A close look at factors in real world,

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

A close look at factors in real world,

Post by larryswedroe »

http://www.etf.com/sections/index-inves ... -work.html

Thought this would be of interest,

Larry
SnowSkier
Posts: 268
Joined: Fri Mar 16, 2012 12:21 pm
Location: Austin

Re: A close look at factors in real world,

Post by SnowSkier »

larryswedroe wrote:Following is a summary of the authors’ findings:
  • Long positions comprise the bulk of the size premium, capture about 60 percent of the value premium, and make up about half of the momentum premium.
  • The value premium is concentrated primarily in small stocks and becomes insignificant in the largest 40 percent of NYSE stocks. This was true in all four 20-year subperiods.
  • The size effect emerges in a significant way only when considering more extreme exposure to small stocks—basically micro-cap stocks.
  • The momentum premium is present and stable across all size groups and the entire 86-year period. It was persistent in all four 20-year subperiods examined, including the most recent two decades following the initial publication of the original momentum studies.
Larry
Thanks Larry! Very interesting.

One question...is there any similar info on Quality (QMJ)? i.e. How much of the QMJ premium is captured by long-only?
countmein
Posts: 653
Joined: Fri Dec 06, 2013 8:10 pm

Re: A close look at factors in real world,

Post by countmein »

I didn't read the paper (costs $40 to download), so I'm wondering, regarding:
The value premium is concentrated primarily in small stocks and becomes insignificant in the largest 40 percent of NYSE stocks. This was true in all four 20-year subperiods.
Is "value" here defined as B/P? Haven't we established that B/P is the only metric for which there is no large-cap value premium; other value metrics do exhibit a large-cap value premium?
Tanelorn
Posts: 2366
Joined: Thu May 01, 2014 9:35 pm

Re: A close look at factors in real world,

Post by Tanelorn »

countmein wrote:I didn't read the paper (costs $40 to download),
Most academic papers can be found for free on the web somewhere, often on the webpage of the authors. Here's a public link to the paper in question.

http://poseidon01.ssrn.com/delivery.php ... 23&EXT=pdf
garlandwhizzer
Posts: 3562
Joined: Fri Aug 06, 2010 3:42 pm

Re: A close look at factors in real world,

Post by garlandwhizzer »

Excellent post, Larry, thanks. Very interesting about the failure of hedge funds and institutions to significantly dilute the premiums up to now.

The more thorny question it seems to me is not do premiums persist in long only portfolios after reasonable estimates of trading costs but how successful have real life funds been in capturing those premiums for investors after costs. Some funds have (notably DFA, although that requires an advisor), but many funds that for example claim to be SCV funds do not achieve outperformance and some underperform small cap blend. It seems that there is a real difference in retrospective academic analysis of factors with estimating trading costs and actually doing that in the real world with funds. F&F's original data showed very significant outperformance of SCV using only P/B as a value parameter. That doesn't seem like a difficult or expensive strategy to follow for SCV indexes or SCV fund managers. Why then have many SCV funds had such spotty performance? What accounts for the gulf between the academic research and the real life experience of many funds that attempt to capture these premiums? I would be very interested in Larry's, Rick's, Robert T's or other's comments on this seeming paradox.

Garland Whizzer
jaab
Posts: 112
Joined: Thu Aug 07, 2014 12:52 pm

Re: A close look at factors in real world,

Post by jaab »

Here is another link, with the finalized layout: http://faculty.chicagobooth.edu/tobias. ... omment.pdf
countmein wrote:Is "value" here defined as B/P? Haven't we established that B/P is the only metric for which there is no large-cap value premium; other value metrics do exhibit a large-cap value premium?
Yes, P/B. Although they have some other metrics in the appendix, page 22 in the PDF above. Ret(-1,-60) means the relative losers of the past 5 years ("LT reversal"). There is still a good decline of "value" from SV to LV. When looking at the diagrams, the long-only / composite metric / size 5 portfolio seems to level around 2-3%/year 1950-2011. Not bad after 1990 either, compare it to the bottom right diagram at page 13 (P/B). No t-stats or more details for page 22, unfortunately. If I understand it correctly they are also using the more concentrated outer 20% of the number of stocks for comparing value/growth (i.e. a 5x5 bucket matrix). Not the outer 30% as HML methodology does.
Last edited by jaab on Wed Oct 08, 2014 4:59 pm, edited 1 time in total.
garlandwhizzer
Posts: 3562
Joined: Fri Aug 06, 2010 3:42 pm

Re: A close look at factors in real world,

Post by garlandwhizzer »

That is a fascinating and thoroughly researched paper, jaab, thanks for the post. Much food for thought.

Garland Whizzer
stlutz
Posts: 5585
Joined: Fri Jan 02, 2009 12:08 am

Re: A close look at factors in real world,

Post by stlutz »

Thanks for the link, jaab.

To me, the #1 conclusion is that momentum is the factor that is most worth pursuing.

For example, for size the paper concludes:
Hence, the size effect emerges when considering more extreme exposure to small stocks, especially extremely mall stocks, through either equal-weighted portfolios or Fama and MacBeth regressions. Less extreme exposure to size, through value weighting, for instance, does not produce a reliable size effect...
Contrast this to momentum where the effect is robust across cap. ranges.
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: A close look at factors in real world,

Post by larryswedroe »

snowskier
haven't seen the long only evidence on QMJ, but most premiums are about half short and half long

But let me add this, on many of the factors the results are really driven by the very bad returns of the other end of the spectrum, not that the first quintile has great returns, much better say than the second or third, but that the fifth is typically awful. That's certainly the case with low vol, and small growth, as just two examples. So one way to run portfolios by simply screening out say high vol (with high short interest) and small extreme growth stocks, and negative momentum stocks,etc. That's why long only funds can do well if they have sufficient exposure to the factor and screen out the extremely poor performers

larry
newbie001
Posts: 438
Joined: Mon Nov 24, 2008 12:50 am

Re: A close look at factors in real world,

Post by newbie001 »

I struggle to understand why the market doesn't incorporate momentum into pricing in light of all the research coming out about its benefits. I think I see why the markets as a whole might not be able to incorporate the value premium into its pricing (stocks where value premium is concentrated are small, the premium may take decades to show up, maybe it's all a risk-story anyhow, etc.), but momentum seems like a factor that would very quickly be exploited by hedge funds and the like and incorporated into market prices. I know Larry says that the momentum premium has persisted, but I can't help wondering if that is going to continue given all the attention it is receiving.

Edit: Shame about the value effect being most pronounced in microcaps, since no ideal ETF for microcaps exists.
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: A close look at factors in real world,

Post by larryswedroe »

newbie
What we can say is that momentum has been well known in many asset classes for decades and it still persists. No good explanation for why on upside (accept market underreaction) but downside there is--like tax loss selling increases as prices go lower and more investors pass their "GMO" point (when stomach screams get me out)
Larry
Freestyle
Posts: 13
Joined: Tue Feb 26, 2013 11:05 am

Re: A close look at factors in real world,

Post by Freestyle »

I had always assumed, without giving it much thought, that a factor's premium was viewed against The Market (e.g. TSM) and not that factor's opposite on the spectrum (e.g. sv v. sg) as Larry explains it in his article. I thought if a factor had a 2% premium that meant it outperformed The Market (TSM) each year over time by 2%.
If one is considering tilting whether to sv or s or whatever away from TSM it would be most useful to see how that factor has compared to TSM over time and not just how it has compared to that factors opposite on the spectrum. Is there data that shows a factors comparison to TSM or are the factors only explained as long/short portfolios?

Ken
User avatar
nedsaid
Posts: 19249
Joined: Fri Nov 23, 2012 11:33 am

Re: A close look at factors in real world,

Post by nedsaid »

Larry Swedroe said,

"The value premium is concentrated primarily in small stocks and becomes insignificant in the largest 40 percent of NYSE stocks. This was true in all four 20-year subperiods."

This is really interesting. I have often said that I don't consider value investing to be cigar butt investing where I get a few last puffs from a discarded butt. I want to buy a box of fine cigars when they go on sale. In my stock picking days, most of the stocks I chose were the large caps. I was looking for good stuff that had temporary problems or was being ignored by Wall Street. In most cases, I wanted a good steady dividend too. So perhaps what I was really doing was going for the quality or profitability factor at a bit of a discount rather than a pure value play. This would be true if there were little or no value premium in the large cap stocks.

The research I have seen shows a Value premium in large-cap stocks but an even larger Value premium in small-cap stocks. So what Larry concluded in the article is a bit at odds with other things I have read. But I agree that the small and value factors can be maximized with a good small cap value fund or index and a good micro-cap fund. Micro-cap might be an area where more active management would work as this gets around traders front running the micro-cap indexes.
A fool and his money are good for business.
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: A close look at factors in real world,

Post by larryswedroe »

nedsaid
What the research seems to be showing is that the large value premium hasn't been as much if looked at through btm lens only, but different story if look at through other value metrics---which give you more exposure to quality/profitability factor (like p/e and p/cf). That's one reason I prefer multiple value factors vs. one.

Freestyle
Factors are always long short, not relative to the TSM, which includes both. You cannot or shouldn't look at annualized returns of factors, as they are annual averages.
What we do know is that roughly half of the factor premium comes from the long and about half the short
Larry
Freestyle
Posts: 13
Joined: Tue Feb 26, 2013 11:05 am

Re: A close look at factors in real world,

Post by Freestyle »

Larry, let's say the sv factor has an average 3% premium then this seems to mean it will beat a sg portfolio, over time, by 3%. Is this correct? I'm guessing the majority of investors who consider factors, even Bogleheads, are as confused as I am on factor analysis and assume the premium relates to beating The Market. I hope you'll write more about this for edification on factors.
Personally I tilt to SV but I do so not for any expected premium but rather for diversification from TSM. I'd like to see the factor discussion bring potential diversification benefits to the forefront rather than expected premium benefits which is where the discussion dominates.

Ken
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: A close look at factors in real world,

Post by larryswedroe »

Freestyle
First, there is no small value factor. There's a small factor and a value factor. But the value factor is bigger in small (because of poor performance of small growth)
Second, 27-13, all ex-utilities
SG 9.22
SV 13.85
LG 942
LV 10.56

You should do it because of the higher expected returns because they are clearly far more risky (SD of SV 30 roughly vs 19 of LG), PLUS the diversification benefit

Larry
User avatar
packer16
Posts: 1488
Joined: Sat Jan 04, 2014 1:28 pm

Re: A close look at factors in real world,

Post by packer16 »

One factor I think that is related to size and value is liquidity. I have seen the largest mispricings (on a % basis) in small illiquid stocks. The French regional banks and the Korean preferred stocks are two current examples I can think of. Each sell at significant discounts to other banks of similar size (few hundred million to a billion market cap) in other countries or common shares (who are the same except the vote and less liquidity) for the Korean preferred stocks.

Packer
Buy cheap and something good might happen
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: A close look at factors in real world,

Post by larryswedroe »

Packer
No question liquidity is a factor, Ibbotson has done studies on this subject, though I had some issues with the paper (equal weighting was used) http://www.etf.com/sections/index-inves ... style.html

Here's another interesting point---the credit/default premium has only been 30bp (before expenses). Now we know that part of the premium yield in below investment grade bonds is a liquidity premium. So given that there was probably virtually no default premium historically even before expenses. Small premium yes (before expenses) but it was liquidity.

Larry
User avatar
oneleaf
Posts: 2562
Joined: Mon Feb 19, 2007 4:48 pm

Re: A close look at factors in real world,

Post by oneleaf »

larryswedroe wrote:Freestyle
First, there is no small value factor. There's a small factor and a value factor. But the value factor is bigger in small (because of poor performance of small growth)
Second, 27-13, all ex-utilities
SG 9.22
SV 13.85
LG 942
LV 10.56

You should do it because of the higher expected returns because they are clearly far more risky (SD of SV 30 roughly vs 19 of LG), PLUS the diversification benefit

Larry
Hi Larry,
Great article.

I have struggled to get the full small and value tilt that I desire in my international stocks, since there is not as much of a variety of offerings for the retail investor. I change my strategy to have a large position in Vanguard's international small blend index fund, and do my value tilting in developed market large stocks.

I have always tried to see a total value and size load on the entire portfolio, doing three factor regression analysis on my funds, but now I am wondering whether it is an accurate portrayal of my exposure to size and value.

If the value premium virtually disappears when you get to higher stocks, then it would seem like looking at the value and size premiums separately is not the best way.
Portfolio A: large value and small blend: HmL 0.4, SmB 0.4
Portfolio B: large blend and small value: HmL 0.4, SmB 0.4

Can we expect similar exposure to size and value factors and expected returns for the two portfolios above? Since they have the same loadings when doing a regression analysis against Fama and French is factors data? Or will portfolio B provide higher expected returns since value is concentrated in the small area?

Thanks!
Post Reply