Larry Swedroe: concentrating risks/minimizing dispersion

[baw703916]

I've seen Larry Swedroe mention in passing on a couple of threads his preference for minimizing dispersion via using a smaller amount of equities with more concentrated risk.

This caught my attention--I'd suggested a "concept portfolio" in thread # 55334 on the M* board to do more or less the same thing, consisting of

30% TIPS
30% Long-term govt. bonds
10% REITs
30% "As risky as you can find" equities (used DFA EM small in running the numbers)

This seemed to give a similar expected return and S.D. over the last 10 years as a 70/30 TSM/TBM portfolio, but do better in severe downturns (as near as I could back test) for .29-32, 73-74.

In that thread it didn't seem to have attracted anybody's notice, but after seeing Larry mention it a few times, now, I wonder if it might warrant a discssion. Especially since he says that his own portfolio actually is constructed around this concept. (My real portfolio looks nothing like the one above).

Questions:

Is this sort of approach practical for most investors, or only those with a relatively low overall risk level (as Larry has)?

Would it actually be possible for any financial advisor to talk a client into using such a portfolio?

Has Larry written at length about this approach in one of his books or elsewhere?

What are others' opinions of this idea?

Thanks in advance.

Best wishes,
Brad

[United]

It's practical and optimal.

According to MPT, if you want less risk/return, you keep the allocation of the tangency portfolio but increase the proportion of your money invested at the risk free rate. In other words, decreasing your risk shouldn't eliminate your exposure to risky assets. These risky assets will always help your risk-adjusted returns, no matter how little risk you want to take.

[MossySF]

I saw a mention of this 2 weeks ago and have been looking at the idea. Using the Simba backtest spreadsheet, I first evaluated 100% stock portfolios:

* 100% Total Stock
* 11.24%, 17.46 std dev

Doing domestic+international equally:

* 50/50 Total Stock/Total Intl Stock
* 12.26%, 17.74 std dev

To produce similar returns using just Treasuries & EM:

* 70% Treasuries, 30% EM
* 12.18%, 9.77 std dev

Standard deviation is way down. Now I suppose someone might not like the idea of only holding 100% EM on the stock side. So let's split it evenly between EM, SCV and REIT -- this will require increasing the stock allocation as we'll be adding stock classes with lesser performance..

* 60% Treasuries, 40% EM+SCV+REIT
* 11.95%, 8.43 std dev

Adding TIPS and commodities for inflation protection:

* 30% Treasuries, 20% TIPS, 10% commodities, 40% EM+SCV+REIT
* 12.10%, 7.81 std dev

This strategy is quite intriguing.

[baw703916]

But I think the concept might be a little bit beyond MPT.

The Efficient Frontier is a curved line on a two-dimensional plot of expected return vs. risk (for which standard deviation is usually used as a proxy).

But the concept, as I understand it, is that by having more concentrated risks, you can minimize the susceptibility to extreme events (the fat tails) while keeping both the expected return and the S.D. constant.

So now there's three variables: expected return, standard deviation, and performance in a depression.

Brad

[stratton]

Larry has mentioned at various times ~25% in equities: SV, EMV, ISV. He used to have REITs, but not now and there is probably a few commodities in there. Approximately 6-7% in TIPS because of tax advantaged limitations and the rest in intermediate munis.

The problem is Simba's spreadsheet doesn't have EMV and ISV. So this is the closest.

Code: Select all

SV:            8%
EM:            8%
IV:            9%
Com:           5%
TIPS:          6%
Inter Munies: 64%

Code: Select all

                    CAGR  Std. Dev.  Sharpe   Sortino   C-US   C-Intl
1972-2006 Nominal   8.99    8.97     0.37       0.57    0.54    0.55
1972-2006    Real   4.10    9.72    -0.16      -0.20    0.49    0.54

1985-2006 Nominal  10.01    7.09     0.77       3.09    0.51    0.76
1985-2006    Real   6.74    6.83     0.32       0.67    0.51    0.80

I copied the 1985-2006 Intermediate Muni numbers to the 1972-2006 and backfilled with data from the Lehman 7 year muni index. The Lehman data compared to the overlapping VG data is more volatile. I should probably scale it back a bit since it makes the earlier data more volatile.

Paul

[Snoopy]

Mossy:

I have been intrigued with this concept, as well, since reading Larry's posts concerning it.

Could you also include Sortino ratios for the sample portfolios you've posted?

Thanks,

Snoopy

[MossySF]

100% Total Stock
= 11.24%, 17.46 std dev, 0.38 sharpe, 0.63 sortino
R 6.25%, 16.97 std dev, 0.09 sharpe, 0.15 sortino

50/50 Total Stock/Total Intl Stock
= 12.26%, 17.74 std dev, 0.43 sharpe, 0.80 sortino
R 7.22%, 17.57 std dev, 0.15 sharpe, 0.22 sortino

70% Treasuries, 30% EM
= 12.18%, 9.77 std dev, 0.67 sharpe, 1.95 sortino
R 7.15%, 10.19 std dev, 0.15 sharpe, 0.28 sortino

60% Treasuries, 40% EM+SCV+REIT
= 11.95%, 8.43 std dev, 0.73 sharpe, 1.35 sortino
R 6.93%, 8.72 std dev, 0.14 sharpe, 0.20 sortino

30% Treasuries, 20% TIPS, 10% commodities, 40% EM+SCV+REIT
= 12.10%, 7.81 std dev, 0.81 sharpe, 1.60 sortino
R 7.07%, 8.09 std dev, 0.16 sharpe, 0.24 sortino

Coffeehouse
= 11.71%, 10.45 std dev, 0.59 sharpe, 1.21 sortino
R 6.70%, 10.45 std dev, 0.58, 0.39 sortino

(Edited to add real return numbers)

[stratton]

Could you also include Sortino ratios for the sample portfolios you've posted?

How useful are the Sharpe and Sortino ratios in cases like this? Look at the real return values to see how much they differ from the nominal.

Paul

[peter71]

hi mossy,

i think it's a very interesting portfolio but it might be that the components providing the bulk of the return (e.g., REITS, EM, SCV and/or commodities) had a particularly good 10 years. one thing i like to do (in addition to checking how a portfolio has done in the year(s) subsequent to a backtest) is to get a broad sense of the performance of each asset over longer time periods and swap out, say, the best one-year return for each with an estimate of that longer term mean.

also, are your figures arithmetic means, and, if so, can you get this spreadsheet to do geomeans as well?

all best,
pete

[AzRunner]

The problem I see with this type of portfolio is expecting the future to resemble the past. Who's to say that EM or ISV is going to outperform with lower standard deviation over the next 5, 10, 20 or 35 years?

In the above portfolio your equity diversification is very limited. Yes, it may work out well, but it does seem to be an anti-Diehard approach. A heavily tilted portfolio requires a very strong stomach and a strong belief in your investment approach. Otherwise you are doomed to bailing out when the risk inevitably shows up.

To me, a more rational approach is to find a diversified equity portfolio and then use your equity/fixed income AA to control your level of risk.

The argument for a heavy concentration in EM and ISV is that you get the value and small-cap expected return premium as well as low correlation with your US fixed income. But this portfolio does not seem to pass the diversification test. If it provides the optimal return then why doesn't everyone use this and vary their equity/fixed income around this portfolio?

Norm

[baw703916]

The only assumption you're really making that the future will resemble the past is that asset classes like EM, ISV, etc. are going to carry higher risk than the U.S. TSM, and that the risk will be compensated. Ideally, you will have a few equity asset classes to add some diversity, provided you manage to concentrate the risk so that in normal circumstances a portfolio with 30% equities behaves more like a portfolio with 60% equities

I see the advantage of this sort of portfolio as that it likely to lose less value in a 1929-32 scenario. Remember, in a major bear market, pretty much all equities worldwide tend to get hit (Bill Bernstein made this point in one of his books). So the diversification is minimized just when you most need it.

But in a 1929-32 scenario, a porfolio consisting of 60-70% government bonds and 30-40% equities wouldn't lose that much of its value. The bonds would actually go up because interest rates would go way down, and the deflation would make the gain even larger in real terms. The stocks, no matter how risky, can't drop below zero.

Then you want some of your bonds to be TIPS, to guard against a 1973-74 scenario. It's kind of hard to do a realistic backtest of such a portfolio, since TIPS were created because of the 1970s experience. Same thing for REITs--they didn't exist until recently.

Best wishes,
Brad

[Erwin]

It seems to me that the reason that people are thinking Larry's way is that the low correlations of the major asset classes which historically have given you shelter, are not any longer doing the job, since with globalization, the asset classes are now highly correlated. Just graph SPY, EFA and EEM for the past 1-2 years and see it for yourself. If this is the case, any back testing is meaningless. Is this not correct? Erwin

[expo62]

Another consideration is how taxes would impact this type of portfolio. The asset classes used (SCV, REIT, TIPs, Commodities, etc.) are not the most tax efficient and would require a lot of tax-deferred space.

I think someone with a mostly taxable portfolio would find it more difficult to implement effectively.

expo62

[Valuethinker]

The only assumption you're really making that the future will resemble the past is that asset classes like EM, ISV, etc. are going to carry higher risk than the U.S. TSM, and that the risk will be compensated. Ideally, you will have a few equity asset classes to add some diversity, provided you manage to concentrate the risk so that in normal circumstances a portfolio with 30% equities behaves more like a portfolio with 60% equities

I see the advantage of this sort of portfolio as that it likely to lose less value in a 1929-32 scenario. Remember, in a major bear market, pretty much all equities worldwide tend to get hit (Bill Bernstein made this point in one of his books). So the diversification is minimized just when you most need it.

But in a 1929-32 scenario, a porfolio consisting of 60-70% government bonds and 30-40% equities wouldn't lose that much of its value. The bonds would actually go up because interest rates would go way down, and the deflation would make the gain even larger in real terms. The stocks, no matter how risky, can't drop below zero.

Then you want some of your bonds to be TIPS, to guard against a 1973-74 scenario. It's kind of hard to do a realistic backtest of such a portfolio, since TIPS were created because of the 1970s experience. Same thing for REITs--they didn't exist until recently.

Best wishes,
Brad

Brad

I'd have to think about this portfolio a lot. It makes sense, in terms of the maths (I think) but

I am very worried about imputing the characteristics of sub asset classes (Emerging Markets within equities, International Small Value within equities) based on historic performance and volatility.

The assumption that the higher risk will be compensated is, I think, quite dangerous. In effect, we have only *one* model run (what actually happened). We don't know all the other infinite number of possible outcomes. There are lots of future states of the universe where the world goes to heck in a handbasket, and Emerging Markets (or whatever) are the casualty.

As I have opined elsewhere, I suspect the long term correlations of these things converge towards 1, the difference in returns being the amount of leverage and the cash flow characteristics of the asset (for example real estate provides a lower return, but a more stable cash flow). Because what drives equities is international economic growth, and the cost of money (real interest rates). These 2 factors drive any financial asset.

When you say stocks can't drop to 0, they can. Dimson and Marsh cover this. I can't remember the exact figure, but of the world's 10 largest stock markets in 1900, effectively several of them (Buenos Aires, Cairo, Berlin, Moscow, Vienna) have dropped to zero. (there was a discussion here about Berlin: apparently your holdings were restored to you after WWII. However: 1. that is still not fully the case for Holocaust survivors, I don't think 2. they, along with others (this is happening in Zimbabwe now) were forced to sell their assets for nothing in the 1930s 3. I don't imagine foreign investors in Germany in the 30s got to keep their assets either).

Which is not to spin apocalyptic scenarios (more than my usual but simply to say US TSM is a plenty risky asset, but also the parameters are fairly known (right to private property, rights of minority external shareholders etc.).

[heyyou]

Preservation portfolios may take different risks than accumulation portfolios. That is what Larry is doing, limiting the down side since he has enough assets.

Emerging markets are diversified geographically. Risk of government action against investors e.g. nationalizing the oil industry or forced redemptions is just another risk of that asset sub-class. Some EM stocks have lower correlation to the market due to lower participation in the world economy. The TV, phone, and energy systems of more populus EM countries are growing regardless of what happens on Wall Street. Financing costs will vary with international interest rates, and that will impact profits, but the market will respond to the demand for those services. EM has extra risks but the diversification is across more countries than other funds. Glass half full or half empty scenario.

[larryswedroe]

Few thoughts

1) With the equities you hold I believe you want to diversify them highly, though concentrate in the risky end. By that I mean I am diversified across all three risk factors--with exposures to beta, small and value, and the important point is that they have low correlation to each other. I am also diversified by geography with about 50% in US and 35% international developed and rest in EM. Then bit of commodities.

2) Now keep in mind re the point about correlations rising, IMO that is wrong. Yes they have risen. But they have risen before and then fallen. The only place I would expect correlations to stay high is intra Euro countries since they have same monetary policy now.

3)Note that the real risk we are concerned about here is that correlations tend to rise when equities get hit hard by some event risk or financial contagion. Well that is exactly when the portfolio holds up best because you have low exposure to beta. Also by holding CCF and high value tilts that allows me to extend duration a bit on bonds (also hold TIPS) and that environment (deflationary type recessions when correlations rise around the world) is when longer bonds turn negatively correlated to stocks--and TIPS also do well as real rates fall. So that is also part of the strategy--little longer bond duration for taxable accounts holding munis (or TIPS for tax advantaged investors). And the other scenario that hits is high inflation. Then the TIPS and CCF help. So you have hedged all of the risks you care about. What you do give up is the good right tail.

4) As pointed out the ONLY thing really counting on is that risk and EXPECTED return are going to be related. If the risks do show up then I did much better having low equity risks. If the risks do not show up then I did fine also because of the high small and value premiums. Remember that it is in good economies that the small and value premiums show up generally (except in bubbles).

5)The real risk to me is discipline--ability to handle the HUGE tracking error that shows up and be able to ignore it totally. That is the key to success of this strategy IMO.


The way to think about the strategy in simplest form is that you have three gas pedals to make your car go faster and one break pedal. By lowering beta exposure you slow your car down from say 60 to 30MPH>But by then stepping on the size and value pedals you once again raise the speed back to 60mph. But you now have the three pedals pressed down a small amount, not one pedal pressed down a lot. The break pedal of course is fixed income

Hope above is helpful

[larryswedroe]

The following was my first attempt at explaining the strategy. Wrote it while ago. Since then have created the tables I have presented here to show various strategies. Hope you find it helpful. I did send it to bunch of the media and was bit surprised that no one picked up on it.

Effective Diversification in a Three-Factor World
In 1530, with the publication of On the Revolution of Celestial Spheres, Copernicus revolutionized the way we think about our Universe. Prior to 1530 the conventional wisdom was that the sun revolved around the Earth. Copernicus showed that it was the other way around. In similar fashion the June 1992 publication of “The Cross-section of Expected Returns” in the Journal of Finance revolutionized the way we think about investing.
Prior to the publication of the study by Professors Eugene F. Fama and Kenneth R. French the conventional wisdom was that we lived in a one-factor world—the risk and return of a portfolio is determined by its beta. Beta is a measure of equity-type risk (or market risk) of a stock, mutual fund, or portfolio, relative to the risk of the overall (U.S.) stock market. An asset with a beta above one has more equity-type risk than the overall market, while an asset with a beta below one has less equity-type risk than the overall market.
Fama and French hypothesized that we actually lived in a three-factor world—the risk and return of a portfolio is explained by not only beta, but also by its exposure to two other risk factors. Those risk factors are the risk of size (small companies) and price (value stocks). Fama and French found that while small and value stocks have higher beta (they have more equity-type risk), they also have additional risk unrelated to beta. Thus small and value stocks are riskier than large and growth stocks and thus they have higher expected returns. Studies have confirmed that the three-factor model explains well over 90 percent of the returns of diversified portfolios.
For the period the 1927–2005, the average annual returns to these three risks factors was:
·Market Factor (The return of the all equity universe minus the return on one-month Treasury bills): 8.08 percent.
·Size Factor (The return of small stocks minus the return of large stocks): 3.17 percent.
·Price Factor (The return of high book-to-market [value] stocks minus the return of low book-to-market [growth] stocks: 5.03 percent.

Independent Risk Factors
It is important to understand that size and price are independent (unique) risk factors in that they provide investors with exposure to different risks than does exposure to market risks. We can see evidence of their independence when we examine the historical correlations of the size and price factors to the market factor. If the correlations are high, the risk factors will be relatively good substitutes for each other. If that is the case then while investors can increase the expected return (and, of course, risk) of the portfolio by increasing their exposure to these risk factors, there is no real diversification benefit. On the other hand, if the correlations are low, not only will investors increase expected returns, but they will also gain a diversification benefit.
For the period 1927–2005, the correlation of the market risk factor to the size risk factor was just 0.408, and its correlation to the price risk factor was 0.091. And the correlation of the size risk factor to the price risk factor was almost zero (.026). In other words, we can effectively diversify equity risks by diversifying across the three independent risk factors. And each of the three risk factors has the potential for increasing investment returns. The following are two recent examples demonstrating that size and price are independent risk factors.
·In 2001, small stocks returned 18 percent and small value stocks returned 40.6 percent while the S&P 500 produced a negative return of 11.9 percent.
·In 1998, while the S&P rose 28.6 percent while small stocks fell 2.3 percent and small value stocks fell 10 percent.

Diversifying Risk
In the one-factor world there were only two ways to increase returns—either increase the allocation to stocks or buy higher beta stocks. In either case you were taking more of the exact same type of risk you were already taking. The work by Fama and French showed investors that there are other ways to increase the expected return of a portfolio. Instead of adding more of the same type of risk, you could add different types of risk. By adding different types of risk you achieve more effective diversification (not all of your eggs are in one risk basket). The following simplified example (it ignores the diversification return) will illustrate the point.
Let’s assume that in the future we expected equities to provide an annualized return of 7 percent. And we have current bond yields of about 5 percent. Now let’s consider an investor with a portfolio that is currently 50 percent bonds and 50 percent stocks. This allocation results in an expected portfolio return of 6 percent. While developing a financial plan our investor determines that in order to achieve his objective he must achieve a return of return of 6.5 percent, greater than the 6 percent expected return of his portfolio. One way to increase the expected return to 6.5 percent is to increase his allocation to stocks from 50 percent to 75 percent.
(75% x 7%) + (25% x 5%) = 6.5%
Let’s now consider an alternative strategy; one that diversifies risk to other risk factors. For the period 1927–2005, small value stocks achieved an annualized return that was 5.4 percent above the market’s annualized return (15.5 percent versus 10.1 percent). Let’s assume that same relationship will continue in the future. (Note that since size and price are risk factors we do not know that this relationship will continue.) Thus if we are forecasting returns to the market of 7 percent, we would also forecast that small value stocks would return 12.4 percent. Using this information we can look at the expected returns for a few different portfolio allocations.
Let’s first consider a portfolio that takes one-half of the equities and allocates to small value stocks. The expected return would now be:
(25% x 7%) + (25% x 12.4%) + (50% x 5%) = 7.35%
By increasing the allocation to riskier stocks we increased the expected return. We did, however, also increase the risk of the portfolio. The result might be more risk than the investor has the ability, willingness, or need to take. So let’s consider another alternative. This time while we will shift some of the equity allocation to small value stocks (increasing risk), we will also lower the overall equity allocation to just 32 percent (lowering risk). The new allocations are 16 percent total market index, 16 percent small value stocks, and 68 percent bonds. The expected return would now be:
(16% x 7%) + (16% x 12.4%) + (68% x 5%) = 6.5%
We now have a portfolio with a 32 percent allocation to stocks that has virtually the same expected return as the portfolio that had a 75 percent allocation to stocks. Consider, however, that in this case, instead of increasing the expected return by taking more of the same type of risk (market risk), we increased returns by adding different types of risk—the risks of small and value stocks. Thus we diversified our equity risks across these two independent factors. We believe that this a more effective form of diversification. Again, it doesn’t place all of our eggs in one risk-factor basket—while the expected returns of the two portfolios are the same, their risks are, in fact, different.

Risk Aversion
There is another consideration that is especially important to risk averse investors (and most investors are risk averse). Since bonds are safer investments than stocks, if we were to experience a severe bear market, the maximum loss the portfolio could experience is far lower with a 32 percent equity allocation than it is with a 75 percent equity allocation. Thus while the expected returns of the two portfolios are the same, the downside risk is much less with the portfolio with the lower equity allocation. Of course, the upside potential is correspondingly lower as well. For an investor for whom the pain of a loss is greater than the benefit of an equal-sized gain, reducing downside risk as the price of reducing upside potential is a good trade-off.

Considerations
There are several factors that should be given careful consideration when deciding on the appropriate portfolio mix. The first is that an investor should consider how their intellectual capital (earning power) correlates with the greater economic cycle risks that small and value stocks have as compared to large and growth stocks. Thus a tenured professor or doctor, with a low correlation to those risks, can prudently take greater small and value risks. On the other, it may not be prudent for a construction or automobile worker, with a high correlation to those risks, to increase exposure to those risk factors.
The second consideration is a psychological one. It is risk called tracking error regret. For equities tracking error is the amount by which the performance of a portfolio varies from that of the total market, or other broad market benchmark such as the S&P 500 Index. By diversifying across risk factors you take on tracking error risk. While very few investors care when tracking error is positive (their portfolio beats the benchmark), it seems that most investors care when the tracking error is negative. To have a chance for positive tracking error, you must accept the almost certainty that negative tracking error will appear from time to time (or there would be no risk). And, unfortunately, the emotions that negative tracking error can lead to causes many investors to throw their well-thought-out plans into the trash heap. And losing discipline is a recipe for failure. Thus only those investors that are willing and able to accept tracking error risk should consider diversifying across the other risk factors.

Summary
Fama and French showed us that there were two additional risk factors that we should consider when constructing portfolios. We can either use those risk factors to increase the expected return (and risk) of a portfolio, or we can maintain the expected return of the portfolio by diversifying across these independent risk factors while lowering the equity allocation. For many investors we believe that diversifying across these independent risk factors is a more effective way to diversify portfolio risk.

[baw703916]

Larry,

Thanks for the explanation about your portfolio. I'm glad we heard from you--I felt a little strange talking behind your back, as it were.

One question I have is the point Norm raised, why don't more people follow this approach?

(In my case it's because most of my tax-advantaged holdings are in the TSP, so no way to concentrate the equity risks w/o tax consequences.)

And, do you think a financial adviser could sell a client on such a portfolio?

Thanks again and best wishes,
Brad

[Valuethinker]

Few thoughts


2) Now keep in mind re the point about correlations rising, IMO that is wrong. Yes they have risen. But they have risen before and then fallen. The only place I would expect correlations to stay high is intra Euro countries since they have same monetary policy now.

I think this is globalisation of capital markets at work. The world is converging, and becoming more interdependent.

That's something of a truism: the world was more globalised in 1890, when 6 Empires controlled most of the planet's surface, than it is now.

But that is what I see happening. I see it in the UK, the portfolios are far more international, and the trading effects on prices are far more international than they were 15 years ago. And the operations of even small and mid-sized companies are far more global than they were: most noticeably in the importance of China, but even Africa is beginning to figure.

Emerging Markets have begun to integrate with global markets (more India than China, at the stock market level). If Mexico crashes, it's going to have an effect on the US, now. And international bond markets are far larger and more global.

In 1990, it would have been slightly inconceivable that the Bank of New Zealand and the Bank of Iceland's main concern is the portfolio movements of Japanese housewives.

An example is the US subprime thing. The US housing/ S&L crash of 1990 had limited effects on world markets (although the Fed's subsequent easing was important). A couple of British banks lost some money. Now hedge funds in Australia and London are closing up shop, because of fiascos lending money in Orange County.

Insurance is obviously another completely globalised market. Katrina happens, and our house insurance rates go up.

Yes currency has had an effect, the world is increasingly divided into dollar/dollar tracking countries, and the Euro. But it extends beyond that.

On monetary policy, who, now, truly has an independent monetary policy? The US soaks up 60% of the world's international savings flows. Inflation rates, have been radically different between developed countries in the 80s, are now very close across the world. No developed country has an inflation rate less than 1%, or greater than 5%, AFAIK.

So the rise in correlation between world markets that has been observed (and this stretches *across* asset classes too) is justified, and likely to continue.

[Indexbroker]

Larry's strategy is very interesting. Basically it represents a logical conclusion of the Three-Factor-Model of French and Fama.

One of the drawbacks IMHO is that the POTENTIAL returns are lower (as it was mentioned) while comparing to a portfolio with a higher exposure to equity risk.

Second: you are simply ignoring large and growth stocks in your AA.

And third(most important): the strategy is right now very tough to implement because the prices of small value stocks are actually very high.

So the conclusion is: The whole strategy must be compared to the "normal" strategies with exposures to (for instance) large cap and growth equities. My first guess is that actually an implementation of Larry's strategy would not produce results which are as good as "normal" portfolios in terms of expected returns.
But if SV are "cheap" from a historical point of view then you should really consider this strategy.

Cheers
Indexbroker

[johnb]

One question I have is the point Norm raised, why don't more people follow this approach?

...

And, do you think a financial adviser could sell a client on such a portfolio?

I'm not Larry, but I think it could be similar to the old saying, "Nobody gets fired for buying from IBM." With financial advisors, nobody gets fired for recommending the S&P 500.

Best regards,
John

[larryswedroe]

Brad
Why dont more people do it.
First, I have never heard anyone else talk about it, nor ever read anything on this. Not saying I "discovered" this, perhaps others have known this but no one I have talked to had ever even thought about this before. So that is my best guess as to why.

Second, I have talked to lots of people about this. But only those that could handle the tracking error risk. All love the idea once they get it. So I think more people will be adopting it once it gets better known.

Indexbroker--yes you are ignoring the large and growth stocks, that is the source of the tracking error risk. But it is hard to imagine a scenario where they do well in long term without small and value also doing well (they might outperform so you lost the far right tail but who cares since you got good returns anyway). But if they get hurt because of beta the more concentrated risk portfolio will far outperform--and that is exactly what the portfolio is designed to do---trade off upside risk which has low marginal utility to get rid of downside risk which is the risk you want to avoid or minimize. Think of it as portfolio insurance with no cost--except upside potential. That to me is great trade for anyone that has low marginal utility of wealth and/or is risk averse (which is most people)

[MossySF]

also, are your figures arithmetic means, and, if so, can you get this spreadsheet to do geomeans as well?

This is the Simba backtest spreadsheet that covers 1972-2006 and also 1985-2006 for more asset classes. The numbers are geometric mean. You can download this spreadsheet and play around with the numbers to your heart's delight. Beware, you will lose days of productivity.

http://passive-investor.googlegroups.com/web/

So the rise in correlation between world markets that has been observed (and this stretches *across* asset classes too) is justified, and likely to continue.

The rise of correlations seems like an argument for Larry's portfolio makeup. If L/M/S+G/V+US/Intl/EM all go up and down in lockstep in the future, why bother holding both? Just hold the riskiest to get the possibility of getting rewarded for the extra risk.

[zalzel]

As one who has lately been more conscious of my daily Portfolio value than I would like to be, an advantage of Larry's type of Portfolio (shall we call it the Swedroe Portfolio) is than you could learn about what "The Market" is doing every day on the news and have no idea what your Portfolio is doing! That is the upside of tracking error.

I'm serious.

zz

[peter71]

[quote="MossySF"]
This is the Simba backtest spreadsheet that covers 1972-2006 and also 1985-2006 for more asset classes. The numbers are geometric mean. You can download this spreadsheet and play around with the numbers to your heart's delight. Beware, you will lose days of productivity.

http://passive-investor.googlegroups.com/web/

Thanks Mossy,

That's a neat program. I've already wasted far too much time doing something similar but more limited in SPSS but perhaps at some point I'll get around to combining them. In terms of the Geomean though (which so far as I know is the same as the CAGR) it looks like that's in a different row than the "Average," yes? So while the slight gap between the two should have the lowest effect on the lowest SD portfolios, that may help to explain a point or so of the gap between your "averages" and another poster's reports of CAGR's based on a similar strategy . . . maybe not that much though, as an SD of 7 or 8 is really low!

Pete

[peter71]

hi again mossy,

alternatively, as i reread your initial post, it wasn't you who used the term "average" at all. so if that 12+%, 7.8 SD portfolio really is from the "CAGR" row, then that really seems (interestingly) in a different ballpark from the two portfolios posted by stratton . . .

pete

[jinvestor]

I wonder how taxes play into this discussion for higher tax bracket investors with limited tax-deferred accounts.

See Rick Ferri's comments from a recent post regarding his companies general recommendations for small and value.

This is from a recent post

"Ferri, Evanson, or Gorlow as my new advisor? Help please!"

.
I'm not going to comment about the strategies of the other two firms. At Portfolios Solutions, a client with a tax-deferred account can expect to own three DFA funds; Small US Value, Small International Value, and Core Emerging Markets. Client with only a taxable account can expect to own the Core Emerging Markets (possible more depending on taxes).
Rick Ferri

[MossySF]

alternatively, as i reread your initial post, it wasn't you who used the term "average" at all. so if that 12+%, 7.8 SD portfolio really is from the "CAGR" row, then that really seems (interestingly) in a different ballpark from the two portfolios posted by stratton . . .

Stratton and I are modelling different portfolios. Stratton is saying based on what Larry Swedroe has posted in the past about his holdings, here are how the numbers might play out for Swedroe's own portfolio. I was using Swedroe's ideas to see how we would match the performance of somebody who held a more typical 50/50 Total Stock/Total Intl. So we should expect to see totally different numbers.

[zalzel]

The equities sub-portfolio of a Concentrated (Swedroe) Portfolio has, by definition, higher risk than that of the equities sub-portfolio of a more traditional Portfolio. That increased risk includes the increased risk of a prolonged and relentless decline. Does not this increased equities risk put the entire Portfolio at increased risk of failure?

Consider- a prolonged and progressive equities collapse (which is a likelier occurence with a Concentrated Portfolio) could wipe out the entire portfolio as the "safe" FI sub-portfolio is rebalanced into equities and is mercilessly destroyed.

So... decreased dispersion with greater risk of failure?


zz

[larryswedroe]

zz, exactly the OPPOSITE

The significantly lower exposure to beta reduces the risk far more than the increased exposure to the other factors--I showed that result in looking at worst years with my type portfolio. Actually while giving up only about 1/3 of upside in best year I gave up 2/3 of downside in worst year.

[MossySF]

I wonder how taxes play into this discussion for higher tax bracket investors with limited tax-deferred accounts.

Taxes definitely make this a harder portfolio to construct. A conventional portfolio can put large growth/small growth -- and large value if needed -- in taxable. Someone with a 60/40 retirement/taxable split would be able to keep a lot of gains sheltered.

The "Swedroe Portfolio" would require a big bond allocation which could easily take up all tax-deferred space -- assuming your employer's 401K offered a low-cost TIPS/treasuries -- leaving rather tax-inefficient equities in taxable. For a slight hit, you could hold muni bonds in taxable which would free up space for equities.

[stratton]

The "Swedroe Portfolio" would require a big bond allocation which could easily take up all tax-deferred space -- assuming your employer's 401K offered a low-cost TIPS/treasuries -- leaving rather tax-inefficient equities in taxable. For a slight hit, you could hold muni bonds in taxable which would free up space for equities.

If you'll notice the portfolio I posted is 6% TIPS and 64% intermediate munis. The other thing to realize is 64% of that portfolio is tax free because of the muni's so those return numbers mostly belong to the investor. The TIPS would be tax-advantaged.

Paul

[MossySF]

If you'll notice the portfolio I posted is 6% TIPS and 64% intermediate munis. The other thing to realize is 64% of that portfolio is tax free because of the muni's so those return numbers mostly belong to the investor. The TIPS would be tax-advantaged.

It is still a small hit. For example, say you hold Treasuries in a Roth IRA because you have plenty of space due to holding LG/SG in taxable. That definitely outperforms munis in taxable since both would be tax free.

[stratton]

For example, say you hold Treasuries in a Roth IRA because you have plenty of space due to holding LG/SG in taxable. That definitely outperforms munis in taxable since both would be tax free.

I probably wouldn't hold treasuries in a Roth. I'd have that space full of TIPS so there wouldn't be any treasuries to throw off better income than munis.

Paul

[richard]

Larry,

A top-down critique:

Fama has said that the market portfolio is always on the efficient frontier, but that one could have higher expected returns at the cost of increased risk. In other words, you can't construct a portfolio with higher expected returns and the same (or lower) risk or the same expected returns with lower risk.

The market portfolio should be considered the entire investment portfolio, not just TSM or equities.

You have created a portfolio that is supposed to have the same expected returns as the market portfolio, but at a lower level of risk.

It would seem that either Fama is wrong or that your portfolio does not achieve its stated goal.

The only other possibility I can think of is that your portfolio is appropriate for someone whose risks are different than the risks of the "representative" holder of the market portfolio. There are many ways to construct better-than-market portfolios for those who have different risks. If this is case, your portfolio could easily work better for some, but would not be a general solution. Tracking error should not be considered a risk for these purposes - economists tend to think of risks as things such as not have having enough money when needed rather than psychological issues such as tracking error.

[zalzel]

The significantly lower exposure to beta reduces the risk far more than the increased exposure to the other factors--I showed that result in looking at worst years with my type portfolio. Actually while giving up only about 1/3 of upside in best year I gave up 2/3 of downside in worst year.

Larry-

Is this correct? You are saying that you can get the same expected return with lower risk (volatility/dispersion) because the Concentrated Portfolio has done a better job of diversifying risk?

zz

[baw703916]

In Larry's defense, I don't really see the approach as being at odds with Fama's premise. In minimizing the loss in the worst year while not being particularly concerned with year to year variance, the portfolio trades one type of risk for another.

Consider an insurance policy with a low deductable but an average limit, vs. one with a high deductable but an extremely high limit. Which one reduces your risk more? Well, it depends if your're talking about hedging the relatively high risk of a fairly bad event, or of the small risk of a catastrophic event.

Brad

[RobertH]

The only other possibility I can think of is that your portfolio is appropriate for someone whose risks are different than the risks of the "representative" holder of the market portfolio. There are many ways to construct better-than-market portfolios for those who have different risks. If this is case, your portfolio could easily work better for some, but would not be a general solution.

But it might be a general solution for an investor whose preference for liquidity is substantially farther out than Mr. Market's, as is the case with people saving for retirement.

Consider drawdown risk: this risk can be separated into two components: how long the asset class stays below its previous peak (drawdown period) and how far it drops from peak to trough. I find it interesting that the drawdown periods of small and value (but not small growth) have historically been shorter than the market's. The peak-to-trough declines have been worse, but for an investor who doesn't need the money until comfortably beyond the likely drawdown period, peak-to-trough risk need not be a major concern, while the shorter drawdown period is an advantage.

I suspect that peak-to-trough risk explains a good part of why Mr. Market grants premia to the small and value asset classes: in down markets there is a "substantial penalty for early withdrawal" so it's really a premium for illiquidity. (Anyone know of research on this?) To the extent this is true, the small and value premia are a free lunch for those who can hold through the drawdown period.

Robert

[larryswedroe]

Richard--the expected return is the same but the dispersion of returns is lower. IMO it is result of more effective diversification. With TSM you only get beta exposure. But with my portfolio you get exposure to two other independent risk factors.

Mossey--still want to hold equities in taxable if can hold them in relatively tax efficient form. So with advent of DFA TM funds and core funds and ETFs you can hold what was once relatively inefficient asset classes in tax efficient form. Especially if they are passive to begin with.

ZZ-YES
'Example-70-06
100% S&P 500 11.2%/16.8% Return and SD
33% S&P 500/ 21% FF Small Value/ 46% 1-Year Treasuries 11.2%/9.41%

Note same return with much lower SD

Here are best and worst years
37.6%/-26.5% 27.9%/-8.6%

[johndcraig]

Some thoughts

This has been compared to Taleb’s barbell strategy. Although there are some similarities, it is very different. Taleb’s strategy is based upon black swans. 90% safe assets (T Bills) to avoid the negative black swans, and 10 % in highly leveraged, options, venture capital, etc. in order to seek out the positive black swans. Taleb is seeking positive black swans and Larry is seeking value and small premiums.

The basic problem I have with Larry’s strategy is that it is too theoretical and history based. It does not fully consider current valuations by asset class. It assumes the historical premium on small and value (skewed by the 90’s bubble), but does not consider the current factors that lead Bernstein and Bogle to conclude that the SV premium does not exist at this time. If B&B (and others such as GMO) are right, then the strategy will do worse than one that has the entire equity portion in LG or quality stocks. Of course, in the event of a major correction, the large fixed holding will be beneficial, but that is independent of the equity piece.

John

[Black Knights]

Larry (and others),

IIUC, Larry's idea is to reduce Beta-risk by lowering the equity allocation, but to maintain expected return by taking on more size- and Btm-risk in the new, lower equity allocation. This raises a larger question on how one allocates Beta-, size-, and Btm-risk.

While I realize that DFA's Core and Vector portfolios are not run for targeted loading factors, they do seem to have fairly stable loading factors over the limited time they have been around. Further, they seem to have loadings close to s = h, that is, the size and Btm loadings are about equal.

My question is this: Since reasonable forward estimates might be SmB = 2 and HmL = 4, why wouldn't an investor want to approximately equalize expected above-market return (and thus risk) from size and Btm by setting s = 2*h? Not knowing the future, a prudent investor might want to equalize Beta-, size-, and Btm-risk.

Bottom line: What is DFA's rationale for having s = h (approximately) in Core and Vector portfolios?

[Random Musings]

It assumes the historical premium on small and value (skewed by the 90’s bubble), but does not consider the current factors that lead Bernstein and Bogle to conclude that the SV premium does not exist at this time. If B&B (and others such as GMO) are right, then the strategy will do worse than one that has the entire equity portion in LG or quality stocks.

I think the difference is that you are looking at a smaller time frame (and hence timing), while Larry is considering the historical long term effects of the small and value premium.

Now, one could be "cute" and use "tactical asset allocation" to shift between asset classes - but timing for the majority usual fails. Even if one does it, it appears that the guru's say that it should be done in moderation.

And when it comes to GMO "quality stocks" - I don't think their corresponding fund has done as well as their theory. Similar to the Value Line fund versus the Value Line survey. Good on paper, hard to implement in the real world.

RM

[larryswedroe]

John
One point I disagree with it. First Bernstein did not say the value premium was gone. The value premium is undoubtedly smaller than it was historically. Davis showed this by showing that the size of the spread between btm or p/e of value vs growth "predicts" the premium. But there is always a premium. Just a question of how big it is. So today you have to tilt bit more to value and small than would have had to do in 2000.

Second, what you miss IMO is that you gain a diversification benefit since the three factors have very low correlations. So you are exposed to three factors not one.

Third, while it is based on historical evidence and historical premiums, it is based on nothing else more than the logic of risk and expected return being related. Of course the risks can show up, especially in short term. but keep in mind that it is very hard to find scenario where small and value do poorly (absolute, not relative) and stocks do well. There really is no even intermediate period where this is true. Only short periods like 98. Thus the strategy also has good logic--remember small and value stocks have exposure to beta, not just size and value. What you have done is protect the downside greatly while only give up bit of upside. Just cannot lose a lot if have low equity exposure.

[4th&Goal]

What you have done is protect the downside greatly while only give up bit of upside. Just cannot lose a lot if have low equity exposure.

Larry, am I on the right track if I take from this that with my high fixed income holdings I may be better served by having my equities in Small Value funds rather than S&P 500 funds?

[zalzel]

'Example-70-06
100% S&P 500 11.2%/16.8% Return and SD
33% S&P 500/ 21% FF Small Value/ 46% 1-Year Treasuries 11.2%/9.41%

Note same return with much lower SD

Larry- of course this is an example of analyzing historical returns, not expected returns. Is it possible to show that the Concentrated Portfolio has the same Expected Return at Lower Risk (Volatility/Dispersion)? Is it possible to calculate a Portfolio Risk Score for the two portfolios in this example?

zalzel (formerly signed as zz)

[Alex Frakt]

Since such a portfolio is going to have wild swings in the value of its equity holdings, you are obviously going to have to regularly rebalance to keep risks in line. Have you factored in the drag of capital gains taxes incurred by this rebalancing?

Of course some of the gains could be offset by assiduous tax loss harvesting. But, as a practical matter, where does the non-DFA investor put the proceeds from TLH sales while you are waiting out the wash sale period? There just aren't that many choices for these classes once you get beyond domestic value.

[Valuethinker]

The basic problem I have with Larry’s strategy is that it is too theoretical and history based. It does not fully consider current valuations by asset class. It assumes the historical premium on small and value (skewed by the 90’s bubble), but does not consider the current factors that lead Bernstein and Bogle to conclude that the SV premium does not exist at this time.

Buying heavily into small value (as opposed to maintaining a weighting) has got to be a dangerous strategy given the degree of outperformance we have seen (and the increased awareness of the small value performance premium).

I am reminded so poignantly of all those articles in 1999 pointing out the superiority of equities v. all other asset classes, although to be fair to Larry I don't think he is saying most investors should pursue this strategy.

[larryswedroe]

few thoughts

First, I am not saying that this is best strategy for all investors. As everyone I hope knows I don't believe there is such a perfect portfolio. Should be tailored to each person.

Second, if you tilt to small and value you add risk and expected return. Thus if you are going to do so without increasing portfolio risk you have to lower beta exposure, or equity risk overall

Third, I think the returns to all equities will be lower in future than in past, And likely the value premium likely to be bit lower also.

Fourth, as to rebalancing--the amounts not large since you only do this at margin. And you can for some use cash flows, dividends, etc. And of course you should manage by lots. And you do have TLH opportunities also. And you can use similar ETFs to go in and out of not DFA investor.

Fifth, as to dangeous, that I totally disagree with. This clearly lowers downside risk as you lower beta exposure a lot. What you do give up is the upside potential--but that is why it is not right for everyone. Depends on the trade offs (different dispersions you are willing to accept). But this is clearly a more conservative strategy--with lots of tracking error.

[richard]

Richard--the expected return is the same but the dispersion of returns is lower. IMO it is result of more effective diversification. With TSM you only get beta exposure. But with my portfolio you get exposure to two other independent risk factors.

TSM is not the market portfolio, just a portion of it. The market portfolio includes bonds and non-US stocks.

Also, TSM gives you exposure to all three FF risk factors (although the regression coefficient of each of the three is zero).

If portfolio A has the same expected returns as portfolio B, but at a lower risk level, investors will buy the components of A (increasing price and decreasing return) until expected returns and risks of A and B equalize (including all diversification, correlation, etc. factors). That's one reason the market portfolio (essentially, cap weighted holding of all traded investable assets) should always be on the general efficient frontier.

Am I misunderstanding that you believe your portfolio has lower risk, not just lower dispersion of returns (unless the terms are synonymous)?

The amount of broad stock ownership you can replace with a concentrated holding of SV, etc. and achieve the same expected returns depends on the size of the future value premium. How should one go about measuring it? Without an accurate gauge, how does one hope to have the same expected return?

Note that we disagree on the meaning of diversification.

As you may recall, I've always thought this is a very clever strategy.

[larryswedroe]

Richard
First, the TSM has no exposure to size and value, by definition its loading is zero. So while you own some stocks that have positive exposure to the factors, you own others with negative exposure, and the net is zero. So hopefully we agree on that point.

Second, the total market is of course as you state. But that has nothing to do what I should hold, especially the bond to equity holdings. And any portfolio should reflect one's unique situation, including what potential dispersion of returns might be not just the mean return expected.

Third, the market may or may not be on the "efficient frontier"--for example if people prefer to hold stocks with low expected returns because they like the potential dispersion of returns that are different--say SG stocks--then the market will take their tastes into account. I don't have to buy those "inefficient" stocks because I don't like their dispersion. DFA itself must not believe the market is perfectly efficient by the way or they would not screen out stocks at all would they? But we digress. As you know I believe the market is highly efficient in general

Fourth, yes I believe it has less risk--with risk defined as lower dispersion and also lower potential for loss. It must have lower potential for loss because the bonds have no risk basically. Own TIPS there is no risk. AT least in tax advantaged accounts (I hope we straightened out that stuff about the risk of high inflation-that is wrong if in tax advantaged accounts where they should be held). Also own very short term Treasuries and almost no risk there either. So max losses much less--as are max gains of course--so you trade off upside for downside protection.

And glad you think it is clever. (:-))

Note we can at least estimate value premiums going forward as we can stock returns. Of course there is no guarantee. But as I said for the risk averse investor this is a no lose strategy. If stocks do well they are okay no matter what happens. Even if value premium turns out bit less than expected (btw-it is historically high when economy good which is when stocks do well), and it turns out great if stocks do poorly. So except for brief tracking error problem in late 90s strategy works very nicely. Even in great depression it works very well because of low beta exposure--and that is when stocks got crushed and value got even more crushed as did size. For example the market fell about 11% a year from 29-33. But a 34% small value portfolio and 5 year bond was about 0. For full period 29-06 the 100% S&P 500 did bit better at 9.7% but with SD of over 19. The other portfolio returned 9.2 with SD of under 11. Which would you have rather lived through? Especially if in 29 you already has sizable assets.