Old myths are hard to kill. Good old myths are nearly impossible to kill. Good old myths with elements of pseudo credibility are like crabgrass in a well-manicured lawn: you can count on it to make an attempt at comeback every season. Think of Area 51 or the conspiracy theories about the faking of the moon landing.
Proponents of those myths tend to grab to isolated facts that, on their own (and in a vacuum), could be correct but when looked in the complete context they just don’t support the myth.
For instance, defenders of the lunar landing hoax claim that the picture of the flag on the moon is proof of their theory (how can it be waived if there is no atmosphere), however, when you read the explanation regarding the malfunctioning telescoping rod, everything makes sense.
What is more likely: a malfunctioning piece of equipment after having suffered hours of extended G-forces or 100,000 people across the space program being part of a massive hoax? Occam’s razor supports the former and not the latter as a plausible explanation, and so do we.
In factor land, one ugly myth, like a zombie that refuses to die in a B Hollywood movie, is that the market capitalisation-weighted benchmark is a good momentum bet. Academics and practitioners have written many articles and notes trying to dispel this notion… and yet, every few years the myth resurfaces – particularly after momentum has had a good run.
One crucial element that can add to the cyclical revival of this question is that, in good times, people like to point out the momentum in the benchmark, which is very different from the benchmark being a good vehicle for momentum.
In this short research note, we will contribute to the efforts done before to discredit this myth, and hopefully add to the growing body of work that is trying to cut as many heads as possible from this hydra.
Momentum in the benchmark
Let us take a few minutes to remind ourselves what is momentum. We know that momentum is a physics term that refers to the quantity of motion in an object, defined as “mass in motion”. In finance, certain stock characteristics exhibit momentum. While the price is the most commonly used variable (specifically using price returns over the past 12 months skipping the most recent month), there are several other fundamental characteristics that exhibit momentum. In this note, we will focus on the momentum factor that goes long or overweight those stocks with recent positive returns and short those with negative stock returns.
Why does this myth persist? Proponents say that as the larger companies in a cap-weighted portfolio post positive returns, their market cap will grow and that will in term increase their weight in the benchmark. This, they say, makes the benchmark a momentum bet. While you might be tempted to agree, it is important to remember that the benchmark is a hodgepodge of factor exposures, having all and none at the same time.
The analysis below shows that since 2002 the benchmark has positive and negative exposure (factor betas) to most popular factors, and the momentum exposure, although positive in 2018 and 2019, is actually barely negative (-0.04) when measuring the average exposure since 2002. It stands then that since 2002 the benchmark has not been a good momentum bet.
Source: FactorResearch
We could rest our case here, but we will dig further. Sceptics could point to the fact that over certain periods, the benchmark did provide positive exposure to momentum, however, why not then make the case that the benchmark is a good low volatility or quality bet? After all, the average exposures to those two factors since 2002 are 0.19 and 0.15, respectively.
For comparison sake, let us look at similar exposures from the largest ETF (MTUM) by assets under management ($10 billion) designed specifically to provide exposure to momentum. The analysis below has these exposures. The data looks smoother because it goes back only to 2014 when this ETF went live.
It is readily noticeable that momentum is by far the largest exposure in most periods, and in fact, the average exposure for momentum is the largest for all factors at 0.31, followed by low volatility (0.22), barely positive for quality (0.06) and consistently negative for value and size (-0.17 each). Paraphrasing our original question: is this momentum ETF a good momentum bet? Our answer: You bet.
Source: FactorResearch
So far, we established two things: a cap-weighted benchmark is not a good momentum bet, and a momentum ETF is a decent momentum bet. Now we look at what is the best possible momentum bet.
To motivate our focus on momentum, we will start by looking at popular ETFs that provide exposure to MSCI factor benchmarks. The analysis below shows the relative performance with respect to the S&P 500 since these are all US-focused. Momentum is clearly the only factor that has delivered in the US relative to the S&P since 2013. We need to caveat these results because we used data as of July 10th. Had we run this analysis a month ago, the relative year-to-date return for momentum drops from 12.2% to barely 1%.
Source: FactorResearch
Some readers may not yet be convinced about the main premise here, so let’s put another nail on the coffin. The analysis below shows the return contribution of the top 25 holdings, by market cap, year-to-date in 2020 of the momentum and market ETFs. For the largest three holdings in each portfolio, results are very close in terms of contribution to returns.
However, after the fourth-largest holding, wheels come off the wagon for the benchmark. We will concede that it could be the case that proponents of the myth focused on the behaviour of the largest three positions and assumed incorrectly this pattern was common to all holdings.
Source: FactorResearch
Getting momentum exposure
We move to the final question, namely, what is the best vehicle to obtain momentum exposure. Long-only smart beta ETFs are limited by construction to expressing negative momentum views to the extent that stocks can be underweighted or simply not held. Every purist quant knows the most efficient way to construct a factor portfolio involves removing the long-only constraint so that short positions can be included.
Institutional investors can buy long-short portfolios packaged as indices from investment banks, however, there are few options for most other investors. There is one ETF (MOM) that provides long-short momentum exposure, but investors have allocated less than $10 million to the product.
It is interesting to note that the performance of the long-only smart beta ETF on a relative basis was approximately comparable to the performance of the long-short market-neutral momentum factor. However, occasionally large discrepancies appear when stock markets crash as beta-hedging and rebalancing are not conducted on a daily basis.
Source: Kenneth R. French data library, FactorResearch
Further Thoughts
Another myth that needs busting is that momentum and growth stocks are the same. The financial media uses these terms synonymously, which makes every quant cringe as these factors are defined differently and portfolios only occasionally overlap.
Even worse, investors have allocated approximately $200 billion to smart beta growth ETFs in the US, almost the same as for value ETFs, compared to less than $15 billion to momentum products. Given that there is almost no empirical evidence that growth is a factor that generates positive excess returns over time while plenty of research supports momentum, this is concerning.
Rodolfo Martell is a former managing director at AQR in the Global Stock Selection Group, and has also held roles at QMA and BlackRock
Sign up to ETF Stream’s weekly email here
