Fama and French: The Five-Factor Model Revisited

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

Seasonal wrote: Sun Jan 16, 2022 9:02 am
nedsaid wrote: Sat Jan 15, 2022 12:51 pm
Seasonal wrote: Sat Jan 15, 2022 12:42 pm
nedsaid wrote: Sat Jan 15, 2022 12:17 pm The Size effect was real, it stopped working when too many people knew about it.
By the time any of us know about a factor, especially if we wait until it seems reliable, too many people know about it. Information disseminates very rapidly these days.
nedsaid wrote: Sat Jan 15, 2022 12:17 pm So you probably need to combine factors a bit, the precise combination is a trade secret or proprietary information, but I don't believe the factors are dead. Eventually, even the professionals who ought to know better will make the behavioral errors that created the factor premiums in the first place.
The market portfolio is a combination of all factors in a way that the market "regards" as appropriate.
The Market Portfolio by definition is the Market Factor. A market portfolio contains Value stocks, contains Small stocks, contains Quality stocks, and so on but does not contain the factors themselves except for the Market factor.

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. If the Market Portfolio contains all of the factors, then all stock portfolios should perform about the same, adjusted for risk. We know that isn't true. This all started when researchers found that high volatility stocks performed less well than predicted and that low volatility stocks performed better than the CAPM model predicted. Clearly something else was at work here.
Why should the that be the case? CAPM is a simplistic model relying solely on mean and variance.
A simplistic model was mostly true, we knew that returns increased as you increased volatility as compared to the S&P 500. However there were limitations to what CAPM could do, researchers found that high volatility stocks performed less well than predicted and that low volatility stocks performed better than expected. Researchers realized a more complex model was need in order to account for the differences in return from stock portfolio to stock portfolio. This model accounted for about 2/3 of the differences. Not bad.

When Size and Value were added, then the new 3 factor model accounted for about 90%. When you account for five factors, they got up to 96%. There is only about 4% of the variation in return that the Academics cannot account for. This is another reason that stock picking has become more difficult.

As far as the math behind this, there are people here with a better grasp of statistics who could better explain it.

Models are imperfect, people have pointed out flaws, and there has been endless debate over whether factor premiums really exist or not.
A fool and his money are good for business.
User avatar
nisiprius
Advisory Board
Posts: 52212
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 »

nedsaid wrote: Sun Jan 16, 2022 9:41 amRight after I bought a Micro-Cap Index ETF, I read a Morningstar article that said investors were better off in a Small-Cap Index. Oh well. I still own the ETF, never sold.
Really? It would be interesting to know more.

I know that Rick Ferri complained that the one year there was finally a really juicy payoff in the micro cap indexes, the micro cap index funds and ETFs he looked at fell ludicrously short of capturing it.

Here it is. By "ludicrous" I mean "The Wilshire US Micro-Cap index was up 47.6%" but his four funds returned 26.7%, 23.7%, 20%, and 18%--in a year when large-caps returned 24.8%.

The first fund offered by Dimensional Fund Advisors (DFA), the DFA US Micro Cap Portfolio (DFSCX)--as far as I know, the first fund intended to capture the supposed small firm premium in a passive way--was specifically designed to capture the CRSP 9th and 10th deciles. Part of DFA's reputation came from was in the techniques they devised for dealing with the poor investibility of the smallest stocks.

I've noted that DFSCX has to date, after forty years, failed to outperform the S&P 500, but I don't know if it failed to capture the 9th-and-10th-decile performance, or whether the 9th-and-10th-decile performance itself has failed to outperform the market.

I don't know if DFSCX is able to do anything that ETFs can't do. Since DFA now offers a "small cap ETF" but not a "micro cap ETF," maybe there is an issue relating to ETF structure.
Last edited by nisiprius on Sun Jan 16, 2022 10:06 am, edited 1 time in total.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
User avatar
jason2459
Posts: 1208
Joined: Wed May 06, 2020 7:59 pm

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

Post by jason2459 »

nedsaid wrote: Sun Jan 16, 2022 9:41 am
jason2459 wrote: Sat Jan 15, 2022 10:28 pm
nedsaid wrote: Sat Jan 15, 2022 9:14 pm
JoMoney wrote: Sat Jan 15, 2022 11:53 am
nisiprius wrote: Sat Jan 15, 2022 11:44 am...
Forty years after launch, despite a great start for the first three years, DFSCX hasn't made a penny more than an S&P 500 index fund...
And that doesn't even account for the fact that for the longest time, the only way to get access to DFA's secret-sauce index like non-index funds was to pay an advisor half a percent (or more) annually.

If one had looked to Vanguard's low cost small-cap index fund results would have been lower. Some people would discount the index fund as being poorly represented by the Russell 2000 over early years (despite the Russell 2000 + Russell 1000 perfectly representing the 'Total Market' Russell 3000) :?
Avantis has solved the problem you discuss above, they offer an excellent suite of ETFs that offer an updated factors approach very similar to DFA. Avantis will soon launch a Small Cap ETF that will invest in Micro-Caps. No need for an Advisor to access.
Even before Avantis, DFA used profitablity as well. And for passive index funds the S&P indices have a set of quality like qualification to get on and stay in the listings. So even though the SP600 value index mentions three specific methods to determine value there's also the initial quality screen to get into the SP 600 universe. I know of three funds that use that index VIOV, SLYV, and IJS.

AVSC Avantis' smaller approach was released this past week. It's avg market cap is 1.62B right now

AVUS avg market cap 2.71B

VIOV avg market cap 1.96B

Just interesting stats.

Because then there's the SP600 pure value index like the RZV ETF with a market cap avg of 1.08B

So is going down into the microcaps really worth it? Is SmB to the extreme going to to pull something out of the hat? I don't think so

SP600 value vs pure value since at least 2006
https://www.portfoliovisualizer.com/bac ... ion2_2=100

Don't know where I'm really going with this. Just something fun to look up.
Right after I bought a Micro-Cap Index ETF, I read a Morningstar article that said investors were better off in a Small-Cap Index. Oh well. I still own the ETF, never sold.

I still own some pink slip stocks from the late 90's that are about worthless. I can't believe they're still around. Just a reminder to me of my Robinhood days.
"In the short run, the stock market is a voting machine; in the long run, it is a weighing machine" ~Benjamin Graham
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 »

Nathan Drake wrote: Sat Jan 15, 2022 10:40 pm
And, as mentioned, even very popular index funds (S&P 500) have quality screens, so this is not uncommon for even "TSM" funds.
That is my understanding as well, one reason I like the S&P Indexes, they screen out the junkier companies. A well constructed index, even a Total Market Index, doesn't contain ALL stocks, many have market caps and trading volumes too small to be investable. I know Vanguard uses a sampling method for Micro-Cap stocks.

I use to wonder about the "anti-factors", two of which would be Value traps and Lottery stocks, good Index construction could filter a lot of the worst companies out. As I recall, one of the Russell indexes gets criticized for being an easy benchmark for active managers to beat because it doesn't filter out the bad companies. I wish someone would comment here. I think it is their Mid/Small-Cap Index, the Russell 2000?

So the well constructed indexes might have a bit of the Quality factor already built in.
A fool and his money are good for business.
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 »

nisiprius wrote: Sun Jan 16, 2022 9:58 am
nedsaid wrote: Sun Jan 16, 2022 9:41 amRight after I bought a Micro-Cap Index ETF, I read a Morningstar article that said investors were better off in a Small-Cap Index. Oh well. I still own the ETF, never sold.
Really? It would be interesting to know more.

I know that Rick Ferri complained that the one year there was finally a really juicy payoff in the micro cap indexes, the micro cap index funds and ETFs he looked at fell ludicrously short of capturing it.

The first fund offered by Dimensional Fund Advisors (DFA), the DFA US Micro Cap Portfolio (DFSCX)--as far as I know, the first fund intended to capture the supposed small firm premium in a passive way--was specifically designed to capture the CRSP 9th and 10th deciles, and part of their fame and brilliance was in the techniques they devised for dealing with the poor investibility of the smallest stocks.

I've noted that DFSCX has to date, after forty years, failed to outperform the S&P 500, but I don't know if it failed to capture the 9th-and-10th-decile performance, or whether the 9th-and-10th-decile performance itself has failed to outperform the market.
One of the problems was with front-running. By the way, the ticker is (IWC) iShares Micro-Cap ETF.
Probably the Dimensional Funds product did a better job as Dimensional doesn't announce to the world beforehand which stocks it will buy and sell.

I saw the article probably not too long after I got really involved in the factor discussions, my guess would be maybe 2013-2014. I bought the ETF in 2007, it was recommended by Merriman.
A fool and his money are good for business.
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 »

nedsaid wrote: Sun Jan 16, 2022 10:01 am
Nathan Drake wrote: Sat Jan 15, 2022 10:40 pm
And, as mentioned, even very popular index funds (S&P 500) have quality screens, so this is not uncommon for even "TSM" funds.
That is my understanding as well, one reason I like the S&P Indexes, they screen out the junkier companies. A well constructed index, even a Total Market Index, doesn't contain ALL stocks, many have market caps and trading volumes too small to be investable. I know Vanguard uses a sampling method for Micro-Cap stocks.

I use to wonder about the "anti-factors", two of which would be Value traps and Lottery stocks, good Index construction could filter a lot of the worst companies out. As I recall, one of the Russell indexes gets criticized for being an easy benchmark for active managers to beat because it doesn't filter out the bad companies. I wish someone would comment here. I think it is their Mid/Small-Cap Index, the Russell 2000?

So the well constructed indexes might have a bit of the Quality factor already built in.

S&P has a very robust governance and methodology process and some can argue that it makes them more of a managed active index.

They have standard reconstitution dates like other passive index funds but they have processes that can kick off at anytime to handle multiple different scenarios like M&A activities or bankruptcy. This really helps drive out the junk quicker. Like look at some of the big funds like VTI they have junk on the bottom that is worthless and sticks around until the next reconstitution date. Some funds it's quarterly and some every six months or more. That could be a long time to drop a zombie company or one that went belly up.

https://www.spglobal.com/spdji/en/governance/
"In the short run, the stock market is a voting machine; in the long run, it is a weighing machine" ~Benjamin Graham
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 »

Another issue with Micro-Cap indexes is that since the trading volumes are much lower for Micro-Caps than for Large-Caps, you get market impact costs when an institution buys and sells Micro-Caps. I suppose sampling techniques are used to try to replicate the index, it probably is not practical to own all of the stocks in the Micro-Cap index. This might be an area of the market where you would actually want a lower turnover active strategy. Front running and market impact costs are probably the reasons that Micro-Cap Index fund performance fell short of the actual Index performance.

What I mean by market impact costs, is that buying large quantities of a thinly traded Micro-Cap stock can push the price of the stock up and that selling large quantities of a thinly traded Micro-Cap stock can push the price down. Market impact costs for Large Caps exist but they are quite small, particularly when you execute patient trading techniques; market impact costs for Micro-Caps can be quite substantial.

This is an example that factor premiums that exist in the data are more difficult to capture in actual practice.
A fool and his money are good for business.
Seasonal
Posts: 3045
Joined: Sun May 21, 2017 1:49 pm

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

Post by Seasonal »

nedsaid wrote: Sun Jan 16, 2022 9:52 am
Seasonal wrote: Sun Jan 16, 2022 9:02 am
nedsaid wrote: Sat Jan 15, 2022 12:51 pm
Seasonal wrote: Sat Jan 15, 2022 12:42 pm
nedsaid wrote: Sat Jan 15, 2022 12:17 pm The Size effect was real, it stopped working when too many people knew about it.
By the time any of us know about a factor, especially if we wait until it seems reliable, too many people know about it. Information disseminates very rapidly these days.
nedsaid wrote: Sat Jan 15, 2022 12:17 pm So you probably need to combine factors a bit, the precise combination is a trade secret or proprietary information, but I don't believe the factors are dead. Eventually, even the professionals who ought to know better will make the behavioral errors that created the factor premiums in the first place.
The market portfolio is a combination of all factors in a way that the market "regards" as appropriate.
The Market Portfolio by definition is the Market Factor. A market portfolio contains Value stocks, contains Small stocks, contains Quality stocks, and so on but does not contain the factors themselves except for the Market factor.

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. If the Market Portfolio contains all of the factors, then all stock portfolios should perform about the same, adjusted for risk. We know that isn't true. This all started when researchers found that high volatility stocks performed less well than predicted and that low volatility stocks performed better than the CAPM model predicted. Clearly something else was at work here.
Why should the that be the case? CAPM is a simplistic model relying solely on mean and variance.
that
A simplistic model was mostly true, we knew that returns increased as you increased volatility as compared to the S&P 500. However there were limitations to what CAPM could do, researchers found that high volatility stocks performed less well than predicted and that low volatility stocks performed better than expected. Researchers realized a more complex model was need in order to account for the differences in return from stock portfolio to stock portfolio. This model accounted for about 2/3 of the differences. Not bad.

When Size and Value were added, then the new 3 factor model accounted for about 90%. When you account for five factors, they got up to 96%. There is only about 4% of the variation in return that the Academics cannot account for. This is another reason that stock picking has become more difficult.

As far as the math behind this, there are people here with a better grasp of statistics who could better explain it.

Models are imperfect, people have pointed out flaws, and there has been endless debate over whether factor premiums really exist or not.
I made two points. One was that if you are right that the size effect stopped working because it become too well known, then that should also be true about other well known factors. You did not dispute that. The second is that the market portfolio is the mix of factors that the market regards as appropriate. Note that the market may regard a net zero tilt of a factor as appropriate. You do not appear to dispute that either.
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 »

jason2459 wrote: Sun Jan 16, 2022 10:16 am
nedsaid wrote: Sun Jan 16, 2022 10:01 am
Nathan Drake wrote: Sat Jan 15, 2022 10:40 pm
And, as mentioned, even very popular index funds (S&P 500) have quality screens, so this is not uncommon for even "TSM" funds.
That is my understanding as well, one reason I like the S&P Indexes, they screen out the junkier companies. A well constructed index, even a Total Market Index, doesn't contain ALL stocks, many have market caps and trading volumes too small to be investable. I know Vanguard uses a sampling method for Micro-Cap stocks.

I use to wonder about the "anti-factors", two of which would be Value traps and Lottery stocks, good Index construction could filter a lot of the worst companies out. As I recall, one of the Russell indexes gets criticized for being an easy benchmark for active managers to beat because it doesn't filter out the bad companies. I wish someone would comment here. I think it is their Mid/Small-Cap Index, the Russell 2000?

So the well constructed indexes might have a bit of the Quality factor already built in.

S&P has a very robust governance and methodology process and some can argue that it makes them more of a managed active index.

They have standard reconstitution dates like other passive index funds but they have processes that can kick off at anytime to handle multiple different scenarios like M&A activities or bankruptcy. This really helps drive out the junk quicker. Like look at some of the big funds like VTI they have junk on the bottom that is worthless and sticks around until the next reconstitution date. Some funds it's quarterly and some every six months or more. That could be a long time to drop a zombie company or one that went belly up.

https://www.spglobal.com/spdji/en/governance/
I actually tested this. I compared a combination of the S&P 500/S&P 400 Mid-Cap Index/S&P 600 Small-Cap index in Portfolio Visualizer vs. the Total Stock Market Index. My theory was that the combination of the 3 S&P Indexes would screen out the "anti-factors" and produce better results than the US Total Stock Market Index. I was right but the difference was very small, hardly worth considering.

You could try this in Morningstar first to get the styleboxes of the 3 S&P Indexes and the Total Stock Market Index to match in order to get the precise percentages for the 3 S&P Indexes to match Total Market. Then run a comparison in Portfolio Visualizer to see the differences in performance.

My conclusion was that the markets were very efficient. It could be that the Total Market Indexes use some sort of a Quality screen too.
A fool and his money are good for business.
Seasonal
Posts: 3045
Joined: Sun May 21, 2017 1:49 pm

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

Post by Seasonal »

Multifactor models do a much better job of explaining returns than a single factor model. The implications of that for portfolio construction are far from clear.
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 »

Seasonal wrote: Sun Jan 16, 2022 10:23 am
I made two points. One was that if you are right that the size effect stopped working because it become too well known, then that should also be true about other well known factors. You did not dispute that. The second is that the market portfolio is the mix of factors that the market regards as appropriate. Note that the market may regard a net zero tilt of a factor as appropriate. You do not appear to dispute that either.
One thing is that you missed is that the calculation of Alpha and Beta have changed. Recall is that Alpha represents excess performance over the benchmark.

It used to be commonly understood that Beta was the volatility of the S&P 500. Today each factor has its own Beta, which does annoy me. So a bond could have a Beta of 1.00 if its volatility matched a bond index, but what I really want to know is how volatile the bond is compared to the stock market. Larry Swedroe and Rick Ferri had a debate about this, Larry was right and Rick was wrong, and a math geek sent me a message and told me Larry was correct. Rick had the old Beta as defined by the S&P 500 set in his mind and it didn't click with him that this all had changed.

So if you run a 3 or 5 factor model, each factor with its own beta, or volatility measurement, then you will get very little Alpha. For a CAPM Model, you could potentially generate a lot of Alpha with active management as the model only predicted 2/3 of the variance in stock returns. Lots of room to generate Alpha. A three factor model accounted for 90% of the variance in stock portfolio returns, an active manager could still generate Alpha but much less. For a 5 factor Model, with each factor having its own Beta, you only have the 4% of variability in stock returns between stock portfolios that the model can't account for, so there is very little Alpha that an active manager can capture.

So instead of one Beta, there are now at least one for each factor in your model. So in a 5 factor model, you have 5 Betas and not just one. You will see this as you analyze portfolios in Morningstar, there is a setting, at least in the professional version, where you can set Beta against the S&P 500 or according to Asset Allocation. So most portfolios, even with a mix of stocks and bonds will have a Beta not too far from one.

You Beta believe it.
A fool and his money are good for business.
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 »

nedsaid wrote: Sun Jan 16, 2022 10:28 am
jason2459 wrote: Sun Jan 16, 2022 10:16 am
nedsaid wrote: Sun Jan 16, 2022 10:01 am
Nathan Drake wrote: Sat Jan 15, 2022 10:40 pm
And, as mentioned, even very popular index funds (S&P 500) have quality screens, so this is not uncommon for even "TSM" funds.
That is my understanding as well, one reason I like the S&P Indexes, they screen out the junkier companies. A well constructed index, even a Total Market Index, doesn't contain ALL stocks, many have market caps and trading volumes too small to be investable. I know Vanguard uses a sampling method for Micro-Cap stocks.

I use to wonder about the "anti-factors", two of which would be Value traps and Lottery stocks, good Index construction could filter a lot of the worst companies out. As I recall, one of the Russell indexes gets criticized for being an easy benchmark for active managers to beat because it doesn't filter out the bad companies. I wish someone would comment here. I think it is their Mid/Small-Cap Index, the Russell 2000?

So the well constructed indexes might have a bit of the Quality factor already built in.

S&P has a very robust governance and methodology process and some can argue that it makes them more of a managed active index.

They have standard reconstitution dates like other passive index funds but they have processes that can kick off at anytime to handle multiple different scenarios like M&A activities or bankruptcy. This really helps drive out the junk quicker. Like look at some of the big funds like VTI they have junk on the bottom that is worthless and sticks around until the next reconstitution date. Some funds it's quarterly and some every six months or more. That could be a long time to drop a zombie company or one that went belly up.

https://www.spglobal.com/spdji/en/governance/
I actually tested this. I compared a combination of the S&P 500/S&P 400 Mid-Cap Index/S&P 600 Small-Cap index in Portfolio Visualizer vs. the Total Stock Market Index. My theory was that the combination of the 3 S&P Indexes would screen out the "anti-factors" and produce better results than the US Total Stock Market Index. I was right but the difference was very small, hardly worth considering.

You could try this in Morningstar first to get the styleboxes of the 3 S&P Indexes and the Total Stock Market Index to match in order to get the precise percentages for the 3 S&P Indexes to match Total Market. Then run a comparison in Portfolio Visualizer to see the differences in performance.

My conclusion was that the markets were very efficient. It could be that the Total Market Indexes use some sort of a Quality screen too.

I agree the difference with something like the total stock market should be negligible as each holding is fairly diluted. VTI/ITOT are my core holdings still.

It would be nice to see how the S&P quality factor influences on a more focused scale. Problems with cross index comparisons though that I came see is that there can be many factors that will skew the results in how the indexes are defined. All things will not being equal.
"In the short run, the stock market is a voting machine; in the long run, it is a weighing machine" ~Benjamin Graham
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 »

Seasonal wrote: Sun Jan 16, 2022 10:29 am Multifactor models do a much better job of explaining returns than a single factor model. The implications of that for portfolio construction are far from clear.
Yep. It has been established that factors work better in combination with other factors, particularly Quality. Hard to know what combination of factors is better than others. One reason that I don't recommend factor based portfolios whenever I do portfolio reviews, too confusing to newbies who need to be convinced to invest in the first place. Another reason is that factor premiums, except for Quality, appear to have shrunk, Larry Swedroe thought they shrunk by about 1/3. Then you get into the discussion of which factor products to buy, I have posted here many times about "good" and "bad" factor products.

It is sort of like picking the winner of the Miss America Pageant. You don't pick the girls you think are prettiest and most talented, you pick the girls who you think the judges believe to be the prettiest and the most talented.

Most people should probably use the simpler portfolios with the Broad Total Stock and Total Bond Indexes.
A fool and his money are good for business.
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 »

nedsaid wrote: Sun Jan 16, 2022 10:42 am
Seasonal wrote: Sun Jan 16, 2022 10:23 am
I made two points. One was that if you are right that the size effect stopped working because it become too well known, then that should also be true about other well known factors. You did not dispute that. The second is that the market portfolio is the mix of factors that the market regards as appropriate. Note that the market may regard a net zero tilt of a factor as appropriate. You do not appear to dispute that either.
One thing is that you missed is that the calculation of Alpha and Beta have changed. Recall is that Alpha represents excess performance over the benchmark.

It used to be commonly understood that Beta was the volatility of the S&P 500. Today each factor has its own Beta, which does annoy me. So a bond could have a Beta of 1.00 if its volatility matched a bond index, but what I really want to know is how volatile the bond is compared to the stock market. Larry Swedroe and Rick Ferri had a debate about this, Larry was right and Rick was wrong, and a math geek sent me a message and told me Larry was correct. Rick had the old Beta as defined by the S&P 500 set in his mind and it didn't click with him that this all had changed.

So if you run a 3 or 5 factor model, each factor with its own beta, or volatility measurement, then you will get very little Alpha. For a CAPM Model, you could potentially generate a lot of Alpha with active management as the model only predicted 2/3 of the variance in stock returns. Lots of room to generate Alpha. A three factor model accounted for 90% of the variance in stock portfolio returns, an active manager could still generate Alpha but much less. For a 5 factor Model, with each factor having its own Beta, you only have the 4% of variability in stock returns between stock portfolios that the model can't account for, so there is very little Alpha that an active manager can capture.

So instead of one Beta, there are now at least one for each factor in your model. So in a 5 factor model, you have 5 Betas and not just one. You will see this as you analyze portfolios in Morningstar, there is a setting, at least in the professional version, where you can set Beta against the S&P 500 or according to Asset Allocation. So most portfolios, even with a mix of stocks and bonds will have a Beta not too far from one.

You Beta believe it.
It seems like there is an awful lot of text and confusion or muddling of the concept of "beta" here. I'm not sure why each factor having it's own coefficient, or beta, should bother anyone. It's just the mathematical formulation of a return-estimating model. You can't have a factor without a coefficient in a linear model.

I think it's most instructive to just talk about this like a linear model that is predicting returns of some portfolio as a linear function of some variables.

Typically, you can write these as something like y_i = beta*x_i + a, where say you're predicting house prices as y (with house y_i as one observed house), and x is a variable, or factor, that represents the number of bedrooms in the house. So a model is that house_price_i = beta * number_of_bedrooms_i + a. a is a term that represents variation in the house prices that is not captured by the number of bedrooms, like age of house, location, etc.

Turning this into stock prices, now you have (R_i - risk_free_rate) = beta * (R_mkt - risk_free_rate) + alpha. Here, R_i is the return of the investment, and R_mkt is the market portfolio's return.

So if you had a portfolio with a beta of 1, that means your portfolio is fully exposed to the market portfolio, and your returns follow the market portfolio, plus/minus alpha, which is some over- or under-performance relative to the market average, just like you describe.

Now, say you want to include more variables in your model, so instead of just the market return of (R_mkt - risk_free_rate), say you think something like P/E matters for value, and size matters, too. Now you have two new variables, HML for value and SMB for size.

(R_i - risk_free_rate) = beta_mkt * (R_mkt - risk_free_rate) + beta_smb * SMB + beta_hml * HML + alpha.

So in this 3-factor model, yes, we have coefficients for the new factors, because we're trying to estimate how they explain the returns of your portfolio. As you mentioned, it means alpha could be smaller, because these new variables can explain some of your portfolio's returns now. (For example now someone is no longer a magical manager with alpa = 5%, they just created a value portfolio that loaded heavily on beta_hml).

I would disagree though that most portfolios would have a market beta of 1. This is true for most stock-only portfolios, but if you are investing in bonds, your portfolio is not 100% exposed to the equity market anymore. For example if you were 100% VTI, your factor expososures in the 3-factor model above would be close to beta_mkt = 1, beta_smb = 0, beta_hml = 0. Now if you change to a 60/40 portfolio, your factor exposures would be beta_mkt = 0.6, beta_smb = 0, beta_hml = 0.

You would then "pick up" fixed-income factor exposures, because in the same way you can make a linear model to estimate equity returns, you can make a linear model to estimate bond returns. In this case, maybe a simple model is "how much term risk is there" (term), and "how much credit risk is there" (corporate bonds vs treasuries), and that model might look something like R_i = beta_term * duration + beta_credit * (some variable that summarizes credit risk) + alpha
Last edited by muffins14 on Sun Jan 16, 2022 11:44 am, edited 1 time in total.
Crom laughs at your Four Winds
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 »

Seasonal wrote: Sun Jan 16, 2022 10:23 am
I made two points. One was that if you are right that the size effect stopped working because it become too well known, then that should also be true about other well known factors. You did not dispute that. The second is that the market portfolio is the mix of factors that the market regards as appropriate. Note that the market may regard a net zero tilt of a factor as appropriate. You do not appear to dispute that either.
The Size affect appears to have stopped working in isolation. If you couple Size with Quality, the Size premium reappears with a vengeance. Value also seems to work better in tandem with Quality.

Has Value stopped working? There is a pretty spirited debate over that right now. The consensus is that the Value premium for Large Stocks is very small and that the Value premium for Mid-Caps and Small-Caps is much larger.

It also depends on whether one believes if factors are risk based only or if there is a behavioral element as well. If there is a behavioral element to factors, one would expect that a return of the premiums would be inevitable, as the bad habits stemming from human nature and human nature would eventually return. My belief that the factors come from the human emotions of enthusiasm that comes with greed and pessimism that comes with fear. It is interesting that Cliff Asness, Gene Fama's brightest student has come around to the belief that there is a big behavioral story behind the factors. Asness is a quant and I am not, I think his opinion here is important.

A market portfolio is not a mix of factors. A market portfolio by definition contains only one factor, that is Market. To get additional factors loaded in the portfolio, you need to tilt away from the Market portfolio in some way. Another way of saying it, is that in a market portfolio, the other factors are set to neutral.
A fool and his money are good for business.
quietseas
Posts: 901
Joined: Fri Dec 27, 2013 3:43 pm

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

Post by quietseas »

Factors and fundamentals at 6 bps to 25 bps (US, up to 35 basis points international) are a whole different matter than factors at 150+ basis points including AUM advisory fees gatekeeping the secret sauce.

Probably not needed, but I view factors as low cost insurance.

I see the path for size, value, quality/profitability/fundamentals.

Momentum complicates matters and I don't think I've ever seen a good explanation of a strategy that encompasses it with the others. I know its not the same but total market with S&P 600 value or Avantis small value tilts is my compromise still.
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 »

muffins14 wrote: Sun Jan 16, 2022 11:11 am
nedsaid wrote: Sun Jan 16, 2022 10:42 am
Seasonal wrote: Sun Jan 16, 2022 10:23 am
I made two points. One was that if you are right that the size effect stopped working because it become too well known, then that should also be true about other well known factors. You did not dispute that. The second is that the market portfolio is the mix of factors that the market regards as appropriate. Note that the market may regard a net zero tilt of a factor as appropriate. You do not appear to dispute that either.
One thing is that you missed is that the calculation of Alpha and Beta have changed. Recall is that Alpha represents excess performance over the benchmark.

It used to be commonly understood that Beta was the volatility of the S&P 500. Today each factor has its own Beta, which does annoy me. So a bond could have a Beta of 1.00 if its volatility matched a bond index, but what I really want to know is how volatile the bond is compared to the stock market. Larry Swedroe and Rick Ferri had a debate about this, Larry was right and Rick was wrong, and a math geek sent me a message and told me Larry was correct. Rick had the old Beta as defined by the S&P 500 set in his mind and it didn't click with him that this all had changed.

So if you run a 3 or 5 factor model, each factor with its own beta, or volatility measurement, then you will get very little Alpha. For a CAPM Model, you could potentially generate a lot of Alpha with active management as the model only predicted 2/3 of the variance in stock returns. Lots of room to generate Alpha. A three factor model accounted for 90% of the variance in stock portfolio returns, an active manager could still generate Alpha but much less. For a 5 factor Model, with each factor having its own Beta, you only have the 4% of variability in stock returns between stock portfolios that the model can't account for, so there is very little Alpha that an active manager can capture.

So instead of one Beta, there are now at least one for each factor in your model. So in a 5 factor model, you have 5 Betas and not just one. You will see this as you analyze portfolios in Morningstar, there is a setting, at least in the professional version, where you can set Beta against the S&P 500 or according to Asset Allocation. So most portfolios, even with a mix of stocks and bonds will have a Beta not too far from one.

You Beta believe it.
It seems like there is an awful lot of text and confusion or muddling of the concept of "beta" here. I'm not sure why each factor having it's own coefficient, or beta, should bother anyone. It's just the mathematical formulation of a return-estimating model. You can't have a factor without a coefficient in a linear model.

I think it's most instructive to just talk about this like a linear model that is predicting returns of some portfolio as a linear function of some variables.

Typically, you can write these as something like y_i = beta*x_i + a, where say you're predicting house prices as y (with house y_i as one observed house), and x is a variable, or factor, that represents the number of bedrooms in the house. So a model is that house_price_i = beta * number_of_bedrooms_i + a. a is a term that represents variation in the house prices that is not captured by the number of bedrooms, like age of house, location, etc.

Turning this into stock prices, now you have (R_i - risk_free_rate) = beta * (R_mkt - risk_free_rate) + alpha. Here, R_i is the return of the investment, and R_mkt is the market portfolio's return.

So if you had a portfolio with a beta of 1, that means your portfolio is fully exposed to the market portfolio, and your returns follow the market portfolio, plus/minus alpha, which is some over- or under-performance relative to the market average, just like you describe.

Now, say you want to include more variables in your model, so instead of just the market return of (R_mkt - risk_free_rate), say you think something like P/E matters for value, and size matters, too. Now you have two new variables, HML for value and SMB for size.

(R_i - risk_free_rate) = beta_mkt * (R_mkt - risk_free_rate) + beta_smb * SMB + beta_hml * HML + alpha.

So in this 3-factor model, yes, we have coefficients for the new factors, because we're trying to estimate how they explain the returns of your portfolio. As you mentioned, it means alpha could be smaller, because these new variables can explain some of your portfolio's returns now. (For example now someone is no longer a magical manager with alpa = 5%, they just created a value portfolio that loaded heavily on beta_hml).

I would disagree though that most portfolios would have a market beta of 1. This is true for most stock-only portfolios, but if you are investing in bonds, your portfolio is not 100% exposed to the equity market anymore. For example if you were 100% VTI, your factor expososures in the 3-factor model above would be close to beta_mkt = 1, beta_smb = 0, beta_hml = 0. Now if you change to a 60/40 portfolio, your factor exposures would be beta_mkt = 0.6, beta_smb = 0, beta_hml = 0.

You would then "pick up" fixed-income factor exposures, because in the same way you can make a linear model to estimate equity returns, you can make a linear model to estimate bond returns. In this case, maybe a simple model is "how much term risk is there" (duration), and "how much credit risk is there" (corporate bonds vs treasuries), and that model might look something like R_i = beta_duration * duration + beta_credit * (some variable that summarizes credit risk) + alpha
What you miss is that Bonds have their own factors too. Going from memory, I recall they are Carry and Term. Too lazy to check Larry Swedroe's factor investing book.

I have done portfolio analysis in Morningstar Professional version and you can set analysis by asset allocation, thus you will get multiple betas. If bonds have their own betas, yes most portfolios will have a Beta not too far from 1.00. This was a surprise to me.

It seemed that in Morningstar that every asset class had its own Beta but I don't know how finely they slice this. I wonder if even High Yield Debt had its own Beta.
A fool and his money are good for business.
DaufuskieNate
Posts: 605
Joined: Wed May 28, 2014 11:53 am

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

Post by DaufuskieNate »

nedsaid wrote: Sun Jan 16, 2022 11:27 am
What you miss is that Bonds have their own factors too. Going from memory, I recall they are Carry and Term. Too lazy to check Larry Swedroe's factor investing book.
Credit and Term. Somewhat analogous to quality and duration.
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 »

DaufuskieNate wrote: Sun Jan 16, 2022 11:36 am
nedsaid wrote: Sun Jan 16, 2022 11:27 am
What you miss is that Bonds have their own factors too. Going from memory, I recall they are Carry and Term. Too lazy to check Larry Swedroe's factor investing book.
Credit and Term. Somewhat analogous to quality and duration.
Thank you.
A fool and his money are good for business.
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 »

I mention bond factors in my last paragraph.

I don’t think you should be “surprised” by the fact that there is a coefficient for some variable with value near 1. Maybe your beliefs are anchored by people talking about “beta” usually referring to equity market beta. So yes, it could be confusing if a bond portfolio had an equity market exposure of 1, but that’s not the case, your bond beta is a totally different variable. Just think of the term “beta” like a generic name for “coefficient” in a model.
Crom laughs at your Four Winds
VTI
Posts: 282
Joined: Sat Mar 13, 2021 11:56 pm

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

Post by VTI »

Nathan Drake wrote: Sat Jan 15, 2022 10:40 pm [...]

A bog standard SCV index fund, RUT 2000 Value, has shown a premium without any significant screens compared to, say, Avantis. And this includes many of the "junk stocks" that Avantis tries to weed out.

[...]
Just like hundreds of active funds, didn't all of the outperformance of Russell 2000 Value happen during a couple of years shortly after it was conceived? It's been a dog for decades.

Regarding value:

It's inevitable for there to be companies that are underpriced compared to future earnings. It's not inevitable that a simple mathematical model will be able to consistently identify those companies ahead of time.
Last edited by VTI on Tue Jan 18, 2022 12:15 am, edited 3 times in total.
Also available as an Admiral™ Shares mutual fund.
lrobb
Posts: 5
Joined: Sun Sep 16, 2012 5:41 am

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

Post by lrobb »

nedsaid wrote: Sat Jan 15, 2022 12:24 pm
nisiprius wrote: Sat Jan 15, 2022 11:44 am
This is accompanied by rhetorical maneuvers that walk back claims in subtle ways. And, of course, to be fair, it is not exactly the same people making the initial claims as the people saying later "we never said that."
I am shocked, just shocked that rhetorical maneuvers are going on in here.
Just a fascinating thread. I'm reminded of the Rational Reminder podcast with Antonio Picca (Vanguard's head of factor investing) where he posited factor investing is just another form of active investing. The arguments here remind me of Wellington & Primecap's investment process - Quantitative screening combined with committee decisions.
afan
Posts: 8193
Joined: Sun Jul 25, 2010 4:01 pm

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

Post by afan »

nedsaid wrote: Sat Jan 15, 2022 6:55 pm
afan wrote: Sat Jan 15, 2022 3:47 pm
nedsaid wrote: Sat Jan 15, 2022 12:51 pm

The Market Portfolio by definition is the Market Factor. A market portfolio contains Value stocks, contains Small stocks, contains Quality stocks, and so on but does not contain the factors themselves except for the Market factor.

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.

Regarding size, I believe it was in their first 5 factor paper where F&F said that size was no longer a significant part of the model.

I don't think Fama ever claimed that factors were a way to achieve increased risk adjusted returns. He has been consistent in his opinion that, to the extent that portfolios that overweight some factors may have increased returns, it is because they come with higher risk.
CAPM, the 3 factor, and 5 factors do not predict future events. I am not sure where you got this idea, no one makes this argument that I am aware of. You are making a classic strawman argument here.

Where did I get this idea? I quoted you.

if the Market portfolio contains all of the factors, then CAPM would explain 100% of the variances in return of stock portfolios,
If factors other than the volatility of the market itself didn't exist, then all stocks would have the same risk adjusted returns

If what you say is true, the risk adjusted returns of Blue Chip stocks and the most speculative stocks (like those that trade on the Vancouver Stock Exchange or on the NASDAQ Bulletin Board) would be the same.

Nothing I have said would imply this. However, it might be true.

Except that "containing all the factors" in no way says that a model explains 100% of the variance. A model cannot explain 100% of the variance of any system that has noise.
You are making a classic strawman argument here.
Again, just quoting you.

If factors other than the volatility of the market itself didn't exist, then all stocks would have the same risk adjusted returns.
I would love to see any study, empirical or theoretical, that supported this statement.

Perhaps we need to go back to what a "factor" is. It is a statistical construct. Hundreds of factors have been proposed. There is an infinity of possible factors. Some have not been promising in initial tests so they are never publicized. Others have not been tested. They are all tried for their potential relationships to stock returns.

Most fail at this, but they are still factors.
All the factors that have some supporters have counter factors. There is a size factor, usually focussing on small stocks. There is just as well a "large" factor. Over some periods, "large" has outperformed "small".

Nothing suggests that portfolios built to overweight certain factors produces excess risk adjusted returns over the market portfolio.

Going from memory, CAPM explained about 2/3 of the variance in returns from stock portfolios. The 3 factor model explained about 90% and the 5 factor model got it up to 96% or so.
This will happen with any multiple variable model. Adding more variables always improves the fit. Stat 101.

You may want to read the academic finance literature itself, rather than apparently, lay summaries. Pay particular attention to the work of Campbell Harvey on factors. To follow it you will need to pick up the information in a couple of undergrad courses on investments and a few courses on statistics. Make sure you cover multiple regression, ANOVA, and factor analysis.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama
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 »

afan wrote: Sun Jan 16, 2022 12:35 pm
This will happen with any multiple variable model. Adding more variables always improves the fit. Stat 101.
Yes, but I imagine adjusted R^2 or metrics like BIC/AIC are also improving here so the effect is more real than artificial. I have not looked at those metrics for factor regressions myself, but we’re going from 1 to 3 or 4 variables here, not from like 50 variables to 51 variables, where I’d be more worried about claiming a “better” model from a marginal increase in R^2
Crom laughs at your Four Winds
Elysium
Posts: 4120
Joined: Mon Apr 02, 2007 6:22 pm

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

Post by Elysium »

nedsaid wrote: Sun Jan 16, 2022 9:41 am
Right after I bought a Micro-Cap Index ETF, I read a Morningstar article that said investors were better off in a Small-Cap Index. Oh well. I still own the ETF, never sold.
Bridgeway Ultra Small Company Market (BRSIX) is a fund recommended by Larry back in the days on the old forum as an alternative to the DFA Microcap fund. It is supposedly close to passive since they use rule based screens to build the portfolio. The fund hasn't done poorly at all since inception in this comparison.

Full disclosure: I used to own this fund for a short period from 2002 to 2005 and had an outstanding 82% growth in 2003. Since then I gave that up to reduce the number funds to opt for simplicity.
Wade Garrett
Posts: 235
Joined: Wed Jul 10, 2013 2:41 pm
Location: Jasper, MO/Double Deuce

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

Post by Wade Garrett »

Elysium wrote: Sun Jan 16, 2022 2:15 pm Bridgeway Ultra Small Company Market (BRSIX) is a fund recommended by Larry back in the days on the old forum as an alternative to the DFA Microcap fund.
Bridgeway is a good shop and BRSIX is a good fund. But it isn't tax friendly. Do not own in a taxable account.
"I'm not an inventor. I'm an improver. I see things that are wrong, and I improve them." - Larry David, Curb Your Enthusiasm
loukycpa
Posts: 735
Joined: Wed Aug 05, 2020 9:52 am

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

Post by loukycpa »

Forgive me if I am as a mere mortal interrupting a conversation between higher beings. Asking a sincere question with the hopes of educating myself. Will confess that I have only a passing knowledge. Just enough to grasp the thesis of what folks are talking about in terms of small cap and value premium, and why some believe in it enough to tilt SCV, etc. Neither a critic or a disciple of it.

So one decides to make a nine box grid and divide the universe of public traded stocks based on size (large, mid, small) and value versus growth (however value is defined, we could argue about that). Then look at historical returns of the boxes over various time frames.

Here is where I disconnect. It seems to me that over any short enough time period that randomly one of the boxes will outperform the other eight. One side of the grid will tend to outperform the other three. I would also expect that the longer the time period, this random outperformance will smooth out to some extent, reversion to the mean.

Isn't it possible that is all this ever was? And the academics that pushed this as a theory just didn't have a long enough time period to know if it was just random or not?

I have gleaned a little bit of wisdom from Warren Buffett on how to understand the financial markets. I don't have the quote readily available, but I recall reading what Warren Buffet said about value versus growth. He sees it as a completely ridiculous distinction. In his mind, everything is about value. Growth expectations (presumably in future cash flows and stock ownership as a claim on those growing future cash flows) are an important piece of the puzzle in determining value. In his mind, all intelligent and rational investing is value investing. He scoffed at those who want to do this nine box thing as essentially know nothings who try to pass themselves off as know somethings to other know nothings. What he said has stuck with me, and I tend to think he is onto something. And I confess this is a big reason why I remain skeptical about factor investing.

I also think the world has changed a lot even since Buffett and Graham, and I think he speaks about this too. The tools and the participants in the market have become more sophisticated over the last 50 years. Much of the advantage Buffett and others had has been squeezed over the years. There was more alpha to be had then than there is today.

My personal hunch is that to truly generate alpha, one needs to be up to the task at hand and also the risk involved. This nine box and overweight one box thing? It just seems like a nothingburger to me.
Last edited by loukycpa on Tue Jan 18, 2022 6:31 am, edited 6 times in total.
"The safe assumption for an investor is that over the next hundred years, the currency is going to zero." - Charlie Munger
Nathan Drake
Posts: 6234
Joined: Mon Apr 11, 2011 12:28 am

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

Post by Nathan Drake »

loukycpa wrote: Mon Jan 17, 2022 10:40 am
Isn't it possible that is all this ever was? And the academics that pushed this as a theory just didn't have a long enough time period to know if it was just random or not?
The long term data shows significantly higher returns for SCV. If it were some mean reverting trend, it would have to mean revert a huge amount going forward…Investing in SCV isn’t generating alpha. It’s simply taking on more risk and additional risk premiums.
20% VOO | 20% VXUS | 20% AVUV | 20% AVDV | 20% AVES
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

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

Post by Random Walker »

loukycpa wrote: Mon Jan 17, 2022 10:40 am Forgive me if I am as a mere mortal interrupting a conversation between higher beings. Asking a sincere question with the hopes of educating myself. Will confess that I have only a passing knowledge. Just enough to grasp the thesis of what folks are talking about in terms of small cap and value premium, and why some believe in it enough to tilt SCV, etc. Neither a critic or a disciple of it.

So one decides to make a nine box grid and divide the universe of public traded stocks based on size (large, mid, small) and value versus growth (however defined, we could argue about that). Then look at historical returns of the boxes over various time frames.

Here is where I disconnect. It seems to me that over any short enough time period that randomly one of the boxes will outperform the other eight. One side of the grid will tend to outperform the other three. I would also expect that the longer the time period, this random outperformance will smooth out to some extent, reversion to the mean.

Isn't it possible that is all this ever was? And the academics that pushed this as a theory just didn't have a long enough time period to know if it was just random or not?

I have gleaned a little bit of wisdom from Warren Buffett on how to understand the financial markets. I don't have the quote readily available, but I recall reading what Warren Buffet said about value versus growth. He sees it as a completely ridiculous distinction. In his mind, everything is about value. Growth expectations (presumably in future cash flows and stock ownership as a claim on those growing future cash flows) is an important piece of the puzzle in determining value. In his mind, all intelligent and rational investing is value investing. He scoffed at those who want to do this nine box thing as essentially know nothings who try to pass themselves off as know somethings. What he said has stuck with me, and I tend to think he is onto something. And I confess this is a big reason why I remain skeptical about factor investing.

I also think the world has changed a lot even since Buffett and Graham, and I think he speaks about this too. The tools and the participants in the market have become more sophisticated over the last 50 years. Much of the advantage Buffett and others had have been squeezed over the years. There was more alpha to be had then than there is today.

My personal hunch is that to truly generate alpha, one needs to be up to the task at hand and also the risk involved. This nine box and overweight one box thing? Forgive me if this sounds too harsh, but it just seems like a nothing burger to me.
I would just make a couple comments. First of all, you do make some very strong points, and it’s hard to go wrong following Buffett advice. Secondly, with regard to the style boxes, understand that there are sound risk based reasons * to believe size and value deserve premiums. A big part of the FF work was to show that market, size, value are unique and independent from one another. So potential strong diversification benefits. And many people don’t get this, but diversification could well be more important over shorter timeframes. Any factor or source of return, including the market factor, can do poorly over extended periods. So if one is looking to avoid a negative outcome over periods of say 1,3,5,10 years, diversification across factors can lessen that risk.

* some people dispute whether small size is truly more risky. I think intuitively it is. Value has both risk based and behavioral based rationale supporting it.

Dave
loukycpa
Posts: 735
Joined: Wed Aug 05, 2020 9:52 am

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

Post by loukycpa »

Nathan Drake wrote: Mon Jan 17, 2022 11:24 am
loukycpa wrote: Mon Jan 17, 2022 10:40 am
Isn't it possible that is all this ever was? And the academics that pushed this as a theory just didn't have a long enough time period to know if it was just random or not?
Investing in SCV isn’t generating alpha. It’s simply taking on more risk and additional risk premiums.
With respect to small cap factor, that actually does make some sense to me. It makes less sense to me on the value side.

Doesn't make me want to tilt. I'm not looking for more risk. More return for the same level of risk? Same return for less risk? That's more my style.
Last edited by loukycpa on Mon Jan 17, 2022 1:19 pm, edited 1 time in total.
"The safe assumption for an investor is that over the next hundred years, the currency is going to zero." - Charlie Munger
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

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

Post by Random Walker »

Nathan Drake wrote: Mon Jan 17, 2022 11:24 am
loukycpa wrote: Mon Jan 17, 2022 10:40 am
Isn't it possible that is all this ever was? And the academics that pushed this as a theory just didn't have a long enough time period to know if it was just random or not?
The long term data shows significantly higher returns for SCV. If it were some mean reverting trend, it would have to mean revert a huge amount going forward…Investing in SCV isn’t generating alpha. It’s simply taking on more risk and additional risk premiums.
And of course, the question arises “What mean are we reverting to?”. If size and value are truly risk premiums, then their mean expected return should add to the mean expected return of market beta in long only funds. Markets price risk.

Dave
User avatar
nisiprius
Advisory Board
Posts: 52212
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 »

Nathan Drake wrote: Mon Jan 17, 2022 11:24 am.…Investing in SCV isn’t generating alpha. It’s simply taking on more risk and additional risk premiums...
Then what, exactly, is the point in investing in it? Why not just take the additional risk and get the additional risk premium by boosting stock allocation?

It's a difficult thing to argue because there are two different theories being put forward by small-cap value advocates. One is that small-cap value does generate alpha, because of persistent behavioral errors. Let's put that aside because you say it doesn't.

To believe that a small-cap value tilt will improve a portfolio more than simply increasing stock allocation, you have to add some fairly strong assumptions to "the additional risk of small-cap value is appropriately rewarded."

These assumptions possibly include:
  • The behavior of small-cap value stocks has not changed following the publication of the Fama-French 1993 paper;
  • Although there are no products or benchmarked to the Fama-French Small Value Research Portfolio, the differences between these indexes and the actual products available to investors is negligible;
  • The correlation between small-cap value and the total market is low; or, there is some other measure of the diversification value of small-cap value, apart from correlation, that makes them powerful diversifiers.
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.
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

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

Post by Random Walker »

nisiprius wrote: Mon Jan 17, 2022 1:09 pm Then what, exactly, is the point in investing in it? Why not just take the additional risk and get the additional risk premium by boosting stock allocation?
[ quote fixed by admin LadyGeek]

Because size and value represent risks that are unique and independent from market beta. More efficient to diversify across different types of risk than simply add more of the same. Historic correlations of size, value, market have been low.

Dave
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 »

nisiprius wrote: Mon Jan 17, 2022 1:09 pm The correlation between small-cap value and the total market is low; or, there is some other measure of the diversification value of small-cap value, apart from correlation, that makes them powerful diversifiers.
This is a core reason I invest in SCV. the returns on average over long time scales should be on par or a little higher for SCV than market alone. However SCV stocks are differently affected by different by different economic conditions compared to large cap stocks and growth stocks. For example inflation and interest rates affect the two differently. Thus the two (a market cap portfolio vs an SCV portfolio) are not perfectly correlated and can be good partners in a portfolio.

My portfolio is about 60% US, 40% international, and 60% market-cap weighted ( like VTI, VXUS) and 40% SCV (like AVUV, AVDV)
Crom laughs at your Four Winds
User avatar
steve r
Posts: 1300
Joined: Mon Feb 13, 2012 7:34 pm
Location: Connecticut

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

Post by steve r »

nedsaid wrote: Sat Jan 15, 2022 11:52 am
....

So I will join you, a lot of the drive for "simplicity" has been good old fashioned performance chasing. Total Market has done great, driven by the FAANG+Microsoft+Tesla. It isn't that Value has done poorly, it actually has done fairly well but not as well as Large Growth.
...
Late to the thread, but I liked this comment a lot. For me, it was not so much performance chasing but resisting change based on performance if that makes sense. (Though I did dable elsewhere).

I do remember when the Morningstar dot was closer to value, and the index was dominated by the likes of Exxon. Fortunately, the M* dot for VT is fairly close to the center.

The one thing I never liked about indexing, though not seeing a reasonable alternative, is the share of the largest handful of holdings. It lead me to min. vol. (since exited) and tempted by AVUS (largest 3 holdings are 11 percent, which is less than VTI), and RSP (equal weight 500).
"Owning the stock market over the long term is a winner's game. Attempting to beat the market is a loser's game. ..Don't look for the needle in the haystack. Just buy the haystack." Jack Bogle
Nathan Drake
Posts: 6234
Joined: Mon Apr 11, 2011 12:28 am

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

Post by Nathan Drake »

nisiprius wrote: Mon Jan 17, 2022 1:09 pm
Nathan Drake wrote: Mon Jan 17, 2022 11:24 am.…Investing in SCV isn’t generating alpha. It’s simply taking on more risk and additional risk premiums...
Then what, exactly, is the point in investing in it? Why not just take the additional risk and get the additional risk premium by boosting stock allocation?

To believe that a small-cap value tilt will improve a portfolio more than simply increasing stock allocation, you have to add some fairly strong assumptions to "the additional risk of small-cap value is appropriately rewarded."
This is not true. SCV could have no premium, but simply exhibit a different dispersion of returns over various periods versus the market, to provide value to the portfolio’s construction.

And we can clearly see that there’s both higher expected returns and different periods of time when those returns manifest.

Simply increasing stock allocation just exposes you more to the same market factor
20% VOO | 20% VXUS | 20% AVUV | 20% AVDV | 20% AVES
rkhusky
Posts: 17764
Joined: Thu Aug 18, 2011 8:09 pm

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

Post by rkhusky »

Random Walker wrote: Mon Jan 17, 2022 1:24 pm
nisiprius wrote: Mon Jan 17, 2022 1:09 pm Then what, exactly, is the point in investing in it? Why not just take the additional risk and get the additional risk premium by boosting stock allocation?
Because size and value represent risks that are unique and independent from market beta. More efficient to diversify across different types of risk than simply add more of the same. Historic correlations of size, value, market have been low.
SmB (Small - Big) and HmL (High - Low) are relatively independent of each other and the market when looking at price movements, but Small and Value are highly correlated with each other and the market.

Investing in small and value is not diversifying risk, but concentrating risk into stocks with similar characteristics (i.e. supposedly undervalued small stocks). Further, adding additional screens like qualify and profitability etc, make the resulting stocks even more similar. You are making a bet that a small sector of the market will out-perform the broader market.
Nathan Drake
Posts: 6234
Joined: Mon Apr 11, 2011 12:28 am

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

Post by Nathan Drake »

rkhusky wrote: Mon Jan 17, 2022 2:54 pm
Random Walker wrote: Mon Jan 17, 2022 1:24 pm
nisiprius wrote: Mon Jan 17, 2022 1:09 pm Then what, exactly, is the point in investing in it? Why not just take the additional risk and get the additional risk premium by boosting stock allocation?
Because size and value represent risks that are unique and independent from market beta. More efficient to diversify across different types of risk than simply add more of the same. Historic correlations of size, value, market have been low.
SmB (Small - Big) and HmL (High - Low) are relatively independent of each other and the market, but Small and Value are highly correlated with each other and the market.

Investing in small and value is not diversifying risk, but concentrating risk into stocks with similar characteristics (i.e. supposedly undervalued small stocks). Further, adding additional screens like qualify and profitability etc, make the resulting stocks even more similar. You are making a bet that a small sector of the market will out-perform the broader market.
You are simply investing in stocks that exhibit different risks than the market with higher expected returns because they are priced by the market in such a way.

It’s not a “bet” anymore than the market is a bet on the largest companies in the world
20% VOO | 20% VXUS | 20% AVUV | 20% AVDV | 20% AVES
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 »

nisiprius wrote: Sat Jan 15, 2022 11:44 am
First, "Quality" is not one of the five Fama-French factors, which are "market," "size," "value," "profitability," and "investment."
Right: it’s two of them. Profitability and Investment jointly are the FF5 analog of “quality”.
"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 »

Nathan Drake wrote: Mon Jan 17, 2022 2:57 pm You are simply investing in stocks that exhibit different risks than the market with higher expected returns because they are priced by the market in such a way.

It’s not a “bet” anymore than the market is a bet on the largest companies in the world
What is the basis for saying that SV stocks have higher expected return? Just because they have more risk? Or based on historical returns?

Betting on the world market is a bet that the world's economy expands, which is much more probable than that a small fraction of the world's companies will consistently out-perform the world market.
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 »

rkhusky wrote: Mon Jan 17, 2022 3:04 pm
Nathan Drake wrote: Mon Jan 17, 2022 2:57 pm You are simply investing in stocks that exhibit different risks than the market with higher expected returns because they are priced by the market in such a way.

It’s not a “bet” anymore than the market is a bet on the largest companies in the world
What is the basis for saying that SV stocks have higher expected return? Just because they have more risk? Or based on historical returns?
It’s that they have MORE risk, that the extra risk is DIFFERENT from general market risk, and SOME investors prefer to avoid that additional type of extra risk.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
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: Mon Jan 17, 2022 3:04 pm
Nathan Drake wrote: Mon Jan 17, 2022 2:57 pm You are simply investing in stocks that exhibit different risks than the market with higher expected returns because they are priced by the market in such a way.

It’s not a “bet” anymore than the market is a bet on the largest companies in the world
What is the basis for saying that SV stocks have higher expected return? Just because they have more risk? Or based on historical returns?

Betting on the world market is a bet that the world's economy expands, which is much more probable than that a small fraction of the world's companies will consistently out-perform the world market.
For one, I think the world economy can expand without stock prices going up. Like how people say not to invest in China because despite growing GDP, foreign investors aren’t participating in that growth by owning VWO.

Two, I don’t think you have to make a bet that SCV will consistently outperform the market. Most SCV funds are 100% exposed to the Market factor.
Crom laughs at your Four Winds
rkhusky
Posts: 17764
Joined: Thu Aug 18, 2011 8:09 pm

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

Post by rkhusky »

vineviz wrote: Mon Jan 17, 2022 3:06 pm
rkhusky wrote: Mon Jan 17, 2022 3:04 pm
Nathan Drake wrote: Mon Jan 17, 2022 2:57 pm You are simply investing in stocks that exhibit different risks than the market with higher expected returns because they are priced by the market in such a way.

It’s not a “bet” anymore than the market is a bet on the largest companies in the world
What is the basis for saying that SV stocks have higher expected return? Just because they have more risk? Or based on historical returns?
It’s that they have MORE risk, that the extra risk is DIFFERENT from general market risk, and SOME investors prefer to avoid that additional type of extra risk.
But there is no guarantee that the investor will get more reward for taking more risk. It is up to the investor to determine (or demand) that there is a sufficiently good chance that they will get the reward and that the potential losses will not be too great. Unfortunately, there is no way to determine the odds with stock investing, like you can with cards and dice.
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: Mon Jan 17, 2022 3:22 pm
rkhusky wrote: Mon Jan 17, 2022 3:04 pm
Nathan Drake wrote: Mon Jan 17, 2022 2:57 pm You are simply investing in stocks that exhibit different risks than the market with higher expected returns because they are priced by the market in such a way.

It’s not a “bet” anymore than the market is a bet on the largest companies in the world
What is the basis for saying that SV stocks have higher expected return? Just because they have more risk? Or based on historical returns?

Betting on the world market is a bet that the world's economy expands, which is much more probable than that a small fraction of the world's companies will consistently out-perform the world market.
For one, I think the world economy can expand without stock prices going up. Like how people say not to invest in China because despite growing GDP, foreign investors aren’t participating in that growth by owning VWO.

Two, I don’t think you have to make a bet that SCV will consistently outperform the market. Most SCV funds are 100% exposed to the Market factor.
Yes, there is a large chunk of the world economy that is not represented in the stock markets. I should have said "the world's investable market expands".

I imagine that SCV will out-perform sometime and under-perform sometime. Whether you profit or lose with SCV will probably depend on whether your investing is in sync with the ups and downs or out of sync.

And, like you say, SCV is correlated with the market to a great extent. So, you aren't going to gain a whole lot or lose a whole lot compared to the world market, if you use a broadly diversified SCV fund.
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 »

rkhusky wrote: Mon Jan 17, 2022 3:27 pm But there is no guarantee that the investor will get more reward for taking more risk.
This is self-evident. Any investment with a guaranteed return would be risk-free: there is NEVER a guaranteed reward when risk is involved.

But with any systematic risk factor there is an expectation that risk will be rewarded, which is what leads people to invest in risky assets like bonds and stocks to begin with.
"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 »

vineviz wrote: Mon Jan 17, 2022 3:39 pm
rkhusky wrote: Mon Jan 17, 2022 3:27 pm But there is no guarantee that the investor will get more reward for taking more risk.
This is self-evident. Any investment with a guaranteed return would be risk-free: there is NEVER a guaranteed reward when risk is involved.

But with any systematic risk factor there is an expectation that risk will be rewarded, which is what leads people to invest in risky assets like bonds and stocks to begin with.
But one needs to do an analysis of the risk to determine if the expected reward is commensurate with size of the risk. I've received mailings for investments in exploratory oil drilling - there is lots of risk, but I doubt I would get much return.
Nathan Drake
Posts: 6234
Joined: Mon Apr 11, 2011 12:28 am

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

Post by Nathan Drake »

rkhusky wrote: Mon Jan 17, 2022 4:44 pm
vineviz wrote: Mon Jan 17, 2022 3:39 pm
rkhusky wrote: Mon Jan 17, 2022 3:27 pm But there is no guarantee that the investor will get more reward for taking more risk.
This is self-evident. Any investment with a guaranteed return would be risk-free: there is NEVER a guaranteed reward when risk is involved.

But with any systematic risk factor there is an expectation that risk will be rewarded, which is what leads people to invest in risky assets like bonds and stocks to begin with.
But one needs to do an analysis of the risk to determine if the expected reward is commensurate with size of the risk. I've received mailings for investments in exploratory oil drilling - there is lots of risk, but I doubt I would get much return.
You can find the historical risk premium. It is fairly significant, especially when valuation spreads are high like they are today.

But factor investing does not mean you necessarily exclude TSM. It’s simply another area to invest in with a long track record of returns consistent with buy and hold passive investing
20% VOO | 20% VXUS | 20% AVUV | 20% AVDV | 20% AVES
rkhusky
Posts: 17764
Joined: Thu Aug 18, 2011 8:09 pm

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

Post by rkhusky »

Nathan Drake wrote: Mon Jan 17, 2022 4:52 pm You can find the historical risk premium. It is fairly significant, especially when valuation spreads are high like they are today.
Perhaps there has been a premium in the past, but will in continue in the future? In an efficient market with rational and knowledgeable investors, one would expect there to be a positive correlation between risk and return. But are the majority of investors rational and/or knowledgeable enough to actually be able to calculate the risk of an investment?
Last edited by rkhusky on Mon Jan 17, 2022 5:16 pm, edited 2 times in total.
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 »

rkhusky wrote: Mon Jan 17, 2022 5:10 pm Perhaps there has been a premium in the past, but will in continue in the future? In an efficient market with rational and knowledgeable investors, one would expect there to be a positive correlation between risk and return. But are the majority of investors rational and/or knowledgeable enough to actually be able to calculate the risk of an investment?
If you think the market is so inefficient or irrational that there is no longer a relationship between risk and reward then we have a bigger problem than which equity index fund to choose.
"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 »

vineviz wrote: Mon Jan 17, 2022 5:15 pm
rkhusky wrote: Mon Jan 17, 2022 5:10 pm Perhaps there has been a premium in the past, but will in continue in the future? In an efficient market with rational and knowledgeable investors, one would expect there to be a positive correlation between risk and return. But are the majority of investors rational and/or knowledgeable enough to actually be able to calculate the risk of an investment?
If you think the market is so inefficient or irrational that there is no longer a relationship between risk and reward then we have a bigger problem than which equity index fund to choose.
I wonder sometimes when I see what is happening with Tesla, crypto-currency and Robinhood. But perhaps the aggregate of all investors averages out to be close to efficient for most things.

Even a positive correlation between risk and return doesn't mean that the relationship is totally rational. But everyone has their own cost function. One investor might require a much higher return for taking a given risk than another investor. The market just displays the average investor.
Post Reply