Asset Alloc - Why not use empirically derived percentages??

[ScaredtoQuit]

I've been following with interest the asset allocation recommendations and tweaks that members of this board have been coming up with for RichinTampa. In other threads, I've also seen quite a number of other members present their own asset allocations. But I've never seen anyone indicate (or recommend) an asset allocation that has supposedly been empirically tested. What I'm talking about is the asset allocations that well respected authors have developed after testing thousands of iterations of historical data. For example, Clyatt in his book, "Work Less, Live More" (a book that I KNOW a good chunk of this board has read) comes up with a theoretical best case asset allocation. In the book, he methodically explains why this allocation is the best, how it was derived and how it was tested. So why doesn't ANYONE seem to just accept it as gospel??

 

[Rich_by_the_Bay]

I've been following with interest the asset allocation recommendations and tweaks that members of this board have been coming up with for RichinTampa. In other threads, I've also seen quite a number of other members present their own asset allocations. But I've never seen anyone indicate (or recommend) an asset allocation that has supposedly been empirically tested. What I'm talking about is the asset allocations that well respected authors have developed after testing thousands of iterations of historical data. For example, Clyatt in his book, "Work Less, Live More" (a book that I KNOW a good chunk of this board has read) comes up with a theoretical best case asset allocation. In the book, he methodically explains why this allocation is the best, how it was derived and how it was tested. So why doesn't ANYONE seem to just accept it as gospel??

It's a very fair question which no doubt will get replies from folks a lot smarter than I am. But a couple of reflections: yes, many AA's have been backtested, I'm sure. What that doesn't reflect is the individual's particular risk tolerance, time scale, presence of SS, pensions, SWR strategy (Bob's reflects his strategy, but not necessarily others), etc. Also, previous classes that were poorly correlated drift over time and become more correlated with each other and vice versa.

I think AAs need to be tailored to the needs of the investor so empiric testing still may not fully determine which one is "best." My goal is to avoid doing something really irrational by accident, and a combination of reading, former advisors, input from posters here, and friends I respect are giving me a comfort level in that regard.

 

[ronin]

My guess is because it isn't empirically foretested.

 

[Spanky]

Simple one:

110 (or 100) - age = % in equity.

For example, age = 50.

Equity = 100 - 50 = 50%.
Fixed income = 50%.

Simple equity allocation: 50% US and 50% International.

 

[audreyh1]

I believe that a financial advisor can use software to derive a very specific allocation mix based on a persons goals, risk tolerance and time frame.

Many articles publish the historical empirical results for various asset allocations. I know the Frank Armstrong "Investing for the 21st Century"? showed graphs for various allocations and showed how adding various asset classes in the mix changed (improved) some characteristics.

The simplest and most important allocation is just the equity to FI ratio, and there are lots of charts and tables around showing average return and volatility for different allocations. An investor can then select the volatility versus average return.

I'm sure some other folks can provide good links - sorry I don't have one handy. I think the Bernstein "Efficient Frontier" articles show some of these curves.

Audrey

 

[bosco]

Many of the asset classes in ESR Bob's recommended portfolio do not have data going back far enough to provide the kind of "empirical test" you describe. Therefore the testing requires a lot of judgment and art (i.e. the opposite of empiricism).

I like Bob's AA for them most part and use it as the starting point for mine, with liberal adjustments made for the fact that I need more protection from a declining dollar than people who will spend their $ in the US. I also skip the hedge fund and private equity parts.

But to take this (or any) suggested AA as some sort of empirically tested fact going forward is wishful thinking at best, and I think Bob would agree. It amounts to a fairly well-hedged, intelligent, betting strategy. One among many possible.

 

[LOL!]

If you go to www.ifa.com you will see 20 asset allocation models labeled 5, 10, 15, ...,95, 100. Presumably, they have all been empirically tested?

But in the end, does it really matter? Someone else will always do better than you did.

 

[unclemick]

William Sharpe - cap weight the world (markets - presumably stock).

Dec 1928 - Walter L Morgan's 'stable investment portfolio' the 70/30 Wellington fund.

Ben Graham's book for boneheads - 50/50 defensive investor postion in The Intelligent Investor. 1949?

60/40 for the 'standard pension fund' - don't know where I picked up that idea.

Vanguard's Target Retirement Series - age ranges and asset allocations - have no idea how they worked that up.

me

heh heh heh - the Norwegian widow says take the asset mix with the 'highest expected growth WITH a current yield you can live on in a hard times mode that can be bought at low expense and doesn't reguire quelling emotion to rebalance.' Target Retirement 2015 is for 58 yr olds(versus my 63/64) so those pesky hormones have me reaching for some lefthanded alpha(live long and prosper) - 2.8% current yield is the epitome of a really cheap person - but only in hard times.

 

[JustCurious]

The answer is what Ronin said:

No matter how much backtesting you do, it's what happens in the future that counts.

If you knew what would happen in the future, there would be no need for asset allocation or diversification, you would simply put all of your money into the one asset class that will perform best in the future. You wouldn't even need to invest across an entire asset class, you would pick the single stock that will appreciate the most and buy it. Obviously, none of us knows the future, so we hedge our bets by diversifying across many stocks, and many asset classes. Therefore, whoever has more money in the right asset class that will do best in the future will then have the "best" asset allocation for that future given time period. However, no amount of backtesting in the world can predict which asset class will peform best in the future. All we can do is make educated guesses using past performance as a guide.

 

[ats5g]

ScaredtoQuit,

I think we can say things about AA like:

1) adding int'l stocks to US stocks may decrease volatility without sacrificing returns
2) adding small stocks to large stocks may decrease volatility without sacrificing returns
3) adding REITs, Commodities, etc. may decrease volatility without sacrificing returns

We can also empirically backtest these portfolios. However, as the other have said, that's a far cry from proving that these trends will continue in the future. One can use tools such as a Mean Variance Optimizer to test portfolios in the future, but the output all depends on the return, risk, and correlation inputs. i.e. Garbage in, Garbage out.

- Alec

 

[cute fuzzy bunny]

Then you have the additional factor that once everyone "knows" something works/helps an AA strategy and they all jump into it, does it still have the same characteristics it had when it was unknown/exotic.

Witness reits, emerging markets and commodities behavior over the last 5 or so years...

On the other hand, if we can convince the unwashed masses that some other, still unknown/underheld asset class is the great big new thing...AFTER we all buy a bunch of it...we'll be RICH....RICH I TELL YOU!

(go ahead Rich, its a slow lob over the plate)

 

[Rich_by_the_Bay]

On the other hand, if we can convince the unwashed masses that some other, still unknown/underheld asset class is the great big new thing...AFTER we all buy a bunch of it...we'll be RICH....RICH I TELL YOU!

(go ahead Rich, its a slow lob over the plate)

I wash my hands 30 times a day.

 

[ScaredtoQuit]

Thanks for all the feedback. For the record, I don't follow ESRBob's recommendations verbatim either. I use his allocation as a starting point and then revise it according to my comfort levels.

I suppose the fact that I was feeling a little guilty over not following the allocations blindly is what prompted me to ask the question. Now I know that I'm not much different than the rest of you.

 

[ronin]

Now I know that I'm not much different than the rest of you.

And you don't find this frightening?

 

[Rich_by_the_Bay]

Thanks for all the feedback. For the record, I don't follow ESRBob's recommendations verbatim either. I use his allocation as a starting point and then revise it according to my comfort levels.

Bob will speak for himself, but I'm guessing that's exactly how he'd want it.

 

[saluki9]

So why doesn't ANYONE seem to just accept it as gospel??

Because as was already pointed out these models are all BACKtested so none of them are going to be gospel. Look at some of those models from "well respected" sources from 1990 - 1995 and see how many of them had an allocation to REITs? Also, you will learn that in academic / professional finance NOTHING is gospel. I could write a paper in the JOF (assuming they would publish it) saying that 1+1=2 and by the next issue there would be 20 pages of letters telling about my lack of research, misunderstanding of risk factors etc...

Also, remember that most of us use the same software (We use Encorr) we use the same data feeds from Bloomberg, and use the same Compustat databases etc. So when we develope a target allocation for a client we have to use our evaluation of the data. (which is for the most part the same data everybody else is using)

 

[sgeeeee]

1) An important question is: What is an asset class?

2) If you decide how to answer 1), then: How many do you want to consider?

3) Correlations between accepted asset classes are not constant though time. What time period do you want to consider? (no matter how you answer this question, there are issues to consider)

4) Will any asset correlations of the past translate into optimal asset allocations over the next several decades?

 

[cute fuzzy bunny]

What time period do you want to consider?

I'll have french toast during the renaissance for $200, Alex.

 

[sgeeeee]

I'll have french toast during the renaissance for $200, Alex.

Oh this should be fun. Who can think of appropriate asset classes for the renaissance?

Lute & clavichord manufacturing?
Painting & sculpture?
Printing presses?
Flying machines (speculative research that did not pay-off for a long time)
. . .

 

[ronin]

I got my Giotto for a song, how about those Medicis...

 

[cute fuzzy bunny]

Oh this should be fun.

Always glad to be of service

Flying machines (speculative research that did not pay-off for a long time)

"Airplanes are interesting toys but of no military value." --
Marechal Ferdinand Foch, Professor of Strategy, Ecole Superieure de Guerre.

A little later than the renaissance, but he also said:

“The most powerful weapon on earth is the human soul on FIRE.”

So, lets just call that first quote a mulligan?

 

[ESRBob]

Thanks for all the feedback. For the record, I don't follow ESRBob's recommendations verbatim either. I use his allocation as a starting point and then revise it according to my comfort levels.

I suppose the fact that I was feeling a little guilty over not following the allocations blindly is what prompted me to ask the question. Now I know that I'm not much different than the rest of you.

Just saw this post -- you mean you don't follow my advice verbatim?

Good, neither do I! (Rich in Tampa was right)

Asset allocation is a personal and dynamic thing. Find targets that feel comfortable (and find as many different points of reference on how they have fared over time) and move toward them at a pace you feel comfortable with, with liberal adjustments for special circumsances.

I like the balance of 40% Equities, 40% bonds and 20% Other, and found lots of other data supporting something like a 50%-50% split giving a nice blend of low volatility and acceptable expected returns. (RetiredInvestor.com has a lot of good studies on different AAs if you're feeling like subscribing).

Then you go the next level into the kinds of equities, or bonds or other asset classes you want -- how much international, small, and what kinds of value/growth blends. With my financial advisor/collaborator for the book we tested a lot of scenarios with as many years of data as we had for this universe, but there will always be longer data series one could want on the more exotic asset classes.

As for my current deviation -- have more private equity than warranted, since I am stuck in an illiquid position in a company that is actually doing well. boo hoo.

Also, we had an investment in a condo which I got sick of (lousy building) and sold for no gain, but haven't felt ready to buy more REITs while I wait around for another good commercial real estate opportunity, so I'm underweight RE for the foreseeable future.

Think of it this way: the day after you rebalance, your AA had drifted from your targets. By a year or two later, when you rebalance again, it will have drifted further. So don't worry about getting AA precisely right -- it will always be off by at least a bit, and maybe more than a bit.

 

[jIMOh]

I've been following with interest the asset allocation recommendations and tweaks that members of this board have been coming up with for RichinTampa. In other threads, I've also seen quite a number of other members present their own asset allocations. But I've never seen anyone indicate (or recommend) an asset allocation that has supposedly been empirically tested. What I'm talking about is the asset allocations that well respected authors have developed after testing thousands of iterations of historical data. For example, Clyatt in his book, "Work Less, Live More" (a book that I KNOW a good chunk of this board has read) comes up with a theoretical best case asset allocation. In the book, he methodically explains why this allocation is the best, how it was derived and how it was tested. So why doesn't ANYONE seem to just accept it as gospel??

Asset allocation is a personal thing. It's almost as personal as underwear, who you sleep with or the kind of beer you drink. There is more than one correct answer to all of above, and even the methods to get to the end result are quite varied.

The purpose of asset allocation is to minimize risk. It is not to maximize return. Therefore back testing gives you "returns", but it's difficult to measure risks because risks can change quickly.

I have my allocation (100% equity; 75% domestic/25% international). The domestic is 45%-15%-15% large-mid-small and the international is 15% large -10% small.

I rebalanced yesterday and sold off 1% into a bond position. I plan to do the same every 6 months until I gradually hit a 10 or 20% bond allocation (I will be close to retiring in about 25 years, maybe less, so I thought I should reduce some risk. I haven't quite figured out which % I'll reduce to accomodate the bonds (I reduced everything 1% to give me the 1% bond position for now).

 

[unclemick]

1966-2006 Screw asset allocation! Having finally arrived at my grand theory:

Chickenheartedness - pick the Target Retirement Series that matches my retirement situation(notice I didn't say age cause I feel young for my age).
I didn't even ask how they arrived/picked/calculated/deduced the 'right' asset allocation - at least it's diversified and cheap.

Went over ten years of early ER age 49- 62 at 60/40ish Lifestrategy moderate - never asked how they selected the weighting - note that it held 'hot rod' Asset Allocation fund which changed based on some quant formula.

BTY - there has never been a year since 1966 (with hindsight) that some allocation/split hasn't blown the barn doors off what I owned at the time. Somewhere along the line - I stopped chasing performance and just worked ongetting the job done for retirement savings/investment. That was exceedingly boring.

Did I mention a few hobby stocks on the side for entertainment purposes?

heh heh heh - - asset allocation - yes - I think I got one!

 

[FinanceDude]

I think it's a personal thing. There are a lot of tools including Monte Carlo simulation to BACKTEST returns...... If I could FORETEST returns, I would be a very rich man..........