Small/Value Tilt question

[Chan_va]

Folks,

This question is directed at those who have a small/value tilt in their portfolio. I am considering tweaking my portfolio to tilt slightly in this direction.

If I understand this right, research suggests that a small/value tilt portfolio has historically outperformed a total stock market portfolio with roughly the same risk/volatility.

My questions are

1. Am I understanding this right?
2. If I am, then why would we expect this to continue in the future? Now that the research is public, shouldn't this be priced into the stocks going forward?

Thanks,

- BC

[am]

You have to be VERY patient and may not see a benefit in your investing lifetime.

[pauliec84]

1. Am I understanding this right?

Yes

2. If I am, then why would we expect this to continue in the future? Now that the research is public, shouldn't this be priced into the stocks going forward?

This is rich debate in finance. One possible explanation is that risk is not completely captured by volatility or covariance with the stock market as a whole.

For example, suppose small cap and value companies are more likely to do badly (compared to large and growth stocks) when there is high unemployment. If so you would need more reward to be persuaded to hold these stocks, as you don't want your financial wealth to crash at the same time you lose your job.

[Aptenodytes]

You should read some books and articles on the topic. The wiki has a good set of recommendations, e.g. at http://www.bogleheads.org/wiki/Fama_and ... ctor_Model.

The short answer to your question is that the small value premium is thought by many people to derive from risk, and therefore will continue.

Larry Swedroe periodically summarizes the academic research on this forum. Not everyone here agrees with him, but I find the evidence pretty compelling. You'll need to read up and think it over and see what you conclude.

[Chan_va]

Thanks Pauli. If true, this seems to offer a "free lunch". I cant understand how there isn't a hidden risk/cost somewhere, but people far smarter than I believe in it (Rick/Larry), so I must be missing something

[pauliec84]

Chan_va

I would say most people (including myself) think it is not a free lunch. It is simply a risk not captured by volatility (or CAPM beta). Such as exposure to unemployment risk etc.

[Chan_va]

Chan_va

I would say most people (including myself) think it is not a free lunch. It is simply a risk not captured by volatility (or CAPM beta). Such as exposure to unemployment risk etc.

Thanks. On one hand, I think I shouldn't really invest in things I don't understand. On the other hand, I invest in bond funds, and who really understands them fully?

[pauliec84]

No need to give up. Take a crack at a little reading.

[stlutz]

1. Am I understanding this right?
2. If I am, then why would we expect this to continue in the future? Now that the research is public, shouldn't this be priced into the stocks going forward?

#1) Not completely. Small/value stocks are more volatile than larger stocks. However, from a mathematical perspective, their historical returns have more than compensated for the additional volatility. Additionally, small/value stocks don't always have horrible years at the same time as the overall market, and therefore adding a tilt has frequently reduced volatility at the overall portfolio level. For example, small/value declined less than the overall market in both of the most recent large bear markets.

Of course, this could change in the future--in future bear markets small/value certainly could decline more, which would make the overall portfolio more volatile.

In general, tilters generally believe that adding small/value does increase the riskiness of the stock portion of your portfolio, and as such, they will usually increase their bond allocation slightly to compensate.

#2) This is generally the most entertaining debate on Bogleheads, IMO. The problem is that nobody really knows why the effect occurred historically. It could be because they are riskier, beyond even their added volatility. It could be that investors have consistently mispriced these securities. It could be that economic circumstances unexpectedly favored these companies a few times that resulted in overall outperformance. It could be a statistical anomaly due to liquidity/trading issues that no longer exist. Reasonable arguments are made in favor of various combinations of these conclusions.

As such, you'll never arrive at a certain conclusion as to whether these stocks will outperform in the future. That's part of the risk of trying to "beat the market." At the end of the day, the argument really comes down to whether you expect historical backtested results to persist in the future--will tomorrow be pretty much like yesterday? Historical backtesting really is the best information we have to go by, but it's limited by the fact that it's a) theoretical (i.e. nobody actually invested like this for the past 100 years) and b) the future is unpredictable.

Regardless, your original question is whether you should consider a modest tilt in a small/value direction? I say, sure, why not, as long as the tilt is modest. If you're not sure, make a plan to move to your desired allocation over a 3 year period. During that time, read more, read some of the old debates here, join in new ones, and review each year whether you still think that a tilt makes sense. If you decide a year or two from now that you don't think so, you won't have made what you have come to believe is a huge mistake. On the other hand, if it makes more sense to you as time goes on, you'll end up with your desired allocation and more importantly, you'll stick with it for the longer term.

[larryswedroe]

There are TWO, not one ways to use the TILT STRATEGY
The first is to get higher EXPECTED returns. While SV stocks have been a lot more volatile than the market (almost 2x) the size and value premiums have very low correlation with the market. So you get some diversification benefit there. This is even more true when you use internationals small value stocks as well

The second way is to TILT and use the higher EXPECTED returns to lower your allocation to stocks appropriately so that you have the same expected return for the portfolio. Doing so has historically cut tail risk, but importantly cutting the left tail far more than the right tail. The benefit comes not only from the diversification but because you are able to hold more high quality bonds which tend to do well when the left tail risk for stocks shows up.

I hope that is helpful

Larry

[Chan_va]

Thanks. I will plan to slowly wade in with VISVX.

[zucckerbugger]

Thanks. I will plan to slowly wade in with VISVX.

Just FYI: You can save 14bps by using the ETF (VBR) instead of the investor shares (VISVX).

To be fair, you can get the admiral shares (same ER as ETF) at 10K as well.....

good luck
-z

[Trev H]

1970-2012


TSM
9.75 = CAGR
17.86 = StDev
0.34 = Sharpe
1.00 = US Market Correlation
0.67 = Intl Market Correlation
54.58 = Growth of 1.00

SV
12.63 = CAGR
20.26 = StDev
0.46 = Sharpe
0.76 = US Market COrrelation
0.47 = Intl Market Correlation
166.23 = Growth of 1.00


50% TSM, 50% SV (annual rebalancing)
11.39 = CAGR
17.89 = StDev <<<<-----(notice this)
0.43 = Sharpe
103.37 = Growth of 1.00

Low correlatoins work nicely in the mix.

It gets even better when you diversify into International (notice how low the correlation of SV is to Intl Market), and well when you add in Intl Small - gets even better.

Trev H

[Rodc]

One thing to note (contained above more or less implicitly), if you have some level of risk you have determined that you can handle and are willing to handle, if you add additional small and or value you need to up your allocation to bonds or use safer bonds.

Say you are 60/40 now, then add 5-10%% in bonds to make it 55/45 or 50/50 depending on how much you tilt. No one can tell you precisely as no one knows precisely how the market works and what will happen in the future. But given the inherent uncertainty that is about right.

Note the post above shows one particular time period and one particular dataset, and yet provides results to 4 (!) significant digits. The data support only about 1 significant digit of understanding of how the markets work. Take these results with a humble dose of salt. YMMV.

[Trev H]

Agree with Rodc...

As you take on asset classes on the equity side with higher expected return (and higher potential volatility)... simply tweak your overall stock/bond allocation a bit to allow for that.

For example

30% Total Stock Market Index, 30% Total International Index, 40% Total Bond Index

vs..

12.5% TSM, 12.5% SV, 12.5% International Large Value, 12.5% Intl Small Market, 50% Total Bond Market Index (or could go with Short Term Bond Index, or a combo of Treasuries and TIPS).

Trev H

[Clive]

As an example 30-70 US Small Caps has been broadly similar to 60-40 TSM since the mid 1920's.

[Clive]

If you combine assets individually having high volatility (risky) but that collectively have low overall portfolio volatility, then the expected rewards will be close to the weighted sum of the ARITHMETIC gains. i.e. you'll uplift the portfolio GEOMETRIC reward closer to the average of the arithmetic gains.

From Simba's backtest spreadsheet, individually measure the yearly arithmetic gain divided by the CAGR (compound average growth rate = annualised or geometric) gain for each asset and rank (sort) those descending and the top 5 since 1972 were VPACX, GLD, BRSIX, VISGX, VFSVX

Blend 10% in each of those with 50% 5 year Treasury (bonds) and the overall portfolio value was relatively stable and produced reasonable overall rewards.

That however is data mined according to the circumstances/events over that period. A similar effect however can be observed when using Small Cap Value and International Small Cap Value, but where that effect is more persistent across time and different markets. Perhaps due to both small and value being riskier investments that generally are expected to reward more for taking on exposure to such higher levels of risk. The international holdings are a reasonable choice of diversification as you are exposed to both stock risk and currency risk (buy one risk and get another non correlated risk thrown in for free).

I suspect that holding small amounts in multiple (different) higher risk assets that have low correlations is better than more exposure to a single similar risk. But that's just a guess. For instance comparison of market cap weighted versus equal weighted indexes has no clear/consistent winner.

[garlandwhizzer]

I don't think there's any question in the academic literature that over very long periods of time, SCV outperforms SCG, TSM, or the S&P 500. The premium does exist but there are qualifiers. First, the outperformance comes with higher volatility. Second, in contrast to the academic literature, many, perhaps most real life SCV funds, do not appear to capture that premium sufficiently to overcome their increased cost and tax inefficiency. Do we include utilities and REITS in SCV? How small is small and how valuey should value be? These are questions that real life SCV funds answer in different ways. Most SCV funds, I suspect, underperform the levels that academic studies suggest that they should attain relative to SCB. Finally, even the strongest proponents of tilting to SCV admit that there are long periods of time when SCV underperforms SCB and SCG as well as the S&P 500. Over the past 10 years Vanguards SCV index fund has underperformed their SCB and SCG index funds, for example.

Having said all that, I have recently taken a position in a SCV fund that is highly volatile, very valuey, and very small cap, microcap in fact, believing that the recent years of underperformance after the financial collapse makes SCV more, not less, likely to outperform going forward.

Garland Whizzer

[nedsaid]

I am a slice-dice investor who tilts his portfolio towards small and value.

It is my belief that small value has higher returns and higher volatility because these stocks are riskier. You are being compensated for taking more risk.

Small outperforms large sometimes and large sometimes outperforms small. One zigs while the other zags. Sometimes they both zig but one more than the other. Value and Growth act in the same way, sometimes one style will outperform the other. Over long periods of time, small outperforms large and value outperforms growth. The fact that these four slices of the market perform differently mitigates the increased risk from owning small value. Adding risky non-correlating asset classes can actually increase return and decrease risk.

I think there are behavioral reasons why the premium for small and value do not disappear even though the effect is widely known. This is a boring strategy because you are over-weighting your portfolio with the most boring, unfollowed, unowned,
and least popular stocks in the market. People tend to gravitate towards what is most popular which are large growth stocks. They want to be where the action is. If small value gets popular, it won't stay popular. As people lose interest, then the small premium and the value premium will reassert themselves. Most people are not into boring.

That being said, a total stock market strategy is a winning strategy. Some here believe in slice and dice and others believe in just buying the indexes.

[Trev H]

On volatility... and this statement..

"First, the outperformance comes with higher volatility."

True - at the individual component level - but not necessarily at the portfolio level (which is all that really matters).

At the portfolio level something else comes into play ---- correlations.

Some facts for the period 1970-2012

StDev for components...

17.86 = TSM
20.26 = SV
22.92 = Total Intl
22.62 = Intl Large Value (including EM LV)
31.25 = Intl Small Market (including EM S)
05.30 = Total Bond

Global Lumper Portfolio

30% TSM, 30% Total Intl, 40% Total Bond (annual rebalancing)

10K Growth
$519,114.84
CAGR
9.62
StDev
11.68
Sharpe
0.43

15% TSM, 15% SV, 15% Intl LargeValue, 15% Intl Small Market, 40% Total Bond (annual rebalancing)

10K Growth
$1,027,190.35
CAGR
11.37
StDev
11.99
Sharpe
0.57

In the S&D mix, (Simplified Ultimate Buy & Hold) equity mix, the return is almost double of the Lumper yet the increase in volatility is hardly noticable.

It increased from 11.68 to 11.99 --- who is going to actually feel or notice that.

And if Volatility is a issue for you, you should adjust your overall stock/bond allocation. Much easier to tame volitility with that.

Shifting from holding nothing but Beta, to holding Large and Small components in equal porpotions can easily be compensated for by reducing equity by 5% increasing bonds by 5%....

Or by shifting to bonds that have lower correlation with equity components than Total Bond does (for example treasuries & tips).

Trev H

[am]

Was the superior performance a result of a short spike or constant? Also, how does one implement such a portfolio if most investable assets are in taxable? And like others have said, can current funds capture the value premium? If it is a no brainer, why are some people not doing this, including some experts?

[dbr]

Was the superior performance a result of a short spike or constant? Also, how does one implement such a portfolio if most investable assets are in taxable? And like others have said, can current funds capture the value premium? If it is a no brainer, why are some people not doing this, including some experts?

I don't do it because I don't have to and I don't want to.

More specific reasons might include the following:

1. Some people might want an even simpler portfolio with only 3 or 4 funds, or even 1 fund, in particular if the assets might be taken over by someone who would not understand why the investment choices are what they are or what to do with the investments.

2. Tilted or S&D portfolios can have what they call "tracking error" meaning results may deviate from the expected benefits by larger and longer lasing degrees than total market portfolios do from their expected behavior. Investors may not wish to experience this or dump that problem on a successor. There is also likely to be more management involved in rebalancing.

3. The actual benefit in optimizing the portfolio may be regarded by some as not worth the bother and the exposure. Remember this is not about some amazing free lunch in risk free return but about optimizing return relative to risk (or vice-versa).

4. Expenses, taxes, and finding funds that truly implement the scheme effectively may take some serious attention. Example: One can do this most effectively through a DFA advisor using DFA funds, but there is a cost for that.

5. The nature of the actual risk is not really understood, and the persistance of the benefit might be in doubt. This reduces the incentive to follow this strategy.

6. Other?

[wintermute]

Chan_va

I would say most people (including myself) think it is not a free lunch. It is simply a risk not captured by volatility (or CAPM beta). Such as exposure to unemployment risk etc.

Return varience only includes risks that have occured, or those whose probability has changed. It's really about how accuratly future risk was priced in the past. The market may somewhat over or under pay for large growth stocks, over periods.

I don't buy that its investors avoiding stocks with downside corellated with their own unemployment risk. Almost no one is that sophisicated. Many sell low and buy high in their 401k's, take loans for trivial reasons, buy fad stocks, etc.

[Khanmots]

Was the superior performance a result of a short spike or constant?

It's not constant, but it's not a single spike either. The excess expected return is by no means guaranteed... that's the nature of risk after all; there's periods where they underperform as well.

[Trev H]

Short Spike or Constant...

Listing the two portfolio's again...and the summary stats for the period of 1970-2012


Global Lumper Portfolio
====================

30% TSM, 30% Total Intl, 40% Total Bond (annual rebalancing)

10K Growth
$519,114.84
CAGR
9.62
StDev
11.68
Sharpe
0.43

Simplified Ultimate Buy & Hold (Slice/Dice) with only 4 equity components.
==============================================================

15% TSM, 15% SV, 15% Intl LargeValue, 15% Intl Small Market, 40% Total Bond (annual rebalancing)

10K Growth
$1,027,190.35
CAGR
11.37
StDev
11.99
Sharpe
0.57


Below shows the results of calculating the CAGR for rolling 5 year periods. There are 39 rolling 5 year periods in that span.
Lumper on the left, SUBH (S&D) on the right.


.LUMP....SUBH
=============
.2.49....4.26..1970-1974
.6.98....9.24
.5.96....8.26
.3.54....7.79
.8.01...14.13
13.23...19.69
12.07...17.64
9.90....15.14
12.06...14.99
12.91...15.24
12.60...14.61
15.24...17.58
20.96...21.85
19.32...20.32
19.37...19.77
21.57...22.23
13.62...13.60
12.09...12.24
11.01...11.14
11.28...11.35
.7.47....7.61
12.66...12.57
10.16...10.56
12.16...11.70
11.47...10.33
14.67...14.21
.9.71...11.52
.6.46....9.08
.2.31....5.62
.3.70....8.13
.2.95....7.08
.5.14....7.85
.9.33...11.40
13.00...13.95
.3.06....3.51
.4.89....5.00
.5.59....5.98
.2.57....2.43
.3.11....3.85..2008-2012
=============

Below shows the difference between the rolling 5 year period CAGR's for the two.

Negative = S&D had higher 5 year CAGR... Positive = Lumper had higher 5 year CAGR

-1.77..1970-1974
-2.26
-2.31
-4.25
-6.13
-6.46
-5.57
-5.24
-2.93
-2.33
-2.01
-2.34
-0.89
-1.00
-0.39
-0.66
.0.02
-0.15
-0.13
-0.08
-0.14
.0.09
-0.40
.0.46
.1.14
.0.46
-1.81
-2.61
-3.31
-4.43
-4.13
-2.71
-2.06
-0.95
-0.45
-0.11
-0.39
.0.14
-0.75..2008-2012


If you were going to implement such a equity mix in Taxable, you would want to use the most tax efficient approach possible (ETF's).


Trev H

[Random Walker]

For a taxable investor, can a number be placed on the potential advantage of the tax managed, tax advantaged, and core funds of DFA compared to a tilted VG portfolio? Can these benefits overcome the increased costs associated with DFA. Thanks,

Dave

[Trev H]

A bit more info on the Lumper vs S&D (SUBH mix)...

If you will notice above in the rolling 5 year CAGR comparison, the simple S&D mix outperformed Lumper in 33 of the 39 (5 year periods).

I had a bit more time this evening and looked at the Volatility (StDev) for those same 39 rolling 5 year periods.


.LUMP....SUBH
==============
14.65...16.79 (1970-1974)
17.46...19.98
16.78...19.19
15.62...18.84
13.93...15.87
.7.38....6.67
.4.99....3.40
.6.44....5.78
.6.18....5.69
.6.56....6.01
.6.73....6.59
11.57...11.63
10.44...10.25
11.89...11.53
11.87...11.55
10.36....9.68
13.30...12.82
10.98...11.20
11.70...11.74
11.89...11.92
11.82...11.56
10.36....9.00
.9.02....8.03
.8.04....7.36
.7.47....5.98
.4.17....4.77
.7.83....5.79
10.62....8.40
11.63...10.77
14.04...14.22
13.31...13.23
12.77...13.13
11.25...12.01
.6.16....8.26
14.97...15.23
16.88...17.04
17.02...17.37
16.71...17.08
16.99...17.57 (2008-2012)
=============

It was almost dead even there...

Lumper was the most volatile in 20 of those periods, where S&D was most volatile in 19 of those periods.

Trev H

[midareff]

Chan_va

I would say most people (including myself) think it is not a free lunch. It is simply a risk not captured by volatility (or CAPM beta). Such as exposure to unemployment risk etc.

Thanks. On one hand, I think I shouldn't really invest in things I don't understand. On the other hand, I invest in bond funds, and who really understands them fully?

LOL... Bond funds are tougher than equities. Small value has historically outperformed and I suspect the quick answer is it has higher risk, thereby a higher reward.