Asset Allocation on Eff. Frontier including TSP G Fund

Have a question about your personal investments? No matter how simple or complex, you can ask it here.
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Asset Allocation on Eff. Frontier including TSP G Fund

Post by tadamsmar »

I would like to have an AA that:

1. Is on the efficient frontier

and:

2. Assumes access to the TSP G Fund

The only AAs available that meet these requirements, that I know of are the TSP L Funds:

http://www.tsp.gov/rates/fundsheet-lfunds.pdf

But I would like to include the Vg REIT Fund in the allocation and the L Funds don't have that option.
diasurfer
Posts: 1855
Joined: Fri Jul 06, 2007 8:33 pm
Location: miami-dade

Post by diasurfer »

I think we need more information. Do you have outside accounts (trad IRA)? If so, what is the relative size of your TSP vs IRA?
User avatar
DaveTH
Posts: 2676
Joined: Thu Apr 05, 2007 7:23 pm

Post by DaveTH »

But I would like to include the Vg REIT Fund in the allocation and the L Funds don't have that option.
REITs are already included in the S fund (~ 6%). Many people include a separate allocation to REITs, but it is not necessary unless you want to overweight that asset class.
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

diasurfer wrote:I think we need more information. Do you have outside accounts (trad IRA)? If so, what is the relative size of your TSP vs IRA?
All accounts are tax defferred (IRA, 401k, TSP) or Roth.

Over 1/4th of the total is in the TSP.

But, you don't think you really need this to answer my question, so please make sure you understand what I am asking for.
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

DaveTH wrote:
But I would like to include the Vg REIT Fund in the allocation and the L Funds don't have that option.
REITs are already included in the S fund (~ 6%). Many people include a separate allocation to REITs, but it is not necessary unless you want to overweight that asset class.
Well, I have been using the Life-Cycle Plan from "Random Walk Down Wall Street". It does overweight REIT.

I realize that I don't need the question answered if I just follow one of the L Fund allocations for my overall allocation.

But, if you search "TSP G Fund" on this board, you will find that people who have a TSP and other accounts are always asking how to allocate while overweighting something or other.

For whatever reason, no everyone is satisfied with an L fund allocation.
dbr
Posts: 46181
Joined: Sun Mar 04, 2007 8:50 am

Post by dbr »

The critical question might be why are YOU not satisfied with an L Fund allocation?

One could well ask why an investor with investment opportunities that represent a good plan would want or need to jump through hoops in hopes (possbibly false hopes) of finding a better plan.
Helot
Posts: 264
Joined: Sat Mar 03, 2007 2:05 pm

Post by Helot »

I'm not sure what it is you are asking.

There is nothing inherently more efficient regarding the L Fund allocations than an allocation you could create on your own using your various accounts. Such an allocation could model itself on various proposed allocations: Malkiel, Swensen, Vanguard Target Retirement, etc..

The "efficient" frontier in regard to risk/return can only be known after the fact as future/expected returns are unpredictable.

A simple way to mimic an allocation proposed by Malkiel using your G fund would be to consider it half of your fixed income allocation with something similar to the F Fund (total bond) representing the other half. There are a number of conversations on this board specific to how to use the G Fund in creating a fixed income allocation.
diasurfer
Posts: 1855
Joined: Fri Jul 06, 2007 8:33 pm
Location: miami-dade

Post by diasurfer »

tadamsmar wrote:
diasurfer wrote:I think we need more information. Do you have outside accounts (trad IRA)? If so, what is the relative size of your TSP vs IRA?
All accounts are tax defferred (IRA, 401k, TSP) or Roth.

Over 1/4th of the total is in the TSP.

But, you don't think you really need this to answer my question, so please make sure you understand what I am asking for.
I only ask because, for example, my wife's retirement is in TSP, and her retirement balance is about 3% of what mine is, so even putting all of hers in G doesn't make much difference to our AA. If it was the other way around, and I wanted VG REIT in our retirement, then that wouldn't make much difference because it could be at most 3%.

It seems to me that you would want to follow the standard posting procedures with accounts and balances and emphasize you want to use G fund. Since you think this is not important, I guess I don't understand what you're asking.

But since I'm no expert even by amateur Boglehead standards I guess I should just leave it alone. [But I am interested in your replies since my wife's account will catch up to mine and everyone talks about how great the G fund is.]
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

dbr wrote:The critical question might be why are YOU not satisfied with an L Fund allocation?

One could well ask why an investor with investment opportunities that represent a good plan would want or need to jump through hoops in hopes (possbibly false hopes) of finding a better plan.
When I took control of my AA in 2000, I adopted the Life-Cycle plan from Random Walk Down Wall Street as my AA schedule. It also represents a good plan, in my opinion. And I don't think the L Funds were even available back in 2000, and I did not know at that time that the G Fund had special characteristics that made it different from a typical MM or bond fund.

Now, I would have to change from that plan and the use an L fund allocation as my overall allocation. But that would be an option. Just a matter of which hoops I jump through.
User avatar
DaveTH
Posts: 2676
Joined: Thu Apr 05, 2007 7:23 pm

Post by DaveTH »

It's not clear what problem you are trying to solve. It's great to want to follow some model portfolio, but obviously you need to consider the funds and accounts that are available to you. Good luck with your investing.
dbr
Posts: 46181
Joined: Sun Mar 04, 2007 8:50 am

Post by dbr »

Whether or not it is helpful I don't know, but an option sometimes mentioned here for a person wanting to hold REITs or TIPS who does not have a tax preferred location for same, is to take out a Vanguard variable annuity.
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

Helot wrote: The "efficient" frontier in regard to risk/return can only be known after the fact as future/expected returns are unpredictable.
I just mean the typical estimation procedure for the efficient frontier. They calculated an efficient frontier to get the L fund allocations, after all. The future is estimated from the past. (Expect for those people you say we are in an oil bubble, they must have time machines :-)).

A simple way to mimic an allocation proposed by Malkiel using your G fund would be to consider it half of your fixed income allocation with something similar to the F Fund (total bond) representing the other half. There are a number of conversations on this board specific to how to use the G Fund in creating a fixed income allocation.
My current allocation came from a similar dark orafice. I was just wondering about the allocation that would come from that orafice that had a Nobel prize stuffed down it in 1990.

Just replacing some bond with G, as you suggest, might just decrease return while increasing risks. That would be my guess.
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

DaveTH wrote:It's not clear what problem you are trying to solve. It's great to want to follow some model portfolio, but obviously you need to consider the funds and accounts that are available to you. Good luck with your investing.
It's not a matter of available funds.

Do you know what the efficient frontier is?
User avatar
DaveTH
Posts: 2676
Joined: Thu Apr 05, 2007 7:23 pm

Post by DaveTH »

Do you know what the efficient frontier is?
Yes. Do you?
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

DaveTH wrote:
Do you know what the efficient frontier is?
Yes. Do you?
Yes. Good that you do too.

Now, The problem I am trying to solve is "What is the efficient frontier for the group of assets that include all the TSP offerings plus the Vg Reit Index?"

I thought maybe someone could calculate that quickly. Or, at least, point me to a way to do it.

That's what I want help doing.
User avatar
DaveTH
Posts: 2676
Joined: Thu Apr 05, 2007 7:23 pm

Post by DaveTH »

The tone of your responses don't really encourage anyone to want to help you any further.
User avatar
ken250
Posts: 1894
Joined: Mon Feb 26, 2007 7:50 pm
Location: US-101

Post by ken250 »

....
Last edited by ken250 on Sat Jul 05, 2008 10:35 pm, edited 1 time in total.
dbr
Posts: 46181
Joined: Sun Mar 04, 2007 8:50 am

Post by dbr »

User avatar
White Coat Investor
Posts: 17409
Joined: Fri Mar 02, 2007 8:11 pm
Location: Greatest Snow On Earth

Post by White Coat Investor »

Helot wrote:I'm not sure what it is you are asking.

There is nothing inherently more efficient regarding the L Fund allocations than an allocation you could create on your own using your various accounts. Such an allocation could model itself on various proposed allocations: Malkiel, Swensen, Vanguard Target Retirement, etc..

The "efficient" frontier in regard to risk/return can only be known after the fact as future/expected returns are unpredictable.
Ditto. The OP doesn't seem to get that there is no way to "get on the efficient frontier" in a prospective manner. You choose your AA, and you hope you're close to the efficient frontier.

As far as the G fund, I would use it in term of a ST treasury fund or another "nominal bond" portion of your allocation.
1) Invest you must 2) Time is your friend 3) Impulse is your enemy | 4) Basic arithmetic works 5) Stick to simplicity 6) Stay the course
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

EmergDoc wrote:
Helot wrote:I'm not sure what it is you are asking.

There is nothing inherently more efficient regarding the L Fund allocations than an allocation you could create on your own using your various accounts. Such an allocation could model itself on various proposed allocations: Malkiel, Swensen, Vanguard Target Retirement, etc..

The "efficient" frontier in regard to risk/return can only be known after the fact as future/expected returns are unpredictable.
Ditto. The OP doesn't seem to get that there is no way to "get on the efficient frontier" in a prospective manner. You choose your AA, and you hope you're close to the efficient frontier.

As far as the G fund, I would use it in term of a ST treasury fund or another "nominal bond" portion of your allocation.
The OP directs you to the information sheet for the L Funds, where they present what they call the efficient frontier.

Many AAs are based on a calculation of the efficient frontiers from past data. Of course they are just estimators relative to the future as you say.
Helot
Posts: 264
Joined: Sat Mar 03, 2007 2:05 pm

Post by Helot »

Well, if you want an allocation that is on the "efficient frontier" as defined by the TSP, then you need to replicate a simulation ala Financial Engines that utilizes the same inputs regarding past risk/return characteristics of the asset classes and future expectations regarding economic factors.

Per the info provided by the TSP, "The asset allocations are based on Mercer’s assumptions regarding future investment returns, inflation, economic growth, and interest rates."

Without that past data and future expectations used by Mercer, there is no way to replicate an L Fund that incorporates a REIT fund.

Taking a step back, this is why many target date funds suggest different allocations (due to the use of different data and expectations).

One idea may be to access Financial Engines. Apparently the service is free to TSP participants. Since this program, I assume, incoportates G Fund data as well as REIT data, you can create your "efficient frontier" portfolio and also have the stamp of approval of Mr. Sharpe, Prize in Economic Sciences in Memory of Alfred Nobel, 1990.
avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Post by avalpert »

I've always found false precision to be more ruinous than imprecise truth - the variance will be greater than the allocation itself so don't sweat it and approximate.
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

avalpert wrote:I've always found false precision to be more ruinous than imprecise truth - the variance will be greater than the allocation itself so don't sweat it and approximate.
Approximate based on what?
avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Post by avalpert »

tadamsmar wrote:
avalpert wrote:I've always found false precision to be more ruinous than imprecise truth - the variance will be greater than the allocation itself so don't sweat it and approximate.
Approximate based on what?
TSP G seems similair to long term treasuries - just displace whatever allocation you have towards that with G fund and keep whatever percent the portfolio you are tracking allocates to REITs in the vanguard fund
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

avalpert wrote:
tadamsmar wrote:
avalpert wrote:I've always found false precision to be more ruinous than imprecise truth - the variance will be greater than the allocation itself so don't sweat it and approximate.
Approximate based on what?
TSP G seems similair to long term treasuries - just displace whatever allocation you have towards that with G fund and keep whatever percent the portfolio you are tracking allocates to REITs in the vanguard fund
Without even bothering to find out how good that approximation is?
avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Post by avalpert »

tadamsmar wrote:
avalpert wrote:
tadamsmar wrote:
avalpert wrote:I've always found false precision to be more ruinous than imprecise truth - the variance will be greater than the allocation itself so don't sweat it and approximate.
Approximate based on what?
TSP G seems similair to long term treasuries - just displace whatever allocation you have towards that with G fund and keep whatever percent the portfolio you are tracking allocates to REITs in the vanguard fund
Without even bothering to find out how good that approximation is?
I'm willing to bet it will be within the margin of error of the models - but if you want to start data mining Financial Engines yourself
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

avalpert wrote:
tadamsmar wrote:
avalpert wrote:
tadamsmar wrote:
avalpert wrote:I've always found false precision to be more ruinous than imprecise truth - the variance will be greater than the allocation itself so don't sweat it and approximate.
Approximate based on what?
TSP G seems similair to long term treasuries - just displace whatever allocation you have towards that with G fund and keep whatever percent the portfolio you are tracking allocates to REITs in the vanguard fund
Without even bothering to find out how good that approximation is?
I'm willing to bet it will be within the margin of error of the models - but if you want to start data mining Financial Engines yourself
So, you look at the yearly returns of the F and G fund here:

Year G F
1998 5.74 8.70
1999 5.99 -0.85
2000 6.42 11.67
2001 5.39 8.61
2002 5.00 10.27
2003 4.11 4.11
2004 4.30 4.30
2005 4.49 2.40
2006 4.93 4.40
2007 4.87 7.09

http://www.tsp.gov/rates/monthly-history.html

And you think they are similar enough so that one can be substituted for the other in an AA?
avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Post by avalpert »

tadamsmar wrote:
avalpert wrote:
tadamsmar wrote:
avalpert wrote:
tadamsmar wrote: Approximate based on what?
TSP G seems similair to long term treasuries - just displace whatever allocation you have towards that with G fund and keep whatever percent the portfolio you are tracking allocates to REITs in the vanguard fund
Without even bothering to find out how good that approximation is?
I'm willing to bet it will be within the margin of error of the models - but if you want to start data mining Financial Engines yourself
So, you look at the yearly returns of the F and G fund here:

Year G F
1998 5.74 8.70
1999 5.99 -0.85
2000 6.42 11.67
2001 5.39 8.61
2002 5.00 10.27
2003 4.11 4.11
2004 4.30 4.30
2005 4.49 2.40
2006 4.93 4.40
2007 4.87 7.09

http://www.tsp.gov/rates/monthly-history.html

And you think they are similar enough so that one can be substituted for the other in an AA?
Well if we are being accurate the F fund is a broader Bond index which includes large pieces of mortages, industrial etc. Like I said above, the G fund behaves more like a safer Treasury fund.
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

avalpert wrote:
tadamsmar wrote:
avalpert wrote:
tadamsmar wrote:
avalpert wrote: TSP G seems similair to long term treasuries - just displace whatever allocation you have towards that with G fund and keep whatever percent the portfolio you are tracking allocates to REITs in the vanguard fund
Without even bothering to find out how good that approximation is?
I'm willing to bet it will be within the margin of error of the models - but if you want to start data mining Financial Engines yourself
So, you look at the yearly returns of the F and G fund here:

Year G F
1998 5.74 8.70
1999 5.99 -0.85
2000 6.42 11.67
2001 5.39 8.61
2002 5.00 10.27
2003 4.11 4.11
2004 4.30 4.30
2005 4.49 2.40
2006 4.93 4.40
2007 4.87 7.09

http://www.tsp.gov/rates/monthly-history.html

And you think they are similar enough so that one can be substituted for the other in an AA?
Well if we are being accurate the F fund is a broader Bond index which includes large pieces of mortages, industrial etc. Like I said above, the G fund behaves more like a safer Treasury fund.
You said:

"TSP G seems similair to long term treasuries"

Did you change your mind?
avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Post by avalpert »

tadamsmar wrote:
You said:

"TSP G seems similair to long term treasuries"

Did you change your mind?
No, I didn't - but Treasuries are not the same as Mortgages, Industiral and Agency debt or do you not understand the difference?
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

avalpert wrote:
tadamsmar wrote:
You said:

"TSP G seems similair to long term treasuries"

Did you change your mind?
No, I didn't - but Treasuries are not the same as Mortgages, Industiral and Agency debt or do you not understand the difference?
Granted they don't behave precisely the same. But closer than the G I think.

Both F and long treasuries have interest rate risk to the principle. G has none. I was figuring F and long treasuries would be pretty similar in their behavior over that 10 year period, particularly during the volatile period.

But I have not looked up the year by year returns for treasuries.
Helot
Posts: 264
Joined: Sat Mar 03, 2007 2:05 pm

Post by Helot »

I'm still a bit confused ...

The G Fund can be used as a substitute for long term treasuries on the return side if not the risk side. Per the TSP fact sheet on G Fund: "The G Fund offers the opportunity to earn rates of interest similar to those of long-term Government securities but without any risk of loss of principal and very little volatility of earnings."

The F Fund, or Total Bond Fund, holds approximately 38% in Treasuries, the rest in other, as Avalpert has suggested.

So, no, the G Fund and the F Fund are not substitutes.

If your question is how to use the G Fund, the default answer is to use it as the L Funds use it. The 2040 fund has it in about a 1:1 ratio with the F Fund, the Income fund with about an 12:1 ratio.

So, simply create a fixed income allocation that matches the appropriate L Fund.

Now, regarding the use of REITs. The simplest option is to create your REIT allocation from your Large Cap US Equity and Developed Markets exposure, since these three asset classes share similar return expectations. How much REIT exposure you want will be somewhere between the market exposure you receive from investing in the C & S funds (hence no need for REIT exposure) to an allocation that attempts to mimic the commerical real estate market that is not held by publicly traded REITs.

I believe Rick Ferri has posted on this topic regarding REIT allocations during the Core Four discussion:
Rick Ferri wrote:
Taylor Larimore wrote:Hi Sunny:
I am especially interested in this problem because this is my personal justification to exclude both SV and REITs from my portfolio.
Just a reminder: You are already holding the market's weight of these two funds in your portfolio.

Taylor
Commercial real estate is a large part of the US economy, but little of it is securitized in REITs. The amount of real estate available to public investors is limited to the financing decisions of companies involved in that industry.

If we want our portfolio to reflect more of the real US "economy" and less the financing decisions of corporate investment boards, we have to overweight REITs in proportion to their weighting in the public markets. You can create a portfolio the mirrors "the market" or "the ecomomy" or a little of both. That decision is up to you.

As Sunny mentions, over-weighting REITs give a US equity portfolio tracking error against the total stock market. Including REITs creates "investment policy risk" for people who do not want tracking error from the shifting correlation between REITs and common stock. If there is tracking error, those investors are prone to switch asset allocations when REITs detract from return, such as in 2007.

Rick Ferri

"Markets may not be efficient, but they are far more efficient than most investors."
Last edited by Helot on Thu Jul 03, 2008 7:34 pm, edited 1 time in total.
User avatar
dave.d
Posts: 936
Joined: Mon Mar 19, 2007 10:30 pm
Location: Richmond, VA

Post by dave.d »

Year by year returns:

G, F, Vg LT Treas
1998 5.74 8.70 13.05
1999 5.99 -0.85 -8.66
2000 6.42 11.67 19.72
2001 5.39 8.61 4.31
2002 5.00 10.27 16.67
2003 4.11 4.11 2.68
2004 4.30 4.30 7.12
2005 4.49 2.40 6.61
2006 4.93 4.40 1.74
2007 4.87 7.09 9.24

So I don't think the G fund is much like LT treasuries at all. It has risk similar to a cash fund (that it, basically none) but a higher expected return (in the amount of the term premium that is offered by long-term treasuries as compensation for the duration risk). In other words, it's a free lunch because there's no risk of say, 1999.

I would treat it as short-term bonds for allocation purposes. If you were trying to set an efficient frontier in a Mean Variance Optimizer program, you would set the expected return on the G fund to the current yield on long-term treasuries, and the standard deviation probably to zero. The result should be a larger allocation to G than you would normally make to short-term bonds.
Value-based allocation.
User avatar
grabiner
Advisory Board
Posts: 35307
Joined: Tue Feb 20, 2007 10:58 pm
Location: Columbia, MD

Efficient frontier depends on the model

Post by grabiner »

tadamsmar wrote:I would like to have an AA that:

1. Is on the efficient frontier

and:

2. Assumes access to the TSP G Fund

The only AAs available that meet these requirements, that I know of are the TSP L Funds:

http://www.tsp.gov/rates/fundsheet-lfunds.pdf
The efficient frontier for a set of investments depends on the model used for their returns and correlations. The L funds are based on one model for estimating the efficient frontier, and this may or may not be a good model. Many people on this board who invest in the TSP avoid the F fund because they believe that the F fund is not even on the efficient frontier of a TSP portfolio; the F fund gets higher returns than the G fund by taking credit and interest-rate risks which are correlated with the stock-market risk. (The F fund isn't a bad fund, as it is equivalent to Total Bond Market with lower expenses; it's just that the G fund is an option.) However, the TSP can't realistically say that an optimal portfolio should omit the F fund entirely.

Realistically, since you don't know the efficient frontier, any reasonable allocation should be about as good as any other, provided that you stick to it.
Wiki David Grabiner
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

dave.d wrote:Year by year returns:

G, F, Vg LT Treas
1998 5.74 8.70 13.05
1999 5.99 -0.85 -8.66
2000 6.42 11.67 19.72
2001 5.39 8.61 4.31
2002 5.00 10.27 16.67
2003 4.11 4.11 2.68
2004 4.30 4.30 7.12
2005 4.49 2.40 6.61
2006 4.93 4.40 1.74
2007 4.87 7.09 9.24

So I don't think the G fund is much like LT treasuries at all. It has risk similar to a cash fund (that it, basically none) but a higher expected return (in the amount of the term premium that is offered by long-term treasuries as compensation for the duration risk). In other words, it's a free lunch because there's no risk of say, 1999.

I would treat it as short-term bonds for allocation purposes. If you were trying to set an efficient frontier in a Mean Variance Optimizer program, you would set the expected return on the G fund to the current yield on long-term treasuries, and the standard deviation probably to zero. The result should be a larger allocation to G than you would normally make to short-term bonds.
That's my thinking.

Of course, I think you can just use the G Fund data directly in MVO.
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

If your question is how to use the G Fund, the default answer is to use it as the L Funds use it.
May I remind you that the way the L Fund uses the G Fund is by calculating the efficient frontier? Now may I refer you back to the OP? And, may I refer you back to the title of the thread:

Asset Allocation on Eff. Frontier including TSP G Fund

:?:
Helot
Posts: 264
Joined: Sat Mar 03, 2007 2:05 pm

Post by Helot »

tadamsmar wrote:
If your question is how to use the G Fund, the default answer is to use it as the L Funds use it.
May I remind you that the way the L Fund uses the G Fund is by calculating the efficient frontier? Now may I refer you back to the OP? And, may I refer you back to the title of the thread:

Asset Allocation on Eff. Frontier including TSP G Fund

:?:
Precisely. You've answered your own non-question. You ask how do I achieve a portfolio on the efficient frontier that incorporates the G Fund. The answer, mimic the L Funds that claim to be on the efficient frontier.

Your question in the OP really relates to the role of REITS in your allocation. Therefore, the title of your post is misleading.

It seems your real question is: How would the proposed L Fund allocations change if the TSP introduced a REIT fund?

If you really desire to know the TSP/Mercer's answer to this question, you should enlist in the lobbying effort to include a REIT fund in the TSP. I believe this was proposed just prior to the current real estate fizzle.

If you want to know how Malkiel's allocations (apparently your preferred "efficient frontier") would change with the addition of a G Fund, I suggest you send him an email and share his response.

I'd be curious to know how many of the experts we rely on here would utilize the G Fund. In fact, perhaps this is a question that should be asked at the Diehards Reunion.
Helot
Posts: 264
Joined: Sat Mar 03, 2007 2:05 pm

Post by Helot »

Here's a quote from a discussion on the TSPTalk site on the issue I just googled:
Financialengines.com, Nobel Laureate William Sharpe's website, recommends 100% G - no F. Incidentally, financialengines is free to government employees trying to set up a TSP asset allocation. In addition, financial author Rick Ferri recommends 100% G. He feels it's better to increase equity risk rather than fixed income risk to achieve higher returns. I don't agree with his argument, but....

Ask your question on the Bogleheads website, http://www.diehards.org/forum/index.php, and you'll get lots of responses. However, 100% G, 100% F, or a mixture are all very defensible. 100% G probably won't hurt your long term returns too much.

Finally, I'm assuming that you're a buy and hold investor. If you're a market timer, ignore everything I've just advised.

Good luck.-----Jim
Last edited by Helot on Thu Jul 03, 2008 8:59 pm, edited 1 time in total.
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

Helot wrote:
tadamsmar wrote:
If your question is how to use the G Fund, the default answer is to use it as the L Funds use it.
May I remind you that the way the L Fund uses the G Fund is by calculating the efficient frontier? Now may I refer you back to the OP? And, may I refer you back to the title of the thread:

Asset Allocation on Eff. Frontier including TSP G Fund

:?:
Precisely. You've answered your own non-question. You ask how do I achieve a portfolio on the efficient frontier that incorporates the G Fund. The answer, mimic the L Funds that claim to be on the efficient frontier.

Your question in the OP really relates to the role of REITS in your allocation. Therefore, the title of your post is misleading.

It seems your real question is: How would the proposed L Fund allocations change if the TSP introduced a REIT fund?

If you really desire to know the TSP/Mercer's answer to this question, you should enlist in the lobbying effort to include a REIT fund in the TSP. I believe this was proposed just prior to the current real estate fizzle.
Precisely!

You are right that I would only need REIT put in.

The title indicates that I was of two minds. I had just finished searching and reading back on recent posts on "TSP G FUND". Lots of people ask for help with allocations that have access to the G Fund, and they all seem to have their pet overweightings as do I.

As a general rule, they are told to put about half of what would otherwise be their bond allocation in the G fund. But is that good advice?

I guess if one did some efficient frontiers one might decide that that general rule is a good rule of thumb.
Wouldn't it be funny if that were true?


Tennesee Willliams, last line of Cat on a Hot Tin Roof
avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Post by avalpert »

tadamsmar wrote:
As a general rule, they are told to put about half of what would otherwise be their bond allocation in the G fund. But is that good advice?

I guess if one did some efficient frontiers one might decide that that general rule is a good rule of thumb.
All they need to do is adjust the assumptions until their allocation was on the effecient frontier - you seem to give a lot of weight to Mercer's (or anyones) models which are only as good as the assumption that go into them. If you like Mercer's model use the L fund, if you have your own pet leanings than it doesn't really matter how you devise the allocation - it's all outside of their model anyway.
User avatar
baw703916
Posts: 6681
Joined: Sun Apr 01, 2007 1:10 pm
Location: Seattle

Post by baw703916 »

One point that Rick Ferri made a while back is that the efficient frontier can only be known in retrospect. So you can never know that your portfolio, however you designed it, is actually one the efficient frontier.

A second issue is which point of the efficient frontier you want to be on. The efficient frontier is just a curved line of the maximum amount of return it's possible to get for a certain amount of year to year fluctuation of returns. How much return are you looking for, and how much risk are you able and willing to take?

One thing you should make sure you include in non-TSP investments, even more than REITs, is emerging markets. Currently, the TSP I fund is only EAFE, which means no emerging markets, and also no Canada. Hopefully at some point the I fund will be changed to follow an "all-world ex-US" index, but that hasn't happened yet. I think it is a pretty safe bet that the efficient frontier doesn't contain 0% emerging markets, so one should put in VWO (VEIEX) or something similar in the non-TSP accounts.

I think the makeup of the L funds is a reasonable place to start in coming up with an allocation across the TSP funds, although I think the equity allocation could use a little less C and a little more S and I.

Best wishes,
Brad
Most of my posts assume no behavioral errors.
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Re: Efficient frontier depends on the model

Post by tadamsmar »

grabiner wrote:Many people on this board who invest in the TSP avoid the F fund because they believe that the F fund is not even on the efficient frontier of a TSP portfolio; the F fund gets higher returns than the G fund by taking credit and interest-rate risks which are correlated with the stock-market risk. (The F fund isn't a bad fund, as it is equivalent to Total Bond Market with lower expenses; it's just that the G fund is an option.) However, the TSP can't realistically say that an optimal portfolio should omit the F fund entirely.

Realistically, since you don't know the efficient frontier, any reasonable allocation should be about as good as any other, provided that you stick to it.
So, let me follow your reasoning:

1. Realistically, I don't know the efficient frontier.

2. Many people on this board know the efficient frontier, or at least know that the F fund is not on the efficient frontier when the G Fund is available.

Did the many people do a calculation, by any chance? Or is it unrealistic for me to hope to know that?
User avatar
PiperWarrior
Posts: 4068
Joined: Fri Dec 21, 2007 3:55 am
Location: right on course

Post by PiperWarrior »

Would you like me to do the calculation?
User avatar
Topic Author
tadamsmar
Posts: 9972
Joined: Mon May 07, 2007 12:33 pm

Post by tadamsmar »

PiperWarrior wrote:Would you like me to do the calculation?
I would.

When I started this thread, I was hoping the first reply would be the calculation.
User avatar
baw703916
Posts: 6681
Joined: Sun Apr 01, 2007 1:10 pm
Location: Seattle

Post by baw703916 »

tadamsmar

I have done calculations for different combinations of the G, F, and C funds over the time the TSP has been in existence.

The caveat, of course, is that this only tells you what has happened historically, not what's going to happen in the future.

What the calculations show, is that for a fairly conservative portfolio (<50% equities), the G fund is much better to use on a risk-adjusted basis.

For a mostly equities portfolio, the F fund has been slightly better to use. Grabiner thinks that's a fluke of the particular time period, see our discussion in this thread. In any event, for say a portfolio of 70% equities, the difference between using 100% G, 100% F, or a 50%/50% combination of the two, has only been about 0.2% different in annual returns. So, if you are holding mostly equities, it probably won't make much difference which bond fund you use. For a mostly bond portfolio, use G.

Best wishes,
Brad
Most of my posts assume no behavioral errors.
joelesposito
Posts: 138
Joined: Fri Sep 21, 2007 1:30 pm

Post by joelesposito »

Dear original poster, As you can see no one on thsi borad can aswer your questions. This is for two reasons:
1. No one knows what the efficient frontier is, except in hindsight.
2. We are all too concerned ith semantics, bickering and showing off how much we know to answer your questions.

While I am victim to 1 & 2, let me offer this, as a fed empliyee that has wrestled with teh same questions:

1. to really answer the question you need to use an MVO and educate your self on its use. That's going to take time, no one will do that for you unfortunately (I want to know the answer too... with TIPS!).
2. but think about this: Why have REITs? I think the answer is to still get equity like returns in an inflationary cycle. I think most studies advocating them are based on what happened in the 70's. So the question is does G fit the bill? partially. G Fund returns are NOT equity like. howveer an often overlooked behavior of G is. Principle is stable but YIELD tracks treasury. What woudl happen if inflation went way up (assuming no simultaneous flight to qualoty). Nominal treasury yeilds go way up (because principle drops like a rock) to maitain a real return. So does G fund yield, but with stable principle. So in that way G fund gives good inflation protection (TIPs like but with no flight to quality benefit).

My answer would be, if you add REIT to slightly decrease your fixed percentage, and slightly decrease G fund (relative to F). because your getting some of the inflation protection with equity like returns. i say slight (5%) though because your probably incraeing your volitility too.

I want to know eff front with Gfund, TIPS REIT Commonidty, plus value stocks.
User avatar
PiperWarrior
Posts: 4068
Joined: Fri Dec 21, 2007 3:55 am
Location: right on course

Post by PiperWarrior »

Then please post a CSV (comma separate values) file containing the returns of funds you would like to consider in the following format.

a1,b1,c1 . .
a2,b2,c2
a3,b3,c3
:
:

where a1 a2 a3 are the annual returns of Fund A, b1 b2 b3 are the annual returns of Fund B, and so on. Of course, aN, bN, and cN must come from the same year for each N.

You can get the historical return of TSP funds from

http://www.tsp.gov/rates/monthly-history.html

including pre-1998 returns of TSP G.

You can get Vanguard funds' historical return from Historical Returns of Vanguard Funds.

For returns prior to the inception of Vanguard funds, you can look at Spreadsheet for backtesting (includes TrevH's data).

By coming up with the CSV file, you are implicitly specifying two things -- the time period to run the efficient frontier calculation and the universe of funds.
tadamsmar wrote:It's not a matter of available funds.
Why? An efficient frontier is calculated by cycling through all possible portfolios given some universe of funds and doing an XY plotting of return and standard deviation. Then you look at points lying on the northwest side of the plot. So, available funds do matter to define the universe.
tadamsmar wrote:All accounts are tax defferred (IRA, 401k, TSP) or Roth.

Over 1/4th of the total is in the TSP.

But, you don't think you really need this to answer my question,
Actually, the relative account size does matter to some extent because some porfolios are not implementable. For example, if the sum of REIT+SCV+TIPS+EM+commodity in one portfolio exceeds 75%, and that's on the efficient frontier, that doesn't do any good for you because you cannot implement such a portfolio. Note that TSP doesn't have REIT, small-cap value, TIPS, emerging markets, or commodity funds.
User avatar
retiredjg
Posts: 54082
Joined: Thu Jan 10, 2008 11:56 am

Post by retiredjg »

Tadamsmar, I love the TSP. I think the TSP folks have done a great job. If I only had exposure to the TSP, I'd probably use an L fund and be pretty happy.

There are, however, certain holes that need to be filled (in my opinion) if you have access to outside locations like a Roth or TIRA.

1) Emerging markets
2) TIPS
3) REIT if you want to overweight it
4) Small cap value if you want to lean that way

Emerging markets, not available in the TSP, is an absolute need in my opinion. TIPS, not available in the TSP, are a probable need in my opinion. REITs are a need only if you want to overweight. SCV is also a need only if you want to overweight.

I'm just wondering if you are putting your efficient frontier interest in the wrong location (REIT). Seems to me that TIPS, EM, and maybe small cap value might be much more important.

By the way, the AA I'm moving to (a la the portfolio goddess Laura) is G, C, and S with international including EM, Reit, and TIPS on the outside. I'm undecided on adding a bit of SCV. Either way, I think it's a good plan.

Even though I think you might be asking the wrong question, I hope you get an answer! jg
User avatar
baw703916
Posts: 6681
Joined: Sun Apr 01, 2007 1:10 pm
Location: Seattle

Post by baw703916 »

Here are my calculations, covering the period 1988-2006

Code: Select all

% Equities	                 C+G                           C+F                         C+F/G 50/50	  
	                    Std. Dev.      Return     	Std. Dev.      Return	     Std. Dev.      Return
     100               16.87          11.63          16.87          11.63          16.87          11.63
      90               15.21          11.23          15.32          11.33          15.26          11.28
      80               13.56          10.81          13.78          11.00          13.66          10.91
      70               11.91          10.36          12.28          10.65          12.07          10.51
      60               10.26           9.89          10.82          10.26          10.49          10.08
      50                8.61           9.39           9.43           9.85           8.94           9.63
      40                6.98           8.87           8.12           9.42           7.41           9.15
      30                5.36           8.31           6.97           8.95           5.95           8.65
      20                3.79           7.74           6.05           8.46           4.61           8.12
      10                2.35           7.13           5.47           7.95           3.52           7.57
       0                1.48           6.50           5.37           7.41           2.98           6.99
Brad
Most of my posts assume no behavioral errors.
Post Reply