I’ve written a few posts about factor investing over the years.  If you’re not familiar with factors, they’re any characteristic of a company or its stock that might have a predictable relationship to that stock’s returns over time.  You can read more about the basics of factors in this post from June 2018.  Some factors have been found to have the potential for increased returns (a so-called “factor premium”) as compared to other segments of the stock market.

But don’t assume that factors are a free ride.  In most cases, factor premiums also involve greater risk in the form of increased routine volatility, and in some cases, factor funds can have higher costs than generic index funds.  Despite these tradeoffs, factor investing is extremely popular because the idea of boosting long-term returns is so seductive.

Quite a few factors have been “discovered” over the years, some of which have more merit than others for reasons I discuss more in my June 2018 post.  Most factors have been discovered by academics using pretty heady statistical methods and datasets that aren’t widely available.  So, imagine my surprise when I stumbled upon my own investing factor just the other day!

Mindfully Investing’s Factor X

I’ll describe my factor in more detail in a moment, but for now, let’s just call it “Factor X”.  I compared the long-term return of $10,000 invested in stocks that meet Factor X criteria to the S&P 500 since 1928 as shown in this graph.

While most of our investing horizons are less than the 93 years shown here, it’s still remarkable that Factor X generated almost $14 million more than investing in the S&P 500.  Expressed another way, Compound Annual Growth Rate (CAGR) or annualized returns represented in the graph are:

• S&P 500 CAGR = 9.79%
• Factor X CAGR = 10.04%

Factor X produced a long-term premium of 0.25% annualized.

Comparing to Established Factors

How does Factor X compare to other factors?  Let’s look at the two most commonly pursued factors: “size” and “value”.  The size factor exists because small-cap stocks have historically outperformed large-cap stocks.  Similarly, the value factor exists because stocks with superior value metrics (things like price-to-book ratio, price-to-earnings ratio, and price-to-free-cash-flow ratio) have outperformed so-called growth stocks with inferior value metrics.

Here’s a similar graph of $10,000 invested in small-cap, large-cap value, large-cap growth, and large-cap stocks starting in 1972, using Portfolio Visualizer data.

In this period, small-caps returned $1.25 million more than large-caps, and large-cap value returned $380,000 more than large-cap growth.  Again, that sounds pretty impressive, but let’s look at the annualized returns (CAGR) over this same timespan:

• Small-cap CAGR = 12.08%
• Large-cap value CAGR = 11.37%
• Large-cap growth CAGR =  10.91%
• Large-cap CAGR = 10.75%

Since 1972, the annualized small-cap premium has been 1.33%, and the large-cap value premium has been 0.45%.  So, the long-term Factor X premium since 1928 was about one-fifth of the small-cap premium and half of the large-cap value premium.

Factor Variability

As I already mentioned, almost no one is likely to invest for 93 years, and very few are likely to invest for 49 years.  So, these long-term CAGRs can be somewhat misleading.  In fact, the actual small-cap or large-cap value premium that any given investor receives over a shorter investing horizon is highly variable.

In this pair of graphs, you can see this variability showing the difference between the annual returns of 1) small-caps minus large-caps and 2) large-cap value minus large-cap growth since 1972.

As you can see, small-cap and value investors have suffered through many years where their “premiums” were actually negative.  Also, note that both small-caps and value stocks had very good runs in the 1970s and early 1980s but have performed erratically since then.  In fact, if we start the CAGR estimates at 1984, the resulting long-term “premiums” are negative:

• Small-cap CAGR “premium” = -0.46%
• Large-cap value CAGR “premium” = -1.42%.

Ouch!  Since 1984, large-cap investors outperformed small-cap investors, and growth investors substantially outperformed value investors.  So, the 0.25% premium from Factor X is starting to look pretty appealing.¹

Investing in Factor X

So, now that I’ve whetted your appetite for some Factor X in your own portfolio, I should tell you how it’s measured.  Similar to the graphs above, here’s the performance of $10,000 invested in 1928 in the S&P 500 as measured by three separate data sources for S&P 500 returns (Shiller, Damodaran, and JST).

I arrived at Factor X by using the Damodaran data for S&P 500 returns and using Shiller S&P 500 data to represent Factor X.  In other words, Factor X is a mirage.  The Factor X premium is just the S&P 500 measured in slightly different ways by two highly esteemed academic researchers.

Which researcher is right?  Essentially, they’re both right.  The differences in these estimates simply reflect the inevitable variability that arises when anyone tries to find, compile, and synthesize 93 years of historical stock data.

For example, these two researchers probably used slightly different constructs to represent the S&P 500, particularly for data before 1957 when the S&P 500 was invented.  And they have may found slightly different data on price growth and/or dividend amounts.  Even further, they may have used slightly different assumptions about reinvesting dividends.  Regardless, they both came up with pretty similar, but not identical, estimates in terms of long-term CAGRs.  But when those slightly different CAGRs are compounded over many years, the difference in the growth of an investment becomes huge.

What Factor X Tells Us

The first concept illustrated by Factor X is that there’s always some noise within the patterns derived from decades of historical data.  We may like to think we know exactly what happened in the past.  But in actual fact, even when looking at the S&P 500, one of the most studied and cataloged indices in the world, there’s considerable uncertainty.

The first concept leads to a second concept.  Given the noise within historical data, we should be cautious about jumping to conclusions based on the compounding of returns over many years.  If a long enough time span is used in the compounding calculation, mere mirages in the data can appear to be reality.

Now consider what these concepts mean for factors like small-cap and value.  The noise in the S&P 500 historical annualized data easily amounts to anywhere from one-fifth to one-half of the small-cap and value premiums.  And I’d wager that historical factor data are most likely far noisier than the noise I just found in the well-established S&P 500 data.

For example, there’s strong evidence to suggest that small-cap data before 1972 are almost useless because of the vagaries of how small-capitalization companies were tracked historically.  That’s one of the reasons that I presented factor statistics starting in 1972.  And of course, it’s nice to assume that all the bugs were ironed out by 1972.  But given what we’ve seen about the S&P 500, it seems implausible that small-cap data became magically pristine in the 1970s.  That view was shared by John Bogle, one of the most famous investors of the last 50 years.  He had severe doubts about the accuracy of factor data in general, which you can read more about here.

Another example is the issue of how to measure value.  It turns out that there’s more than one way to measure “value”, and the measure (or combined measures) used can have a drastic effect on the outcome of value strategies.  Price-to-book used to be the standard measure of stock value, but this measure has failed to work well recently.  You can read more about some of these details in this article at New Constructs, but I don’t share their conclusion that “improved” valuation metrics necessarily solve all the problems with value investing.

To head off potential comments, these examples don’t prove that small-cap and value factors are a fantasy.  Rather, they’re simply some of the many uncertainties that can plague the development and consistent tracking of factors.  And those additional uncertainties lead me to the conclusion that, if anything, the noise in factor data is much louder than the noise in S&P 500 data.

Put another way, the smaller the advertised factor premium, the more likely the results could be mostly a product of data noise and the less likely the premium will continue.  Based on the S&P 500 noise, I would say that any advertised factor premium that is less than 0.5% could be largely a mirage.  And as we’ve seen, over the last 37 years, the negative premiums for small-caps and value have been well below half a percent.

An Imperfect World

Ignoring for a moment the uncertainties in factor data, let’s assume that past factor data are indeed predictive of future positive long-term premiums.  In that case, the factor investor still has to wrestle with the details of implementing a factor strategy that will reap the expected premium in the future.  Here are some examples of the realities of investing in an imperfect world that could erode that factor premium.

Fund Costs – You can easily invest in a low-cost index fund tracking the S&P 500 at rock-bottom prices.  For example, Vanguard’s S&P 500 exchange-traded fund has an annual cost of 0.03%.  The same brand of small-cap and value funds cost 0.05% and 0.04%, respectively.  That’s not a lot of erosion.  But you might instead be attracted to a different brand of factor fund.  For example, I keep reading rave reviews claiming a superior track record for Dimensional Fund’s factor offerings.  The costs of Dimensional’s small-cap and value funds are both 0.34%.  That represents pretty significant erosion when we’re talking about factor premiums that under ideal conditions are only in the 0.5% to 1.3% range.²

Adviser and Plan Costs – Or perhaps you’re pursuing factors because a financial adviser suggested it.  In 2019, the average total cost for a retail investment adviser was 1.17%.  Alternatively, perhaps you’re investing through your company’s 401K plan, the average cost of which ranged between 0.37% to 1.42% in another 2019 study.  Either of these costs is enough to mostly or entirely wipe out the small-cap or value premium you might hope to gain.  Maybe that’s one of the reasons advisers seem to like factors so much.  It’s one of the few ways they might potentially offset their own cost erosion.

Reinvesting Dividends and Taxes – All these statistics assume that dividends are reinvested regularly.  But what if you’re not completely assiduous about how and when you reinvest?  Or what if you occasionally use some dividends for something else?  And what if you’re dividends are taxed because you’re not working in a tax-deferred account?  All of this will help to slowly but surely erode your returns as compared to the ideal situations assumed by these factor statistics.

Consistency – Finally, what if you take a small amount out of your 401K for an emergency, and then put it back a year later?  Or even worse, what if you panicked during the pandemic crash in March 2020, sold your factor funds, and then bought them back again in December?  Any one of these maneuvers, by itself, could ruin any expected premium for years to come.

Beyond all that, recall that small-cap and value factors produced negative premiums over the last 37 years.  So, there’s the ultimate issue that the factor premium may fail to materialize regardless of how carefully you invest.

I hear some readers complaining that all these issues apply equally to investing in the S&P 500 and other broad-market portfolios.  That’s absolutely true.  But how wise is it to pursue a level of premium that can easily be eroded away by implementation realities?  Why not simply pursue a cost/tax minimization and consistency premium instead?  Such an approach certainly gives you greater personal control over your returns performance than historically fickle factors.

Conclusions

This is not the first post that I’ve written where I appear to pound on factor investors, but that’s not my intent.  I’m really only using factors here as an example of a larger issue.  The same data and implementation uncertainties apply to almost any strategy that’s intended to generate a premium (with or without greater risks) over super-simple, broad-market, and low-cost investing.

That’s because we can only measure the past with a certain degree of accuracy.  And since that’s true, why would we assume we can predict the future with even better accuracy by favoring certain factors, market segments, or even individual companies?  If historical factor or other premiums were consistently reported in the 2% to 5% range, then I’d say the data and implementation noise is too small to be relevant.  But as we’ve seen with the small-cap and value examples, that’s usually not the case.  I prefer to minimize costs/taxes, invest/reinvest with diligent consistency, and watch my boring broad-market index funds meet all my investing goals over the long term.

---

1 – Some readers may protest that this statement is comparing apples to oranges given that the Factor X premium was calculated since 1928, while small-cap and value premiums were calculated since 1984.  So, for the record, the Factor X premium since 1984 was 0.14%.  But that doesn’t matter for reasons that will become clear if you keep reading.

2 – Further, I used Portfolio Visualizer to compare CAGRs for Dimensional’s (DFSTX) and Vanguard’s (NAESX) small-cap mutual funds going back to 1992, which marks the inception of DFSTX.  It turns out that the CAGR for Dimensional was 10.79% as compared to Vanguard’s 10.74%.  Portfolio Visualizer statistics include fund costs.  So, this shows that, at least over this period, higher costs do not necessarily guarantee a substantially better return.