper. The paper’s conclusion is:

“Using expense ratios to choose funds helped in every asset class and in every quintile from 2010 to 2015. For example, in U.S. equity funds, the cheapest quintile had a total-return success rate of 62% compared with 48% for the second-cheapest quintile, then 39% for the middle quintile, 30% for the second-priciest quintile, and 20% for the priciest quintile. So, the cheaper the quintile, the better your chances.”

The predictive power also holds up in the other areas we tested. It points investors to a better outcome for investor returns and for load-adjusted returns. That makes some sense, as both are fairly closely tied to total returns.”

The chart below does a good job at presenting the data visually, showing that the cheapest funds (the left-most bars of each colour) had higher success ratios than the most expensive funds (right-most bars).

NB: Success ratios show the percentage of funds which managed to both survive and outperform their category group. Only funds which did both count toward the success ratio, as it’s hard to argue that funds which no longer exist or underperformed were successful (more on that later).

Source: Morningstar

According to their whitepaper, lower fees do predict better performance across each of the asset classes they tested. Comparing the left bars to the right bars, it’s interesting to contrast the lowest quintiles’ success ratios (all over 50%) to the highest quintiles’ (all below 25%).

Morningstar also periodically produce a report called the ‘Active/Passive Barometer’, which, similarly to the SPIVA Scorecard, is an excellent and regularly-updated source of information on the success of active funds.

This post isn’t about the overall performance of active vs passive – that’s been covered extensively in the ‘Active vs Passive’ section of this blog. If you’re a regular reader, you’ll know how convincingly the statistics look in favour of passive performances vs active. Here, however, we’re focussing solely on how much of that outperformance is due to lower fees (as opposed to other variables which might cause passives to outperform, like concentration, active share, size, skewness, style drift, etc).

Luckily, the Barometer has some data for us.

The table above is taken from the report, and shows that for the US Large Cap Blend category of funds, the cheapest quintile of active funds had a success rate of 17.2% over the last 10 years, compared to the most expensive quintile which had a success rate of only 4.1%.

(Remember, success rate is the percentage of funds which both survived and outperformed)

Neither of those stats are particularly encouraging for active investors, but all else equal the cheaper funds had the better chance of success.

Sticking only with the active funds for now, by looking at the detail of the table, we can see that this result was a combination of two things. Firstly, the cheaper funds were more likely to survive than the expensive funds. 49% of the cheapest quintile active funds survived, compared to 39% of the most expensive quintile. Secondly, the cheaper funds had better performance than the expensive funds. On an asset-weighted basis, the cheapest quintile active funds produced an annualised return of 12.5%, compared to the return from the most expensive quintile of only 8.9%.

The combination of the expensive funds being more likely to be closed/merged and being more likely to underperform results in the much lower success ratios for the expensive quintiles.

If we compare those stats to the passive funds, we can see the passives have done much better. The 49%-55% survivorship rates for the active funds compare very unfavourably against the 48%-83% survivorship ratios for the passive funds. Passive funds, especially the cheapest ones, are much less likely to close or be merged. On the performance side, passives come out ahead too, with passives outperforming actives across all 5 fee quintiles – by 1.3% per year in the cheapest quintile to 3.1% per year in the most expensive quintile.

The table above is only one category examined in the report. The findings are the same for plenty of other asset classes included in their analysis, including all combinations of large cap/mid cap/small cap and blend/value/growth, as well as for European funds, emerging market funds, real estate funds, and bond funds.

The table below shows the full list of asset classes covered, as well as the high-level success ratios of the lowest cost versus highest cost quintiles for each category:

The two coloured columns are the ones we’re interested in here as they’re the ones which contain fee data. The 10-Year (Lowest Cost) column shows the success rates of the lowest cost quintiles for each category, and the 10-Year (Highest Cost) column does the same for the highest cost quintiles. A green number represents an above average success rate, and a red column represents a below average success rate.

The first line of the table shows the stats we were looking at earlier for the US Large Blend category – 17.2% versus 4.1% active success rates.

It’s pretty clear even from just eyeballing the table that the lowest cost quintiles across all categories had a substantially higher chance of success compared to the highest cost quintiles.

Bringing it all together into one statistic on the predictive nature of fees, Morningstar found that :

“The cheapest funds succeeded about twice as often as the priciest ones (a 34% success rate versus a 17% success rate) over the 10-year period ended Dec. 31, 2020. This not only reflects cost advantages but also differences in survival, as 63% of the cheapest funds survived, whereas 49% of the most expensive did so.”

Turning now to the final piece of Morningstar research on the predictive nature of fees, titled ‘How Expense Ratios and Star Ratings Predict Success.’

This is one of my favourite articles when it comes to why fees are so important. It’s an article written by Morningstar, comparing the extent to which both fees and Morningstar’s own fund star ratings are able to predict future performance. It ultimately comes to the conclusion that while both investment fees and Morningstar’s own star ratings are useful, fees came out on top as the better predictor of future returns:

“Perhaps the most compelling argument for expenses is that they worked every time–because costs always are deducted from returns regardless of the market environment. The star rating, as a reflection of past risk-adjusted performance, is more time-period dependent. When the market swings dramatically, the star rating is going to be less effective.

Investors should make expense ratios a primary test in fund selection. They are still the most dependable predictor of performance. Start by focusing on funds in the cheapest or two cheapest quintiles, and you’ll be on the path to success.”

The article was written in 2010, so it’s quite dated and Morningstar may well have updated their star-rating methodology by now. But the crux of why fees come out on top still stands – and always will. Costs are always deducted from returns, regardless of market environment. High fees are a continuous, persistent drag on performance.

Morningstar’s findings are confirmed by research from Vanguard, whose 2019 paper ‘The case for low-cost index-fund investing’ is probably the most concise paper I’ve read on an arguments-per-page basis for the merits of indexing. It’s only 15 pages long, but covers the major advantages of passive investing, accompanied by some great charts.

One of such charts is the one below. It show the correlations between the level of fees and ten-year annualised excess return. While it’s not presented in traditional academic terms with t-stats and R²s, it’s useful for highlighting the high-level idea that the higher the fees are, the lower future returns are.

Source: Vanguard

The 9 charts show higher fees indicating lower returns across all three styles (value, blend, and growth) and across large, mid, and small cap funds.

This also holds true for bond funds, where the effect seems even more pronounced:

Source: Vanguard

Whether you’re investing in stocks or bonds, and regardless of your style, cap bias, or duration, fees matter.

The final chart from the paper I think which is worth highlighting is on fund closures.

The ‘success rate’ used in Morningstar analysis above included only funds which both survived and went on to outperform their benchmark. The idea behind using only funds which survived for the whole period is that funds which closed – or, more likely, merged with another fund – are usually closed/merged because they underperformed.

Fund managers are less able to market an underperforming fund, so it’s much easier to merge the assets into a fund with a better track record. The flows into the new fund also make it more attractive for new investors who might have concerns about fund size or liquidity. This is one of the reasons why it pays to be cautious when looking at graphs showing outperformance.

It makes sense to exclude closed funds from the analysis from a theoretical standpoint, but the chart from Vanguard below provides the confirmatory data behind the idea that closed/merged funds are usually underperformers:

Source: Vanguard

Across all 18 categories tested, the annualised excess returns were negative before being closed.

I touched on survivorship bias in my last post, and this is why it’s so important. Potential investors only ever see funds which have survived, masking an huge invisible graveyard of funds which closed due to poor performance.

The pervasive power of fees

Lower fees correlating with better performance is such a fundamental principle in investing that it shows up almost everywhere you look.

The evidence indicates that, on an aggregate level, lower fees predict higher performance. But even at a more granular level, when digging into the different investment styles and vehicles, high fees are almost always found to be correlated to lower returns:

For example, it’s been shown that cost matters in the performance of endowments:

“Both advisory fees and other investment expenses are negatively correlated with performance. In other words, the cost of the advice does matter, and higher costs don’t necessarily translate into superior performance.”

It’s also shown up in ESG investing:

“While the majority of sustainable funds across the seven categories considered here have beaten their average traditional peer over the past 10 years, success rates have varied depending on fees.

Selecting a fund in the lowest-fee quartile five years ago would have improved the odds of picking a winner in all but one category — Global large-cap growth, where just 33% of the cheapest sustainable funds beat the cheapest traditional funds.

In some categories, for example the global and US large-cap blend categories, where sustainable funds’ success rates have the highest, the percentage of passive sustainable funds has been higher than that of traditional funds. Lower fees offered by passive funds have contributed to the higher success rates.”

And in private equity investing:

“CEM also examined returns for pension plans that ran in-house private equity portfolios, a lower cost model pioneered by large Canadian funds such as CPPIB and Omers. These outperformed CEM’s benchmark by 144bp annually between 1996 and 2018. Pension plans that used fund of PE funds, a high-cost model with additional layers of fees, underperformed CEM’s benchmark by an average of 227bp a year.”

And in foundations:

“As a group, the NHCF’s advisors and managers have delivered below average performance, at least lately. Returns on the core investment fund in NHCF’s endowment have lagged the broader markets, year in and year out, since 2009. For the five year period ending in 2017, the NHCF’s Combined Investment Fund generated average annual returns of 7.7 percent, compared with average annual returns of 8.2 percent for a 70/30 mixed of global stocks and bonds. That half a percentage point is a significant gap, representing about $3 million in 2017 alone.”

And in pension funds (commentary here):

“Over a nine-year period (2008–2016), state and municipal pension funds embarked on a grand experiment. They boosted their commitments to alternative assets, spending tens of billions of dollars per year on additional third-party money management fees.

This paper compares aggregate public pension fund returns and volatilities, over differing time periods, with a series of institutional benchmarks and replicating indexes. We conclude that the states and municipalities obtained neither lower risk nor higher returns with the higher level of active management and diversification implied by alternative assets. The experiment is thus a failure.”

No matter which way you slice it, fees matter.

Is diversification worth higher fees?

It’s clear that higher fees correlate with lower returns. But how much should investors pay for a diversifier in their portfolio? If the goal of an investment isn’t to maximise returns, but to provide uncorrelated returns with the rest of the portfolio, should investors be paying more attention to the correlations rather than the fees? Does it make sense to pay higher fees for genuine diversification?

An interesting 2018 paper in the Financial Analysts Journal tackles this question. Annoyingly it’s behind a paywall, but I’ve snipped what I thought were the most relevant and interesting parts.

The authors examine 11 diversifying asset classes, and examine how much of their diversification benefit is eroded by fees. They examine the diversification benefit to three types of institutions – small endowments, state pensions, and quality foundations. The three institutions are all charged varying levels of fees on each diversifier, depending on their size. Foundations, for example, tend to be able to negotiate lower fees than small endowments.

The key chart is below. It shows how much of the risk-adjusted incremental returns above the market are consumed by fees for each of the asset classes. A white circle represents fees having no impact on the diversification benefit, a black circle represents fees eroding 100% of the diversification benefit.

Source: Financial Analysts Journal

For the more numeric-minded, the accompanying table with underlying data is below:

Source: Financial Analysts Journal

The authors conclude:

“By comparing the incremental benefit of diversification with the incre­mental cost, we have shown that many seemingly attractive investments lose their luster as diversifiers. We have also shown that fees rearrange the relative attractiveness of many diversifying asset classes. Although it might seem obvious that diversifying asset classes have higher investment management fees, we think readers will be surprised by the mag­nitude of the problem—by how much of the diversi­fication benefit is absorbed by higher fees.”

Although all investors are in search of genuinely uncorrelated returns, the paper shows that the more you pay for such diversifiers, the less diversification benefit they’re likely to provide.

It’s useful to remember that all three types of institutions in this analysis will be being charged far less than a retail investor. For us mere mortals of the investing landscape, funds are likely to consumer far more of the diversification benefit.

It reminds me of my favourite quotes from AQR’s legendary Cliff Asness:

“There is no investment product so good gross, that there isn’t a fee that could make it bad net.”

What about cheap active funds?

Thinking back to Sharpe’s arithmetic, the main reason active funds on the whole tend to underperform their passive counterparts is higher fees. But what about cheap active funds? Are index funds just as good as equivalently priced active funds?

Enter ‘The Historical Record on Active vs. Passive Mutual Fund Performance’ from David Nanigian of San Diego State University.

The author looks at the performance of US funds between 1991 and 2019, and finds that when the performance of a value-weighted portfolio of passively managed funds is compared to that of a value-weighted portfolio of actively managed funds, the actively managed funds underperformed the passively managed funds by a statistically significant 83 basis points per year (based on Carhart Four-Factor Model alpha). However, when the sample is restricted to only those funds that rank in the bottom quintile of expense ratio, the actively managed funds underperformed their passively managed counterparts by only 19 basis points per year:

Note the lack of statistical significance on the 0.19% alpha difference for the bottom-quintile fund comparison. This is the crucial part of the paper, because it implies the difference in performance between passive funds and equivalently-priced active funds is statistically indistinguishable from zero.

In the authors own words:

“When passively managed funds are compared to these competitively priced actively managed funds, there is no statistically significant difference in performance between passively managed funds and actively managed funds. The central practical implication of this study is that, setting tax considerations aside, as long as investors are cost-conscious in their mutual fund selection process, investing in passively managed funds does not meaningfully impact the performance of one’s portfolio.”

This is consistent with Sharpe’s arithmetic – cheap active, in aggregate, is just as good as cheap passive.

It’s not the approach that matters, but fees.

Are fees all that matter?

It’s true, fees should be an important part of an investor’s decision-making process.

But it makes no sense to base a decision on fees alone. A terrible investment process can’t be transmuted into a robust one by reducing its fee. There’s no fee so low (short of the fund manager paying me) that would make me want to invest in a portfolio managed by dart-throwing monkeys.

This is a trickier problem for active investors, thanks to passive investing’s much simpler investment process (another one of the reasons it’s my preferred approach). Active management requires more analysis on the interaction between the investment process, people, philosophy, price, parent, performance, and one other thing beginning with P which I’ve forgotten. It’s much more difficult to place a value on an active manager with so many interconnected variables.

But when it comes to passive, usually – but not always – cheaper is better.

Having said that, even for passive investors fees are still subject to diminishing returns. Once fees get into the sub 20-30 basis point range, the amount paid in fees is no longer going to be one of the major levers which determines an investor’s financial outcomes.

In my post on the best Vanguard index tracker funds for UK investors, I ended by saying that once you’d made it to the decision on precisely which vehicle you’re going to use to achieve your low-cost, transparent, liquid, globally diversified equity exposure, you’ve already won the investing game. At this point, other factors such as your spending habits, housing decisions, children, and choice of spouse will have a far larger impact on your eventual portfolio size than choosing between any of the funds mentioned in the article.

This is equally true when it comes to fees.

A move from fees of 2% to 0.5% is a huge one. A move from 0.5% to 0.2% isn’t.

Of course there’s no harm in reducing fees further, and if you can, you probably should. But once you’ve made it to the sub 20-30 basis point mark, reducing fees won’t move the needle as much.

Once an investor has been successful in creating a low-cost portfolio, rather than scrabbling around for further cost savings, their time is likely better spent focussing on the other higher leverage areas.

What about active share?

One argument which seems to occasionally rear its head as a counterpoint to the idea that fees are the best predictor of future performance is that the level of active share (i.e. the extent to which a portfolio differs from its benchmark) has more predictive power.

This will be the topic of my next post, as there’s too much evidence to go into here without this post becoming excruciatingly long. But I can safely summarise here by saying that the answer is fairly conculsively: no.

Active share is not a better predictor of returns than fees. In fact, rather than being a potentially helpful statistic for investors looking to improve their portfolio, active share can oftentimes be actively harmful to investors who place too much reliance on it.

Full details in my next post.

Conclusion

There are no such thing as laws in investing. When it comes to markets, we can never share the same level of certainty as we do in Newton’s laws of motion.

The absence of counterfactuals makes proving anything impossible. We’ll never know exactly how much each causal factor contributed to the last market crash, because it’s impossible to re-run the experiment and alter each variable.

To make matters even more difficult, the environment in which we’re operating is always changing. Newton was able to prove gravity existed because the laws of physics never changed – he was able to run experiments while keeping everything else constant. But markets are always changing.

Trying to prove something in investing is like Newton trying to prove gravity exists in a world where sometimes things are pulled towards each other, sometimes they aren’t, sometimes the opposite happens, and sometimes something invisible comes out of nowhere and throws everything around a bit.

Investors can never really be sure of anything – we’re left to make the best of unprovable theories and confidence levels while navigating an environment in constant flux. But no matter how much changes in markets, no matter how many theories you choose to place confidence in, one thing will remain true regardless of approach. All else equal, lower fees will result in better performance.

And although all else isn’t always equal, both the theory and the evidence show that the best and most consistent way to increase returns is to reduce fees.

This is a powerful conclusion for investors, because while so much of what happens during our investing lifetimes is outside of our control, how much we pay for our investments is very much inside our control.

Given that the amount paid in fees is a great predictor of performance in investing, focussing on reducing fees is the most reliable way investors have to increase their odds of investing successfully.

—

PS: I’ve only included a smattering of research on the topic of fees here. It has, understandably, received a huge amount of scrutiny over the years. For those wanting to dive into the detail, the following papers also deal with the impact of fees on performance: Jensen (1968), Gruber, Das, and Hlavka (1993), Malkiel (1995), Gruber (1996), Carhart (1997), Harless and Peterson (1998), Chevalier and Ellison (1999), Wermers (2000), Elton, Gil-Bazo and Ruiz-Verdu (2009), Fama and French (2010), Ferson (2010), Musto (2011), Wermers (2011), and Cremers, Ferreira, Matos, and Starks (2016).