Asset Allocation

[J6o All In]

So fine, a lot of "invest in index funds" advice gets thrown around here, and that's good stuff. But index funds are only part of the story. Once you decide to invest in them, there is still the question of "what percentages in what funds". This is the question of ASSET ALLOCATION, and it is responsible both for your risk AND return.

This is because, EVEN IF YOU WANT TO MAXIMIZE EV, investing isn't just a matter of finding the index fund (or stock, or whatever) with the highest EV. By diversifying you can increase EV with the same risk, or decrease risk with the same EV. This is why so many academics agree that asset allocation is the most important portfolio decision you will make.

So, I'm going to post my asset allocation, and then you guys either post yours or critique mine (or both).

About me: I am 23, and so this is very aggressive. I created it by reading "a random walk down wall street", "unconventional success", "the four pillars of investing", and "all about asset allocation."




Cool,
Tom

[mwgr5]

First, it would be helpfull to know what types of accounts these funds are in (ie taxable, IRA, or Roth IRA). Also, I think you real estate allocation is quite high, possible performance chasing. What is your rational for that high of a REIT allocation?

[J6o All In]

Yes, good point. Everything is in a taxable account, except maybe 15% of the REIT which is in my 401k. I put the REIT in there because I thought it was the least tax-efficient of the funds that my 401k offered (their bond funds didn't look so great).

Why so much in real estate? Well, the truth is that I feel like the exact figures are really touch and go. Swenson recommends 20% real estate, but also 30% bonds. Since I'm only going 10% bonds, if he's right it should be pretty easy to justify 22.5% real estate. Malkiel's recommendations are similar.

With regard to hard data. Ferri has a risk-return chart between total stock market and real estate from 1978-2004 which indicates that even MORE real estate has been a good idea historically. According to the chart the sweet spot is more or less 50-50 stocks and real estate.

But I don't claim to be a genius at this. How would you go about making this decision?

(Suffice it to say though that the decision wasn't motivated by performance chasing--I don't even keep track of where the market is. I'm in for the long haul.)

Tom

[almostbusto]

i don't see how diversifying can increase EV. i am pretty sure it can maintain or nearly maintain EV at less risk though.


let x,y have Ev=3
Ex=3
Ey=3
E(.5x)+E(.5y)= .5Ex+.5Ey=3

so it doesn't matter how you split up your money across x and y EV is the same. however variance MAY (in the real world, definitely) be reduced.

let varx=vary=4
Var(.5x + .5y)=.25varx+.25vary+.5cov(x,y)

if x and y are independent then cov(x,y)=0
if they are actually the same variable then cov(x,y)=cov(x,x)=varx and variance is maintained
if they are negatively correlated then cov is negative which is even better than them being independent.

so, there you go.

EDIT: i would also consider your portfolio to be conservative FWIW. you are very diversified and long term returns from bonds and reits (the reit part might be arguable, i am not sure) will be lower than a diversified portfolio consisting entirely of stocks.

personally i believe you can be totally diversified and still have all your assets in one class. this isn't true for all asset classes however.

finally, just to clarify if anything is unclear. if you only care about maximizing EV you would invest all your money into the one asset with the highest EV. if there are two asset choices that tie for highest EV (ie in the real world, the difference between the two asset's estimated EV is not statistically significant) then of course you would split up your money between the two.

[jively]

I agree that the REIT allocation is high. I like 10% of the stock portion be in REITs. REITs are stocks, not fixed income, even though they have a higher yield.

I don't like High-yield (junk) bonds at all. Along a DFA-type philosophy, the only reason for fixed income is to reduce the risk of the entire portfolio. If you want higher returns, have a higher allocation to stocks and a lower allocation to bonds. Thus, fixed income should be short- or intermediate-term and high quality. The premiums you get for term (long bonds) and credit quality (junk) are not high enough to justify the risk.

So, instead of the junk bond fund, use Short-Term Bond Index fund. (But for 5% it really doesn't make that much difference.)

Everything else is pretty good.

-Tom

[jively]

[ QUOTE ]
i don't see how diversifying can increase EV. i am pretty sure it can maintain or nearly maintain EV at less risk though.

[/ QUOTE ]
Alright math geek. The arithmetic mean would stay the same but the geometric mean can be raised using diversification. Thus the ending wealth will be increased.

If you have investment A that has a 20% return in even years and a 0% return in odd years, the arithmetic return is 10%. Then you have investment B that has a 0% return in even years and a 20% return in odd years, with the same arithmetic return is 10%. However, the correlation is -1 - completely uncorrelated.

The geometric average of investment A and investment B is 8.15%. However, a portfolio of 50% A and 50% B has a geometric average of 8.25%

After 4 years of investment A or investment B, $1 would grow to $1.44. 4 years of 50% A 50% B grows to $1.46.

Booyah!

-Ton

[almostbusto]

[ QUOTE ]
[ QUOTE ]
i don't see how diversifying can increase EV. i am pretty sure it can maintain or nearly maintain EV at less risk though.

[/ QUOTE ]
Alright math geek. The arithmetic mean would stay the same but the geometric mean can be raised using diversification. Thus the ending wealth will be increased.

If you have investment A that has a 20% return in even years and a 0% return in odd years, the arithmetic return is 10%. Then you have investment B that has a 0% return in even years and a 20% return in odd years, with the same arithmetic return is 10%. However, the correlation is -1 - completely uncorrelated.

The geometric average of investment A and investment B is 8.15%. However, a portfolio of 50% A and 50% B has a geometric average of 8.25%

After 4 years of investment A or investment B, $1 would grow to $1.44. 4 years of 50% A 50% B grows to $1.46.

Booyah!

-Ton

[/ QUOTE ]ummmmm no...

i could be wrong but i don't think i am. you are making some mistakes in your math. first of all investment A and B don't have the same EV. if you end on year 2N+1, the A portfolio is greater than the B portfolio. if you end on year 2N, then the portfolios are equal in value. (nottice invoking that this is an infinite series doesn't help your case, the holding period is finite, people's lives are finite)


furthermore, unless i doing the math wrong a half A half B portfolio really would be worth 1.44 after 4 years.

if i am wrong, i would honestly like you to point it out because i have an interest in this kind of math.


i think ultimately you are confusing the concept of expected value and you are using some 'fuzzy math' it appears.

also, 0 correlation means completely uncorrelated. and in fact these two investments are correlated, postively.


what's wrong with this?

invest 1 dollar in A and 1 dollar in B (50-50 allocation you mentioned).

YEAR 1
A is worth 1.2
B is worth 1

YEAR 2
A is worth 1.2
B is worth 1.2

Year 3
A is worth 1.44
B is worth 1.2

Year 4
A is worth 1.44
B is worth 1.44

so after 4 years, 44% return from A, 44% return from B, 44% return from .5A and .5B

[jively]

Ok, I forgot to mention that I have annual rebalancing. So, at the end of year 1, I sell some of investment A and buy some of investment B so I have 1.1 of each.

Also, if you just say that 50% of the years are 20% return, 50% of the years are 0% return, the arithmetic mean for both is 10%.

I realize I did do some math wrong. 100% A and 100% B have geometric means of 9.54%, and the blended 50% A 50% B portfolio has a geometric mean of 10%. 50% A 50% B has a standard deviation of 0%.

Finally, I agree that 0 correlation is uncorrelated, but -1 is negatively correlated and 1 is positively correlated.

I apologize to OP; we have hijacked the thread.

-Tom

[almostbusto]

the annual rebalancing maximizes EV yes. however your suggestion doesn't actually maximize EV. The best strategy would be 100% of funds in A on odd years and 100% of funds in B on even years. so your system is leaving money on the table.

the increased returns don't come from diversification, they came from investing in the highest returning vehicle at the time. (something that can be very hard to do practically)


so if you were to employ this strategy, which many people try, you would be trying to time the market. buy low and sell high. a day trader who invests all of his money in microsoft on day A then on day B invests everything in cisco, then puts everything in intel on day C isn't practicing diversification.

so... i don't know what else to say at this point other than diversification is a way of lowering you EV slightly for significantly reducing risk. being over-diversified is dangerous imo and is something to keep in mind.

[chaz64]

[ QUOTE ]
being over-diversified is dangerous imo and is something to keep in mind.

[/ QUOTE ]

I have been reading about this recently, and various authors have said that to be properly diversified, you need anywhere from 3 to 11 different funds (depending on the author). I looked at different Vanguard index funds over the last 10 years and how they correlated to the S&P 500. If you reduce risk by diversifying into funds that are uncorrelated I'm thinking it could be done with only four funds - S&P 500, Small Cap Value, Emerging Markets, and TIPS. I'm leaving REITs out because I own a house. Assuming you rebalance periodically, is there any reason to think you really need more funds to be diversified?