September 1, 2014

Does the Small Cap Value Premium Exist?

Does the small-value premium exist? Do you believe in Santa? The tooth fairy? All important questions, but I'm focusing on the first one for now. Fama and French famously found that value stocks generally outperform growth stocks and that small companies tend to do better than larger ones. One the second point about size, I agree to a degree. If we're looking at market cap broken out by decile, the largest companies have pretty consistently underperformed their smaller cousins. They're mostly below average decade-after-decade as you can see in the chart below. Even in decades when large and mega cap stocks (roughly the top three deciles) had above average performance, companies that fall into the mid cap range sometimes performed better. For example, the 80s and 90s are often thought as somewhat of a golden age for large caps. But investors would have done a bit better in the 80s focusing on the middle of the market. In the 90s and waaaayyy back in the last part of the 20s, large caps outperformed more convincingly.

Average stock returns by decade, broken out by market cap.

But across the course of 87 years, investors would have been well-served mostly ignoring the top end of the market. You can see a lot more green above in the lower half of market cap size. Looking at size alone though (and ignoring value vs. growth), I can't agree with the notion that small always outperforms large. Micro caps have had some long stretches of weak performance, as have other deciles within the small cap range. The erratic behavior at the smaller end of the spectrum would have resulted in lower total returns than mid caps over the past 30 years.
Over the past 20, it's roughly a draw between small and mid. Just looking at the last decade, the lowest couple of deciles were some of the worst performers (outside of mega caps):

Erratic stock returns of smaller companies make the size premium hard to justify.

In short, the size premium isn't convincing once you look past the very largest stocks. But what about when you add growth and value factors into the mix? Specifically, we're talking about the BE/ME (book-to-market) ratio as a measure of value. The higher the book value of a company is in relation to its market value, the more it falls on the value end of the spectrum.

Using Ken French's data looking at size quintiles against value quintiles, we get 25 different portfolios with which to examine if the small value premium exists. I previously stated I was pretty unconvinced that value does any better than growth when it comes to large caps, referencing evidence from John C. Bogle spanning 63 years that shows value as a whole outperforming growth by a mere 0.1%.

However, I believe the underlying data used to draw the above conclusion include P/E as a measurement of value in addition to P/B. French-Fama look at P/B only. Using P/B as the only measurement of growth vs. value, and breaking average and median returns out by market cap quintile, we get:

Evidence of small value premium: Comparing French-Fama factors across market cap and growth/value

Over the course of the 87-year data set, there is evidence contrary to what I previously stated at the large cap end of the spectrum. The quintile of large cap stocks with the lowest (value) P/B ratio outperformed the highest (growth) quintile by over 3.3% on average!

As we go from right-to-left, that gap grows for each market cap segment with small value more than doubling the returns of small growth on average! This holds true decade-by-decade for smaller companies. The most value-heavy quintile is almost always the best performer.

The problem is figuring out how to best capture this value premium. While French-Fama used book value in relation to price as the only measurement of value, many fund families use entirely different formulas. Dimensional Fund Advisors stick with this form of measurement and the Russell indexes (available as iShares ETFs) use low P/B as 50% of their criteria. But others factor P/E, cash flow, dividends, sales-to-price, and forward earnings metrics much more heavily - some to the point of making P/B a quarter or less of the overall equation.

This may explain why in my comparison of Russell 1000 Growth and Value ETFs (IWF and IWD), the two have ended up with identical returns in more than a decade. While the value variant did far better when you consider the few years before that include the aftermath dot-com bubble, there are no guarantees that we'll see another bubble that will behave the same way. After all, during the run-up of value stocks during the housing bubble, they came crashing back down to growth stock levels in the Great Recession and the two have moved mostly in lockstep ever since.

The Russell 2000 small cap ETFs (IWN for value, IWO for growth) look much the same. Value crushed growth from July of 2000 to August, 2014. For the past 12 years though, the two have had nearly identical total returns, becoming even more closely linked in the past 5.

For the value indexes that do use P/B as a primary criteria, I'm not sure if it's companies that fall into an extreme end of value (like the quintiles above), or just the top/bottom halves, for example (meaning the difference between value and growth would be lower than we see above). And unfortunately, Dimensional Fund Advisors appears to require investors to work with a financial advisor to invest in their US Small Cap Value Portfolio (DFSVX) mutual fund. An advisor charging 1-2% of assets, plus a 0.52% expense ratio would eat into any additional premium and, in fact, the iShares Russell 2000 ETF (IWM) has performed equally to DFSVX over the past 13 years (again, post dot-com bubble).

In short, the small value premium appears to be quite real, but difficult to capture in practice through traditional ETFs and mutual funds. There is clearly a huge overlap in both growth and value funds in many fund families, so most appear to be defining growth or value by picking stocks that are above or below average in chosen metrics (not lower or upper quintiles at each end of the scale). We'll look at individual small cap value ETFs later to see if any stand out as doing a better job capturing the small value premium.

No comments:

Post a Comment