The Five Dimensions of Risk

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Novine
Posts: 1164
Joined: Mon Nov 17, 2008 9:07 pm

Re: The Five Dimensions of Risk

Post by Novine » Wed Jul 25, 2012 9:09 pm

David Marotta has taken the view that there are times to tilt to growth when small cap value is relatively expensive. Knowing that there have been times when growth has outpaced value, is this kind of approach worth considering or does this veer too far off the path of what most would consider prudent?

Bongleur
Posts: 2066
Joined: Fri Dec 03, 2010 10:36 am

Re: The Five Dimensions of Risk

Post by Bongleur » Wed Jul 25, 2012 9:49 pm

>Because each of the 5 risk factors is largely unique (if they weren't, they would be explained by one of the other factors), you can diversify across multiple sources of risk/expected return that are largely independent of one another -- in effect spreading your risks or not putting all your risk in one bucket in an effort to get the necessary returns. -- Jerrylee
>

So the least risky portfolio is 20% of each? Any examples of that, and what returns it generates?
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.

User avatar
LH
Posts: 5490
Joined: Wed Mar 14, 2007 2:54 am

Re: The Five Dimensions of Risk

Post by LH » Thu Jul 26, 2012 2:24 am

How is "risk" defined here?

One gets the sense that it is somewhat circularly defined as, that which generates return?

Also, the data that FF sampled, is not statistically significant, so just right there, to start talking about precise and such...
If a financial lifetime is say 40 years, it's not significant.

Comparisons between physics misses the point. Brownian motion is brownian motion. That is predictive. The stock market
Is a human behavioral construct. It can cease to exist due to war, system change, etc. and everything in between, the sample slice again, is small. Brownian motion of particles is of a different order.



take options, and the use of heat diffusion, well prior to the equation for heat diffusion, heat still defused according to the equation(with a nod to the tree falling in the forest sound or no, and schroedingers cat, entanglement etc), prior to black schol Merton etc, option price most Likely did NOT follow the equations, they were arbitrarily valued. Only AFTER the equations were created, did option prices start to follow them.

That is KEY, it's fundamental principle of all this, the friggin particles performing brownian motion or heat diffusion, don't look at the equation generated and CHANGE! Markets do.

So basically
1) risk will be very poorly defined, in a circular fashion with reward
2) the sample size used to generate FF is not statistically significant
3) the system is a human behavioral one, which reads the equations, then changes its behavior.

Portfolio insurance, LTCM, et al all had "precision"


LH

This is done on iPhone, sorry if it's worse than normal

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

Re: The Five Dimensions of Risk

Post by richard » Thu Jul 26, 2012 4:49 am

Jerry_lee wrote:That is a good question. Fama and French are macroeconomists, and their research relates to market behavior from a 30 thousand foot view. To my knowledge, they have never authored a paper that provides investment advice or explicitly covers guidelines on portfolio/asset class selection.
They have given interviews.

Fama has explicitly said that TSM is on the efficient frontier and is a fine equity strategy, but that if one wants to try for more return (at the cost of more risk), tilt a bit towards small value.

French has said that a barbell approach, that is overweighting small and therefore underweighting midcaps, is a decrease in diversification.

None of the material you cite is anywhere close to as explicit as these.
Jerry_lee wrote:Part of the confusion (I think) on the part of TSM investors (or "one factor" investors) is they view stocks vs. bonds as the dominate portfolio decision, and tilting to small and value as optional or tertiary consideration -- or an "add on".
If this is confusion, it's a confusion Fama shares.

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

Re: The Five Dimensions of Risk

Post by richard » Thu Jul 26, 2012 5:00 am

LH wrote:So basically
1) risk will be very poorly defined, in a circular fashion with reward
2) the sample size used to generate FF is not statistically significant
3) the system is a human behavioral one, which reads the equations, then changes its behavior.
1) Correct as the the Fama-French three factor model. Risk essentially means that which explains returns rather than being anything one would normally associate with risk, such as increased chance of bankruptcy or even leverage. Others have shown correlations between the FF "risk factors" and "real risk" but it's not in the original research and is not an exact match. FF have refused to identify the economic risks for which their factors proxy.

They are strong believers in efficient markets. In an efficient market, higher expected return has to go with higher risk, which is why they label their factors risk factors.

Even if we don't have a circularity problem, increasing risk has the effect of giving you a riskier portfolio, which means the real possibility of lower returns. There's no free lunch.

3) It's amazing how widespread it is to ignore the fact that investors change their behavior in light of new information. Changing investor behavior means things that worked in the past may not work in the future. This can be true even for genuine risk factors, as new information may lead to a reassessment of the trade-offs between risk and return.

As a general pattern, if some strategy appears to work, people increase their use of that strategy. This raises prices of the relevant investment, decreasing future returns, which decreases the efficacy of the strategy.

hafius500
Posts: 217
Joined: Sat Sep 10, 2011 4:31 pm

Re: The Five Dimensions of Risk

Post by hafius500 » Thu Jul 26, 2012 5:30 am

Jerry_lee wrote:Because each of the 5 risk factors is largely unique (if they weren't, they would be explained by one of the other factors), you can diversify across multiple sources of risk/expected return that are largely independent of one another -- in effect spreading your risks or not putting all your risk in one bucket in an effort to get the necessary returns. And because each risk is less than perfectly correlated, a multi-risk/multi-factor portfolio is not as risky as the weighted average of each of the stand alone risk factors.
Bolded = free lunch. Fama/French do not believe in free lunches.

Please refute this argument: --If 'factor diversification' exists, the factor model must be wrong--
If you think this is wrong, you must be in a position to demonstrate that the regression equation indicates what factor diversification means. Please explain how the equation predicts factor diversification.

R(t) - RF(t) = a + b[RM(t) - RF(t)] + sSMB(t) + hHML(t) + e(t)

green = compensated, systematic risks
red = uncompensated risks (different country or sector weights, different security selection, noise)

We diversify away diversifiable uncompensated risks. A higher diversification means a higher expected return without higher risk or the same expected return with lower risk.

If we increase SMB and/or HML (on the right side), the expected return (on the left side) must increase too. The expected return increases in line with higher exposure to compensated risks. That's not an increased diversification.

To observe an increased diversification we must maintain that total risk DEclines despite the higher compensated risks. This happens when the exposure to uncompensated risk (error term) DEcreases. = higher expected return without the higher risks predicted by the higher exposure to compensated risks.

This means factor diversification exists if the error term has an expected value different from zero.
But the error term must be random (zero) if the regression equation and the model are valid.
The FF model assumes all uncompensated risks have been diversified away so that the impact of the error term is zero (random)..That explains why FF write about "diversified portfolios of growth stocks".
Factor diversification means the market portfolio is still exposed to uncompensated risks because it is under- or over-exposed to particular sectors or securities..

Thus, if factor diversification exists, the model is invalid. If the model is valid, factor diversification cannot exist.
A non-mathematical example:
Assume morbidity risk (comparable to portfolio risk) is determined by
- hereditary risk factors ( gender, race, = beta risk because all SMB and HML portfolios have simular expected beta risks)
- lifestyle risk factors ( smoking = value risk )
- social risk factors ( size risk ) and that these risks are "independent of one another".
Factor diversification would mean:

A smoking Mexican priest who lives in Mexico and who is a notorious drug dealer has a lower morbidity risk than a non-smoking Mexican priest who lives in Rome and who is not a criminal because the former is exposed to three different morbidity risks while the latter is exposed to one risk factor.
prior username: hafis50

Jerry_lee
Posts: 915
Joined: Tue Mar 06, 2012 1:55 pm

Re: The Five Dimensions of Risk

Post by Jerry_lee » Thu Jul 26, 2012 4:01 pm

richard wrote:
Jerry_lee wrote:That is a good question. Fama and French are macroeconomists, and their research relates to market behavior from a 30 thousand foot view. To my knowledge, they have never authored a paper that provides investment advice or explicitly covers guidelines on portfolio/asset class selection.
They have given interviews.

Fama has explicitly said that TSM is on the efficient frontier and is a fine equity strategy, but that if one wants to try for more return (at the cost of more risk), tilt a bit towards small value.

French has said that a barbell approach, that is overweighting small and therefore underweighting midcaps, is a decrease in diversification.

None of the material you cite is anywhere close to as explicit as these.
Jerry_lee wrote:Part of the confusion (I think) on the part of TSM investors (or "one factor" investors) is they view stocks vs. bonds as the dominate portfolio decision, and tilting to small and value as optional or tertiary consideration -- or an "add on".
If this is confusion, it's a confusion Fama shares.
Fama has said that a TSM portfolio simply represents one of an almost infinite number of options in a mutli-factor market (sorting portfolios on any combination of size and value). And if one is comfortable deriving all of their expected returns (in excess of fixed income) from a LG dominated portfolio, then it is a fine choice. A self employed small business owner, for example, may choose a TSM portfolio to offset their small/value human capital. An XOM employee may choose a TSM/Small Cap allocation with no additional value tilt. A Google employee may choose an all-value equally weighted large and small mix. In every instance, the model guides the way, and the one-factor CAPM is useless.

French said breaking up the market into combinations of S&P and micro cap that mirror the size tilt of TSM doesn't increase diversification, as it excludes mid cap stocks. He didn't say 1/3 each in large cap, mid cap, and small cap is less diversified than 100% in TSM, which is where you are trying to go with your comment.

I cannot say it any simpler than this: if there is more than one risk/return behavior that is impacting portfolio returns, then you already know that those risk/return dimensions are distinct from one another (or there would be only one risk). And knowing that it is a risk/expected return factor, the return could fail to materialize. So if your portfolio isolates only one risk/return dimension, that's fine, but if it fails to materialize, you have effectively concentrated your risk budget in a way that you have nothing to show for it.
The most disciplined investor in the world.

Jerry_lee
Posts: 915
Joined: Tue Mar 06, 2012 1:55 pm

Re: The Five Dimensions of Risk

Post by Jerry_lee » Thu Jul 26, 2012 5:31 pm

hafius500,

There is a lot there, and most of it I disagree with.

First, lets dispense with this "free lunch" nonsense. Its cliche and really means nothing as it relates to the discussion. Assets with high returns and less than perfect correlations can be combined in a way so that the portfolio of assets is not as risky as each of the components. Is that that hard to understand? I mean, modern portfolio theory is a pretty easy to understand and universal principle. We can replace risk/return dimension for security under the MPT framework (given positive expected returns and less than perfect correlations) and draw relevant parallels.

Next, the regression equation you produced simply provides us a measuring stick for how a portfolio loads on the various risk/return dimensions. It says nothing about why the factors have positive expected returns, or the interaction between the sources of the returns themselves.

Lets assume two portfolios and historical risk/return relationships hold. One holds an allocation to stocks, with a small and value tilt, along with fixed income kept reasonably short and high quality. Another investor has levered up a long term treasury bond portfolio so that the expected return is the same as the previous allocation. Which portfolio is "riskier" in the practical sense? Is it the portfolio that assumes a bit of each risk (but doesn't bet everything on anyone in particular?), or is it the levered long bond investor? They only have a single risk in their portfolio....they must have safer portfolio, right?

You are saying (lets use "disappoint" as a proxy for risk) "each risk/return could disappoint, so more factors = more chances to disappoint". I am saying "each risk/return is (largely) independent of the others, so by having multiple exposures, you reduce the total impact when one or more inevitably disappoint."

Finally, let me give you a non-mathematical example:

If your goal is to increase physical fitness, but don't know for sure which activities will have the greatest impact, you don't begin a 2 hour/day cardio routine. Optimally, you reduce calories, eat a more balanced diet, go to the gym for 30 minutes, do 30 minutes of cardio, and 30 minutes of Pilates.

You are saying this "diversified" workout routine won't work as well because it increases our odds of performing activities that won't work as well to improve our physical fitness. I am saying, I don't know which efforts our body will respond to best, so I prefer to diversify my efforts in hopes that I don't waste all my time on one activity that doesn't work (what if my 2hr cardio increases my appetite so much that it offsets the benefits, or I develop an over-use injury).

Thats all I can contribute on this topic.
The most disciplined investor in the world.

User avatar
LadyGeek
Site Admin
Posts: 46056
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: The Five Dimensions of Risk

Post by LadyGeek » Thu Jul 26, 2012 9:35 pm

I took a stab at updating the wiki: Fama and French Five-Factor Model

There's probably more to say here, but this is as far as I can go on my own. I included the references to Robert T's posts, suggested by Ranger.

Comments / questions / suggestions are welcome.
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

User avatar
ObliviousInvestor
Posts: 3272
Joined: Tue Mar 17, 2009 9:32 am
Contact:

Re: The Five Dimensions of Risk

Post by ObliviousInvestor » Fri Jul 27, 2012 9:13 am

Jerry_lee wrote:Assets with high returns and less than perfect correlations can be combined in a way so that the portfolio of assets is not as risky as each of the components.
True. But let's talk for a minute about just US stocks. A TSM fund has already combined all the assets in question. So your point must be in the words "in a way." That is, you want to combine the same stocks in a different way, with the goal of achieving a lower overall level of risk. OK fine.

Please permit a brief tangent.

I've seen people make statements in favor of tilting (toward small, toward value, toward REITs, toward anything) that seem to imply that by definition you can get a better risk-adjusted return from a group of assets by overweighting a sub-group that has similar return characteristics (i.e., return and standard deviation) to the rest of the group, yet less than perfect correlation to the rest of the group. But I do not think that can be true.

Consider the stocks in the S&P 500. Let's break them up into five different groups:
- Group 1 includes stocks 1 and 10, 11 and 20, 21 and 30, etc.
- Group 2 includes stocks 2 and 9, 12 and 19, 22 and 29, etc.
- Group 3 includes stocks 3 and 8, 13 and 18...
- Group 4 includes stocks 4 and 7, 14 and 17...
- Group 5 includes stocks 5 and 6, 15 and 16...

Presumably, the five groups should have fairly similar risk/return characteristics to each other. And each group should have a less than perfect correlation to the S&P 500 as it stands now.

So can we double the allocation to any of the five groups, and end up with a better risk-adjusted return? I don't think so. If it did work that way, we could then double the allocation to one of the other groups using the same argument. And we could keep going (theoretically improving risk-adjusted return each time) until we've doubled the allocation to each of the five groups, thereby ending up exactly where we started (meaning we obviously didn't improve a darned thing).

In other words, just the fact that one sub-group of stocks has a less than perfect correlation to the rest of the group isn't necessarily a reason to overweight that sub-group.

So, without using historical data, what's the argument in favor of most people tilting toward small/value? What is it that's special about a tilt toward small/value stocks?

On another note, I asked previously about what "particular circumstances" would lead a person to tilt toward small/value. You weren't replying to me directly, but you later stated:
A self employed small business owner, for example, may choose a TSM portfolio to offset their small/value human capital. An XOM employee may choose a TSM/Small Cap allocation with no additional value tilt. A Google employee may choose an all-value equally weighted large and small mix.
That type of thing makes sense to me. I understand that small-cap stocks are different from large-cap stocks, and value stocks are different from growth stocks. And if your job is very clearly tied more closely to one corner of the Morningstar style box, it seems reasonable to tilt your portfolio toward the opposite corner.

However, this seems to me to be quite different from a suggestion that most people should tilt toward small/value.
Mike Piper, author/blogger

yobria
Posts: 5978
Joined: Mon Feb 19, 2007 11:58 pm
Location: SF CA USA

Re: The Five Dimensions of Risk

Post by yobria » Fri Jul 27, 2012 12:03 pm

ObliviousInvestor wrote:
Jerry_lee wrote:Assets with high returns and less than perfect correlations can be combined in a way so that the portfolio of assets is not as risky as each of the components.
True. But let's talk for a minute about just US stocks. A TSM fund has already combined all the assets in question. So your point must be in the words "in a way." That is, you want to combine the same stocks in a different way, with the goal of achieving a lower overall level of risk. OK fine.

Please permit a brief tangent.

I've seen people make statements in favor of tilting (toward small, toward value, toward REITs, toward anything) that seem to imply that by definition you can get a better risk-adjusted return from a group of assets by overweighting a sub-group that has similar return characteristics (i.e., return and standard deviation) to the rest of the group, yet less than perfect correlation to the rest of the group. But I do not think that can be true.

Consider the stocks in the S&P 500. Let's break them up into five different groups:
- Group 1 includes stocks 1 and 10, 11 and 20, 21 and 30, etc.
- Group 2 includes stocks 2 and 9, 12 and 19, 22 and 29, etc.
- Group 3 includes stocks 3 and 8, 13 and 18...
- Group 4 includes stocks 4 and 7, 14 and 17...
- Group 5 includes stocks 5 and 6, 15 and 16...

Presumably, the five groups should have fairly similar risk/return characteristics to each other. And each group should have a less than perfect correlation to the S&P 500 as it stands now.

So can we double the allocation to any of the five groups, and end up with a better risk-adjusted return? I don't think so.
Right, we *can* however, do just the opposite - end up with a worse risk adjusted return because we failed to diversify. If the unique risk to Group 1 can be (mostly) diversified away by owning the other groups, why would the market compensate you for it?

Portfolio concentration is an appealing idea, as long as you don't look too closely.

hafius500
Posts: 217
Joined: Sat Sep 10, 2011 4:31 pm

Re: The Five Dimensions of Risk

Post by hafius500 » Fri Jul 27, 2012 12:56 pm

Jerry_lee wrote:hafius500,

There is a lot there, and most of it I disagree with.

First, lets dispense with this "free lunch" nonsense. Its cliche and really means nothing as it relates to the discussion. Assets with high returns and less than perfect correlations can be combined in a way so that the portfolio of assets is not as risky as each of the components. Is that that hard to understand? I mean, modern portfolio theory is a pretty easy to understand and universal principle. We can replace risk/return dimension for security under the MPT framework (given positive expected returns and less than perfect correlations) and draw relevant parallels.

Next, the regression equation you produced simply provides us a measuring stick for how a portfolio loads on the various risk/return dimensions. It says nothing about why the factors have positive expected returns, or the interaction between the sources of the returns themselves.

Lets assume two portfolios and historical risk/return relationships hold. One holds an allocation to stocks, with a small and value tilt, along with fixed income kept reasonably short and high quality. Another investor has levered up a long term treasury bond portfolio so that the expected return is the same as the previous allocation. Which portfolio is "riskier" in the practical sense? Is it the portfolio that assumes a bit of each risk (but doesn't bet everything on anyone in particular?), or is it the levered long bond investor? They only have a single risk in their portfolio....they must have safer portfolio, right?

You are saying (lets use "disappoint" as a proxy for risk) "each risk/return could disappoint, so more factors = more chances to disappoint". I am saying "each risk/return is (largely) independent of the others, so by having multiple exposures, you reduce the total impact when one or more inevitably disappoint."

Finally, let me give you a non-mathematical example:

If your goal is to increase physical fitness, but don't know for sure which activities will have the greatest impact, you don't begin a 2 hour/day cardio routine. Optimally, you reduce calories, eat a more balanced diet, go to the gym for 30 minutes, do 30 minutes of cardio, and 30 minutes of Pilates.

You are saying this "diversified" workout routine won't work as well because it increases our odds of performing activities that won't work as well to improve our physical fitness. I am saying, I don't know which efforts our body will respond to best, so I prefer to diversify my efforts in hopes that I don't waste all my time on one activity that doesn't work (what if my 2hr cardio increases my appetite so much that it offsets the benefits, or I develop an over-use injury).

Thats all I can contribute on this topic.
You didn't answer my question: Is "factor diversification" consistent with a factor model? Look at the model ( the equation ) and tell us what happens when we increase HML or SMB and why would you call it "factor diversification" and why would it reduce portfollo risk. It's a pure mathematical model. Therefore, any reasonable assertion must be formulated in mathematical language. I did it.

Obviously you can't. (BTW, a regression searches for dimensions that are unrelated. This means there are no diversification benefits by definition. If I could improve portfolio efficiency by overweighting X and underweighting Y, X and Y would not be independent dimensions in a regression equation and in a rational factor model that is in equilibrium).

I could have used another argument:
You advise how the individual investor should invest.
But no rational capital market/factor model tells the individual investor how she should invest. A model that could do it would need perfect foresight of all future economic conditions and markets would be unnecessary. The FF model doesn't say investors who want a higher return "should" overweight small or value stocks. It's nothing more than a data-mined picture of the past. Nor does the (FF) model predict the same factors will persist in the future. In fact, today investors use more than a dozen (instead of three) factors.

This means market models are quite unreliable or useless for individual investment decisions (and we haven't yet touched the adaptivity of markets...). We can just use models to estimate expected returns and risks. But there are always competing models. And we observe our interpretations of these models are imperfect because our investments don't perform as expected and as predicted by our models. We have a simple name for this: active management. You just say your (tilted = active) portfolio is better.

William Sharpe suggested a simple test for good investment advice:
Obviously your recommendation everybody should deviate from the market portfolio in the SAME direction because the FF model suggests this is sensible fails this test --Because markets would not clear--
Reject market efficiency and the 3F model and your thesis might be a valid working hypothesis.

Bobcat2 and jlnxx (J. Norstad) on: "William Sharpe on Personal Investing"
"2004 Princeton Lectures in Finance
.
He proposes that investment advisers must ask themselves if they are "macro-consistent" in the following sense:
"If I advised everyone in the world, would markets clear?"
...
Here's an example Sharpe gives:

"It is impossible to give macro-consistent advice without explicitly taking into account the current market values of the various asset classes. If European stocks have a current value equal to 20% of the value of the world market portfolio, the overall portfolio that would be recommended to all world investors must have 20% of its value allocated to European stocks to be consistent with market-clearing (that is for demand to equal supply). If the advisor would recommend 30% be invested in European stocks he or she is assuming that European stocks are undervalued. This may be correct, warranting the corresponding bet on European stocks and against other asset classes. But it is a bet nonetheless and should be recognized as such......
The cure for these and other ills is to incorporate current market values in one's forecasts so that every asset allocation analysis is either macro-consistent or intentionally departs from macro-consistency in explicitly desired ways."
prior username: hafis50

stlutz
Posts: 4543
Joined: Fri Jan 02, 2009 1:08 am

Re: The Five Dimensions of Risk

Post by stlutz » Fri Jul 27, 2012 4:46 pm

These discussions always ignore what the math of a risk model is telling you. A risk model measures the covariance of one curvy line (your portfolio) to other lines (beta, size, valuation) historically. The predictive aspect of a risk model is simply to estimate how much various "factors" will drive your returns going forward. It does not tell you which stocks will do best; it doesn't even tell you which stocks will be riskier in the objective sense (e.g. a high factor loading on "low volatility" would also be low risk).

If you believe that smallcap value stocks will perform the best in the future, then you'd want to look for high small & value loadings. If you think they will do worse, then you'd look for loadings on big and growth. If you don't have a prediction, then the F/F models simply tell you what will be the drivers of your returns, not what those returns will be.

Repeat, a risk model does not forecast future returns or return premiums. Historically, it doesn't even tell you much about average returns over time. Rather, they measure how one price chart fits vs. other price charts over whatever time you were measuring.

If you make a lot of market bets and tactical asset allocation, they are very helpful with portfolio construction. If you're just trying to build a decent diversified portfolio, a risk model really doesn't do anything for you--it's just fun stuff for finance/math geeks. :D

User avatar
SVariance1
Posts: 1187
Joined: Mon Jun 20, 2011 11:27 am
Location: Philadelphia Area

Re: The Five Dimensions of Risk

Post by SVariance1 » Sat Jul 28, 2012 5:25 pm

Jerry_lee wrote:The CAPM, before it was replaced by the FF 5F model, didn't say anything about the size of the equity risk premium or the magnitude of the risk free rate either, but it remained the model for determining how to take risk and expect return in a portfolio (vary the allocation to stocks and t-bills) and what the composition of the risky asset should be (the market).
When was CAPM replaced by the FF 5F model?
Last edited by SVariance1 on Tue Jul 31, 2012 12:21 pm, edited 1 time in total.
Mike

User avatar
nisiprius
Advisory Board
Posts: 36025
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: The Five Dimensions of Risk

Post by nisiprius » Tue Jul 31, 2012 11:49 am

Jerry_lee wrote:First, lets dispense with this "free lunch" nonsense.
Why, since exponents of multi-asset investing almost always make that claim in those very words. Possibly Markowitz himself.

"The only free lunch in investing is diversification."--Larry Swedroe, The Quest for Alpha, 2011, p. 157

"Generations of economics students who learned that 'there ain't no such thing as a free lunch' may be surprised to discover that Nobel laureate Harry Markowitz called diversification one of the economic world's rare 'free lunches.'"--David Swenson, Unconventional Success, p. 17

Is it a free lunch, or isn't it?
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

yobria
Posts: 5978
Joined: Mon Feb 19, 2007 11:58 pm
Location: SF CA USA

Re: The Five Dimensions of Risk

Post by yobria » Tue Jul 31, 2012 12:01 pm

nisiprius wrote:
Jerry_lee wrote:First, lets dispense with this "free lunch" nonsense.
Why, since exponents of multi-asset investing almost always make that claim in those very words. Possibly Markowitz himself.

"The only free lunch in investing is diversification."--Larry Swedroe, The Quest for Alpha, 2011, p. 157

"Generations of economics students who learned that 'there ain't no such thing as a free lunch' may be surprised to discover that Nobel laureate Harry Markowitz called diversification one of the economic world's rare 'free lunches.'"--David Swenson, Unconventional Success, p. 17

Is it a free lunch, or isn't it?
Since you can define "free lunch" a million ways, it would depend on your specific definition.

User avatar
SVariance1
Posts: 1187
Joined: Mon Jun 20, 2011 11:27 am
Location: Philadelphia Area

Re: The Five Dimensions of Risk

Post by SVariance1 » Tue Jul 31, 2012 12:26 pm

yobria wrote:
nisiprius wrote:
Jerry_lee wrote:First, lets dispense with this "free lunch" nonsense.
Why, since exponents of multi-asset investing almost always make that claim in those very words. Possibly Markowitz himself.

"The only free lunch in investing is diversification."--Larry Swedroe, The Quest for Alpha, 2011, p. 157

"Generations of economics students who learned that 'there ain't no such thing as a free lunch' may be surprised to discover that Nobel laureate Harry Markowitz called diversification one of the economic world's rare 'free lunches.'"--David Swenson, Unconventional Success, p. 17

Is it a free lunch, or isn't it?
Since you can define "free lunch" a million ways, it would depend on your specific definition.
I actually agree agree with Yobria on this. It totally depends on the definition. Is value investing a free lunch? I would say that it is not because you get what you expect to get. In contrast, I would describe growth investing as getting less than what you expect to get
Mike

User avatar
GregLee
Posts: 1748
Joined: Wed Oct 27, 2010 3:54 pm
Location: Waimanalo, HI

Re: The Five Dimensions of Risk

Post by GregLee » Tue Jul 31, 2012 2:14 pm

I'm puzzled by the slogan that one hears from Larry Swedroe: you're not compensated for diversifiable risk. But if you can reduce risk by diversifying, that should be worth money to you, because you can then increase your risk to the level it would have had without the diversification by buying securities with higher risk and higher return. So you are compensated by those increased returns, after all.
Greg, retired 8/10.

User avatar
Taylor Larimore
Advisory Board
Posts: 27099
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

Risk and Diversification

Post by Taylor Larimore » Tue Jul 31, 2012 2:32 pm

GregLee wrote:I'm puzzled by the slogan that one hears from Larry Swedroe: you're not compensated for diversifiable risk. But if you can reduce risk by diversifying, that should be worth money to you, because you can then increase your risk to the level it would have had without the diversification by buying securities with higher risk and higher return. So you are compensated by those increased returns, after all.
GregLee:

This article may solve your "puzzled":

The Risk of a Non-Diversified Portfolio

Best wishes
Taylor
"Simplicity is the master key to financial success." -- Jack Bogle

User avatar
GregLee
Posts: 1748
Joined: Wed Oct 27, 2010 3:54 pm
Location: Waimanalo, HI

Re: Risk and Diversification

Post by GregLee » Tue Jul 31, 2012 2:53 pm

Taylor Larimore wrote: This article may solve your "puzzled":

The Risk of a Non-Diversified Portfolio
The article doesn't seem to deal with my puzzle. Since any reduction in risk lets you make more money by buying riskier securities which have higher expected returns, without increasing risk above the level it had before the reduction, why should diversifiable risk be any different in this regard?
Greg, retired 8/10.

yobria
Posts: 5978
Joined: Mon Feb 19, 2007 11:58 pm
Location: SF CA USA

Re: The Five Dimensions of Risk

Post by yobria » Tue Jul 31, 2012 2:54 pm

GregLee wrote:I'm puzzled by the slogan that one hears from Larry Swedroe: you're not compensated for diversifiable risk. But if you can reduce risk by diversifying, that should be worth money to you, because you can then increase your risk to the level it would have had without the diversification by buying securities with higher risk and higher return. So you are compensated by those increased returns, after all.
Right, diversification is not a free lunch because your expected return doesn't change all equal, but risk decreases, so "return adjusting", it's a free lunch.

Post Reply