Why does CAPM make sense?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
zmaqoptyxbglp
Posts: 110
Joined: Wed May 16, 2018 10:17 pm

Why does CAPM make sense?

Post by zmaqoptyxbglp » Tue Jun 12, 2018 10:16 pm

I'm fairly new on this forum, and from my previous post, it seems like virtually everyone here is a very strong believer in CAPM, talking repeatedly of "risk adjusted" returns. I, however, don't get it at all. I will repost something I said in a previous post.
(For an arbitrarily chosen stock) Thirty years from today, assuming some fixed net profit margin, the revenue growth is going to be the major factor determining the price then. If the company grows by 10x in revenue over the next thirty years, sure the stock will do great. If it goes down by 50% in revenue, the stock probably will tank a fair bit. Does the amount of wiggling seen on the chart tell you that? Nope.

Twenty years ago, looking at how much Apple's stock was wiggling would you have been able to tell where it would end up? Unlikely. Would you have been able to see IBM's fall from the 1990s looking at its wiggliness? I don't think so.
What part exactly is incorrect in my reasoning above? What exactly is so obviously intellectually alluring about CAPM that seems to just pass me by? If you wanted to convince me that I was wrong, what would be your main argument? I would highly appreciate answers that use logical reasoning rather than ones that talk about models that try to fit historical data, and insights purportedly gleaned therefrom.

stlutz
Posts: 4740
Joined: Fri Jan 02, 2009 1:08 am

Re: Why does CAPM make sense?

Post by stlutz » Tue Jun 12, 2018 10:27 pm

Views about CAPM tend to be highly nuanced amongst folks who like to talk investing theory around here. Because there is actually a lot to the model (even though it is mathematically very simple), it might be helpful if you identified your main concerns?

I haven't read the thread that you took your quote from, so I really don't know the context. Is your concern with extrapolating long-term risk from shorter-term volatility?

zmaqoptyxbglp
Posts: 110
Joined: Wed May 16, 2018 10:17 pm

Re: Why does CAPM make sense?

Post by zmaqoptyxbglp » Tue Jun 12, 2018 10:43 pm

stlutz wrote:
Tue Jun 12, 2018 10:27 pm
Views about CAPM tend to be highly nuanced amongst folks who like to talk investing theory around here. Because there is actually a lot to the model (even though it is mathematically very simple), it might be helpful if you identified your main concerns?

I haven't read the thread that you took your quote from, so I really don't know the context. Is your concern with extrapolating long-term risk from shorter-term volatility?
I would change risk to returns. My concern is with extrapolating long term returns from short term volatility. I'm not sure that risk is something that can be mathematically assessed, returns sure can be. I don't think short term volatility provides any insights into long term returns. I am basically looking for a logical refutation for the stuff I quoted.

I would think that my quote is pretty self contained. Please ask me questions if you feel like anything is unclear.

stlutz
Posts: 4740
Joined: Fri Jan 02, 2009 1:08 am

Re: Why does CAPM make sense?

Post by stlutz » Tue Jun 12, 2018 11:05 pm

My concern is with extrapolating long term returns from short term volatility.
I agree with you.

One of the unusual things about CAPM vs. other models that get promoted in the world of finance is that with CAPM, they started with the hypothesis first, which allowed for it to be tested with actual data.

When actual stock returns were plotted, the "security market line" ended up being pretty flat. That is, higher volatility stocks did not in fact produce higher returns. So, in that sense the model has been demonstrated to be false, so we can move on.

That said, I think it *does* have validity in terms of stocks as a group being riskier and likelier to produce higher returns that bonds as a group. For huge amount of money in the marketplace, the risk of stocks makes no sense. So, even though these investors can believe that stocks will beat bonds over the long run, bonds are still the more rational choice for them. As such, stocks can be priced to offer superior returns to bonds. In that sense, I think the CAPM has validity.

That's not the only reason I can think of for stocks to beat bonds, but it is one at least.

bigred77
Posts: 1999
Joined: Sat Jun 11, 2011 4:53 pm

Re: Why does CAPM make sense?

Post by bigred77 » Tue Jun 12, 2018 11:16 pm

Your quote said to assume a fixed net profit margin and consider a revenue multiplier of 10x and 0.5x over a couple of decades.

You are right, in that case revenue growth is going to correlate pretty darn well with stock returns. Nobody would say your wrong.

The problem is profit margins aren’t constant, market perception about an individual stocks future prospects isn’t constant, the risk of bankruptcy isn’t constant, valuations (P/Es) aren’t constant, etc.

The CAPM equation (which I think you may be using to reference the much broader concept of Modern Portfolio Theory) simply states expected returns = the risk free rate + (beta x the equity risk premium). All the of those variables are subject to interpretation (is the risk free rate the 10 year treasury yield? Overnight rate? 3 month yield? Is beta calculated by using the previous 6 months or 10 years? Daily volatility or weekly?)

you referenced short term volatility predicting long term returns. Nobody is (or should) be doing that. I do think many here would agree the CAPM equation better explains returns than revenue growth. I think many here would agree an individual stock bears a certain amount of increased risk that the investor is not compensated for taking over a broad based index fund.

zmaqoptyxbglp
Posts: 110
Joined: Wed May 16, 2018 10:17 pm

Re: Why does CAPM make sense?

Post by zmaqoptyxbglp » Tue Jun 12, 2018 11:28 pm

stlutz wrote:
Tue Jun 12, 2018 11:05 pm
My concern is with extrapolating long term returns from short term volatility.
I agree with you.

One of the unusual things about CAPM vs. other models that get promoted in the world of finance is that with CAPM, they started with the hypothesis first, which allowed for it to be tested with actual data.

When actual stock returns were plotted, the "security market line" ended up being pretty flat. That is, higher volatility stocks did not in fact produce higher returns. So, in that sense the model has been demonstrated to be false, so we can move on.

That said, I think it *does* have validity in terms of stocks as a group being riskier and likelier to produce higher returns that bonds as a group. For huge amount of money in the marketplace, the risk of stocks makes no sense. So, even though these investors can believe that stocks will beat bonds over the long run, bonds are still the more rational choice for them. As such, stocks can be priced to offer superior returns to bonds. In that sense, I think the CAPM has validity.

That's not the only reason I can think of for stocks to beat bonds, but it is one at least.
I don't agree with the reasoning there on bonds. I don't think bonds NEED to offer less returns than stocks by their nature. In my view, they just happen to in the past given the relatively low real interest rates, compared to the real return on productive assets. But this fundamentally need not be the case.

Consider this hypothetical scenario: The real prime interest rate is raised to 15% tomorrow. Now let's say stocks have an earnings yield (1/PE) of 6% (around current forward earnings yield for the US), and marked to market price to book ratio of 2 (roughly where we're at right now in the US). Now what will the market do? It will adjust the prices to reflect this, of course. But what happens after it slashes the price in half? The return on the bonds are still higher, but the market cannot under most normal circumstances sell for less than the salvage value of its assets. In this case, the bonds WILL return more than stocks. We just have not had this in recent US history.

zmaqoptyxbglp
Posts: 110
Joined: Wed May 16, 2018 10:17 pm

Re: Why does CAPM make sense?

Post by zmaqoptyxbglp » Tue Jun 12, 2018 11:39 pm

bigred77 wrote:
Tue Jun 12, 2018 11:16 pm
Your quote said to assume a fixed net profit margin and consider a revenue multiplier of 10x and 0.5x over a couple of decades.

You are right, in that case revenue growth is going to correlate pretty darn well with stock returns. Nobody would say your wrong.

The problem is profit margins aren’t constant, market perception about an individual stocks future prospects isn’t constant, the risk of bankruptcy isn’t constant, valuations (P/Es) aren’t constant, etc.

The CAPM equation (which I think you may be using to reference the much broader concept of Modern Portfolio Theory) simply states expected returns = the risk free rate + (beta x the equity risk premium). All the of those variables are subject to interpretation (is the risk free rate the 10 year treasury yield? Overnight rate? 3 month yield? Is beta calculated by using the previous 6 months or 10 years? Daily volatility or weekly?)

you referenced short term volatility predicting long term returns. Nobody is (or should) be doing that. I do think many here would agree the CAPM equation better explains returns than revenue growth. I think many here would agree an individual stock bears a certain amount of increased risk that the investor is not compensated for taking over a broad based index fund.
So the fact that profit margins aren't constant explain what exactly? Not sure I follow. That's just a simplifying assumption. But you can definitely consider the realistic case where everything is variable. But how is the bottom line 30 years from today reasonably predicted using the past wiggliness of the stock price? I know what MPT says. I'm just asking for a rationale for accepting it.

The CAPM formula says that return is given as some constant added to another constant times the wiggliness of the stock price. You're not saying still why I should consider the wiggliness of the stock price.

lack_ey
Posts: 6613
Joined: Wed Nov 19, 2014 11:55 pm

Re: Why does CAPM make sense?

Post by lack_ey » Tue Jun 12, 2018 11:49 pm

CAPM is not even that widely understood to be that accurate for what it was intended. It's not. First focusing on how well CAPM models returns for a given asset...

Most importantly, you're not analyzing things that make much sense under CAPM. Single stocks are highly driven by undiversified, idiosyncratic risk, especially some stocks. It's a large collection of stocks (where the undiversified components start averaging each other out) that is better modeled by CAPM in terms of how good the fit is. Furthermore, you're focusing on long-term returns. CAPM's market factor explains the variability of short-term returns. The market goes up and down a lot, more so than it goes in any direction. An asset can move substantially with the market while returning over the long term a significantly different amount because returns over time care about net movement, not all the wiggles. Most of the movement is comprised of the wiggles over time.

Furthermore, there's the distinction between backwards-looking analysis and forward predictions. If we estimate an asset's beta over past returns, we get a noisy estimate that is an approximation of the true value over the period, if we want to think of it as that. But that true value is not even what the beta will be going forward, as it likely changes over time. There's further slippage if we assume our computed beta over the past sample will equal the beta in forward returns.

It's just a model, and a lot of its underlying assumptions are probably not even really right anyway, but you're not interpreting the results and using it in an appropriate manner regardless.


If we think more about what CAPM means for risk pricing, it basically just says that the more undiverisifiable risk an asset has (the component that moves with the market, and thus isn't able to be diversified by adding the market) , the more it should return (on average). Otherwise risk and return would be kind of out of whack and there would be some free lunches. It uses a single variable for one type of undiversifiable risk. This is accurate to some extent, but a lot of asset pricing anomalies and perhaps other risk factors add a whole lot of exceptions, and of course it's far from perfect.

JBTX
Posts: 4043
Joined: Wed Jul 26, 2017 12:46 pm

Re: Why does CAPM make sense?

Post by JBTX » Wed Jun 13, 2018 12:03 am

Been decades since I studied it, but simplistically -

More volatile groups of stocks, like tech go go growth stocks, small stocks, etc, tend to have higher betas. They tend to rise faster and fall harder, and tend to have higher returns - long term, as a group. To a large degree, market movements account for the movement of specific groups of stocks, and beta is an indicator of their sensitivity to market movements.

Of course trying to fit any individual stock into it is going to give you garbage. I think it is really supposed to be applicable to a diversified portfolio.

zmaqoptyxbglp
Posts: 110
Joined: Wed May 16, 2018 10:17 pm

Re: Why does CAPM make sense?

Post by zmaqoptyxbglp » Wed Jun 13, 2018 12:07 am

lack_ey wrote:
Tue Jun 12, 2018 11:49 pm
Most importantly, you're not analyzing things that make much sense under CAPM. Single stocks are highly driven by undiversified, idiosyncratic risk, especially some stocks. It's a large collection of stocks (where the undiversified components start averaging each other out) that is better modeled by CAPM in terms of how good the fit is.
That's something you claim.

https://www.investopedia.com/terms/c/capm.asp

"The CAPM model says that the expected return of a security or a portfolio equals the rate on a risk-free security plus a risk premium."

"Using the CAPM model and the following assumptions, we can compute the expected return for a stock:

The risk-free rate is 2% and the beta (risk measure) of a stock is 2. The expected market return over the period is 10%, so that means that the market risk premium is 8% (10% - 2%) after subtracting the risk-free rate from the expected market return. Plugging in the preceding values into the CAPM formula above, we get an expected return of 18% for the stock:

18% = 2% + 2 x (10%-2%)"

https://www.investopedia.com/exam-guide ... -model.asp

"Example: CAPM model
Determine the expected return on Newco's stock using the capital asset pricing model. Newco's beta is 1.2. Assume the expected return on the market is 12% and the risk-free rate is 4%.

Answer:
E(R) = 4% + 1.2(12% - 4%) = 13.6%.

Using the capital asset pricing model, the expected return on Newco's stock is 13.6%."

So you're saying that the authors here are not correct?

zmaqoptyxbglp
Posts: 110
Joined: Wed May 16, 2018 10:17 pm

Re: Why does CAPM make sense?

Post by zmaqoptyxbglp » Wed Jun 13, 2018 12:10 am

JBTX wrote:
Wed Jun 13, 2018 12:03 am
Been decades since I studied it, but simplistically -

More volatile groups of stocks, like tech go go growth stocks, small stocks, etc, tend to have higher betas. They tend to rise faster and fall harder, and tend to have higher returns - long term, as a group. To a large degree, market movements account for the movement of specific groups of stocks, and beta is an indicator of their sensitivity to market movements.

Of course trying to fit any individual stock into it is going to give you garbage. I think it is really supposed to be applicable to a diversified portfolio.
Yes, and? Tech stocks rose, fell and then rose again, and hence do have pretty high beta. According to MPT, they would be expected as a group to return well above average returns. Do you think that tech stocks are going to continue to outperform for the next 30 years?

JBTX
Posts: 4043
Joined: Wed Jul 26, 2017 12:46 pm

Re: Why does CAPM make sense?

Post by JBTX » Wed Jun 13, 2018 12:15 am

zmaqoptyxbglp wrote:
Wed Jun 13, 2018 12:10 am
JBTX wrote:
Wed Jun 13, 2018 12:03 am
Been decades since I studied it, but simplistically -

More volatile groups of stocks, like tech go go growth stocks, small stocks, etc, tend to have higher betas. They tend to rise faster and fall harder, and tend to have higher returns - long term, as a group. To a large degree, market movements account for the movement of specific groups of stocks, and beta is an indicator of their sensitivity to market movements.

Of course trying to fit any individual stock into it is going to give you garbage. I think it is really supposed to be applicable to a diversified portfolio.
Yes, and? Tech stocks rose, fell and then rose again, and hence do have pretty high beta. According to MPT, they would be expected as a group to return well above average returns. Do you think that tech stocks are going to continue to outperform for the next 30 years?
My crystal ball says.....

Who knows. If long term the markets are up, mostly likely tech stocks will lead the way. if the markets go in the toilet long term, tech stocks will probably circle the drain faster than any of them.

CAPM is just a conceptual framework. I don't think it is meant to be a highly accurate predictor of the future.

zmaqoptyxbglp
Posts: 110
Joined: Wed May 16, 2018 10:17 pm

Re: Why does CAPM make sense?

Post by zmaqoptyxbglp » Wed Jun 13, 2018 12:22 am

JBTX wrote:
Wed Jun 13, 2018 12:15 am

CAPM is just a conceptual framework. I don't think it is meant to be a highly accurate predictor of the future.
Eh, I don't follow why the axioms of the framework make sense, and I have not seen a good reason so far noted here either.

But even if the former is somehow the case, why do people keep quoting risk adjusted return metrics like the Sharpe ratio and Sortino ratio and such on this board, kind of assuming that it is a very mathematically precise figure?

JBTX
Posts: 4043
Joined: Wed Jul 26, 2017 12:46 pm

Re: Why does CAPM make sense?

Post by JBTX » Wed Jun 13, 2018 12:39 am

zmaqoptyxbglp wrote:
Wed Jun 13, 2018 12:22 am
JBTX wrote:
Wed Jun 13, 2018 12:15 am

CAPM is just a conceptual framework. I don't think it is meant to be a highly accurate predictor of the future.
Eh, I don't follow why the axioms of the framework make sense, and I have not seen a good reason so far noted here either.

But even if the former is somehow the case, why do people keep quoting risk adjusted return metrics like the Sharpe ratio and Sortino ratio and such on this board, kind of assuming that it is a very mathematically precise figure?
So is it your assertion that there is no incremental reward for risk - however you choose to define risk?

Finridge
Posts: 498
Joined: Mon May 16, 2011 7:27 pm

Re: Why does CAPM make sense?

Post by Finridge » Wed Jun 13, 2018 12:42 am

I don't think most people on this forum are familiar with or refer to CAPM formula--it's not typically discussed in the "Boglehead" books. Nor is any familiarity necessary for Boglehead investing. I would submit that knowing it isn't even useful, because it does not predict the future.

That said, after I learned the CAPM formula, I used it to look at a municipal bond in my portfolio, and the behavior of the bond price was a perfect match. It was very interesting and very cool. I can see why Markowitz and Sharpe were awarded Nobel Prizes for it. But knowing it didn't change my investment strategies at all.

User avatar
JoMoney
Posts: 5878
Joined: Tue Jul 23, 2013 5:31 am

Re: Why does CAPM make sense?

Post by JoMoney » Wed Jun 13, 2018 12:47 am

zmaqoptyxbglp wrote:
Tue Jun 12, 2018 10:16 pm
I'm fairly new on this forum, and from my previous post, it seems like virtually everyone here is a very strong believer in CAPM, talking repeatedly of "risk adjusted" returns. I, however, don't get it at all. I will repost something I said in a previous post.
(For an arbitrarily chosen stock) Thirty years from today, assuming some fixed net profit margin, the revenue growth is going to be the major factor determining the price then. If the company grows by 10x in revenue over the next thirty years, sure the stock will do great. If it goes down by 50% in revenue, the stock probably will tank a fair bit. Does the amount of wiggling seen on the chart tell you that? Nope.

Twenty years ago, looking at how much Apple's stock was wiggling would you have been able to tell where it would end up? Unlikely. Would you have been able to see IBM's fall from the 1990s looking at its wiggliness? I don't think so.
What part exactly is incorrect in my reasoning above? What exactly is so obviously intellectually alluring about CAPM that seems to just pass me by? If you wanted to convince me that I was wrong, what would be your main argument? I would highly appreciate answers that use logical reasoning rather than ones that talk about models that try to fit historical data, and insights purportedly gleaned therefrom.
To start with, you're making assumptions about what you think other people think, then trying to generalize that as being the opinion of a broad group of people. Which are both bad ideas... People quite frequently have very nuanced beliefs that will be based on many experiences different then yours.

I'm not a "believer" in many pieces of CAPM, but there are parts of it that make a lot of sense to me.
For one, the more volatile the portfolio the more sensitive it will be to dramatic changes in potential end point values. If an asset happens to have an extra 1% return over some specific time period, it really isn't saying much if it could just as likely be -1% or more. If a coin flip is just as likely to end up heads as tails, there is no advantage to betting on one or the other (other than taking an unnecessary, unrewarded -no advantage- risk). I don't think stocks follow a normal distribution or random walk path, but it is a model to think about if you don't have other information and are otherwise making bets with no particular advantage.

I am a believer in the idea that markets are very competitve. They're not giving away free lunches, and to some extent, are actually likely to take advantage of those playing a weak game. If you're not bringing an informational advantage to your trading, you're the sucker at the poker table. To that extent, unless you believe you know more (have better information) than "the market" then deviating from the market consensus is more likely to expose you to risks that are unnecessary.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

zmaqoptyxbglp
Posts: 110
Joined: Wed May 16, 2018 10:17 pm

Re: Why does CAPM make sense?

Post by zmaqoptyxbglp » Wed Jun 13, 2018 12:59 am

JoMoney wrote:
Wed Jun 13, 2018 12:47 am
To start with, you're making assumptions about what you think other people think, then trying to generalize that as being the opinion of a broad group of people.
I have just seen people talk about risk adjusted returns, with no one questioning or apparently doubting it (again, my interpretation of what I saw), which made me think that. My goal was definitely not to mischaracterize others' views - my sincere apologies to anyone that felt like I did that.
JoMoney wrote:
Wed Jun 13, 2018 12:47 am
I am a believer in the idea that markets are very competitve. They're not giving away free lunches, and to some extent, are actually likely to take advantage of those playing a weak game.
And so am I (particularly so with US multi-billion dollar caps). I don't disagree on that at all. All I disagree on is the claim that past volatility of any stock or class of stocks predicts future returns.

User avatar
JoMoney
Posts: 5878
Joined: Tue Jul 23, 2013 5:31 am

Re: Why does CAPM make sense?

Post by JoMoney » Wed Jun 13, 2018 1:21 am

zmaqoptyxbglp wrote:
Wed Jun 13, 2018 12:59 am
... And so am I (particularly so with US multi-billion dollar caps). I don't disagree on that at all. All I disagree on is the claim that past volatility of any stock or class of stocks predicts future returns.
It's been pretty well documented that higher volatility stocks don't necessarily earn more than lower volatility stocks, especially when they are discounted by the margin of error caused by the more volatile stocks. The debunking of CAPM as a pricing model is what led Fama-French to look for other "factors" and added 'size' and 'value' as factors that need to be taken into consideration along with the markets 'beta'. When you combine all those factors together, they believe it's a better explanation of how the market is efficiently pricing risk/return... Since that research was popularized, mutual funds were developed specifically to target those other "risk factors", and a lot of of the discussions on this board center around the idea that people think they can increase there exposure to that measure of "risk" on that 3 factor spectrum (sometimes including more risk factors) and increase their returns. I think they're wrong... It makes for a lot of debate on the forum, as far as I can tell there is no consensus.
As far as I know though, nobody has refuted the idea that the market in aggregate won't earn anything extra because people trade, if there is a portfolio that's earning higher returns, there's an equivalent amount under-performing the market somewhere else. You can take a neutral position on that, our you can make bets or tilts in some particular direction based on some specific belief (but most of those tilts will have higher expenses than a cheap broad market index), and when someones bet pays off, and they beat the markets return, there will likely always be a question as to whether that was through skill or luck or maybe a "risk premium".
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

User avatar
telemark
Posts: 2309
Joined: Sat Aug 11, 2012 6:35 am

Re: Why does CAPM make sense?

Post by telemark » Wed Jun 13, 2018 1:34 am

CAPM is not an iron law that none can escape. It's a model, and as George E. P. Box famously said, all models are wrong. He went on to say that some models are useful, and the interesting question is when and how can CAPM be useful.

Risk-adjusted return is certainly a useful concept but difficult to measure in practice. Using the standard deviation is mathematically convenient but is necessarily an oversimplification of all the things we lump under the term "risk". Maybe the problem is in the name.

lack_ey
Posts: 6613
Joined: Wed Nov 19, 2014 11:55 pm

Re: Why does CAPM make sense?

Post by lack_ey » Wed Jun 13, 2018 1:51 am

zmaqoptyxbglp wrote:
Wed Jun 13, 2018 12:07 am
lack_ey wrote:
Tue Jun 12, 2018 11:49 pm
Most importantly, you're not analyzing things that make much sense under CAPM. Single stocks are highly driven by undiversified, idiosyncratic risk, especially some stocks. It's a large collection of stocks (where the undiversified components start averaging each other out) that is better modeled by CAPM in terms of how good the fit is.
That's something you claim.

https://www.investopedia.com/terms/c/capm.asp

"The CAPM model says that the expected return of a security or a portfolio equals the rate on a risk-free security plus a risk premium."

"Using the CAPM model and the following assumptions, we can compute the expected return for a stock:

The risk-free rate is 2% and the beta (risk measure) of a stock is 2. The expected market return over the period is 10%, so that means that the market risk premium is 8% (10% - 2%) after subtracting the risk-free rate from the expected market return. Plugging in the preceding values into the CAPM formula above, we get an expected return of 18% for the stock:

18% = 2% + 2 x (10%-2%)"

https://www.investopedia.com/exam-guide ... -model.asp

"Example: CAPM model
Determine the expected return on Newco's stock using the capital asset pricing model. Newco's beta is 1.2. Assume the expected return on the market is 12% and the risk-free rate is 4%.

Answer:
E(R) = 4% + 1.2(12% - 4%) = 13.6%.

Using the capital asset pricing model, the expected return on Newco's stock is 13.6%."

So you're saying that the authors here are not correct?
Maybe I misunderstood your intent. Are you criticizing the estimate of expected return of the model, the fact that the actual return may be very different from the predicted return, or basically the R^2 of the CAPM model for fitting a single stock's returns?

I'm saying that CAPM's market factor doesn't fit any given stock's returns series very well because most of the behavior is outside market beta and is specific to that stock (in addition to a number of other problems already brought up). Under CAPM, the expected return of a stock is determined by the market beta, the expected return of the market, and the risk-free rate, sure. This model is not actually correct, as evidenced by all the actual empirical data and many of the refinements (which still don't cover everything) in the decades since. But regardless, it may be somewhat in the realm of reasonable in a number of cases.

Stocks with higher market beta will tend to move more when the market moves more, more likely in the same direction. But there is a lot more going on other than that. Under CAPM the "everything else" component is assumed to have zero mean.

Were CAPM absolutely accurate, then we would still have some stocks returning much less than the expected return based on what CAPM says, and some returning much more. After all, the expected value is just a mean of a distribution (probability-weighted average over possible outcomes). CAPM itself is not inconsistent with your example in the original post.

The concept behind CAPM is that stocks should effectively be priced by their undiversifiable risk (the market risk), with expected return thus popping out from that risk exposure. A stock with high sensitivity to the market might be more leveraged, more reliant on good conditions, liable to lose more if things go poorly for stocks in general, etc.; its higher beta and higher undiversifiable risk means that if there's rational pricing based on the market's aggregate trades, it should be discounted and have a higher expected return. Otherwise there wouldn't be as much reason to own it. That's just the basic intuition.


I'm also not clear on how you're tying this into Sharpe and Sortino ratios. Measures of past return relative to some volatility or downside volatility are applicable whether or not CAPM is true to any extent. They have some kind of meaning but need to be understood properly. CAPM doesn't directly relate to these. The explanatory variable has to do with the market, not volatility. The stock's expected return under CAPM relates to the degree of the linear relationship with the market. One stock could have higher beta than another but lower volatility than it.
Last edited by lack_ey on Wed Jun 13, 2018 1:58 am, edited 2 times in total.

User avatar
JoMoney
Posts: 5878
Joined: Tue Jul 23, 2013 5:31 am

Re: Why does CAPM make sense?

Post by JoMoney » Wed Jun 13, 2018 1:55 am

Measuring the market's "beta" and comparing it to another portfolios "beta", is a way to see how much that other portfolio is deviating away from the market (at least in terms of it's being more volatile).
The problem is, in the past, you could have defined and constructed a portfolio on other "risk factors" with the same market beta (1.0) and had it earn higher returns.
So CAPM was debunked as a pricing model, and people started chasing these other "risk factors" believing they would lead to higher returns, and those higher returns start to disappear... and once again those portfolios that are deviating from the market show up even under the CAPM model as not being anything more than a deviation that could just as likely go up as down.

There isn't any style of stocks you can pick that are going to "persistently" earn higher returns, CAPM may not be a good pricing model, but the "broad market" is always going to be the center of gravity as the aggregate of all the market participants.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

User avatar
nisiprius
Advisory Board
Posts: 36655
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: Why does CAPM make sense?

Post by nisiprius » Wed Jun 13, 2018 7:04 am

I quote Sharpe ratios a lot.

1) It makes sense to me that you absolutely need to take risk into account as well as return when you are comparing two investments. Personally, regardless of how any other investors act, I am not willing to take extra risk unless I have a conviction that there is a likelihood of extra reward. Furthermore, failing to allow for risk is one of the biggest avenues for deception e.g. by active funds.

I was personally burned by this in small way circa 2000; I had a part of my portfolio invested in an actively managed fund that seemed to be meaningfully outperforming its benchmark, despite about a 1% expense ratio. I didn't notice, or care, that it was taking extra risk in several small ways--it had a high weight in midcaps and a significant exposure to international stocks. While the market was going up it outperformed. When the market began to fall in 2000-2002, it underperformed. It was just a magnified copy of the market. It wasn't a big deal, but I had misjudged the situation. I was taking more risk than I thought, and the "outperformance" wasn't alpha, it was just due to taking more risk. To put it another way, the fund had a 60/40 allocation but because of the extra risk in the stocks it was acting more like a 70/30 allocation. It was not a big problem, but the outperformance I thought I was getting was an illusion, and I wouldn't have been fooled if I'd paid more attention to risk.

For better or worse, the Sharpe ratio is a measure of risk-adjusted reward, and it's one that can be easily obtained. Despite various potshots, I believe that standard deviation is a good proxy for risk in two senses. First, when I've done some reality checks, it basically has paralleled other measures of risk. You can define risk various ways, but I believe that most kinds of risk go along with other kinds of risk.

2) The Sharpe ratio and CAPM make sense to me in terms of equivalence. Let's say that over some time period the risk-free return was 1%, than fund A and B both had Sharpe ratios of 0.5, and that fund A had a return of 5% and a standard deviation of 8%. That's the same Sharpe ratio, (5% - 1%) / 8% = 0.5, (9% - 1%) / 16% = 0.5. You might prefer fund A or fund B by itself based on how much risk you want to take, but the point is that fund A is exactly equivalent to using equal allocations to fund B and to cash. If B is too risky for us, we can mix 50% B with 50% cash and the mix will have exactly the same return, standard deviation, and Sharpe ratio as fund A.

If two funds have the same Sharpe ratio, we can turn one into the other by mixing in cash.

This works better for decreasing risk than increasing it because of the cost of leverage, but the point is that the Sharpe ratio is a useful, workmanlike measure of something meaningful.

3) Notice that it is still useful even if CAPM is wrong. If, in the real world, we find that risk is not rewarded, then the Sharpe ratio will tell us that, because we will find that the less-risky funds and portfolios have higher Sharpe ratios. If risk is disproportionately rewarded, then the Sharpe ratio will tell us that the riskier funds and portfolios have higher Sharpe ratios.

4) Certainly I'm interested in other measures of risk-adjusted reward, but I'm not aware of many that are easily obtained, except the Sortino ratio. I'm not sure I understand math behind the Sortino ratio--for example, in the MPT/CAPM formulation I can visualize how the tangent line maximizes the Sharpe ratio, but I don't know what, graphically, maximizing the Sortino ratio amounts to. So far I haven't yet seen a case where something with a higher Sharpe ratio had a lower Sortino ratio.

So to me there are two options. a) Use the Sharpe ratio, because it is one reasonable way to put a number on "risk-adjusted reward," or b) focus only on reward, ignoring risk.

I would ask the question: if not Sharpe ratios, and if you do not want to ignore risk, what measurements would you use to decide which of two alternatives had had better risk-adjusted reward... using your own measure of risk, and your own chosen way of adjusting for it?

In real life I pay very little attention to Sharpe ratios in my own investments. That's because I decided a long time ago that I'm a "satisficer," not an "optimizer." I mostly use Sharpe ratios to, uh, confirm my bias that differences from simple total market portfolios are so tenuous that I might as well stick with what's simple, because the alternatives are unconvincing (to me).
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.

bgf
Posts: 609
Joined: Fri Nov 10, 2017 9:35 am

Re: Why does CAPM make sense?

Post by bgf » Wed Jun 13, 2018 8:27 am

CAPM and other similar models like Black-Scholes option pricing formula all come with explicitly unrealistic assumptions about how real markets and securities behave. This is just straight up undisputed. In fact, Black wrote an article titled something like "The Holes in Black-Scholes" that lists every single assumption, and there are several, and why they are unrealistic but necessary to enable one "to do math," which is what economists have been wanting to do for the past 100 years or so. volatility of securities is not static; it changes from period to period, substantially, and is wholly dependent on whatever period you happen to be observing. its measure can also suffer from the quality of the data set used.

i think part of the problem is that these models are referred to so often on the internet, in books, and even in academic papers while their assumptions and inherent limitations are rarely mentioned. this isn't unreasonable, as it would be impractical to have a lengthy footnote of all the technical shortcomings of the Sharpe ratio every single time it was mentioned. i certainly had the impression that these models were far more representative of markets than they actually are, as i was never formally educated in this stuff and just learned it from reading books and articles.

at this point, i kind of take them all with a grain of salt, in much the same way i do when someone tries to determine the over or undervaluation of a company based on its PE ratio.
“TE OCCIDERE POSSUNT SED TE EDERE NON POSSUNT NEFAS EST"

User avatar
camontgo
Posts: 324
Joined: Tue Oct 27, 2009 1:46 pm
Location: Chicago, IL
Contact:

Re: Why does CAPM make sense?

Post by camontgo » Wed Jun 13, 2018 10:44 am

(For an arbitrarily chosen stock) Thirty years from today, assuming some fixed net profit margin, the revenue growth is going to be the major factor determining the price then. If the company grows by 10x in revenue over the next thirty years, sure the stock will do great. If it goes down by 50% in revenue, the stock probably will tank a fair bit. Does the amount of wiggling seen on the chart tell you that? Nope
At a high level, there are a couple steps involved in determining a stock's price.

Step1: Estimate future cash flows
Step2: Determine present value of these cash flows using an appropriate discount rate (based on riskiness of the cash flows estimated in Step 1).

These discounted future cash flows determine today's price for the stock.

In theory, an efficient market is constantly doing both of these things. You are right that the future cash flows we expect the business to generate are extremely important for determining an appropriate price for the stock, and the CAPM tells us nothing about that. However, how do we estimate the discount rate for Step 2? The CAPM is only about this discount rate in Step 2.

The CAPM says that only the non-diversifiable risk is important, and the stock's beta is used to estimate this "market" risk and determine a discount rate. There are plenty of valid criticisms (maybe CAPMs simplifying assumptions are extreme?)...but I think the basic idea that non-diversifiable risk is key is reasonable.

So, assume the market is, on average, doing a good job with the two steps above. If so, we only need to estimate the discount rate to know the expected return of a portfolio. Of course, the cash flows matter, and the price discovery process in the market is certainly trying to estimate them (that's why high growth stocks have higher PEs, stocks prices move on news about future, etc)...but if the market is estimating well, the discount rate = expected return. In practice, we can't predict the return with a CAPM based discount rate estimate, but we can use tools like the CAPM to determine that some portfolios have more exposure to market risk than others and these tend to have a higher return on average....albeit with higher risk of disaster if market conditions get ugly.
"Essentially, all models are wrong, but some are useful." - George E. P Box

Post Reply