No "Value Premium" in real mutual funds

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
JoMoney
Posts: 7016
Joined: Tue Jul 23, 2013 5:31 am

No "Value Premium" in real mutual funds

Post by JoMoney » Sun Feb 08, 2015 5:10 am

Not sure if it's been posted before, but I stumbled onto this paper:
Portfolio Constituency Rules and the Value Premium in the Small-Cap Space
and thought it might be interesting to others here.
... While our findings do not explicitly overturn the risk thesis to the value premium, our results may go a long way in explaining why the returns of market-based growth fund managers are statistically equal to returns of value fund managers over time. ...
It seems to imply that when actual mutual-funds (index or otherwise) are implemented, that the most illiquid stocks are often excluded. The effect of this improves the measured performance of actual funds (relative to FF Factor data), and the improvement because of this within the "growth" style space is much larger than the improvement in the "value" style space. They look specifically at DFA's Small-Value fund, and although they find that the value premium is " still economically large and statistically significant " relative to the FF data, when instead it's computed using a more realistic model of a growth portfolio as is typical of a growth mutual fund that " the opportunity set of average monthly premium returns completely disappears, both economically and statistically ".
...We suggest these results might go a long way in explaining why market-based growth fund returns generally equal those of their value fund counterparts over time...
I think it gives cause for more consideration of Mr.Bogle's data that he's discussed in various presentations like The Telltale Chart regarding Value vs. Growth in actual mutual funds over time...
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

stlutz
Posts: 5105
Joined: Fri Jan 02, 2009 1:08 am

Re: No "Value Premium" in real mutual funds

Post by stlutz » Sun Feb 08, 2015 12:37 pm

Thanks for the link. Some of the findings in the paper have been discussed before--this puts a lot of information into a consolidated place, which is helpful. Several comments:

--The prior work in this area has been problematic from a couple of perspectives. One is the selection of start/end dates. As nisiprius has demonstrated in other threads, the selection of start and end dates makes a big difference on the returns shown by a strategy. One problem with some of the various comparisons of actual growth/value funds is the fact that the studies end right around 2000. Anyone should expect growth to look better when such a date is selected as the end. The second problem has been the use of arithmetic vs. geometric averaging. When you compare compound rates of returns, the value funds actually do better.

--the authors show that "real world" smallcap growth funds are never as bad as the F/F data suggests they should be. When you apply realistic holding criteria for institutional funds (whether active or passive), the "negative" aspect of SG is much reduced. The SG stocks that are held only by individuals are a "black hole"; for those that are held by institutions--no so much.

--These authors also demonstrate the value premium most happens in down markets. What they also show is that SV outperforms SG when smallcap stocks as a whole are underperforming. It seems that relative to the market and to the size factor overall, value does best when the larger market, whether that be the whole market or just smallcap stocks, is doing poorly. This is consistent with my own view that value outperformance is because these stocks have historically been less risky, not more. (Note: That's my conclusion, not the authors'.)

It would seem based on what they are showing is that there is a real value premium, but the real world portfolio impact is probably measured more on the right side of the decimal point than the left.

lazyday
Posts: 3307
Joined: Wed Mar 14, 2007 10:27 pm

Re: No "Value Premium" in real mutual funds

Post by lazyday » Mon Feb 09, 2015 7:01 am

stlutz wrote:It would seem based on what they are showing is that there is a real value premium, but the real world portfolio impact is probably measured more on the right side of the decimal point than the left.
Haven't DFA funds done much better than S&P 500 funds?

User avatar
nisiprius
Advisory Board
Posts: 38316
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: No "Value Premium" in real mutual funds

Post by nisiprius » Mon Feb 09, 2015 7:53 am

lazyday wrote:
stlutz wrote:It would seem based on what they are showing is that there is a real value premium, but the real world portfolio impact is probably measured more on the right side of the decimal point than the left.
Haven't DFA funds done much better than S&P 500 funds?
Well, let me be a little snarky here. Why, no, certainly not, DFEQX (blue) is an example of a DFA fund that has done much worse than "an S&P 500 fund (orange)."

Source: Morningstar
Image

The reason is that it's not an apples-to-apples comparison because DFEQX, DFA Short-Term Extended Quality, is a short-term bond fund that takes far less risk than an S&P 500 stock fund. The point here is that simply comparing returns is a fool's game. It is very hard to be sure you are making a true apples-to-apples comparison.

The claim that's made for DFA funds that I think is probably true is that they are well-focussed on delivering high loadings on whatever factors their Nobel laureate consultants think are worthwhile.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

User avatar
nisiprius
Advisory Board
Posts: 38316
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: No "Value Premium" in real mutual funds

Post by nisiprius » Mon Feb 09, 2015 8:07 am

Here's an interesting detail that relates directly to the Scislaw paper. It's been observed over and over again that performance of Vanguard Small-Cap Value Index and Vanguard Small-Cap Growth Index has actually been an almost perfect tie, but factor believers dismiss this on the grounds that Vanguard's funds are no darn good.

Here then is the comparative performance of small growth and small value in two DFA funds, the connoisseur's choice for the best factor exposures available in the real world. Blue, small-cap growth, the "black hole" of the factor universe. orange, small-cap value, the fashion queen of the factor universe (or maybe not any more--Fama and French have declared her to be "redundant" in their five-factor model). The point is: which outperformed? Small growth (blue) or small value (orange)? OK, the time period is so short as to be meaningless. But it's interesting because it's DFA.
Source: Morningstar
Image

To me, it all comes down to this. I have a choice. I can just take the simple route and buy a broad cap-weighted total market index fund and find that in real life, by golly, I really get exactly what I expected to get. Or, I can go in for intriguing theories, buy into trickier strategies, not get in practice what I expected to get in theory, and enjoy reading papers like Scislaw's after the fact that will tell me the fascinating reasons why I didn't.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

garlandwhizzer
Posts: 2291
Joined: Fri Aug 06, 2010 3:42 pm

Re: No "Value Premium" in real mutual funds

Post by garlandwhizzer » Mon Feb 09, 2015 4:15 pm

Good post and good comments. The small value premium is robust in the literature but tends to be less impressive with real life funds especially in recent years. Choosing a heavily tilted SV portfolio is not a slam dunk to achieve outperformance. You pay your money and take your chances. TSM is in my opinion a rational stand alone US portfolio particularly these days when full time professionals set market prices. I tilt modestly (20% of portfolio) but the biggest reason for that is simply to get more exposure to mid and small cap stocks which make up only 28% of TSM.

Garland Whizzer

thx1138
Posts: 929
Joined: Fri Jul 12, 2013 2:14 pm

Re: No "Value Premium" in real mutual funds

Post by thx1138 » Mon Feb 09, 2015 4:51 pm

Interesting paper, thanks for posting it! I believe some others on this forum (Swedroe for example) have pointed out that certain selection criteria in a fund can mean reducing one factor exposure into an orthogonal one in real world funds. And that seems to be part of the effect outlined in the paper.

One oddity is that they seem to say removing the more illiquid securities improves return when theory might predict you should actually be getting an illiquidity premium from those securities.

Personally I only have SV exposure because through a strange subsidization in one 401k account the ER for one of the DFA SV funds is actually lower than that for the SP500 fund! So my SV exposure return will partly really be a lower ER return :)

comeinvest
Posts: 198
Joined: Mon Mar 12, 2012 6:57 pm

Re: No "Value Premium" in real mutual funds

Post by comeinvest » Tue Feb 10, 2015 6:14 pm

Would this whole thing mean that active managers, after fees, outperform the index, for small growth?

User avatar
Topic Author
JoMoney
Posts: 7016
Joined: Tue Jul 23, 2013 5:31 am

Re: No "Value Premium" in real mutual funds

Post by JoMoney » Wed Feb 11, 2015 2:36 am

comeinvest wrote:Would this whole thing mean that active managers, after fees, outperform the index, for small growth?
The way I read it, no... But it may depend on what index you use for the "small growth" benchmark. Most indexes have constituency requirements regarding liquidity. They find no significant premium when value/growth are measured against indexes.
... Market-based value style equity portfolios do not systematically outperform market-based growth style
equity portfolios, despite considerable academic research that suggests that they should. ...


... Houge and Loughran [2006] conclude that a value premium does not exist in managed mutual funds or in the passive indexes they also examine.
This latter finding was recently confirmed by Scislaw and McMillan [2012] for a wider collection of US and international indexes. Scislaw and McMillan find the premium statistically non-existent in the benchmark opportunity sets of stocks, as represented by market indexes. They also find a non-existent premia regardless of whether index returns are observed at the top level or within the traded constituency of, at least, the S&P 500, 600 and 1500 indexes. ...
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

comeinvest
Posts: 198
Joined: Mon Mar 12, 2012 6:57 pm

Re: No "Value Premium" in real mutual funds

Post by comeinvest » Wed Feb 11, 2015 8:03 pm

I'm not quite understanding your last post. So where is the value premium supposed to exist, if not in active portfolios nor the indexes you mentioned?

stlutz
Posts: 5105
Joined: Fri Jan 02, 2009 1:08 am

Re: No "Value Premium" in real mutual funds

Post by stlutz » Wed Feb 11, 2015 10:31 pm

I'm not quite understanding your last post. So where is the value premium supposed to exist, if not in active portfolios nor the indexes you mentioned?
It exists in most significant form in stocks that are held almost entirely by individual investors.

Much of the "value premium" is really a "growth penalty", and this penalty is most significant in stocks that aren't in the indexes or in institutional portfolios for various reasons. Just by starting out with the "base screen" of companies that index providers or fund managers like DFA use eliminates much of the growth penalty.

User avatar
nisiprius
Advisory Board
Posts: 38316
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: No "Value Premium" in real mutual funds

Post by nisiprius » Wed Feb 11, 2015 10:48 pm

Whoa. stlutz... have I got this right...

Theoretically small growth stocks are supposed to be the "black hole of investing," but in real life, small growth mutual funds fail to capture the "black hole" darkness, because the worst of the small growth stocks, the holey of holeys, aren't really investible?

Therefore, small growth mutual funds, "alas," fail to exhibit the theoretical underperformance the academics say they should exhibit?

I love it!
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

stlutz
Posts: 5105
Joined: Fri Jan 02, 2009 1:08 am

Re: No "Value Premium" in real mutual funds

Post by stlutz » Wed Feb 11, 2015 10:51 pm

Whoa. stlutz... have I got this right...
Yes, that is what the paper from the OP argues (although with less exciting language. :) )

User avatar
Ketawa
Posts: 2140
Joined: Mon Aug 22, 2011 1:11 am
Location: DC

Re: No "Value Premium" in real mutual funds

Post by Ketawa » Wed Feb 11, 2015 11:12 pm

nisiprius wrote:Here then is the comparative performance of small growth and small value in two DFA funds, the connoisseur's choice for the best factor exposures available in the real world. Blue, small-cap growth, the "black hole" of the factor universe. orange, small-cap value, the fashion queen of the factor universe (or maybe not any more--Fama and French have declared her to be "redundant" in their five-factor model).
According to a post by Bill Bernstein, DFA hopes for market-like returns from the growth funds by offsetting a potential growth penalty with a focus on profitability. They "fill the style boxes" for institutional investors.

countmein
Posts: 405
Joined: Fri Dec 06, 2013 9:10 pm

Re: No "Value Premium" in real mutual funds

Post by countmein » Wed Feb 11, 2015 11:36 pm

stlutz wrote:
I'm not quite understanding your last post. So where is the value premium supposed to exist, if not in active portfolios nor the indexes you mentioned?
It exists in most significant form in stocks that are held almost entirely by individual investors.

Much of the "value premium" is really a "growth penalty", and this penalty is most significant in stocks that aren't in the indexes or in institutional portfolios for various reasons. Just by starting out with the "base screen" of companies that index providers or fund managers like DFA use eliminates much of the growth penalty.
Could be remembering it wrong, but I think the black hole phenom is only true for the P/B value sort (the DFA way), but not so much for other value sorts or a blended sort.

User avatar
nedsaid
Posts: 11602
Joined: Fri Nov 23, 2012 12:33 pm

Re: No "Value Premium" in real mutual funds

Post by nedsaid » Thu Feb 12, 2015 1:10 am

nisiprius wrote:Whoa. stlutz... have I got this right...

Theoretically small growth stocks are supposed to be the "black hole of investing," but in real life, small growth mutual funds fail to capture the "black hole" darkness, because the worst of the small growth stocks, the holey of holeys, aren't really investible?

Therefore, small growth mutual funds, "alas," fail to exhibit the theoretical underperformance the academics say they should exhibit?

I love it!
Yes, Nisiprius you have it right. There are some real paradoxes to investing. Value works great and works mainly for behavioral reasons. But then momentum works great too, and also for behavioral reasons. One style works because it is unpopular and the other style works because it is popular.

Quality seems to work great too. But isn't quality or profitability another name for growth? If a company has been around long enough and generated consistent earnings growth for enough time to become a large cap stock, isn't that the ultimate measure of quality? Shouldn't we be putting all our marbles into large cap growth?

What it comes down to is that the factors all seem to work even if some of them seem contradictory to each other. But it seems that the factors work best at different times. In theory, if you are invested across factors you should achieve market gains with less volatility. So perhaps, you get a diversification benefit.

This is what is so frustrating to those of us who really try to do the right thing by our portfolios. I am wondering that maybe we shouldn't be trying so hard. It is frustrating because the extra effort is not always rewarded. Or we do what is the right thing but get told we are using the wrong funds or ETF's.
A fool and his money are good for business.

lack_ey
Posts: 6675
Joined: Wed Nov 19, 2014 11:55 pm

Re: No "Value Premium" in real mutual funds

Post by lack_ey » Thu Feb 12, 2015 1:58 am

The worst small growth investments may also not meet the "can this be readily shorted?" screen. There is the hypothesis that companies that are expensive and scarce to short may end up with inflated prices and worse performance. I've seen some data on the effect before somewhere, I think.

That's heavily related to liquidity in general, so nothing that amazing.

User avatar
Robert T
Posts: 2559
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Re: No "Value Premium" in real mutual funds

Post by Robert T » Thu Feb 12, 2015 2:48 am

JoMoney wrote:... Houge and Loughran [2006] conclude that a value premium does not exist in managed mutual funds or in the passive indexes they also examine.
IMO the Houge and Loughran paper shows the opposite - or at least you can draw the opposite conclusions. The paper looks at monthly average returns. Using the same indexes, managed mutual fund returns, and time period presented as in the Houge and Loughran paper, but presenting annualized returns (and implied growth in $1000) over the same time periods (more relevant to investors) – gives the following result.

Index returns

1975-2002 Annualized Return (%)/Implied growth in $1,000
S&P 500/Barra Value = 14.1/$40,152
S&P 500/Barra Growth = 12.2/$25,036

1979-2002 Annualized Return (%)/Implied growth in $1,000
Russell 3000 Value = 14.0/$23,151
Russell 3000 Growth = 11.6/$13,850

Equity mutual funds

1965-2001: Annualized returns/implied growth in $1,000
Small Value = 12.6%/$79,570
Small Growth = 12.0%/$67,264

Large Value = 10.4%/39,093
Large Growth = 9.6%/$29,608

A different conclusion from the paper - yet using the same data.

Robert
.

BigJohn
Posts: 1708
Joined: Wed Apr 02, 2014 11:27 pm

Re: No "Value Premium" in real mutual funds

Post by BigJohn » Thu Feb 12, 2015 9:07 am

nedsaid wrote: Value works great and works mainly for behavioral reasons.
nedsaid, I thought from what I've read from Larry and others that small and/or value tilting was more of a risk story than a behavioral story. I'm not a fan of tilting but I can at least see the logic for tilting if its a higher risk with higher reward. Why do you believe that behavior will persist in a fast changing world? For example, haven't people's behaviors been fundamentally changed due to easy access to a broad range of mutual funds that let then buy tilts at relatively low cost?

User avatar
nedsaid
Posts: 11602
Joined: Fri Nov 23, 2012 12:33 pm

Re: No "Value Premium" in real mutual funds

Post by nedsaid » Thu Feb 12, 2015 12:40 pm

BigJohn wrote:
nedsaid wrote: Value works great and works mainly for behavioral reasons.
nedsaid, I thought from what I've read from Larry and others that small and/or value tilting was more of a risk story than a behavioral story. I'm not a fan of tilting but I can at least see the logic for tilting if its a higher risk with higher reward. Why do you believe that behavior will persist in a fast changing world? For example, haven't people's behaviors been fundamentally changed due to easy access to a broad range of mutual funds that let then buy tilts at relatively low cost?
I haven't yet seen evidence that the value premium is a risk story. I have read about the "bad company" story that value companies tend to have less predictable cash flows, carry more debt, etc. Somebody lectured me about value being a risk story and the poster included links to articles. But when I read the actual article, the standard deviation of large value was less than large growth and small value was less than small growth. The actual statistics didn't bear out the argument. So it appeared to me you got higher return for less risk.

To me, it is a matter of expectations. If a stock has low expectations built into the price, it has a higher likelihood of beating earnings expectations that a stock priced to perfection. I will see if I can find the post and my response.

Larry Swedroe says the risk versus behavior is hotly debated among academics. Many think it is a mix of both.
A fool and his money are good for business.

User avatar
nedsaid
Posts: 11602
Joined: Fri Nov 23, 2012 12:33 pm

Re: No "Value Premium" in real mutual funds

Post by nedsaid » Thu Feb 12, 2015 1:00 pm

I found a post of mine, a "Nedsaid greatest hit", that addresses your question.

Here is a link to the article that I refer to below:
http://www.efficientfrontier.com/ef/999/risk.htm


Dbr, I did read the efficient frontier article on the "riskiness" of value stocks compared to their growth brethren. However the math in the article did not bear this out. According to the article, Fama and French settled on a "sick company" theory that explained that value stocks really were riskier. But there is no proof offered that Value stocks really are riskier. The math says otherwise. Pretty much what they said was, "Return is higher but if return is higher then the risk has GOT TO BE higher." The article offered no mathematical proof but only a narrative.

From the period 7/63-3/99 Small Value had 17.47% annualized return and a 18.63% standard deviation versus Small Growth had a 10.18% annualized return and 23.04% standard deviation. Large Value had 15.16% annual return and a 14.86% standard deviation versus 11.75% and 16.35% standard deviation for Large Growth. Ye Gads, that looks like more return with less risk indeed. The free lunch or at least a pretty darned good snack.

So yes I believe Value stocks to be less risky than Growth stocks. The reason why is not rocket science. It is a matter of how much expectation is built into the stock. Stocks with less expectations built into their price have better odds of beating earnings estimates than the popular growth stocks priced to perfection.

I will wrap up my argument with this example of GE Commercial Paper. What could be safer than AAA short term paper issued by the bluest of blue chip companies? When the financial crisis hit in 2008 and the credit markets dried up, there were no buyers for GE paper!! If Warren Buffett hadn't stepped in, GE could possibly have defaulted on one of the safest investments imaginable. So a "safe' asset turned "dangerous" almost on a dime. All because of unique circumstances that no one could have foreseen. Everyone "knew" that GE paper was safe but when the time of crisis hit there were no buyers.

Fama and French's argument that "everyone knows" that value stocks are riskier than growth stocks and base their argument on narrative rather than math is not convincing to me. I am not really much of a quant. But I need to see better evidence than a "sick company" narrative.

Below is a response from Larry Swedroe where he addresses the risk versus behavior argument. He elaborates on the "bad company" theory.

nedsaid
Well, this isn't black or white one
I think it's very clear that value stocks are riskier, but the premium is too big

Value stocks tend to have some common traits which are intuitively risky---higher SD of earnings, dividends and more leveraged. Hard to argue such companies are safer. They also importantly have BAD BETA risk. They tend to do worse in bad times, when bad returns are most risky because of risks to labor capital. Value companies tend to have less capital options and thus more at risk in financial crisis. Assets that do poorly in bad times should carry big risk premiums. So IMO very difficult to buy the story there isn't risk there. But I do think premium too big for the extra risk. So my own personal conclusion is it's not free lunch but it is free stop at dessert tray.

Larry
A fool and his money are good for business.

User avatar
nedsaid
Posts: 11602
Joined: Fri Nov 23, 2012 12:33 pm

Re: No "Value Premium" in real mutual funds

Post by nedsaid » Thu Feb 12, 2015 1:13 pm

Big John,

In answer to your question, I think the small and value premiums are likely to go away for a while. The reason being is that small/value tilting is becoming more popular and as you mentioned there are more low cost vehicles being developed to capitalize on this. It is like the Yogi Berra quote about a certain restaurant being so crowded that no one goes there anymore. I am concerned that the small/value restaurant is getting pretty crowded and perhaps there is better fare to be had elsewhere at lower prices.

My enthusiasm for small/value tilting has waned. This is probably not the best time to go rushing into small value stocks though valuations seem to have improved compared to the broad market. About a year ago, the Vanguard Small Value Index traded at about the same P/E as the broad market. Last I looked, the Small Value P/E was about 16 compared to the market P/E of about 18. These are forward P/E's based on estimated earnings.

My take is that people who are looking for the quick kill and don't get it will get bored at some point and start pursuing large growth again. I believe that at some point the strategy will get less popular and start to work again. No good investment strategy works 100% of the time.

Indeed, the 1990's was all about large cap growth. Value was passé and Warren Buffett was a doddering old fool. The 2000's were the mirror image as they were the decade of small cap stocks and value stocks. Warren Buffett and his value oriented strategies became popular again. I think we could be in a stretch where the growth stocks do better for a while.
A fool and his money are good for business.

User avatar
Epsilon Delta
Posts: 8090
Joined: Thu Apr 28, 2011 7:00 pm

Re: No "Value Premium" in real mutual funds

Post by Epsilon Delta » Thu Feb 12, 2015 1:46 pm

It is highly likely there are types of risk that are not measured by standard deviation. For example:
nedsaid wrote: I will wrap up my argument with this example of GE Commercial Paper. What could be safer than AAA short term paper issued by the bluest of blue chip companies? When the financial crisis hit in 2008 and the credit markets dried up, there were no buyers for GE paper!! If Warren Buffett hadn't stepped in, GE could possibly have defaulted on one of the safest investments imaginable. So a "safe' asset turned "dangerous" almost on a dime. All because of unique circumstances that no one could have foreseen. Everyone "knew" that GE paper was safe but when the time of crisis hit there were no buyers.
Where does that show up in an easily computed statistic? Yet it could clearly have lost a lot of money, so it is risk. (Or uncertainty, or whatever you want to call it) If you believe in efficient markets the market will price these different types of risk and they will not show up in mean variance statistics. Even if you don't believe in efficient markets you should still acknowledge that the markets could be pricing this. Perhaps you want to call this "behavioral" instead of "risk", but that's just word games. It's not a free lunch just because you don't call it risk.

User avatar
nisiprius
Advisory Board
Posts: 38316
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: No "Value Premium" in real mutual funds

Post by nisiprius » Thu Feb 12, 2015 1:51 pm

So if, in the real world, small value doesn't actually outperform small growth, does it outperform small blend?
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

User avatar
nisiprius
Advisory Board
Posts: 38316
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: No "Value Premium" in real mutual funds

Post by nisiprius » Thu Feb 12, 2015 1:55 pm

Epsilon Delta wrote:....It is highly likely there are types of risk that are not measured by standard deviation....
My working hypothesis is that risk is risk, and that all honest measures of risk are highly correlated and any one of them is a decent proxy for all the others. And I tested this by taking the Simba backtest data and plotting 1985-2007 standard deviation on the X axis and year-2008 return on the Y axis. It doesn't matter whether you care about drawdown in a crash or statistical uncertainty--the same mutual funds that have high standard deviations are the same ones that crash.

Drawdown risk in 2008 isn't measured by standard deviation but it is highly correlated with standard deviation.

Image
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

User avatar
Epsilon Delta
Posts: 8090
Joined: Thu Apr 28, 2011 7:00 pm

Re: No "Value Premium" in real mutual funds

Post by Epsilon Delta » Thu Feb 12, 2015 2:21 pm

nisiprius wrote:
Epsilon Delta wrote:....It is highly likely there are types of risk that are not measured by standard deviation....
My working hypothesis is that risk is risk, and that all honest measures of risk are highly correlated and any one of them is a decent proxy for all the others.
If risk is two (or more) dimensional this cannot be true. Think of it like latitude and longitude. How far north you are tells you nothing about how far east you are. If you then say okay lets use distance from Greenwich (a function of both latitude and longitude) you find you have no information on direction. If different people care about two measures in different ways there is no way to reduce it to a single measure.

The mean variance framework cannot capture non-linearities. For example if you care about skew or kurtosis (both plausible measures of risk) they cannot be measured by standard deviation. Or if one investment is a function of the absolute value of another (which can be approximated with options) it is not captured in mean variance framework -- the covariance is zero, despite one result being an explicit function of the other.

richard
Posts: 7961
Joined: Tue Feb 20, 2007 3:38 pm
Contact:

Re: No "Value Premium" in real mutual funds

Post by richard » Thu Feb 12, 2015 2:31 pm

Epsilon Delta wrote:
nisiprius wrote:
Epsilon Delta wrote:....It is highly likely there are types of risk that are not measured by standard deviation....
My working hypothesis is that risk is risk, and that all honest measures of risk are highly correlated and any one of them is a decent proxy for all the others.
If risk is two (or more) dimensional this cannot be true. Think of it like latitude and longitude. How far north you are tells you nothing about how far east you are. If you then say okay lets use distance from Greenwich (a function of both latitude and longitude) you find you have no information on direction. If different people care about two measures in different ways there is no way to reduce it to a single measure.

The mean variance framework cannot capture non-linearities. For example if you care about skew or kurtosis (both plausible measures of risk) they cannot be measured by standard deviation. Or if one investment is a function of the absolute value of another (which can be approximated with options) it is not captured in mean variance framework -- the covariance is zero, despite one result being an explicit function of the other.
Yes. In a multi-factor world (for example, the Fama-French 3 factor rather than CAPM single factor), risk is multi-dimensional. See, for example, http://www.nber.org/papers/w7170

BigJohn
Posts: 1708
Joined: Wed Apr 02, 2014 11:27 pm

Re: No "Value Premium" in real mutual funds

Post by BigJohn » Thu Feb 12, 2015 3:50 pm

nedsaid wrote:But when I read the actual article, the standard deviation of large value was less than large growth and small value was less than small growth. The actual statistics didn't bear out the argument. So it appeared to me you got higher return for less risk.
nedsaid, thanks for all your responses. I'm not an expert in these things but is it possible that value is higher risk but doesn't show up in the standard deviation measure? If the risk you are being compensated for is that the value companies have a higher probably of disappearing altogether the SD might not reflect that loss. This could skew the SD based on a survivor bias and would seem most likely in the small value slice of the market.
nedsaid wrote:It is like the Yogi Berra quote about a certain restaurant being so crowded that no one goes there anymore. I am concerned that the small/value restaurant is getting pretty crowded and perhaps there is better fare to be had elsewhere at lower prices.
I love the analogy :D

comeinvest
Posts: 198
Joined: Mon Mar 12, 2012 6:57 pm

Re: No "Value Premium" in real mutual funds

Post by comeinvest » Fri Feb 13, 2015 8:10 pm

stlutz wrote:
I'm not quite understanding your last post. So where is the value premium supposed to exist, if not in active portfolios nor the indexes you mentioned?
It exists in most significant form in stocks that are held almost entirely by individual investors.

Much of the "value premium" is really a "growth penalty", and this penalty is most significant in stocks that aren't in the indexes or in institutional portfolios for various reasons. Just by starting out with the "base screen" of companies that index providers or fund managers like DFA use eliminates much of the growth penalty.
I don't invest in "stocks that are held almost entirely by individual investors". I've been trying lately to tilt toward small value via various global stocks, ETFs, etc, based on the hope that the SV premium, that according to numerous posts in this forum and numerous papers that I've read, has been consistently observed for many decades in most domestic and international markets over long time periods, might persist in the future. I'm confused. Was my effort futile? Has it no basis not even extrapolating the past [if I don't short obscure "stocks that are held almost entirely by individual investors"]?

User avatar
nedsaid
Posts: 11602
Joined: Fri Nov 23, 2012 12:33 pm

Re: No "Value Premium" in real mutual funds

Post by nedsaid » Fri Feb 13, 2015 8:34 pm

BigJohn wrote:
nedsaid wrote:But when I read the actual article, the standard deviation of large value was less than large growth and small value was less than small growth. The actual statistics didn't bear out the argument. So it appeared to me you got higher return for less risk.
nedsaid, thanks for all your responses. I'm not an expert in these things but is it possible that value is higher risk but doesn't show up in the standard deviation measure? If the risk you are being compensated for is that the value companies have a higher probably of disappearing altogether the SD might not reflect that loss. This could skew the SD based on a survivor bias and would seem most likely in the small value slice of the market.
nedsaid wrote:It is like the Yogi Berra quote about a certain restaurant being so crowded that no one goes there anymore. I am concerned that the small/value restaurant is getting pretty crowded and perhaps there is better fare to be had elsewhere at lower prices.
I love the analogy :D
You are right, Standard Deviation is only one measure of risk but it is the one most commonly cited.

The "bad company" argument has a big grain of truth to it, but I only pointed out that the article gave no statistical proof for the assertion. Pretty much what they said was greater risk means greater reward. If you get greater reward from value, there MUST be more risk. Not a bad argument but again no real proof. Wouldn't one expect more volatility with a riskier asset class?

My argument is that higher expectations come with built in higher risks. Once those high expectations aren't met, stocks priced to perfection can and do get hammered pretty hard.

Sometimes I get a bit annoyed with the survivor bias argument. Shoot, the indexes themselves are a product of survivor bias, I haven't heard of an index yet that includes deceased companies. There gets to be a point where one could make a survivor bias argument against almost anything. It is a good point to make but I think it is overdone sometimes.
A fool and his money are good for business.

User avatar
nedsaid
Posts: 11602
Joined: Fri Nov 23, 2012 12:33 pm

Re: No "Value Premium" in real mutual funds

Post by nedsaid » Fri Feb 13, 2015 8:46 pm

Epsilon Delta wrote:It is highly likely there are types of risk that are not measured by standard deviation. For example:
nedsaid wrote: I will wrap up my argument with this example of GE Commercial Paper. What could be safer than AAA short term paper issued by the bluest of blue chip companies? When the financial crisis hit in 2008 and the credit markets dried up, there were no buyers for GE paper!! If Warren Buffett hadn't stepped in, GE could possibly have defaulted on one of the safest investments imaginable. So a "safe' asset turned "dangerous" almost on a dime. All because of unique circumstances that no one could have foreseen. Everyone "knew" that GE paper was safe but when the time of crisis hit there were no buyers.
Where does that show up in an easily computed statistic? Yet it could clearly have lost a lot of money, so it is risk. (Or uncertainty, or whatever you want to call it) If you believe in efficient markets the market will price these different types of risk and they will not show up in mean variance statistics. Even if you don't believe in efficient markets you should still acknowledge that the markets could be pricing this. Perhaps you want to call this "behavioral" instead of "risk", but that's just word games. It's not a free lunch just because you don't call it risk.
What I was pointing out was that in extreme market conditions that a lot of things can get turned on its head. The situation with GE Commercial Paper will probably never happen again in our lifetimes. But what was regarded as AAA paper all the sudden had no market. No easily computed statistic could have predicted this event. It is also a way of showing the limitations of risk measurements. In market panics, no one gives a rip about standard deviation. Stuff will go down whether it is supposed to or not. Markets do not have to obey our expectations even if those expectations are expressed beautifully with math.

I also agree with efficient markets but the GE Commercial Paper example shows that markets can become very irrational in the short term. When no one wants to buy your security, the market value is essentially zero. At the point markets lost confidence to the extreme, the market valuation of GE Commercial Paper was zero. Was this an efficient price? Warren Buffett didn't think so and he provided GE with the liquidity it needed. Markets acted like GE didn't generate the cash flows to meet its debt obligations. Of course that was ridiculous.
A fool and his money are good for business.

User avatar
Topic Author
JoMoney
Posts: 7016
Joined: Tue Jul 23, 2013 5:31 am

Re: No "Value Premium" in real mutual funds

Post by JoMoney » Fri Feb 13, 2015 9:58 pm

nedsaid wrote:...At the point markets lost confidence to the extreme, the market valuation of GE Commercial Paper was zero. Was this an efficient price? Warren Buffett didn't think so and he provided GE with the liquidity it needed. Markets acted like GE didn't generate the cash flows to meet its debt obligations. Of course that was ridiculous.
This reminds me of an analogy that I believe was made by Buffett, but I can't seem to find the reference right now.
A good value hunter buys Christmas cards the day after Christmas when the markets have all the Christmas products marked half off the price. For someone who understands the intrinsic nature of Christmas cards, they know full well the value those cards continue to hold despite the temporary mark-down in the marketplace, and is likely to have a quite satisfactory result from their purchase come next Christmas season.
On the other hand, if you don't understand Christmas products very deeply, you might confuse the prior result and go out purchasing Christmas trees the day after Christmas and hoping to achieve the same result. Someone buying a half-price Christmas tree the day after Christmas and expecting a pleasant result next Christmas is likely to be disappointed.
The problem is, most of us aren't Warren Buffett, and we often tend to think we "know" more than we do. I have a friend that insisted he "knew" General Motors wasn't going to go away during the financial crisis, essentially 'too big to fail', and thought it was a great opportunity at a bargain. He was correct that they didn't "go away", but they still went bankrupt and under government conservatorship managed to re-emerge, but he lost all of his "investment".

There's another Buffett'ism along the lines of "It's far better to buy a wonderful company at a fair price than a fair company at a wonderful price.", I think a similar statement can be drawn that even paying too much for a good company is better than paying anything for a failing company. At least with the former, you have time and the ability of the still operating company to make something back.

For those of us who don't know the difference between Christmas cards and Christmas trees, a great strategy is to diversify buying both. If they're both half off and one goes to zero and one doubles we're at a zero sum. If we diversify across time we're buying some as the prices rise and some as prices fall and attain an average price. We have a wonderful strategy that can eliminate Mr.Markets crazy sentimental price movements and attain a long-run "average" result. This doesn't allow us to gain anything extra by out smarting others playing a speculative trading game, but it allows us to get closer to eliminating the speculative price changes, and closer to reaping the full investment return that comes from the growth and dividends yielded by the companies we own.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

User avatar
nedsaid
Posts: 11602
Joined: Fri Nov 23, 2012 12:33 pm

Re: No "Value Premium" in real mutual funds

Post by nedsaid » Sun Feb 15, 2015 10:13 pm

JoMoney, you are correct. None of us knows as much as we think we know and I think that includes Mr. Buffett. This is why successful investing takes a certain amount of courage. You never fully know what you think you know and even if you know a lot, you realize there are things out there you don't know. The military strategists call it the fog of war. Sometimes there is confusion out there.
A fool and his money are good for business.

Post Reply