Grantham v. Fama-French on risk and reward

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.

Grantham v. Fama-French on risk and reward

Postby Robert T » Mon Apr 26, 2010 9:20 am

.
The second part of Grantham’s latest quarterly letter has some interesting insight into the source of his diverging views with Fama-French on risk and reward - and it deserves some attention from investors IMO.
http://www.gmo.com/websitecontent/JGLetter_ALL_1Q10.pdf

On what they agree on: P/B and size are risk factors

Grantham wrote:I think P/B and yield and price-to-earnings (P/E) are risk factors. They have less fundamental quality and are therefore more prone to failure in rare crashes. I think this is the one thing Fama and French got right – for the wrong reasons.

Exhibit 3 in Grantham’s letter shows both low p/b and small caps to be ‘lower quality’ companies (lower levels and stability of profits and higher levels of debt).

On what they disagree on: The reliability of the relationship between risk and reward

As Grantham doesn’t seem to believe risk and reward are reliably related he seems to think it was sheer luck that Fama-French came to the conclusion that P/B and size were risk factors based solely on analysis explaining the past variability of returns. As Grantham says:

Grantham wrote:“The real behavioral market is perfectly happy not rewarding “risk” when it feels like it, as is shown by the 70-year underperformance of high beta stocks. But this time it worked.”

This is where Grantham looses me. Data shows a long-term premium for low p/b and small cap stock portfolios for US and international markets as far back as data is available. This suggests to me risk and reward are related. And the Fama-French research show this to be the case by the fact that the variability in past returns can be explained by P/B (HmL) and size (SmB), which Grantham shows in his quarterly letter to be risk factors (?).

To supplement Grantham's arguments he points to the ‘alpha’ on GMOs intrinsic value series (returns not due to exposure to p/B and size risk).

Gantham wrote:“What this means is that any outperformance on our intrinsic value is pure alpha, where for P/B, etc., and for small cap it is a risk premium, and a risk that definitely comes to bite you every so often.”

But if we look at the FF3F regression on the GMO Intrinsic Value, alpha disappears after accounting for exposure to the risk factors.
Code: Select all
August 1999 - March 2010

                            Alpha (t-stat)  Beta   Size   Value   R^2

GMO Intrinsic value        -0.06  (-0.47)   0.86  -0.17   0.37   0.91


Finally, and more specifically on quality. Grantham says:

Grantham wrote:But what about “Quality?” This factor has outperformed forever. (The S&P had a High Grade Index that started in 1925 and handsomely outperformed the S&P 500 to the end of 1965 when our data starts.) Since the market is efficient, to Fama and French quality must be a risk factor!

On ‘quality outperforming forever”:
    - From 1966 to 1993 (from Exhibit 6 in the article) – that’s 27 years – its cumulative performance was slightly below the S&P500.
    - From 2004 to date, since inception of the GMO Quality fund, it has underperformed the S&P 500 by about 0.5% per year.
So wouldn’t call this forever. And not sure what happened from 1994 to 2003. If high quality does outperform (if this is indeed the case), is this an anomaly, or an impending future correction? I don’t know.

So as investors what should we focus on?

The reasonably strong relationship between risk and reward (as reflected in the historical data across countries, for which there are reasonable explanations), or the ‘small set’ of apparently anomalous relationships for which there is little explanation? Personally, I go with the former over the long-term.

Robert
.
User avatar
Robert T
 
Posts: 1824
Joined: Tue Feb 27, 2007 10:40 pm
Location: 1, 0.2, 0.4, 0.5

Postby edge » Mon Apr 26, 2010 11:17 am

Thanks Robert. Very interesting. I find it odd that Grantham would make some of these easily refutable claims.
edge
 
Posts: 1865
Joined: Mon Feb 19, 2007 8:44 pm
Location: Great Falls VA

Postby BlueEars » Tue Apr 27, 2010 12:34 am

I've not completely read the Grantham letter yet but this stood out for me:
Everybody wanted to be a value manager by 1983 because it had done so
dazzlingly well since 1974. It had beaten the market by over 100 percentage points! The growth managers were
hiding under the table. Yet from 1984, because value investing became so trendy, you made no extra money in the cheapest P/B (value stocks) for 19.5 years! Now that takes patience! You were paid absolutely nothing extra for carrying the lower fundamental quality that P/B
represents.

So I checked one data analysis of mine and got this for Jan 84 to Jan 04 CAGR's:
LV 14.1%
LG 11.3%
SV 15.3%
SG 13.2%

the smb/hml values I used were (LV & SV were courtesy of Robert from some time ago which cover the period July 95 to Feb 09):
LV -0.17/0.43 (based on MSCI index)
LG -0.10/-0.33
SV 0.39/0.75 (based on MSCI index)
SG 0.87/0.12

So assuming the CAGR's are roughly correct on my part (hopefully somebody will check this), these results are contrary to Grantham's assertions on value versus growth. Could Grantham be defining value & growth differently? What smb/hml values would he use?
User avatar
BlueEars
 
Posts: 3159
Joined: Sat Mar 10, 2007 1:15 am
Location: West Coast

Postby Robert T » Tue Apr 27, 2010 7:12 am

.
Les,

Code: Select all
Grantham is about right on the 19 yrs.

Annualized return (%)

                 1974-83      1984-02 
                            (19 years)

S&P500            10.6         12.2 
FF LV(exU)        17.2         11.0 
CRSP6-10          23.8         10.0
CRSP9-10          25.8          8.9
FF SV(exU)        28.5         12.1   


.
User avatar
Robert T
 
Posts: 1824
Joined: Tue Feb 27, 2007 10:40 pm
Location: 1, 0.2, 0.4, 0.5

Postby BlueEars » Tue Apr 27, 2010 11:26 am

Hi Robert,

Looking at another source, the Simba backtest data, which uses a combo of the Russel indexes and Vanguard indexes for 1984 to 2002 we see:
Code: Select all
LV     11.8%
LG     11.6%
SV     11.2%
SG      7.0%
SP500  12.0%

I tend to view the SP500 as a blend. This data looks pretty favorable to value for the period. At least it wasn't a bust but I suppose you could agree with Grantham that for this period you did not get paid to take on the added risk. That risk is discussed in connection with Grantham's exhibit #1.

Grantham's exhibit #1 where he shows value taking an extremely long period of time to recoup from the 1929 crash should make any value tilted investor a little anxious. About exhibit #1 he says:
And when there is a depression or a crash,
then the companies start to get cut – to go out
of business – and “value” companies get to take serious
pain. We sent someone into the stacks to get data from
1929 to 1932 (he nearly died of dust inhalation). This
data (Exhibit 1) is completely proprietary and it must be
said that some contradictory data has also been dug out of
the archives. If this data is correct, as we believe, then it
certainly shows the hidden risk of low P/B.

That is what made me look at the more available value/growth data for the later 1984-2002 period. Also wonder about the "contradictory data".

Any thoughts on Grantham's exhibit #1 and/or his definitions of value & growth?
User avatar
BlueEars
 
Posts: 3159
Joined: Sat Mar 10, 2007 1:15 am
Location: West Coast

Postby Robert T » Tue Apr 27, 2010 1:53 pm

.
Les,

Les wrote:Any thoughts on Grantham's exhibit #1 and/or his definitions of value & growth?

Ken French has quintile data for p/b sorts on his website. His corresponding multiple for the ‘cheapest’ quintile is closer to 10 than 14 – if I recall (don’t have it in front of me). In addition – the 41 years is not what actually happened. Its calculated using the ‘expected’ risk premium of 2% per year. Again, if I recall, it took about 11 years to catch (pass in a significant way) the highest p/b quintile of stocks. i.e. the actual risk premium was much higher than 2% for the 11 years post 1932.

It still puzzles me why he uses the 2% expected return and 41 years (when the actual is markedly different). Swensen quotes the same numbers in his first (1998) book, if I recall correctly, quoting an earlier analysis from Grantham (which I think is what is used in the article).

Robert
.
User avatar
Robert T
 
Posts: 1824
Joined: Tue Feb 27, 2007 10:40 pm
Location: 1, 0.2, 0.4, 0.5

Postby BlueEars » Tue Apr 27, 2010 4:24 pm

Robert T wrote:...(snip)...
It still puzzles me why he uses the 2% expected return and 41 years (when the actual is markedly different). Swensen quotes the same numbers in his first (1998) book, if I recall correctly, quoting an earlier analysis from Grantham (which I think is what is used in the article). ...

So I gather that you are not taking the exhibit #1 seriously. Although we see that value tends to do worse in a decline like 1929, it still feels like there is not a severe penalty for value tilting other then tracking error over a few years. Or one might say the penalty has not shown up yet.
User avatar
BlueEars
 
Posts: 3159
Joined: Sat Mar 10, 2007 1:15 am
Location: West Coast

Postby Robert T » Tue Apr 27, 2010 5:05 pm

Les wrote:So I gather that you are not taking the exhibit #1 seriously. Although we see that value tends to do worse in a decline like 1929, it still feels like there is not a severe penalty for value tilting other then tracking error over a few years. Or one might say the penalty has not shown up yet.

Les,

I take the exhibit 1 chart seriously, but less so what’s written under it (re: the 41 years). If it really did take 41 years to catch up IMO the (expected) value premium would likely be higher than the equity premium or no-one would own value stocks (risk/reward). But over the great depression it’s a fact that low p/b declined by more than high p/b, just at a time when unemployment spiked. Risk showing up just at the wrong time IMO, a characteristic which contributes to the long-term ‘expected premium’. So I am in the Grantham and Fama-French camp that the value premium is a risk premium – not a free lunch.

Robert
.
User avatar
Robert T
 
Posts: 1824
Joined: Tue Feb 27, 2007 10:40 pm
Location: 1, 0.2, 0.4, 0.5

Postby BlueEars » Tue Apr 27, 2010 6:20 pm

Robert, thanks for your thoughts on this.
User avatar
BlueEars
 
Posts: 3159
Joined: Sat Mar 10, 2007 1:15 am
Location: West Coast

Postby tms » Wed Apr 28, 2010 9:23 am

It seems like Grantham is saying yes, P/B and size are risk factors, but that there won't necessarily be a reward in the form of higher returns.

Usually, the critique goes in the opposite direction. Yes, the higher returns have been there, but they are not risk factors, it's just data mining.
tms
 
Posts: 400
Joined: Mon Dec 15, 2008 12:44 pm

Postby maxfax » Wed Apr 28, 2010 11:40 am

Excellent article. Always good to challenge the doctrine.

I had a problem with the analysis that measured "# of years to recover to starting point". Seems to me he was analyzing risk of losses, so measuring the percent drop from the top would have been a more appropriate measure. Holding the same asset while it recovers says nothing about its risk of loss.

But his main point that 'cheapness' has a negative fundamental value accords with my POV.
maxfax
 
Posts: 422
Joined: Thu Mar 11, 2010 1:28 pm

Postby Kenster1 » Wed Apr 28, 2010 11:59 am

Good discussion guys and interesting perspective.

Just wanted to throw this out there since we're on the topic of Grantham and his liking now for big high-quality stocks....

http://www.bloomberg.com/apps/news?pid= ... 34U&pos=15

With big stocks trading at the steepest discount to small companies since at least 1982, the list of investors who share Perkins’s view includes Jeremy Grantham of Grantham Mayo Van Otterloo Co. and Donald Yacktman of Yacktman Asset Management Co. They are among the best-performing bargain hunters of the stock world, value managers who scoop up companies they think others have overlooked.

The premium investors are paying today to own small- capitalization stocks versus their large counterparts is the biggest in at least 27 years, said James Floyd, senior analyst at Leuthold Group LLC, a research firm based in Minneapolis. Leuthold defines large stocks as those with a market value of more than $9 billion and small stocks as those from $300 million to $1.4 billion.

Grantham, 71, wrote at the end of March that “high- quality” U.S. stocks would return 6.1 percent a year above the rate of inflation for the next seven years compared with a loss of 1.2 percent for small-cap stocks. Quality stocks have high, stable returns on capital and low debt, Grantham wrote on GMO’s Web site.

Microsoft, Johnson & Johnson

The largest holdings in the GMO Quality Fund include Redmond, Washington-based Microsoft Corp. and New Brunswick, New Jersey-based Johnson & Johnson. Both have higher returns on equity and a lower ratio of total debt to equity than the average for the S&P 500, Bloomberg data show.

Grantham, chief investment strategist at Boston-based GMO, is best known for his gloomy, and often accurate, long-term forecasts. In 2000, he predicted that U.S. stocks would lose money in the coming decade.

Grantham and his colleagues come up with their value targets by assuming a long-term average price-to-earnings ratio for stocks and applying it to a long-term average for profit margins.

“Quality stocks are pretty damn cheap right now,” said Ben Inker, director of asset allocation at GMO in a telephone interview. Because they are more stable and better able to withstand shocks, blue chips will outperform by an especially wide margin in turbulent times, said Inker.
Last edited by Kenster1 on Wed Apr 28, 2010 12:06 pm, edited 1 time in total.
SURGEON GENERAL'S WARNING: Any overconfidence in your ability, willingness and need to take risk may be hazardous to your health.
User avatar
Kenster1
 
Posts: 2871
Joined: Wed Feb 28, 2007 11:50 pm

Postby Rob't » Wed Apr 28, 2010 12:05 pm

It is a very stimulating article.

I was interested in his speculations on the reasons why the benefit of low Price-to-Book disappeared for long intervals. He introduces the idea of "valuation" of P/B affect. Did anyone make anything of Exhibit 4 where the benefit of holding the riskier value stocks decreases as the difference between their BtM and that of the total market declines. Is it possible that the risk of Small Value is not beneficial when the total market has a less disparate Price to Book?
Rob't
 
Posts: 177
Joined: Thu Apr 19, 2007 9:33 pm

Postby BlueEars » Wed Apr 28, 2010 12:31 pm

This is a plot using French-Fama data and Robert's smb/hml numbers from recent year's MSCI index. The data is also CPI adjusted to account for the deflation in the 1930's. If I did this right then it looks like it took roughly 8 years to recover to the 1929 peak. It wasn't until the 1940's when the value tilted portfolio would have started to outperform the market. Of course, all this is a bit theoretical since nobody back then had an MSCI index and there are probably other objections one could make. Still somewhat interesting torturing of the data :).

Image
User avatar
BlueEars
 
Posts: 3159
Joined: Sat Mar 10, 2007 1:15 am
Location: West Coast

Postby Blue » Thu Apr 29, 2010 1:43 am

Robert T wrote:.

It still puzzles me why he uses the 2% expected return and 41 years (when the actual is markedly different).


Excellent point. I didn't pick up on this distinction on my first read.

Seems like the expected small-value premium should be highest when small-value is at the lowest, analagous to a higher expected ERP in a deep bear market. Not sure why a bright guy like Grantham would take the approach that he did of assuming a flat 200 bp premium.
User avatar
Blue
 
Posts: 1017
Joined: Sat Jul 12, 2008 11:18 pm


Return to Investing - Theory, News & General

Who is online

Users browsing this forum: clast, Exabot [Bot], FAST Enterprise [Crawler], ftobin, Google [Bot] and 51 guests