People have often asked me what I think about value and small-cap equity portfolios. So, this is a post I always wanted to write but kept postponing because I never knew how to best frame it. But now I have the perfect excuse to write it; last week, I listened to the ChooseFI podcast and they had Paul Merriman as a guest in episode 130. Paul Merriman is one of the big proponents of small-cap and value stocks. Of course, they talked about a variety of topics and I thoroughly enjoyed most of the discussion. I’m completely on the same page with Paul, Jonathan and Brad on a wide range of issues. For example:
- Choose index funds over actively managed funds
- Take emotions, especially fear and greed out of investment decisions
- Young investors benefit from Dollar Cost Averaging. Specifically, a market crash early in your investing life can even be beneficial, at least under specific assumptions, as I wrote earlier this year in “How can a drop in the stock market possibly be good for investors?“
- Young investors shouldn’t even think about bonds in the portfolio when starting out. Use 100% equities, use as much risk as possible.
But there’s one thing I vehemently disagreed with! Paul Merriman seems to suggest that by using a “better” or “smarter” equity allocation, specifically, overweighting the value and small-cap styles – both domestically and internationally – we can increase our expected return by two full percentage points a year. And just to be sure, this is not stock picking but simply asset class/style picking, all implemented with passive ETFs. Two percentage points of extra return? Let that sink in! 2% a year can compound to a large sum over the years and then you retire and you get extra 50% of withdrawals when you add 2 percentage points to your 4% safe withdrawal rate. Pretty impressive! There’s only one problem: Merriman’s recommended portfolio didn’t return those two extra percentage points over the past few decades. And there are good reasons to believe that you will not gather those 2% extra returns going forward either. Let me explain why…
Side note 1: Brad and Jonathan are obviously good buddies of mine. Rest assured that I sent them a courtesy email to give them the heads-up before publishing this. So there should be no hard feelings (I hope) from their side.
Side note 2 (August 2, 2018): The ERN blog was nominated for the 10th Annual Plutus Awards in two categories: Best Financial Independence/Early Retirement Blog and Best Series: Blog, Podcast, or Video (specifically, the Safe Withdrawal Rate Series). Thanks to the growing community of all the amazing readers here who nominated my small blog!
Background: The value and small-cap factors 1926-2019
Value stocks and small-cap stocks have been studied in finance for a long time. The most famous research comes compliments of Ken French and Eugene Fama. Actually, Fama’s contributions to the Econ/Finance field were so substantial that he was one of the 2013 Economics Nobel Prize winners. Along with Robert Shiller (yes, that Shiller who came up with the CAPE ratio) and Lars Peter Hansen, a fellow Minnesota Econ Ph.D., by the way!
In any case, back to the Fama-French Factors, I looked them up on Prof. French’s website, and here’s the cumulative return of the size and value style premium, see below. Wow! Small-cap and Value did indeed have a wonderful outperformance run over the last 90+ years. And remember, this would have been the performance over and on top (!!!) of your equity index return. We’re now looking at some astronomical outperformance! No wonder this is so popular!!!
But notice something else in the cumulative return chart? The performance has started to sputter a little bit over the last few decades. The size premium has been flat since the early 1980s. The Value premium started a nasty decline in December 2006. So, let’s look at the return stats over the entire history and then also since 1984 (the peak of the SML premium) and 2006 (the peak of the HMP premium), see the table below.
Over a long history, the two styles provided significant excess returns. The t-stats for the mean return were 2.15 and 3.44, respectively and the annualized CAGR excess returns were 1.89% and 3.68%, respectively. But this all looks a lot less impressive now. Small-caps were flat, even slightly down since 1984 and even though Value still performed well, its mean return is no longer statistically significant. Since 2006, Value got hammered (though still not significantly!) and Small-caps were very slightly positive. In light of the recent performance, I’d be cautious about taking the mean style premium results over the entire 91+ year period and extrapolating them forward.
Side note: I’m aware of the irony. In my post a few weeks ago I pointed out the dangers of looking at short horizons and picking convenient start/end points. But to every rule, there’s an exception. I think here we have an example of a structural break in the data where you want to be cautious about keeping the old and irrelevant data in the sample.
How did the Merriman Ultimate Portfolio perform?
Given the underwhelming recent performance of the two styles, I wasn’t expecting much. Well, checking out Merriman’s website I was hoping for at least some numbers. But I couldn’t even find any detailed simulation results. Strange! So, if someone could help me out and find a Merriman-approved return simulation with industry-standard return stat tables please let me know. Until then, I’ll have to use what I found on the web. Here’s a site “Portfolio Einstein” that put out 1-year, 10-year and 30-year simulated returns (all windows ending in 2018). But the Ultimate and Best-in-class portfolios were really not that great. They underperformed the simple passive benchmark over all three horizons. And this writer seems to be a fan of Merriman and writes a glowing review about this approach despite the underperformance, so it doesn’t look like this site wants to make the returns look bad…
Update 6/13/2019: As someone pointed out, the underperformance of the current Merriman portfolio relative to the S&P 500 index comes mainly because of the different country allocation. If we compare the Merriman portfolios to benchmarks with 50% U.S., 50% non-US or 40% US total market or alternatively – even closer to the 10-ETF portfolio – 40% US Total Market, 40% non-US-developed, 10% US REITs, 10% Emerging then the Merriman portfolio is still much worse over the last year (2018), about on par over 10 years but much better than the US+international benchmark over 30 years. But still: the outperformance came only during the early period.
Where does the 2% excess return number come from?
Honestly, I don’t know. It’s the first time I ever see an actual hard number from the small-cap-value crowd. Again, maybe it’s information overload with so much material on the Merriman website and I didn’t find the exact source and careful, understandable and replicable calculation. Please provide a link to the calculation if you know where the number came from.
Until then I can only speculate. The source of the 2% figure could be this page where they introduce the new March 2019 Merriman recommendation, also mentioned in the podcast. If you scroll to the end of the page you can spot the table with a “Factor Analysis” and the 10-fund ETF portfolio has estimated factor betas: 0.37 for small-cap and 0.23 for Value. If we apply the SMB and HML average excess returns since 1926 (1.89% and 3.68%) and multiply with the appropriate factor loadings (0.37 and 0.23) then you get 1.89%x0.37+3.68%x0.23=1.5457%. The portfolio also had a higher than 1.00 market beta (1.08, to be precise), so if we add another 0.08 times the equity risk premium (say, 5%) you’re right there at around 1.95%.
But this calculation is really, really troublesome. I don’t think that even Fama and French would argue that the 90+ year average return is to be blindly applied going forward, almost like a physics constant. There is a very active discussion in finance, both in academia and among practitioners, about whether the size and value abnormalities have disappeared or not, but even the staunchest supporters of the idea that the premia haven’t disappeared would concede that you’ll no longer earn 3.68% from value over growth going forward. So, if anyone extrapolates the long-term average alphas my “bullshit alarm” goes off. But again, maybe there’s a good explanation for the 2% excess returns.
Explanations for the style return pattern
What explains the recent underperformance of the Fama-French factors and the Merriman portfolio? Here are two possible explanations:
- Bad luck. This is obviously the explanation that Merriman and his crew hope for. Over short horizons, but even over several decades, a bona fide alpha source can have a streak of bad luck. If this is the case, don’t worry about the poor performance over the last 30 years. A return to consistent outperformance in the value and size premium should be right around the corner!
- The party is over. There was a market anomaly that created the excess returns for both small-cap and value stocks during the early years. But this anomaly has now been detected, it has been studied in detail, it has been widely publicized in both academic and practitioner journals and it’s now even widely known to the average retail investor. Thus, the arbitrage opportunity, in the best case, has been watered down or, in the worst case, has been completely arbitraged away. Curb your expectations going forward!
Which one of the two explanations do I find more compelling? Merriman disciples will hate me for it but I’m afraid I’m leaning toward option 2. The small-cap value party is likely over. Now, I am not trying to say that SCV will underperform going forward. I am merely saying that a 2% excess return seems unrealistic. Going forward, I’d still expect a little bit of extra return, just a lot less than the historical averages – more on that later!
Why the style premium party might be over (or at least a lot less profitable going forward)
A lot of institutional money has been flowing into exactly the styles Merriman favors. And all the other styles that are en vogue right now: momentum, quality over junk, low volatility, minimum volatility, low beta over high beta, just to name a few. Once these style premia go mainstream and there’s an ETF for it (plus hundreds of billions of dollars of institutional money) some or most or even all of the style alpha will be arbitraged away. It’s how efficient markets work. So, I just have trouble rationalizing that a 2% extra return from a completely static/strategic bias to passive indexes will persist forever.
More on the size premium
Imagine you had known about the size premium before Eugene Fama and Ken French published their path-breaking research. You’d now kick yourself that Eugene Fama “stole” your Nobel Prize 🙂 but at least you would have made a lot of money, right? Well, maybe not! It would have been hard to “harvest” this premium. Remember, before the advent of ETFs and internet trading platforms you’d call up your broker on the phone. Mutual funds were available but often had loads and high expense ratios. How exactly would a retail investor have constructed a portfolio replicating the Russell 2000 index, a popular small-cap benchmark with 2000 constituents? Hey, I have $1,000 to invest this month, can you please buy the 2000 stocks in the index? Hard to do as an average Joe retail investor.
Side note 1: Actually, before the famous Fama-French work, Rolf Banz is credited with pointing out the size premium for the first time (to my knowledge) in 1981. And right after the publication, the whole thing stopped working. A coincidence?
Side note 2: The comment that Fama stole “your Nobel Prize” was tongue-in-cheek. In fact, Fama didn’t even win the Prize for the Fama-French research but for his contributions on the “Efficient Market Hypothesis”
Institutional investors would have had fewer technical constraints than the average retail investor harvesting the size premium. But many had compliance/legal/regulatory constraints. Endowments and pension funds were often constrained about how exotic they could go in their domestic equity allocation. Right around the time when some of those constraints abated and the market became less segmented is when the size premium lost its luster. Coincidence? You be the judge!
But just for the record, I certainly concede that even today there is a justification for at least some size premium due to less transparency, less analyst coverage, higher macroeconomic risk (small stocks often get hammered more during recessions/bear markets), etc. I simply don’t believe that the size premium can be as large as it has been historically. More on that later!
Update 6/14/2019: Someone pointed out – quite correctly – that the 20-year return of small-cap looks really good. The reason is easy to spot, see that chart below. Small-caps initially underperformed during the dot-com bust, but then didn’t drop as badly. But over the last 15 years, the two indexes tracked each other extremely well again, including the Global Financial Crisis.
More on the value premium
One thread going through Merriman’s media appearances and especially the ChooseFI podcast was the idea that value stocks deserve a premium because they also involve more risk. More risk means more expected return. Every investor knows that, right? Of course, the truth is a little bit more nuanced. Shorting a stock market index has the exact same standard deviation but much lower expected returns than being long the index. So, the expected return is not just tied to risk but also when the big drawdowns happen. (side note for the finance geeks: for expected returns, it’s not so much the risk but the market beta that matters for expected returns)
Historically, the value over growth premium had very sharp drawdowns right around the big recessions and bear markets. Look at the cumulative return chart above; value stocks underperformed growth stocks during the Great Depression, during the 1970s, 1980s recessions, and 1991 recession. So, because value stocks underperform when it hurts the most, one can justify the extra return. So, just for the record, this return pattern most definitely justifies a return premium!
It even looks like the value over growth style continued this pattern during the dot-com bust and the Great Recession. But nothing could be farther from the truth! Value stocks underperformed before the year 2000 market peak by a whopping 44%. Actually, the worst underperformance of the Value style happened during the craziest part of the internet bubble! And then Value rallied again by 80% during the dot-com bear market, so the whole risk premium story goes out of the window. It’s the other way around: value stocks were a hedge against the 2001-2003 bear market! Instead of demanding a risk premium you should pay an insurance premium.
The experience during the Great Recession (Dec 2007 – June 2009) and the associated bear market (October 2007 to February 2009, when using month-end data) is also very atypical. Sure, you lost 18% during the 2007-09 bear market but the decline of the Fama-French HML factor was a long slog, a decline of 36% since the peak that spanned an entire business cycle (the late expansion before the recession, the recession and bear market and 10+ years of recovery. Very different from the return patterns before.
So, in other words, if this fundamental “more risk, more return” story was the explanation for the excess returns it went up in smoke over the last two major market events.
Another problem with the value/growth classification: sector biases
I looked at the sector weights of the iShares Core S&P U.S. Value ETF (IUSV) and the iShares Core S&P U.S. Growth ETF (IUSG) and noticed that value and growth stocks have wildly different sector weights. Value stocks have a large overweight to financials and energy but much smaller weights in IT, Health Care, Consumer Discretionary and Communications. Why does this matter? If we believe that we’re in the middle of another industrial revolution that will boost certain sectors and threaten the decline – even demise – of others then it becomes dangerous pinning your hopes on out-of-favor sectors and industries coming back in favor.
Notice how this sector story also points to a potential “solution” for how to make the value over growth premium “work” again. I would certainly like to invest in a value fund if it didn’t have those sector biases. So, instead of looking for low price-to-book stocks over the entire universe, why not construct an ETF that has the exact same sector weights as the index, but then within each sector (and maybe even subsector) you do the value bias and buy the out-of-favor stocks measured by the P/B ratio. And you do it for the exact reason that Merriman uses. So, don’t get me wrong: my core investing philosophy is actually close to Merriman’s, i.e., relying on value and valuation. I just believe that the reversion to the mean argument, i.e., out-of-favor stocks going back in favor, is a lot easier to make within sectors than across sectors!
What would be a reasonable excess return estimate for Merriman’s Ultimate Portfolio?
There is a saying in economics (and I’m sure in many other fields as well): “it takes a model to beat a model!” So, I can’t just criticize others and walk away. I certainly want to understand how much extra return you can expect to make from implementing the Merriman portfolio tilts. In other words, if we start with a portfolio comprised of 40% U.S. Total Market, 10% U.S. REITs, 40% ex-U.S. developed and 10% Emerging Markets without any size or value tilts and we move to the Merriman 10-fund portfolio with all the size/value mumbo jumbo, how much can I expect in extra returns?
- Size: BNYM uses a premium of 0.6% p.a. of small stocks over the total market (6.9% Russell 2000 over 6.3% Russell 3000). JPM doesn’t report a separate category for the Total Stock Market but if we weight the values for large/med/small (5.25%, 5.50%, 6.00%) and use a 5.4% blended return for the overall market then we get – miraculously – the same 6.00%-5.40%=0.60% premium of small-cap over the total stock market. Also, quite intriguingly, BNYM reports separate figures for MSCI World ex U.S., both the broad market and small-cap and estimates zero small-cap premium there!
- Value: BNYM doesn’t report an expected return for value stocks. JPM uses 6.00% for value stocks, which translates into the same 0.60% (=6%-5.4%) value premium over the total market (which then implies 1.20% for value-over-growth).
Well, these are quite a bit smaller than the 90+ year averages from the Fama-French data! Of course, I can already hear the Merriman disciples complain: these are the evil Wall Streeters. But why would Wall Street try to hide the small/value premium? If anything, wouldn’t they have an incentive to overstate those style premia? After all, they charge higher fees managing the more exotic styles. Well, in any case, if we now assume 0.6% extra return from small-cap and value tilts, what kind of an impact would that have on the Merriman portfolio? That’s what I do in the table below. And that’s assuming I can translate the same small-cap alpha to international stocks even though BNYM assumes a premium only for domestic stocks but I’m feeling generous today.
So, I get a 0.54% extra expected return per year. But that portfolio would also have a 0.28% expense ratio. I could construct a 40% U.S. Total Market portfolio plus 50% world ex-U.S. (both developed and emerging) using the zero-expense Fidelity Funds. The 10% allocation to Vanguard’s VNQ would contribute 0.012% to the overall expense ratio (12 basis points times 10% weight), so we would cut in half the excess return to about 0.27%. Not quite the 2% that Merriman promised. With a quarter percentage point annual alpha, this small-cap and value stuff makes it onto my maybe pile. At most!
You want the value premium to work again? Careful what you wish for!
You know what it would take to bring doubters like me on board again? During the next (!) recession and bear market, value stocks get totally clobbered! Just as Paul Merriman said on the podcast: More risk implies more return. Value stocks are more exposed to macroeconomic shocks. But do you see the irony in that? In order for value to be considered a bona fide risk premium again, we’d have to see a big drawdown to undo the damage from that strange behavior of the value premium during the last two recessions. Then why would I want to invest in value stocks now? Wouldn’t it be safer to wait until after the next recession and bear market? Why would I want to take that risk? After a 10+ years bull market and the longest macroeconomic expansion on record, the next recession plus bear market may not be so far away, so why should I be in a rush and take the risk? For a measly 0.27% annual alpha? I got time on my side and I’ll decide whether and when to dip my toes in value stocks right around the time the total stock market is down 30%+!
Let’s all become market timers?!
Here’s another irony in the whole Value stock discussion. Notwithstanding the fact that I don’t see much of a reason to strategically shift into value stocks (yet), I totally admit that there’s a case to tactically shift into value stocks. What do I mean by that? Your strategic asset allocation is the very long-term target allocation, while the tactical allocation would be the short to medium-term shifts in the allocation due to market timing. And thanks to the very long drawdown in the value premium, value stocks now look really cheap. In other words, forget the very long-term return assumptions that the strategic allocators would use (see the BNY Mellon and JP Morgan numbers above) but instead use the estimates that the short- to medium-term asset allocators would use – the market timers. This expected return gap between value and growth is indeed quite large today, compared to history. So, just from a purely short- to medium-term perspective, there would be a rationale to dip your toes in value stocks because the valuation of value stocks (pun intended) looks attractive now. But you better make sure that you get out again before the next bear market, see the point I made above. But I’m not sure Merriman and his followers will like this rationale. Remember, it involves market timing (!), which he considers a big no-no!
I’ve encountered several Merriman followers while blogging here and they are certainly emotional about this approach. I don’t think I wrote anything too controversial today but I’m definitely bracing for a bit of a sh!tstorm from the small-cap-value crowd. Some will call me one an evil, dishonest Wall Streeter. Some might call me one of those dogmatic VTSAX investors, which I’m certainly not. Quite the opposite, I’m pretty open-minded; 35% of our portfolio is in my options trading strategy and 10% in real estate. If there’s an opportunity to beat the market, I’ll be all ears!
Both sides, the totally passive equity indexers (much of the FI community, Bogelheads, etc.) and the small-cap-value crowd will probably not agree on anything anytime soon. That’s because we have different portfolios to start with. And being in this “statistical limbo state” where I don’t see enough evidence to start a Merriman Ultimate Portfolio and they don’t see enough evidence to dump their Merriman Ultimate Portfolio we’ll probably coexist for a while. Deal with it! But I like to pose a question to the Merriman crowd:
What would it take for you to change your portfolio back to just the passive broad index ETFs?
It’s a question that Jonathan and Brad maybe should have asked Paul on the podcast. I told you what it would take me to change my mind: I’d like to see a sector-neutral value fund (haven’t found that one yet, please let me know if you know of one). I’ll shift to small-cap and value (and even small-cap-value) after those styles get clobbered in the next bear market. And then, hopefully, I’ll make a ton of money in the bull market that follows. But my question for the people on the other side of the argument: How many more years of underperformance can you take? 5 years? 10 years? 30 years? Have you ever even thought about that question? It’s what academics do, right? They try to challenge their theories not look for confirmation. So, that’s what I want the style premium investors to think about: What would it take for you to throw in the towel? Looking forward to the comments on this one!
Hope you enjoyed today’s post! We’re traveling right now and might be late replying, but I will read every single comment here and I will reply to every comment. But please keep it civil! 🙂
Featured Image from Pixabay.com