Fama and French: The Five-Factor Model Revisited

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
Fremdon Ferndock
Posts: 1181
Joined: Fri Dec 24, 2021 11:26 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Fremdon Ferndock »

Bill Sharpe said that we have the greatest amount of historical data for the "market" and, even so, we can't be 100% sure that the model is correct. "Even if the Gods are kind, and distributions never change and even if you have lots of data you can still be far off for the premium for the whole market."

You can guess what he has to say about subsector "factors": "Anyone who thinks that looking at empirical data will produce a resolution to the question of whether the premium of small growth stocks is different from large value stocks with any degree of precision is just kidding himself."
"Risk is what’s left over when you think you’ve thought of everything." ~ Morgan Housel
rkhusky
Posts: 17764
Joined: Thu Aug 18, 2011 8:09 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by rkhusky »

muffins14 wrote: Tue Jan 18, 2022 11:04 am I take issue with the "same returns" comment -- maybe over a very long time period, but over shorter ones, the returns would certainly be different.

Even if the long-term average returns of SCV and LCG were equal, they are not 100% correlated, so they may produce their returns at different points in time, and under different market conditions. Therefore an SCV fund may not have the same returns as a TSM fund, even if there is no premium, because over any finite observation period, the two are exposed to different risks.

Market = Market Risk
SCV = Market Risk + Small-company risk + value-company risk

Even if premium you get for taking on small and value is zero, the non-zero correlations between them all may mean that you have a smoother ride than being only exposed to market risk. My thinking is that diversification across those risks would help prevent bad outcomes, like running out of money in retirement if the market does poorly (but small and value do OK, for example)
I was speaking about the output of factor models, which don’t model reality with 100% accuracy.

The correlations between components would show up in the alpha term or the noise/error term. They are not part of the main components.

If one neglects the alpha and noise/error terms, the returns of a factor model will be exactly the market returns if all the other premiums are zero.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Fama and French: The Five-Factor Model Revisited

Post by willthrill81 »

jeffyscott wrote: Tue Jan 18, 2022 12:01 pm
willthrill81 wrote: Tue Jan 18, 2022 11:33 am
burritoLover wrote: Tue Jan 18, 2022 11:27 am I think factor investing is probably an imperfect model and I know my SCV tilt may not deliver a premium going forward (or the premium may be reduced) but I'm not willing to place my entire retirement savings into a MCW equities allocation which is really a huge bet on large caps even if you invest in total market funds.
This is a key point that I believe is lost on many. Some state that SCV is a 'bet' on a small portion of the market (despite SCV funds typically being exposed to 600-1,000 stocks), but they have no problem 'betting' on large-caps, which are currently very concentrated among the biggest tech companies.
For me, this is a bigger reason for not using just a TSM fund, rather than any factor theory. I didn't like the idea back in the 1990s and I don't like it now, because I don't like the idea of, for example, putting something like 25% of my money in 10 stocks. And that those 10 stocks are typically those that have had their prices run up the most just adds to my discomfort.
It's even worse than that. In October of 2021, this source stated that the 10 largest companies in the S&P 500 represented 30% of the index, up from nearly 16% at one point in 2015. I don't believe that so much of the S&P 500 has ever been in just 10 companies.
The Sensible Steward
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by Apathizer »

wenchleaf wrote: Tue Jan 18, 2022 11:57 am
Apathizer wrote: Sat Jan 15, 2022 11:55 am
Fremdon Ferndock wrote: Sat Jan 15, 2022 9:19 am
So how well has Fama and French’s five-factor model explained returns over the decades? According to our analysis, only one factor has truly held up over all time periods.
hint: it isn't value and it isn't small

https://blogs.cfainstitute.org/investor ... revisited/
drumboy256 wrote: Sat Jan 15, 2022 10:37 am This goes to show that Avantis is ahead of the curve again in terms of quality. Not surprising really.
Size is especially weak in ex-US and emerging markets.
Any more info (sources or discussion) on this?
Here are links to a detailed paper and a video summarizing the research.
https://www.pwlcapital.com/resources/fi ... with-etfs/
https://www.youtube.com/watch?v=jKWbW7Wgm0w
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
Topic Author
Fremdon Ferndock
Posts: 1181
Joined: Fri Dec 24, 2021 11:26 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Fremdon Ferndock »

Using Portfolio Visualizer, I found that the annualized compound return from TSM from 1972 was 10.93% and for SCV is was 14.01%. If you look at the year by year returns, TSM and SCV pretty much go up and down together. I would have liked that juicy 14% from SCV but, you know what? I would have been happy enough with the 11% from TSM. I don't want to be the richest stiff in the cemetery anyway.

As they say: "don't let the search for the perfect distract you from the good."
"Risk is what’s left over when you think you’ve thought of everything." ~ Morgan Housel
User avatar
jason2459
Posts: 1208
Joined: Wed May 06, 2020 7:59 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by jason2459 »

Fremdon Ferndock wrote: Tue Jan 18, 2022 12:27 pm Using Portfolio Visualizer, I found that the annualized compound return from TSM from 1972 was 10.93% and for SCV is was 14.01%. If you look at the year by year returns, TSM and SCV pretty much go up and down together. I would have liked that juicy 14% from SCV but, you know what? I would have been happy enough with the 11% from TSM. I don't want to be the richest stiff in the cemetery anyway.

As they say: "don't let the search for the perfect distract you from the good."
That same argument can be the same for holding SCV the past decade. With LCG outperforming but SCV still providing a nice return.
"In the short run, the stock market is a voting machine; in the long run, it is a weighing machine" ~Benjamin Graham
muffins14
Posts: 5528
Joined: Wed Oct 26, 2016 4:14 am
Location: New York

Re: Fama and French: The Five-Factor Model Revisited

Post by muffins14 »

rkhusky wrote: Tue Jan 18, 2022 12:11 pm
muffins14 wrote: Tue Jan 18, 2022 11:04 am I take issue with the "same returns" comment -- maybe over a very long time period, but over shorter ones, the returns would certainly be different.

Even if the long-term average returns of SCV and LCG were equal, they are not 100% correlated, so they may produce their returns at different points in time, and under different market conditions. Therefore an SCV fund may not have the same returns as a TSM fund, even if there is no premium, because over any finite observation period, the two are exposed to different risks.

Market = Market Risk
SCV = Market Risk + Small-company risk + value-company risk

Even if premium you get for taking on small and value is zero, the non-zero correlations between them all may mean that you have a smoother ride than being only exposed to market risk. My thinking is that diversification across those risks would help prevent bad outcomes, like running out of money in retirement if the market does poorly (but small and value do OK, for example)
I was speaking about the output of factor models, which don’t model reality with 100% accuracy.

The correlations between components would show up in the alpha term or the noise/error term. They are not part of the main components.

If one neglects the alpha and noise/error terms, the returns of a factor model will be exactly the market returns if all the other premiums are zero.
I don’t think your construction is true.

0) best not to just ignore noise/error. That masks times when they return differently. I think it’s useful to talk about observed outcomes rather than just the expectation value. Or better yet, think about a range of possible expected outcomes across different time windows, which again relies on using the error term for inference.

1) in any regression, there can be correlation among the predictors. This is especially valid since as investors we typically use long-only funds, and there is correlation between long value and long small with the market, even if the long/short factors are independent

2) Even if there is no value premium on average over 100 years, the value factor can explain some variation in returns in short time windows like 0-5 years or 10-20 years, like when value does well or poorly relative to market.

Think of a model that ignores the value factor. Now part of the error in the price estimate from your model is because you’ve excluded that variable from your model. The error term will be correlated with the value factor if the premium changes over time, as we observe

Now include the value factor. You will be able to better explain returns in time periods when value differs from market, even if the value premium is zero on average, it is clearly not zero in finite time windows

I think that inference means that the factor is explaining why some stocks have different returns than others in some time periods

Maybe another analogy is that I could have treasuries for bonds only. There is some expected return from that.

Maybe I could instead invest in corporate bonds of shorter duration. Those two bond portfolios might have the same expected return, but they are exposed to different risks. One has credit risk and one does not. I think of SCV-correlated risk in the same way. Maybe the expected value is the same as market over the very long term, but the risks are different and show up at different times under different conditions
Last edited by muffins14 on Tue Jan 18, 2022 12:56 pm, edited 7 times in total.
Crom laughs at your Four Winds
Topic Author
Fremdon Ferndock
Posts: 1181
Joined: Fri Dec 24, 2021 11:26 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Fremdon Ferndock »

willthrill81 wrote: Tue Jan 18, 2022 12:12 pm
jeffyscott wrote: Tue Jan 18, 2022 12:01 pm
willthrill81 wrote: Tue Jan 18, 2022 11:33 am
burritoLover wrote: Tue Jan 18, 2022 11:27 am I think factor investing is probably an imperfect model and I know my SCV tilt may not deliver a premium going forward (or the premium may be reduced) but I'm not willing to place my entire retirement savings into a MCW equities allocation which is really a huge bet on large caps even if you invest in total market funds.
This is a key point that I believe is lost on many. Some state that SCV is a 'bet' on a small portion of the market (despite SCV funds typically being exposed to 600-1,000 stocks), but they have no problem 'betting' on large-caps, which are currently very concentrated among the biggest tech companies.
For me, this is a bigger reason for not using just a TSM fund, rather than any factor theory. I didn't like the idea back in the 1990s and I don't like it now, because I don't like the idea of, for example, putting something like 25% of my money in 10 stocks. And that those 10 stocks are typically those that have had their prices run up the most just adds to my discomfort.
It's even worse than that. In October of 2021, this source stated that the 10 largest companies in the S&P 500 represented 30% of the index, up from nearly 16% at one point in 2015. I don't believe that so much of the S&P 500 has ever been in just 10 companies.
But these companies represent an even larger percentage of the net income and profits than is reflected in their weighting in the S&P500. They aren't big for no reason, and nobody knows if they are getting even bigger. Or, we could just become active investors and see how that works out. I'm nervous too, so I decided to diversify into global ex-U.S. to achieve a better sector balance and diversify more broadly. At least that's consistent with a capital markets approach and avoids playing around with different pieces of the U.S. market.
"Risk is what’s left over when you think you’ve thought of everything." ~ Morgan Housel
rkhusky
Posts: 17764
Joined: Thu Aug 18, 2011 8:09 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by rkhusky »

muffins14 wrote: Tue Jan 18, 2022 12:40 pm
I don’t think your construction is true.

0) best not to just ignore noise/error. That masks times when they return differently. I think it’s useful to talk about observed outcomes rather than just the expectation value. Or better yet, think about a range of possible expected outcomes across different time windows, which again relies on using the error term for inference.

1) in any regression, there can be correlation among the predictors. This is especially valid since as investors we typically use long-only funds, and there is correlation between long value and long small with the market, even if the long/short factors are independent

2) Even if there is no value premium on average over 100 years, the value factor can explain some variation in returns in short time windows like 0-5 years or 10-20 years, like when value does well or poorly relative to market.
0) error/noise and alpha contain the unexplained parts of the actual return. Some can be reduced by using a better model.

1) the standard FF model doesn’t include correlation or cross terms

2) if the premiums are zero over 100 years, the return over 100 years is the market return. If the premiums over 10 years are nonzero, the return over 10 years will be different from the market, apart from a chance cancellation between factors.
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Northern Flicker »

nedsaid wrote: The thing is, if the Market portfolio contains all of the factors, then CAPM would explain 100% of the variances in return of stock portfolios, it only explains about 2/3 of the variances.
The market portfolio does not contain all of the factors. It is only exposed to the market factor. Factors are defined as exposure offsets from the market portfolio, not as market segments.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Fama and French: The Five-Factor Model Revisited

Post by willthrill81 »

Fremdon Ferndock wrote: Tue Jan 18, 2022 12:40 pm
willthrill81 wrote: Tue Jan 18, 2022 12:12 pm
jeffyscott wrote: Tue Jan 18, 2022 12:01 pm
willthrill81 wrote: Tue Jan 18, 2022 11:33 am
burritoLover wrote: Tue Jan 18, 2022 11:27 am I think factor investing is probably an imperfect model and I know my SCV tilt may not deliver a premium going forward (or the premium may be reduced) but I'm not willing to place my entire retirement savings into a MCW equities allocation which is really a huge bet on large caps even if you invest in total market funds.
This is a key point that I believe is lost on many. Some state that SCV is a 'bet' on a small portion of the market (despite SCV funds typically being exposed to 600-1,000 stocks), but they have no problem 'betting' on large-caps, which are currently very concentrated among the biggest tech companies.
For me, this is a bigger reason for not using just a TSM fund, rather than any factor theory. I didn't like the idea back in the 1990s and I don't like it now, because I don't like the idea of, for example, putting something like 25% of my money in 10 stocks. And that those 10 stocks are typically those that have had their prices run up the most just adds to my discomfort.
It's even worse than that. In October of 2021, this source stated that the 10 largest companies in the S&P 500 represented 30% of the index, up from nearly 16% at one point in 2015. I don't believe that so much of the S&P 500 has ever been in just 10 companies.
But these companies represent an even larger percentage of the net income and profits than is reflected in their weighting in the S&P500. They aren't big for no reason, and nobody knows if they are getting even bigger. Or, we could just become active investors and see how that works out. I'm nervous too, so I decided to diversify into global ex-U.S. to achieve a better sector balance and diversify more broadly. At least that's consistent with a capital markets approach and avoids playing around with different pieces of the U.S. market.
It's possible that these companies will just keep getting bigger and bigger, but it seems far more likely to me that they will eventually fall like all their predecessors have. Just look at how many of the largest companies were either comparatively tiny 20 years ago or didn't even exist at all. And historically, there's been a very strong tendency for such high concentration in a small number of companies in the S&P 500 to be followed by dramatic outperformance of SCV, as shown here. Will that persist in the future? Nobody knows, but if history rhymes at all, I know where I would put my money.
The Sensible Steward
garlandwhizzer
Posts: 3565
Joined: Fri Aug 06, 2010 3:42 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by garlandwhizzer »

Northern Flicker wrote:

The market portfolio does not contain all of the factors. It is only exposed to the market factor. Factors are defined as exposure offsets from the market portfolio, not as market segments.
This is true if you choose to use the definitions of factor models. I do not accept their definitions. Factor models describe beta as "just another single factor," but they don't treat beta like the non-beta factors. Non beta factors get cost-free long/short portfolios, beta gets long only. To be fair the beta return should be doubled by adding the returns from index that shorts beta. That automatically doubles the returns of non-beta factors relative to beta. Beta's long only return also suffers because the risk-free rate (presumably T-Bills) gets subtracted from it. I do not know whether the same occurs with non-beta factors or not. Furthermore, non-beta factors get cost free, friction free, illiquidity free unlimited trading. Beta does essentially no trading because it is the buy and hold market portfolio. To get a level playing field comparison of non-beta factors relative to beta, estimates of trading costs and trading frictions/illiquidity constraints should be subtracted from all non-beta factor returns, this done after you cut returns in half for no long/short. Trading frictions and illiquidity restraints are massive in the SC space, one reason why the harvestable small factor which was reputed to be substantial in factor models has in fact been zero in real funds since the 1982 (DFA micro-cap fund versus S&P 500) as nisi's post points out. But it's actually worse than zero because the DFA micro-cap fund suffered much more volatility to produce that zero premium for 40 years.

Why is beta not treated like the other factors? The answer is simple. Downgrading beta's return relative to non-beta factors makes the others look more impressive, makes a bigger splash in academic circles, more promotions and career advancements in academia, more high paying consulting jobs from funds companies who employ factor investing approaches. Quite simply the models do not create a level playing field when it comes to beta versus other factors. The models are designed to look impressive and if you accept their unrealistic assumptions and well-chosen rules and you allow them to choose factor definitions that optimize their returns in the rear view mirror, they succeed in doing this. You can also abandon P/B when it no longer works and switch to something else like Avantis funds have done. Problem is in another 10 or 20 years, those new Avantis factor approaches may also be abandoned because they too no longer work. When SCV factor definition itself is in a constant state of flux 30 years after its discovery what that tells you is that the secret sauce remains illusive.

How well the real factor funds translate into putting dollars in the pockets of investors is another question entirely. For the last 17 years they have produced substantial opportunity costs for true believing investors whose faith has not been deterred. Will that reverse in the future and non-beta factor funds make up all the lost ground relative to beta and actually outperform? That IMO is uncertain, although there are strong opinions on both sides of this question. What is clear is that so far factor investing in real funds has not been a gravy train relative to "one factor beta."

Garland Whizzer
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Fama and French: The Five-Factor Model Revisited

Post by vineviz »

garlandwhizzer wrote: Tue Jan 18, 2022 2:46 pm Non beta factors get cost-free long/short portfolios, beta gets long only.
Not true. Beta is a long/short portfolio also: long equities and short the risk-free asset.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
rkhusky
Posts: 17764
Joined: Thu Aug 18, 2011 8:09 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by rkhusky »

The FF factor model is set up the way it is because it works pretty well for a variety of diverse portfolios (doesn't work for everything though). I am sure FF tried all sorts of combinations before hitting on one that gave good results. Of course, they were guided by historical anomalies with value and small stocks.
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Northern Flicker »

garlandwhizzer wrote: This is true if you choose to use the definitions of factor models. I do not accept their definitions.
Whether or not you accept the formulation by Fama & French is a separate issue. I was responding to a posting that was referring to the Fama & French formulation.
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Northern Flicker »

garlandwhizzer wrote: Non beta factors get cost-free long/short portfolios, beta gets long only. To be fair the beta return should be doubled by adding the returns from index that shorts beta. That automatically doubles the returns of non-beta factors relative to beta. Beta's long only return also suffers because the risk-free rate (presumably T-Bills) gets subtracted from it.
Why does this matter? Nobody is claiming that all factors have premiums of equal magnitude, or that they should. Whatever part of a stock's expected return is accounted for by market return, various factor models attempt to explain the part that is not accounted for by market return.

Factors can be defined statically without long/short portfolios. There certainly are problems with the models, not the least of which is that stock returns do not appear to be linear functions of the properties that have been used to model expected returns. Maybe stock returns are linear functions of other, yet to be discovered variables, but linear/factor models seem to be limited at least some in their ability to fully model expected return by this behavior.

This applies even to CAPM, the original factor model. I don't think it is disputed that there is drift in the betas of individual stocks.
garlandwhizzer wrote: Furthermore, non-beta factors get cost free, friction free, illiquidity free unlimited trading.
There are legitimate questions concerning how well a given factor premium can be harvested. These are separate from a basic theoretical question of how well a given factor model models expected return.
User avatar
nedsaid
Posts: 19275
Joined: Fri Nov 23, 2012 11:33 am

Re: Fama and French: The Five-Factor Model Revisited

Post by nedsaid »

Northern Flicker wrote: Tue Jan 18, 2022 1:22 pm
nedsaid wrote: The thing is, if the Market portfolio contains all of the factors, then CAPM would explain 100% of the variances in return of stock portfolios, it only explains about 2/3 of the variances.
The market portfolio does not contain all of the factors. It is only exposed to the market factor. Factors are defined as exposure offsets from the market portfolio, not as market segments.
Not sure I stated it too well, somewhere I explained it like this:

By definition a market portfolio has only one factor: Market. If you want factor exposure in a portfolio, you have to tilt away from a market portfolio. For example, if you want Value exposure, you need to tilt the portfolio towards stocks that have Value characteristics, you favor Value stocks over other kinds of stocks. Yes, Value is not a market segment but I used the example of industry sectors to explain that stocks perform differently from each other, the stock market is not a monolith.

Utility stocks will behave differently from Tech stocks. Stocks with Value characteristics will behave differently than stocks with Growth characteristics. "Value" and "Growth" are more like ranges within a continuum, these stock characteristics are a matter of degree and not an exact definition. You hear such terms as "Deep Value", "Growth at a Reasonable Price," "Slow Growth", "Fast Growth", etc. These definitions can be in the eye of the beholder.
A fool and his money are good for business.
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Northern Flicker »

You are describing ways to construct portfolios with factor exposures. At the most fundamental level, a factor model is a model for the expected return of a single stock.
rkhusky
Posts: 17764
Joined: Thu Aug 18, 2011 8:09 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by rkhusky »

Northern Flicker wrote: Thu Jan 20, 2022 12:35 pm You are describing ways to construct portfolios with factor exposures. At the most fundamental level, a factor model is a model for the expected return of a single stock.
Since the factor model return is a linear combination of the returns of collections of stocks, I would expect that the model would not work well for most individual stocks.
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Northern Flicker »

rkhusky wrote: Thu Jan 20, 2022 12:56 pm
Northern Flicker wrote: Thu Jan 20, 2022 12:35 pm You are describing ways to construct portfolios with factor exposures. At the most fundamental level, a factor model is a model for the expected return of a single stock.
Since the factor model return is a linear combination of the returns of collections of stocks, I would expect that the model would not work well for most individual stocks.
It is defined for single stocks, and then extended to portfolios of stocks. That does not mean the model suggests holding a single stock.
Last edited by Northern Flicker on Thu Jan 20, 2022 7:23 pm, edited 1 time in total.
User avatar
nisiprius
Advisory Board
Posts: 52215
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: Fama and French: The Five-Factor Model Revisited

Post by nisiprius »

Fremdon Ferndock wrote: Tue Jan 18, 2022 12:04 pm Bill Sharpe said that we have the greatest amount of historical data for the "market" and, even so, we can't be 100% sure that the model is correct. "Even if the Gods are kind, and distributions never change and even if you have lots of data you can still be far off for the premium for the whole market."

You can guess what he has to say about subsector "factors": "Anyone who thinks that looking at empirical data will produce a resolution to the question of whether the premium of small growth stocks is different from large value stocks with any degree of precision is just kidding himself."
In 2013 I posted Endpoint sensitivity on "historic" data.

Image

The point, obvious but I haven't seen it presented this way elsewhere, is that even something as basic as "the historic return of the US stock market" varies from 9% to 11%, even when measured over very long periods of time, when you make fairly small changes in endpoints.

Instead of saying "the historic return of the stock market has been 10%," we should be saying "the historic return of the stock market has been 10%±1%."
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.
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Northern Flicker »

Fremdon Ferndock wrote: Bill Sharpe said that we have the greatest amount of historical data for the "market" and, even so, we can't be 100% sure that the model is correct. "Even if the Gods are kind, and distributions never change and even if you have lots of data you can still be far off for the premium for the whole market."
Having an outcome for the equity risk premium that differs from its historical average does not imply incorrect models.
Topic Author
Fremdon Ferndock
Posts: 1181
Joined: Fri Dec 24, 2021 11:26 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Fremdon Ferndock »

Bill Sharpe has argued that the findings of Fama and French, to oversimplify somewhat, say that relative to their beta values small cap stocks outperform large cap stocks and value stocks outperform growth stocks. Even over a long historical period he argues that these findings are an artifact of relatively small stock performance in a small corner of the market.

My take is this: Many people "tilt" based on the implicit assumption that this will result in higher portfolio returns going forward as has been the case looking backward. Overweighting any particular slice of the market will only improve your overall absolute equity returns if that slice has a higher absolute returns than the overall market consistently over a given period of time. That, in fact, was what was observed by Fama and French in regard to small cap and value stocks. If you expect, for example, that the returns for small cap or value stocks will be dependably higher than that for the market then it makes sense to overweight these "factors" in one's equity portfolio. If you are less certain, then it might make more sense to increase the relative weighting non-equity risk with similar volatility to stocks, such as commodities and long bonds, in order to "de-risk" your overall portfolio. In this case, there would be a weaker argument for "diversifying" into equity factors than for diversifying into assets that more clearly embody non-equity risk.

This translates into my belief that the speculative argument for "tilting" is strongest for investors who choose to maintain a very large allocation to equities and a small allocation to non-equities. That would apply to younger investors and perhaps to wealthier investors. The rest of us mere mortals might be better served by not investing in equity factor products.
"Risk is what’s left over when you think you’ve thought of everything." ~ Morgan Housel
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Fama and French: The Five-Factor Model Revisited

Post by vineviz »

Fremdon Ferndock wrote: Fri Jan 21, 2022 8:50 am Bill Sharpe has argued that the findings of Fama and French, to oversimplify somewhat, say that relative to their beta values small cap stocks outperform large cap stocks and value stocks outperform growth stocks.
I'd say there's no need to "argue" this point, since relative to their beta values is an explicit part of the Fama-French methodology.

Even over a long historical period he argues that these findings are an artifact of relatively small stock performance in a small corner of the market.
Fremdon Ferndock wrote: Fri Jan 21, 2022 8:50 amMy take is this: Many people "tilt" based on the implicit assumption that this will result in higher portfolio returns going forward as has been the case looking backward.
People "tilt" (which is a loaded word in an of itself) for all sorts of reasons, sometimes based on implicit assumptions and sometimes explicit assumptions. It's hard for me to know what MOST people are doing, but my experience has been that a significant fraction of factor-aware investors are operating with an explicit assumption that their multi-factor strategy is only exposing them to higher expected returns because it is also exposing them to more risk.
Fremdon Ferndock wrote: Fri Jan 21, 2022 8:50 am This translates into my belief that the speculative argument for "tilting" is strongest for investors who choose to maintain a very large allocation to equities and a small allocation to non-equities. That would apply to younger investors and perhaps to wealthier investors. The rest of us mere mortals might be better served by not investing in equity factor products.
I think the use of the word "speculative" is inflammatory and inaccurate, for the most part, but as a matter of portfolio construction the relative factor exposures of the equity portion of a portfolio do not depend on the ratio of stocks to bonds. If an investor finds it desirable to have half of their stocks in SCV when equity is 90% of the portfolio, they would rationally keep half their stocks in SCV when equity is 50% of the portfolio. And to the extent that investors with more bonds tend to closer to retirement age, these investors likely benefit more from diversification within the equity portfolio than younger investors because of the pernicious risk of a poor sequence of returns.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Topic Author
Fremdon Ferndock
Posts: 1181
Joined: Fri Dec 24, 2021 11:26 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Fremdon Ferndock »

People "tilt" (which is a loaded word in an of itself) for all sorts of reasons, sometimes based on implicit assumptions and sometimes explicit assumptions. It's hard for me to know what MOST people are doing, but my experience has been that a significant fraction of factor-aware investors are operating with an explicit assumption that their multi-factor strategy is only exposing them to higher expected returns because it is also exposing them to more risk.
Even if a "significant fraction" of investors assume this (but I'll wager that another large fraction think they're reducing risk by diversifying), the Fama-French model argues that the small and value factors affect expected returns, but they are agnostic with regard to whether this reflects risk premiums or not.
"Risk is what’s left over when you think you’ve thought of everything." ~ Morgan Housel
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Fama and French: The Five-Factor Model Revisited

Post by vineviz »

Fremdon Ferndock wrote: Fri Jan 21, 2022 11:08 am
People "tilt" (which is a loaded word in an of itself) for all sorts of reasons, sometimes based on implicit assumptions and sometimes explicit assumptions. It's hard for me to know what MOST people are doing, but my experience has been that a significant fraction of factor-aware investors are operating with an explicit assumption that their multi-factor strategy is only exposing them to higher expected returns because it is also exposing them to more risk.
Even if investors assume this (and I'm not sure I agree - some think they're reducing risk by diversifying), the Fama-French model argues that the small and value factors affect expected returns, but they are agnostic with regard to whether this reflects risk premiums or not.
That’s true of the asset pricing model itself, but there’s clearly a strong consensus belief that the factors are at least partly (and probably mostly) risk-based.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Topic Author
Fremdon Ferndock
Posts: 1181
Joined: Fri Dec 24, 2021 11:26 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Fremdon Ferndock »

vineviz wrote: Fri Jan 21, 2022 11:52 am
Fremdon Ferndock wrote: Fri Jan 21, 2022 11:08 am
People "tilt" (which is a loaded word in an of itself) for all sorts of reasons, sometimes based on implicit assumptions and sometimes explicit assumptions. It's hard for me to know what MOST people are doing, but my experience has been that a significant fraction of factor-aware investors are operating with an explicit assumption that their multi-factor strategy is only exposing them to higher expected returns because it is also exposing them to more risk.
Even if investors assume this (and I'm not sure I agree - some think they're reducing risk by diversifying), the Fama-French model argues that the small and value factors affect expected returns, but they are agnostic with regard to whether this reflects risk premiums or not.
That’s true of the asset pricing model itself, but there’s clearly a strong consensus belief that the factors are at least partly (and probably mostly) risk-based.
Yes true. But many factor proponents advocate that the additional risk factors diversify market risk and thus lower overall risk. Larry Swedroe has discussed this from a "risk parity" perspective. If I wanted to increase overall equity risk to boost expected returns, I could just use leverage to do that but clearly factor investors think they're doing something better than that.
"Risk is what’s left over when you think you’ve thought of everything." ~ Morgan Housel
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Fama and French: The Five-Factor Model Revisited

Post by willthrill81 »

Fremdon Ferndock wrote: Fri Jan 21, 2022 12:00 pm
vineviz wrote: Fri Jan 21, 2022 11:52 am
Fremdon Ferndock wrote: Fri Jan 21, 2022 11:08 am
People "tilt" (which is a loaded word in an of itself) for all sorts of reasons, sometimes based on implicit assumptions and sometimes explicit assumptions. It's hard for me to know what MOST people are doing, but my experience has been that a significant fraction of factor-aware investors are operating with an explicit assumption that their multi-factor strategy is only exposing them to higher expected returns because it is also exposing them to more risk.
Even if investors assume this (and I'm not sure I agree - some think they're reducing risk by diversifying), the Fama-French model argues that the small and value factors affect expected returns, but they are agnostic with regard to whether this reflects risk premiums or not.
That’s true of the asset pricing model itself, but there’s clearly a strong consensus belief that the factors are at least partly (and probably mostly) risk-based.
Yes true. But many factor proponents advocate that the additional risk factors diversify market risk and thus lower overall risk.
Data from the last 50+ years seems to support that. SCV has had about 40% lower start date sensitivity than TSM.
The Sensible Steward
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Fama and French: The Five-Factor Model Revisited

Post by vineviz »

Fremdon Ferndock wrote: Fri Jan 21, 2022 12:00 pm Yes true. But many factor proponents advocate that the additional risk factors diversify market risk and thus lower overall risk. Larry Swedroe has discussed this from a "risk parity" perspective. If I wanted to increase overall equity risk to boost expected returns, I could just use leverage to do that but clearly factor investors think they're doing something better than that.
Speaking as a "factor proponent" I think this misrepresents both the viewpoints of the people you're criticizing and the nature of diversification.

On the second point, it is entirely possible to both diversify AND increase the aggregate amount of risk at the same time. And this is a fundamental premise of the "Larry Portfolio": increasing the allocation to SCV diversifies the equity risk and increases equity risk, so Treasury bonds are added to de-risk the portfolio back to the original level.

Leveraging a market cap weighted equity portfolio accomplishes part of that process (increasing equity risk, leaving room to add bonds to de-risk), but not all of it. The diversification piece is omitted, since leveraging a fund does not -by itself - diversify the risk exposure of the portfolio.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by Apathizer »

willthrill81 wrote: Tue Jan 18, 2022 2:35 pm
Fremdon Ferndock wrote: Tue Jan 18, 2022 12:40 pm
willthrill81 wrote: Tue Jan 18, 2022 12:12 pm
jeffyscott wrote: Tue Jan 18, 2022 12:01 pm
willthrill81 wrote: Tue Jan 18, 2022 11:33 am

This is a key point that I believe is lost on many. Some state that SCV is a 'bet' on a small portion of the market (despite SCV funds typically being exposed to 600-1,000 stocks), but they have no problem 'betting' on large-caps, which are currently very concentrated among the biggest tech companies.
For me, this is a bigger reason for not using just a TSM fund, rather than any factor theory. I didn't like the idea back in the 1990s and I don't like it now, because I don't like the idea of, for example, putting something like 25% of my money in 10 stocks. And that those 10 stocks are typically those that have had their prices run up the most just adds to my discomfort.
It's even worse than that. In October of 2021, this source stated that the 10 largest companies in the S&P 500 represented 30% of the index, up from nearly 16% at one point in 2015. I don't believe that so much of the S&P 500 has ever been in just 10 companies.
But these companies represent an even larger percentage of the net income and profits than is reflected in their weighting in the S&P500. They aren't big for no reason, and nobody knows if they are getting even bigger. Or, we could just become active investors and see how that works out. I'm nervous too, so I decided to diversify into global ex-U.S. to achieve a better sector balance and diversify more broadly. At least that's consistent with a capital markets approach and avoids playing around with different pieces of the U.S. market.
It's possible that these companies will just keep getting bigger and bigger, but it seems far more likely to me that they will eventually fall like all their predecessors have. Just look at how many of the largest companies were either comparatively tiny 20 years ago or didn't even exist at all. And historically, there's been a very strong tendency for such high concentration in a small number of companies in the S&P 500 to be followed by dramatic outperformance of SCV, as shown here. Will that persist in the future? Nobody knows, but if history rhymes at all, I know where I would put my money.
I agree, except for the highlighted portion. I don't necessarily expect FAANG companies to fail, but am concerned they're so large and expensive as to have an unreasonably high market-cap. I don't expect these companies to fail, but think their future returns will be much lower. I still hold these companies but have shifted my allocation towards less expensive companies, both US and ex-US, with strong fundamentals relative to price.

As I've said before, maybe I'm wrong. Maybe US large-growth will continue to have disproportionately stellar returns. If that happens I'll still capture some of them. Allocating more to smaller, less expensive companies should serve nicely as hedge if LG is overvalued and future returns lag the market.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Fama and French: The Five-Factor Model Revisited

Post by willthrill81 »

Apathizer wrote: Fri Jan 21, 2022 12:54 pm
willthrill81 wrote: Tue Jan 18, 2022 2:35 pm
Fremdon Ferndock wrote: Tue Jan 18, 2022 12:40 pm
willthrill81 wrote: Tue Jan 18, 2022 12:12 pm
jeffyscott wrote: Tue Jan 18, 2022 12:01 pm

For me, this is a bigger reason for not using just a TSM fund, rather than any factor theory. I didn't like the idea back in the 1990s and I don't like it now, because I don't like the idea of, for example, putting something like 25% of my money in 10 stocks. And that those 10 stocks are typically those that have had their prices run up the most just adds to my discomfort.
It's even worse than that. In October of 2021, this source stated that the 10 largest companies in the S&P 500 represented 30% of the index, up from nearly 16% at one point in 2015. I don't believe that so much of the S&P 500 has ever been in just 10 companies.
But these companies represent an even larger percentage of the net income and profits than is reflected in their weighting in the S&P500. They aren't big for no reason, and nobody knows if they are getting even bigger. Or, we could just become active investors and see how that works out. I'm nervous too, so I decided to diversify into global ex-U.S. to achieve a better sector balance and diversify more broadly. At least that's consistent with a capital markets approach and avoids playing around with different pieces of the U.S. market.
It's possible that these companies will just keep getting bigger and bigger, but it seems far more likely to me that they will eventually fall like all their predecessors have. Just look at how many of the largest companies were either comparatively tiny 20 years ago or didn't even exist at all. And historically, there's been a very strong tendency for such high concentration in a small number of companies in the S&P 500 to be followed by dramatic outperformance of SCV, as shown here. Will that persist in the future? Nobody knows, but if history rhymes at all, I know where I would put my money.
I agree, except for the highlighted portion. I don't necessarily expect FAANG companies to fail, but am concerned they're so large and expensive as to have an unreasonably high market-cap. I don't expect these companies to fail, but think their future returns will be much lower. I still hold these companies but have shifted my allocation towards less expensive companies, both US and ex-US, with strong fundamentals relative to price.

As I've said before, maybe I'm wrong. Maybe US large-growth will continue to have disproportionately stellar returns. If that happens I'll still capture some of them. Allocating more to smaller, less expensive companies should serve nicely as hedge if LG is overvalued and future returns lag the market.
Everything we've seen in capitalistic markets indicates that the incumbents eventually fall from their position of dominance at the very least. How long it takes for that to happen has varied tremendously though.

Heck, we've already seen this happen in the last few years. Netflix was part of FAANG for years but has fallen tremendously in the ranks, being now replaced by Tesla, which has surpassed Facebook in market cap.
The Sensible Steward
investyoumust
Posts: 142
Joined: Thu Jan 20, 2022 3:55 am

Re: Fama and French: The Five-Factor Model Revisited

Post by investyoumust »

willthrill81 wrote: Fri Jan 21, 2022 1:53 pm
Everything we've seen in capitalistic markets indicates that the incumbents eventually fall from their position of dominance at the very least. How long it takes for that to happen has varied tremendously though.

Heck, we've already seen this happen in the last few years. Netflix was part of FAANG for years but has fallen tremendously in the ranks, being now replaced by Tesla, which has surpassed Facebook in market cap.
Exactly. Warren Buffett talked about this at the last annual meeting: https://www.youtube.com/watch?v=yFcESoqtF_Y
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Fama and French: The Five-Factor Model Revisited

Post by willthrill81 »

investyoumust wrote: Fri Jan 21, 2022 1:56 pm
willthrill81 wrote: Fri Jan 21, 2022 1:53 pm
Everything we've seen in capitalistic markets indicates that the incumbents eventually fall from their position of dominance at the very least. How long it takes for that to happen has varied tremendously though.

Heck, we've already seen this happen in the last few years. Netflix was part of FAANG for years but has fallen tremendously in the ranks, being now replaced by Tesla, which has surpassed Facebook in market cap.
Exactly. Warren Buffett talked about this at the last annual meeting: https://www.youtube.com/watch?v=yFcESoqtF_Y
To further demonstrate the point, I just found this information below.

In Q2 of 1999, the largest companies in the U.S. by market cap were as follows:
Microsoft
GE
Cisco
Exxon
Walmart
Intel
NTT
Lucent Technologies
Nokia
BP

Just 5 years later, look at how much the same list changed (in Q2 of 2004):
GE
Exxon
Microsoft
Pfizer
Citi
Walmart
BP
AIG
Intel
Bank of America

Jump forward to 2014:
Apple
Exxon
Alphabet
Microsoft
Berkshire Hathaway
Johnson & Johnson
Shell
GE
Wells Fargo
Roche

And now:
Apple
Microsoft
Alphabet
Amazon
Tesla
Facebook
Berkshire Hathaway
NVIDIA
Visa
United Health

Believing that the companies on top right now will be dominant over the long-term seems more like 'this time, it's different' than anything based on knowledge of history.
The Sensible Steward
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by Apathizer »

willthrill81 wrote: Fri Jan 21, 2022 1:53 pm
Apathizer wrote: Fri Jan 21, 2022 12:54 pm
willthrill81 wrote: Tue Jan 18, 2022 2:35 pm
Fremdon Ferndock wrote: Tue Jan 18, 2022 12:40 pm
willthrill81 wrote: Tue Jan 18, 2022 12:12 pm

It's even worse than that. In October of 2021, this source stated that the 10 largest companies in the S&P 500 represented 30% of the index, up from nearly 16% at one point in 2015. I don't believe that so much of the S&P 500 has ever been in just 10 companies.
But these companies represent an even larger percentage of the net income and profits than is reflected in their weighting in the S&P500. They aren't big for no reason, and nobody knows if they are getting even bigger. Or, we could just become active investors and see how that works out. I'm nervous too, so I decided to diversify into global ex-U.S. to achieve a better sector balance and diversify more broadly. At least that's consistent with a capital markets approach and avoids playing around with different pieces of the U.S. market.
It's possible that these companies will just keep getting bigger and bigger, but it seems far more likely to me that they will eventually fall like all their predecessors have. Just look at how many of the largest companies were either comparatively tiny 20 years ago or didn't even exist at all. And historically, there's been a very strong tendency for such high concentration in a small number of companies in the S&P 500 to be followed by dramatic outperformance of SCV, as shown here. Will that persist in the future? Nobody knows, but if history rhymes at all, I know where I would put my money.
I agree, except for the highlighted portion. I don't necessarily expect FAANG companies to fail, but am concerned they're so large and expensive as to have an unreasonably high market-cap. I don't expect these companies to fail, but think their future returns will be much lower. I still hold these companies but have shifted my allocation towards less expensive companies, both US and ex-US, with strong fundamentals relative to price.

As I've said before, maybe I'm wrong. Maybe US large-growth will continue to have disproportionately stellar returns. If that happens I'll still capture some of them. Allocating more to smaller, less expensive companies should serve nicely as hedge if LG is overvalued and future returns lag the market.
Everything we've seen in capitalistic markets indicates that the incumbents eventually fall from their position of dominance at the very least. How long it takes for that to happen has varied tremendously though.

Heck, we've already seen this happen in the last few years. Netflix was part of FAANG for years but has fallen tremendously in the ranks, being now replaced by Tesla, which has surpassed Facebook in market cap.
I think we might be arguing over semantics. Falling isn't the same as failing. It seems likely they'll fall (drop), but I don't necessarily expect them to fail, at least not most of them.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
investyoumust
Posts: 142
Joined: Thu Jan 20, 2022 3:55 am

Re: Fama and French: The Five-Factor Model Revisited

Post by investyoumust »

Apathizer wrote: Fri Jan 21, 2022 2:19 pm I think we might be arguing over semantics. Falling isn't the same as failing. It seems likely they'll fall (drop), but I don't necessarily expect them to fail, at least not most of them.
Some of them will undoubtedly fall into the steps of Coca Cola. Great company, no doubt, but bid up too high and subsequent multi-decade low returns.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Fama and French: The Five-Factor Model Revisited

Post by willthrill81 »

Apathizer wrote: Fri Jan 21, 2022 2:19 pm I think we might be arguing over semantics. Falling isn't the same as failing. It seems likely they'll fall (drop), but I don't necessarily expect them to fail, at least not most of them.
I agree, but they don't have to fail in order to be terrible investments. For instance, every one of the top ten stocks by market cap in the year 2000 went on to trail the market for the next 18 years, as discussed here. Only Microsoft finally came back to outperform the S&P 500 over the entirety of the period. Five of those ten companies had negative returns, and two completely failed.
The Sensible Steward
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Northern Flicker »

Fremdon Ferndock wrote: Fri Jan 21, 2022 11:08 am
People "tilt" (which is a loaded word in an of itself) for all sorts of reasons, sometimes based on implicit assumptions and sometimes explicit assumptions. It's hard for me to know what MOST people are doing, but my experience has been that a significant fraction of factor-aware investors are operating with an explicit assumption that their multi-factor strategy is only exposing them to higher expected returns because it is also exposing them to more risk.
Even if a "significant fraction" of investors assume this (but I'll wager that another large fraction think they're reducing risk by diversifying), the Fama-French model argues that the small and value factors affect expected returns, but they are agnostic with regard to whether this reflects risk premiums or not.
The risk premium discounted into a stock by the market based on whatever systematic exposures it has probably does not cover the risk measures that would be the focus of a prudent retirement saver.
km91
Posts: 1388
Joined: Wed Oct 13, 2021 12:32 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by km91 »

willthrill81 wrote: Fri Jan 21, 2022 2:29 pm
Apathizer wrote: Fri Jan 21, 2022 2:19 pm I think we might be arguing over semantics. Falling isn't the same as failing. It seems likely they'll fall (drop), but I don't necessarily expect them to fail, at least not most of them.
I agree, but they don't have to fail in order to be terrible investments. For instance, every one of the top ten stocks by market cap in the year 2000 went on to trail the market for the next 18 years, as discussed here. Only Microsoft finally came back to outperform the S&P 500 over the entirety of the period. Five of those ten companies had negative returns, and two completely failed.
How should an investor factoring this in to their asset allocation decisions? The fact that the incumbents that dominate the market cap of the S&P will eventually be surpassed or that they go on to underperform the index doesn't seem to be relevant, by definition they will be surpassed by the companies who outperform. Look at the composition of the S&P 10 or 20 years ago. The index did just fine even though companies like Exxon and GE fell out of the top holdings. And I definitely wouldn't be tilting small cap value based on this. I look at the holdings of AVUV and I see a lot of once large cap names that fell out of the S&P; Alcoa, Macy's, Ryder. SCV investors are going to be buying XOM in 2050 when we're all driving around in electric cars
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Fama and French: The Five-Factor Model Revisited

Post by willthrill81 »

km91 wrote: Fri Jan 21, 2022 4:25 pm
willthrill81 wrote: Fri Jan 21, 2022 2:29 pm
Apathizer wrote: Fri Jan 21, 2022 2:19 pm I think we might be arguing over semantics. Falling isn't the same as failing. It seems likely they'll fall (drop), but I don't necessarily expect them to fail, at least not most of them.
I agree, but they don't have to fail in order to be terrible investments. For instance, every one of the top ten stocks by market cap in the year 2000 went on to trail the market for the next 18 years, as discussed here. Only Microsoft finally came back to outperform the S&P 500 over the entirety of the period. Five of those ten companies had negative returns, and two completely failed.
How should an investor factoring this in to their asset allocation decisions? The fact that the incumbents that dominate the market cap of the S&P will eventually be surpassed or that they go on to underperform the index doesn't seem to be relevant, by definition they will be surpassed by the companies who outperform. Look at the composition of the S&P 10 or 20 years ago. The index did just fine even though companies like Exxon and GE fell out of the top holdings. And I definitely wouldn't be tilting small cap value based on this. I look at the holdings of AVUV and I see a lot of once large cap names that fell out of the S&P; Alcoa, Macy's, Ryder. SCV investors are going to be buying XOM in 2050 when we're all driving around in electric cars
I'm not knocking the S&P 500. Rather, much of this discussion was predicated on the stock market having reached unprecedented concentration in the top 10 largest companies by market cap, and we know from the past that there has been a strong relationship between such concentration and the subsequent outperformance of SCV.

But large-caps (which is what the S&P 500 is) have had very erratic performance. Many have crowed about how LC has outperformed SCV for over a decade, but they seldom refer to the significant real losses it encountered from 2000-2009 when SCV still turned out a respectable +5% annualized real return. This is precisely what we expect from a single factor like market beta, which is all that the S&P 500 is exposed to; it's going to behave more erratically over time than a combination of multiple, independent factors.

Some of today's SCV firms will probably become one of tomorrow's market cap leaders. Some of them are yesterday's winners that are today's 'losers'. That's the nature of indexing.
The Sensible Steward
km91
Posts: 1388
Joined: Wed Oct 13, 2021 12:32 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by km91 »

willthrill81 wrote: Fri Jan 21, 2022 4:35 pm
km91 wrote: Fri Jan 21, 2022 4:25 pm
willthrill81 wrote: Fri Jan 21, 2022 2:29 pm
Apathizer wrote: Fri Jan 21, 2022 2:19 pm I think we might be arguing over semantics. Falling isn't the same as failing. It seems likely they'll fall (drop), but I don't necessarily expect them to fail, at least not most of them.
I agree, but they don't have to fail in order to be terrible investments. For instance, every one of the top ten stocks by market cap in the year 2000 went on to trail the market for the next 18 years, as discussed here. Only Microsoft finally came back to outperform the S&P 500 over the entirety of the period. Five of those ten companies had negative returns, and two completely failed.
How should an investor factoring this in to their asset allocation decisions? The fact that the incumbents that dominate the market cap of the S&P will eventually be surpassed or that they go on to underperform the index doesn't seem to be relevant, by definition they will be surpassed by the companies who outperform. Look at the composition of the S&P 10 or 20 years ago. The index did just fine even though companies like Exxon and GE fell out of the top holdings. And I definitely wouldn't be tilting small cap value based on this. I look at the holdings of AVUV and I see a lot of once large cap names that fell out of the S&P; Alcoa, Macy's, Ryder. SCV investors are going to be buying XOM in 2050 when we're all driving around in electric cars
I'm not knocking the S&P 500. Rather, much of this discussion was predicated on the stock market having reached unprecedented concentration in the top 10 largest companies by market cap, and we know from the past that there has been a strong relationship between such concentration and the subsequent outperformance of SCV.

But large-caps (which is what the S&P 500 is) have had very erratic performance. Many have crowed about how LC has outperformed SCV for over a decade, but they seldom refer to the significant real losses it encountered from 2000-2009 when SCV still turned out a respectable +5% annualized real return. This is precisely what we expect from a single factor like market beta, which is all that the S&P 500 is exposed to; it's going to behave more erratically over time than a combination of multiple, independent factors.

Some of today's SCV firms will probably become one of tomorrow's market cap leaders. Some of them are yesterday's winners that are today's 'losers'. That's the nature of indexing.
Can we capture the SCV premium by holding a total market index, or is it just that relative to the weighting of large caps in the index it would be barely noticeable? Dumb question that I should probably know, what is the reason that SCV is a non diversifiable risk factor?
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Fama and French: The Five-Factor Model Revisited

Post by willthrill81 »

km91 wrote: Fri Jan 21, 2022 5:03 pm
willthrill81 wrote: Fri Jan 21, 2022 4:35 pm
km91 wrote: Fri Jan 21, 2022 4:25 pm
willthrill81 wrote: Fri Jan 21, 2022 2:29 pm
Apathizer wrote: Fri Jan 21, 2022 2:19 pm I think we might be arguing over semantics. Falling isn't the same as failing. It seems likely they'll fall (drop), but I don't necessarily expect them to fail, at least not most of them.
I agree, but they don't have to fail in order to be terrible investments. For instance, every one of the top ten stocks by market cap in the year 2000 went on to trail the market for the next 18 years, as discussed here. Only Microsoft finally came back to outperform the S&P 500 over the entirety of the period. Five of those ten companies had negative returns, and two completely failed.
How should an investor factoring this in to their asset allocation decisions? The fact that the incumbents that dominate the market cap of the S&P will eventually be surpassed or that they go on to underperform the index doesn't seem to be relevant, by definition they will be surpassed by the companies who outperform. Look at the composition of the S&P 10 or 20 years ago. The index did just fine even though companies like Exxon and GE fell out of the top holdings. And I definitely wouldn't be tilting small cap value based on this. I look at the holdings of AVUV and I see a lot of once large cap names that fell out of the S&P; Alcoa, Macy's, Ryder. SCV investors are going to be buying XOM in 2050 when we're all driving around in electric cars
I'm not knocking the S&P 500. Rather, much of this discussion was predicated on the stock market having reached unprecedented concentration in the top 10 largest companies by market cap, and we know from the past that there has been a strong relationship between such concentration and the subsequent outperformance of SCV.

But large-caps (which is what the S&P 500 is) have had very erratic performance. Many have crowed about how LC has outperformed SCV for over a decade, but they seldom refer to the significant real losses it encountered from 2000-2009 when SCV still turned out a respectable +5% annualized real return. This is precisely what we expect from a single factor like market beta, which is all that the S&P 500 is exposed to; it's going to behave more erratically over time than a combination of multiple, independent factors.

Some of today's SCV firms will probably become one of tomorrow's market cap leaders. Some of them are yesterday's winners that are today's 'losers'. That's the nature of indexing.
Can we capture the SCV premium by holding a total market index, or is it just that relative to the weighting of large caps in the index it would be barely noticeable? Dumb question that I should probably know, what is the reason that SCV is a non diversifiable risk factor?
No, you have no exposure to SCV by holding TSM because in addition to SCV companies you are also holding the antithesis to SCV, which is LCG, as well as SCG, LCV, etc.

I don't know quite what you mean by your last question.
The Sensible Steward
km91
Posts: 1388
Joined: Wed Oct 13, 2021 12:32 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by km91 »

willthrill81 wrote: Fri Jan 21, 2022 5:09 pm
km91 wrote: Fri Jan 21, 2022 5:03 pm
willthrill81 wrote: Fri Jan 21, 2022 4:35 pm
km91 wrote: Fri Jan 21, 2022 4:25 pm
willthrill81 wrote: Fri Jan 21, 2022 2:29 pm

I agree, but they don't have to fail in order to be terrible investments. For instance, every one of the top ten stocks by market cap in the year 2000 went on to trail the market for the next 18 years, as discussed here. Only Microsoft finally came back to outperform the S&P 500 over the entirety of the period. Five of those ten companies had negative returns, and two completely failed.
How should an investor factoring this in to their asset allocation decisions? The fact that the incumbents that dominate the market cap of the S&P will eventually be surpassed or that they go on to underperform the index doesn't seem to be relevant, by definition they will be surpassed by the companies who outperform. Look at the composition of the S&P 10 or 20 years ago. The index did just fine even though companies like Exxon and GE fell out of the top holdings. And I definitely wouldn't be tilting small cap value based on this. I look at the holdings of AVUV and I see a lot of once large cap names that fell out of the S&P; Alcoa, Macy's, Ryder. SCV investors are going to be buying XOM in 2050 when we're all driving around in electric cars
I'm not knocking the S&P 500. Rather, much of this discussion was predicated on the stock market having reached unprecedented concentration in the top 10 largest companies by market cap, and we know from the past that there has been a strong relationship between such concentration and the subsequent outperformance of SCV.

But large-caps (which is what the S&P 500 is) have had very erratic performance. Many have crowed about how LC has outperformed SCV for over a decade, but they seldom refer to the significant real losses it encountered from 2000-2009 when SCV still turned out a respectable +5% annualized real return. This is precisely what we expect from a single factor like market beta, which is all that the S&P 500 is exposed to; it's going to behave more erratically over time than a combination of multiple, independent factors.

Some of today's SCV firms will probably become one of tomorrow's market cap leaders. Some of them are yesterday's winners that are today's 'losers'. That's the nature of indexing.
Can we capture the SCV premium by holding a total market index, or is it just that relative to the weighting of large caps in the index it would be barely noticeable? Dumb question that I should probably know, what is the reason that SCV is a non diversifiable risk factor?
No, you have no exposure to SCV by holding TSM because in addition to SCV companies you are also holding the antithesis to SCV, which is LCG, as well as SCG, LCV, etc.

I don't know quite what you mean by your last question.
Wouldn't that imply that the only way to be truly exposed to SCV would be to hold 100% of equities as SCV?

What I'm trying to ask is what is the underlying economic rationale that explains SCV as a systematic premium. Why does the small cap value premium exist and why should we believe that it is a systematic risk factor that the market compensates investors for holding, and not just an anomaly in the data?
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Fama and French: The Five-Factor Model Revisited

Post by willthrill81 »

km91 wrote: Fri Jan 21, 2022 5:44 pm Wouldn't that imply that the only way to be truly exposed to SCV would be to hold 100% of equities as SCV?
No. You would merely need to tilt toward SCV in order to be exposed to it. About 3% of TSM is in SCV, so as long as your SCV holdings represent more than that 3%, you're tilted toward it.
km91 wrote: Fri Jan 21, 2022 5:44 pm What I'm trying to ask is what is the underlying economic rationale that explains SCV as a systematic premium. Why does the small cap value premium exist and why should we believe that it is a systematic risk factor that the market compensates investors for holding, and not just an anomaly in the data?
That's a topic that I'll leave for others. Volumes have been written about it.
The Sensible Steward
km91
Posts: 1388
Joined: Wed Oct 13, 2021 12:32 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by km91 »

willthrill81 wrote: Fri Jan 21, 2022 5:57 pm
That's a topic that I'll leave for others. Volumes have been written about it.
What should I read? My understanding of the small cap premium is that it's to compensate for the liquidity risk of holding small cap names vs blue chip large caps. Not sure if that's been proven or disproven but it makes sense to me. The value premium I have a bit more trouble grasping. Maybe it has to do with more operating risk/uncertainty in value names, but maybe this is already captured by the market risk premium. To me value seems like a market bias. Investors over paying for growth and under pay for value is a tale as old as markets and the affect is amplified in small caps because the mispricing might not get arbitraged away as quickly. SCV performed well after severe recessions in the 80's and 2000 - 2010 so it's probably sensitive to domestic macro factors but I'm not sure that suggests a systematic risk premium as it seems like an investor could diversify a strong exposure to the business cycle by holding large cap and international stocks
User avatar
vanbogle59
Posts: 1314
Joined: Wed Mar 10, 2021 7:30 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by vanbogle59 »

willthrill81 wrote: Fri Jan 21, 2022 5:09 pm Can we capture the SCV premium by holding a total market index, or is it just that relative to the weighting of large caps in the index it would be barely noticeable? Dumb question that I should probably know, what is the reason that SCV is a non diversifiable risk factor?
No, you have no exposure to SCV by holding TSM because in addition to SCV companies you are also holding the antithesis to SCV, which is LCG, as well as SCG, LCV, etc.

I don't know quite what you mean by your last question.
[/quote]

OK, this is getting way too hand-wavy for me.
It is nonsensical to say that you have no exposure to stuff you own.

That can only be asserted in the view that SCV is a thing and the actual companies are not things.
I accept that the Factor model shows no deviation if all you own is TSM. But the model can't come first, the actual stocks have to have epistemological priority. :shock:

So, of course owners of TSM are exposed to SCV stocks (albeit very little).
But the model attempting to describe the behavior of a TSM portfolio cannot invoke the SCV factor if the composition of the portfolio is pure market weight.
km91
Posts: 1388
Joined: Wed Oct 13, 2021 12:32 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by km91 »

vanbogle59 wrote: Fri Jan 21, 2022 8:23 pm
OK, this is getting way too hand-wavy for me.
It is nonsensical to say that you have no exposure to stuff you own.

That can only be asserted in the view that SCV is a thing and the actual companies are not things.
I accept that the Factor model shows no deviation if all you own is TSM. But the model can't come first, the actual stocks have to have epistemological priority. :shock:

So, of course owners of TSM are exposed to SCV stocks (albeit very little).
But the model attempting to describe the behavior of a TSM portfolio cannot invoke the SCV factor if the composition of the portfolio is pure market weight.
The factor portfolios constructed by FF are long/short. To earn the size premium for example the investor is long a basket of small stocks and short a basket of large stocks and the spread between the two is the factor premium. The small premium only exists relative to large stocks. Holding just TSM doesn't capture the premium because you need to be relatively overweight SCV vs TSM to have a net exposure to the risk factor, or at least that is how I am understanding it
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Fama and French: The Five-Factor Model Revisited

Post by willthrill81 »

vanbogle59 wrote: Fri Jan 21, 2022 8:23 pm
willthrill81 wrote: Fri Jan 21, 2022 5:09 pm No, you have no exposure to SCV by holding TSM because in addition to SCV companies you are also holding the antithesis to SCV, which is LCG, as well as SCG, LCV, etc.
OK, this is getting way too hand-wavy for me.
It is nonsensical to say that you have no exposure to stuff you own.
I wasn't precise in my above response. Yes, you have exposure to the SCV companies when you own TSM, but you also own all of the 'anti-SCV' companies that effectively cancels out the potential premiums for size and value. That's why the long-term return differential between S&P 500, which is only large-caps, and TSM has been close to zero.

If you want exposure to the size and value premiums, then you must own more of the small and value companies than the overall market cap weighting.

Further, SCV only represents about 3% of TSM, which wouldn't be enough to move the needle even if it weren't for the 'anti-SCV' companies in TSM.
km91 wrote: Fri Jan 21, 2022 8:50 pmHolding just TSM doesn't capture the premium because you need to be relatively overweight SCV vs TSM to have a net exposure to the risk factor, or at least that is how I am understanding it
Correct.
The Sensible Steward
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Fama and French: The Five-Factor Model Revisited

Post by Northern Flicker »

km91 wrote: The factor portfolios constructed by FF are long/short. To earn the size premium for example the investor is long a basket of small stocks and short a basket of large stocks and the spread between the two is the factor premium.
I think the main purpose of the long/short portfolios is to try to measure the historical premia and their correlations empirically. I don't think F & F intended for these portfolios to be held by investors, and they are not how factors are defined. You first have to define a factor before you can specify a long/short portfolio to try to measure the factor premium.

My view is that the market factor is historically the most rewarded and most significant driver of equity returns. You don't want to throw the baby out with the bathwater by shorting out the market return from your portfolio.

You can see the factor exposures in long only portfolios when doing factor regressions:

https://www.portfoliovisualizer.com/fac ... sion=false

or:

https://www.portfoliovisualizer.com/fac ... tion2_1=20
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by Apathizer »

willthrill81 wrote: Fri Jan 21, 2022 9:05 pm
vanbogle59 wrote: Fri Jan 21, 2022 8:23 pm
willthrill81 wrote: Fri Jan 21, 2022 5:09 pm No, you have no exposure to SCV by holding TSM because in addition to SCV companies you are also holding the antithesis to SCV, which is LCG, as well as SCG, LCV, etc.
OK, this is getting way too hand-wavy for me.
It is nonsensical to say that you have no exposure to stuff you own.
I wasn't precise in my above response. Yes, you have exposure to the SCV companies when you own TSM, but you also own all of the 'anti-SCV' companies that effectively cancels out the potential premiums for size and value. That's why the long-term return differential between S&P 500, which is only large-caps, and TSM has been close to zero.

If you want exposure to the size and value premiums, then you must own more of the small and value companies than the overall market cap weighting.

Further, SCV only represents about 3% of TSM, which wouldn't be enough to move the needle even if it weren't for the 'anti-SCV' companies in TSM.
km91 wrote: Fri Jan 21, 2022 8:50 pmHolding just TSM doesn't capture the premium because you need to be relatively overweight SCV vs TSM to have a net exposure to the risk factor, or at least that is how I am understanding it
Correct.
I wouldn't phrase non-SCV stocks as 'anti-SCV', though I guess that's sort of correct. But I guess I'm just being picky about semantics.

Yes, to capture the SCV risk-premium you must hold them in higher weights than a cap-weighted total market index like VTI. This is a good vid explaining the SCV premium. He discusses weights beginning at about the 13:20 time-point. Since it predates Avantis, the Funds he discusses are a little dated, but the general principles of SCV are still relevant.
https://www.youtube.com/watch?v=2MVSsVi1_e4
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
km91
Posts: 1388
Joined: Wed Oct 13, 2021 12:32 pm

Re: Fama and French: The Five-Factor Model Revisited

Post by km91 »

Apathizer wrote: Fri Jan 21, 2022 10:23 pm
willthrill81 wrote: Fri Jan 21, 2022 9:05 pm
vanbogle59 wrote: Fri Jan 21, 2022 8:23 pm
willthrill81 wrote: Fri Jan 21, 2022 5:09 pm No, you have no exposure to SCV by holding TSM because in addition to SCV companies you are also holding the antithesis to SCV, which is LCG, as well as SCG, LCV, etc.
OK, this is getting way too hand-wavy for me.
It is nonsensical to say that you have no exposure to stuff you own.
I wasn't precise in my above response. Yes, you have exposure to the SCV companies when you own TSM, but you also own all of the 'anti-SCV' companies that effectively cancels out the potential premiums for size and value. That's why the long-term return differential between S&P 500, which is only large-caps, and TSM has been close to zero.

If you want exposure to the size and value premiums, then you must own more of the small and value companies than the overall market cap weighting.

Further, SCV only represents about 3% of TSM, which wouldn't be enough to move the needle even if it weren't for the 'anti-SCV' companies in TSM.
km91 wrote: Fri Jan 21, 2022 8:50 pmHolding just TSM doesn't capture the premium because you need to be relatively overweight SCV vs TSM to have a net exposure to the risk factor, or at least that is how I am understanding it
Correct.
I wouldn't phrase non-SCV stocks as 'anti-SCV', though I guess that's sort of correct. But I guess I'm just being picky about semantics.

Yes, to capture the SCV risk-premium you must hold them in higher weights than a cap-weighted total market index like VTI. This is a good vid explaining the SCV premium. He discusses weights beginning at about the 13:20 time-point. Since it predates Avantis, the Funds he discusses are a little dated, but the general principles of SCV are still relevant.
https://www.youtube.com/watch?v=2MVSsVi1_e4
Lol I listen to his podcast, this definitely wasn't the face I would picture when I'd hear his voice. At the 10 minute mark he talks about the persistence of the small and value factors, but seems to point the the historical premiums in the back test as evidence of their continued existence. This doesn't explain to me why value in particular is a systematic risk that the market compensates investors for. Couldn't an investor just hold growth stocks in combination with value and diversify away whatever risk there might be. It's not clear to me what the actual risk being taken is when overweighting to towards value.
Post Reply