Monday, July 14, 2014

Does the Value Premium Exist?

Many investors take it as fact that small-cap stocks earn higher long-term returns than large-cap stocks, and that value stocks earn higher long-term returns than growth stocks. This belief originates with work by Fama and French on their 3-factor model of stock returns. The evidence that a small-cap premium exists is compelling, but not so for the value premium. Whether or not these premiums exist affects the choice of ETFs for an indexed portfolio.

John Bogle discusses growth vs. value stocks in the final third of an excellent speech in 2001. In it he observes that the Fama and French conclusions are based on stock market data from 1963 to 1990. It turns out that the size and even the existence of a value premium is period dependent.

Bogle extended the period of study to 1937-2000 and found the average annual compound returns to be 11.8% for growth stocks and 11.9% for value stocks, hardly a significant edge. You may ask what happened from 2001 to 2013. According to Standard and Poors, value stocks won by a little over half a percent per year. So, there was no statistically significant value premium in U.S. stocks from 1937 to 2013.

There certainly were periods where either value or growth stocks had a large advantage. However, unless an investor can predict whether a given future period will favour growth or value stocks, these swings are of no use.

Some might object that investors like Warren Buffett proved that value stocks are superior. This is a different situation altogether. Bogle, Fama, and French are comparing all value stocks to all growth stocks. But Buffett sought to own only those stocks he believed would outperform. Value investing through stock-picking is very different from an index strategy where the investor chooses to own all value stocks.

Implications for index ETF portfolios

In my own portfolio, I’ve chosen not to tilt toward either value stocks or growth stocks. I do have a modest tilt to small-cap stocks over large-cap stocks. I prefer Vanguard Canada’s fund VCN over the iShares fund XIU because VCN contains some small-cap stocks.

In the U.S., I prefer Vanguard’s VTI, which represents the entire U.S. stock market, over funds that hold just the S&P 500. In addition, I own some of Vanguard’s VB fund, which holds U.S. small caps, to add an extra small-cap tilt.

As always, I don’t think anyone should follow my choices blindly; think for yourself. That said, hopefully this discussion of how I bet my own money carries more weight than some pundit’s “investing ideas.” Comments and constructive criticism are most welcome. There is too much at stake when investing to ignore useful ideas.

14 comments:

  1. is it a typo or I miss something?
    1. " I prefer Vanguard Canada’s fund VCN over the iShares fund XIU because VCN contains some small-cap stocks."
    2. I own some of Vanguard’s VB fund, which holds U.S. small caps, to add an extra small-cap tilt.

    So, do you add small caps or avoid them?

    ReplyDelete
    Replies
    1. @AnatoliN: I've added small caps to give my portfolio a small-cap tilt. What I haven't added is a value tilt.

      Delete
  2. First of all, I have great respect for Jack Bogle. Secondly, a total market index approach is a good investing strategy.

    Nevertheless, I'm now a value investor, although I used to use a total market index approach. I haven't looked at the speech link you provided. When others have presented it, it looks like Jack Bogle compared active growth funds to active value funds, and not growth stocks to value stocks. So it's relevant for those using active funds, but otherwise not.

    Why am I a value investor? To sum it up in one word, bubbles. Index investing tends to invest more in overpriced companies and less in underpriced companies. If one completely followed an indexing approach, one would have had nearly 50% of one's stocks portfolio in the Japanese stock market around the start of 1990. The PE10 of that market was around 90. Off the top of my head, I think Larry Swedroe stated that the Japanese stock market from Jan 1990 to Dec 2013 had an annual return of minus 1.9%.

    ReplyDelete
    Replies
    1. @Anonymous: Bogle's results are consistent with those of Fama and French for the 1963-1990 period that Fama and French examined. If it's really the case that nobody has examined the relative performance of growth and value stocks over long periods, then I'm happy to say that the existence of a value premium is undecided. But the onus is on those who believe a value premium exists to prove it. 28 years of data is not enough. I find Bogle's longer-term results to be more compelling than Fama and French's 28 years.

      Bubbles certainly exist, but this fact is of little use unless you can find a way to outperform the market. Those who invest in value stocks avoid many of the stocks that end up crashing, but they also miss many of the stocks that rise quickly. I'm skeptical that mechanical methods of choosing value stocks will outperform over the long run. If this is correct, then it all comes down to stock-picking, which is a very competitive game that depends on the existence of players who do it poorly.

      Delete
    2. @anonymous: Indexers would not likely have more than about a third of their equities in international stocks. I'm not sure how much of that would have been in Japanese stocks in 1990, but obviously it was not nearly 50% of ones stock portfolio.

      Delete
  3. According to Larry Swedroe in his latest book "Reducing the risk of black swans", using the CRSP data from the University of Chicago for 1927 to 2013, the annual average (not compound or annualized) size premium has been 3.6% per year and the value premium 4.9%. Of course, there has been lengthy periods when both premiums have been negative. I'm not sure that the debate is so much about the existence of the value premium, but whether it is due to risk or behaviour.

    ReplyDelete
    Replies
    1. @Grant: I've heard Larry speak about this subject before. I'd like to hear an explanation of why his analysis appears to differ so greatly from Bogle's. I found an article by Swedroe that gives some insight into how the premiums are defined:

      http://www.cbsnews.com/news/how-the-four-stock-premiums-work/

      The value premium is based on ranking stocks by book-to-market ratio, and comparing the returns of the top 30% to the bottom 30%. This is quite different from comparing value stocks to the overall market. If we toss in some trading rules to avoid excessive trading while stocks move into and out of the 30% ranges, maybe that explains the difference between what Swedroe says and Bogle says -- I'm not sure.

      Delete
  4. That's true, although in the same book Swedroe says that small value outperformed the S&P by an annualized 3.8% over the same period. He doesn't mention figures for value or small separately.


    I came across this thread on Bogleheads where Larry suggests that the difference between him and Bogle is that Bogle is using live funds that do not access the premiums properly.

    ReplyDelete
    Replies
    1. @Grant: Many smart people have said Bogle is wrong over the years, but Bogle was often the one who was right. Even so, if we assume Swedroe is correct, the next question is whether there is enough value premium to make up for higher transaction costs, and whether this situation will persist into the future.

      I think some of the small value premium is attributable to higher risk, unless the 3.8% already takes this into account.

      Delete
  5. Sorry, here's the thread

    http://www.bogleheads.org/forum/viewtopic.php?f=10&t=142910&newpost=2122750

    ReplyDelete
    Replies
    1. @grant: Interesting thread. Larry knows his stats, so I believe him when he says that by some definition a value premium exists. There could be some amount of data mining effect, but even if we assume not, I'm not convinced enough that any premium can be captured to try betting my money on value stocks.

      Delete
  6. It's certainly true that the factor loadings of the funds matter as do costs, but since 2001 the Vanguard funds have done well - the total market fund returned 6.48% and the small value fund returned 11.6%. They follow the CRSP indexes. Of course there is no guarantee that any of the premiums will persist (for that matter the equity premium, too), but we only have historical data to go on. Certainly the small value outperformance is due to higher risk, giving the higher return of 3.8%. I'm curious as to why you tilt to small but not value (or small value) given the returns (according to Larry in the embedded ETF.com article in the thread) of small 1.4% over the S&P, large value 0.5%, and small value 3.8% in the period 1927-2013. I tilt to small value in my US and international equities, but not Canadian as there is not an appropriate ETF to do so in the Canadian market.

    ReplyDelete
    Replies
    1. @Grant: I can't see a reason why a small value premium (beyond the risk premium) should exist going forward. I may change my mind at some point, but I find the evidence mixed for now and am still thinking about it all. An added complication is that I've been trying to figure out how traditional investing theory is affected if we throw out the assumption of lognormally-distributed returns and adopt some other stable distribution whose tails more closely match actual returns. I intend to mostly hug the index until I come to some sort of conclusion. It may be that VBR will outperform VB over time. Perhaps I'm just biased by all the nonsense I see about the magic of dividend and value investing from people who just buy Canadian banks.

      Delete
  7. Michael, I'm with you on the nonsense on the magic of dividends from dividend growth stock pickers. I thought you'd enjoy Larry's "magic pants" article. http://seekingalpha.com/article/2093953-dividends-and-the-magic-pants

    ReplyDelete