Another SWR Question?

[mb]

In a recent post Brewer, ESRBob, etc. discussed (1) the fact that a few "worst case" scenarios determined SWR and (2) asset classes that may perfom well during stagflation.

My questions is has anyone looked at the effect of a wide range of (hopefully non-correlated) asset classes on SWR?

In a FIRECalc run I got the following data for portfolio value at the end of a 50 year period (starting with $1M):

  $11M average value
  $14M standard deviation
  $68M maximum value
-$277k mimimum value (small in magnitude compared to other values)
  91% success rate

At the risk of being obvious, this demonstrates ESRBob's worst case comments.  Any ER should be happy with the average.  It is the low end of the distribution that causes the problem.  Furthermore the standard deviation is very large with respect to the average.  (Although a lot of the SD comes from the long tail on the high end not the low end of the distribution.  It is highly skewed.)

If you look at the year-by-year results it shows that most of the failures are clustered around the '30s (the Depression) and the 60-70s (Stagflation) which shows, again at the risk of being obvious, that the failures are caused by excessive withdrawals durings long periods of underperformance.  (Brewer was this the source of the stagflation question?) Unfortunately as ESRBob commented you don't know how long the underperformance will last until it is too late!

It seems to me that one possible way to address this problem and increase SWR is to try to decrease the variability in the end-of-life (a bit morbid, sorry) portfolio value, possibly at the expense the average value by holding a wide range of (hopefully non-correlated) asset classes.  i.e. Our old friend diversification.

For example I got the r2 (w/the SP500) and 10 year returns of several example funds from the Vanguard web site:

Fund                                           r2                 10-Year Return
SP500                                         1.0                     9.0
TRP Em Mrkt                               0.72                  10.1
TRP New Era (Nat Res)              0.52                   5.05
TRP Sml Cap Val                         0.72                  14.6
TRP Int Bond                              0.95                    5.3
Fid Infl Prot                                0.79                   6.8 (3 year return)
Fid Sel Gold                                0.08                   7.4
TRP Rel Estate                            0.18                  16.0

I believe that in the 60-70s stagflation era that a healthy dose of real estate, gold, and natural resources would have helped your retirement portfolio survive.  I think that they did well historically during that time and the above table shows that the r2 with the SP500 is low.

Also notice the r2 for the Int Term Bond fund.  It is highly correlated with the SP500.  To me this says that bonds help "diversify" the porfolio because they dampen the losses because of the low beta (they're not as volatile as stocks) but this also means, according to efficient market theory, that their long term return will be less) compared to stocks not because they are "zigging while stocks are zagging."  I think that perhaps a better way to diversify is to try to find something that tends to zig when stocks are zagging (low r2) but with a higher beta and hopefully a higher long term return.  (Disclosure:  My portfolio is about 60% stocks, mostly total market index, and 30% bonds so I can't claim to be practicing this.)

I guess that my conclusion to this ramble is that I think that this suggests that a good strategy for maximizing SWR is to diversify with a handfull of asset classes with low r2 with respect to the market and also with respect to each other.  For example nat'l resouces, real estate, emerging markets, gold, TIPS as well as stocks and conventional bonds.

Of course diversification of this sort is nothing new and I recall that based on other posts many of you already hold a wide array of asset classes.  Since I'm relatively new to the topic I'm sure that many others have thought along these lines but I don't recall seeing any studies on the effect on SWR?

I suspect a "history based" calculator like FIREcalc would be best for this sort of work because as SG stated in another post it would probably be difficult to get relationships for all of the asset classes for a Monte Carlo calculator.

MB     

 

[sgeeeee]

. . . My questions is has anyone looked at the effect of a wide range of (hopefully non-correlated) asset classes on SWR

It's tough to do this with a historical simulator because we don't have good performance data for very many asset types that date back to 1871. People have done this using monte carlo simulators. This is one of the major strengths of monte carlo. You might check out raddr's board. That's something he has experiemented with quite a bit. Several of the posters on that board are obsessed with finding the perfect asset allocation.

In a FIRECalc run I got the following data for portfolio value at the end of a 50 year period (starting with $1M):

$11M average value
$14M standard deviation
$68M maximum value
-$277k mimimum value (small in magnitude compared to other values)
91% success rate

Another weakness of historical simulators is that the reliability of the predictions is questionable as the length of retirement becomes long. One problem with a 50 year period is that the most recent period that can be analyzed started in 1951 and ended in 2002. The reason this is problematic is because one of the worst times in history to retire started in the mid-60's to early-70's. A 50 year simulation does not capture the importance of this period. It is more meaningful to run simulations of increasing length from 25 to 35 years and plot the SWR vs period. Then extrapolate your curve to longer periods. SWR assymptotically approaches a value of about 3.7% for long time periods.

It seems to me that one possible way to address this problem and increase SWR is to try to decrease the variability in the end-of-life (a bit morbid, sorry) portfolio value, possibly at the expense the average value by holding a wide range of (hopefully non-correlated) asset classes. i.e. Our old friend diversification.

Some people recommend investing some money in an annuity to reduce the variability for long retirement periods. There are disadvantages to this approach. Namely fees tend to be high and you lose the opportunity to pass the money on to heirs, but it does provide some stability for very long retirement periods.

I believe that in the 60-70s stagflation era that a healthy dose of real estate, gold, and natural resources would have helped your retirement portfolio survive. I think that they did well historically during that time and the above table shows that the r2 with the SP500 is low.

That's what asset allocators are looking for. There are some problems. People don't always agree even on what an asset class is. Over time, the definition of the asset class tends to evolve. And over a 50 year retirement, what are the chances that asset class correlations remain the same?

I guess that my conclusion to this ramble is that I think that this suggests that a good strategy for maximizing SWR is to diversify with a handfull of asset classes with low r2 with respect to the market and also with respect to each other. For example nat'l resouces, real estate, emerging markets, gold, TIPS as well as stocks and conventional bonds.

That makes sense. Again, I recommend you check out raddr's board. There are a number of asset allocation fanatics over there that are always looking for another uncorrelated asset class.

I suspect a "history based" calculator like FIREcalc would be best for this sort of work because as SG stated in another post it would probably be difficult to get relationships for all of the asset classes for a Monte Carlo calculator.

Actually, I think this is one place where monte carlo is better suited. You just can't find the historical record you need to do it with FIREcalc. The only problem with the monte carlo approach is that you have to define the performance distribution with a small amount of actual data. So you look at a few decades worth of performance and then assume that the asset class will perform similarly for many decades. There is no perfect crystal ball.

 

[amt]

Have you considered commodity futures?

Raddr has done a great analysis here:

http://raddr-pages.com/research/CommodityFutures.htm

Cheers,

 

[brewer12345]

I tend to look at the issue based on correlations, not r2.

That aside, 10 years of return data is not enough to really have the kind of long term view that an early retiree will be taking (at least at the start of retirement). So I am somewhat disinclined to place to much emphasis on such a short period.

I approach diversification based on an attempt to A) smooth out the humps and bumps without surrendering much potential return and B) insulate myself from some of the worst case scenarios thatwould otherwise impact my domestic equity-heavy portfolio. As such, I hold 8 to 10% of commodities, 8 to 10% of unhedged foreign bonds, 15 to 20% of foreign stock and about 5% CDs and cash. Most of the rest of my portfolio is small cap value individual stocks, although I have started tip-toeing into exchange-traded bonds and preferred stocks that appear to be trading cheap. The big difference i would make if I had to start living on withdrawals today would be to reduce the amount I am willing to hold of any individual stock, and to reduce my US equity position by about 15% and put the proceeds in high quality, limited duration bonds and cash.

 

[Cut-Throat]

Buying an asset class that zigs while stocks zag is only possible when done in hindsight. Berstein discusses this in his 4 pillars book when he laments 'Oh if it were only so easy!'

I think the best approach in these 'bad' times like the 30's and 70's, is to slash spending. That would seem to have the biggest impact on portfolio survival. I think most everyone here has 30% of possible cuts in their budget.

 

[unclemick]

Yep

My all time personal best - never to be repeated was 12k for one year.

Look at it this way - if I were still working - or got layed off during a period of stagflation - what would I do to slash spending. During 1966- 1982, the jobs I held didn't have automatic COLA adjustments.

We have more control over our spending - aka Mr Market doesn't care about individuals.

heh heh heh

 

[Mr._johngalt]

Yep

My all time personal best - never to be repeated was 12k for one year.


heh heh heh

We (couiple- 4 dogs) could do this today, but not under our present circumstances of course. 12K would be "possum living" in my
book. Not a pretty sight, but it could be done. How you ask?

Very small, paid off living quarters.
No debt.
Heat with wood.
One rustbucket vehicle. or maybe none at all
Hunt, fish, garden, raise small livestock.
One phone with no extras.
No cable or computer
Sell odds and ends for pocket money.
Control your income so that you pay no taxes and qualify for all
available government assistance (it's a lot).
No med. insurance. Use Medicaid/free services.
Dumpster dive.
Look for freebies and shop almost 100% at resale shops.

There are millions who live this way right now. Not all feel miserable and deprived. I'll bet a lot are perfectly content.

JG

 

[Mr._johngalt]

Buying an asset class that zigs while stocks zag is only possible when done in hindsight. Berstein discusses this in his 4 pillars book when he laments 'Oh if it were only so easy!'

I think the best approach in these 'bad' times like the 30's and 70's, is to slash spending. That would seem to have the biggest impact on portfolio survival. I think most everyone here has 30% of possible cuts in their budget.

Good post C-T. Still my favorite liberal.

JG

 

[Jarhead*]

I think most everyone here has 30% of possible cuts in their budget.

Cut: Green-Fees, Tournament fees, fly-fishing,
(The important things) are non-negotiable.
(The unimportant things), like basic shelter, food, etc. can always be adjusted.

Jarhead, who agrees with ReWahoo that Texas is the best college football team in the country.

 

[SteveR]

All this seems to reinforce the concept of having a stash of cash for just such market periods.  It would seem appropriate to have 3-5 years worth of cash (CDs, MM or I-bonds) at hand to cover spending during the down years to prevent chewing into principle.  That is my hedge anyway.  I have between 3-4 years of basic living expenses in various places that I can get to and not lose the principle while still earning a modest amount on the stash.  Having a small pension helps keep this amount a little smaller.  Post retirement medical insurance with shared premiums from my DW's employer will also keep the total amount needed in the stash a bit lower.  

I have some heavy investments in equities in my after tax portfolio.  It may take a beating (has already in the 1999-2001 market) in a down market.  I tend to me heavy in a couple of sectors that are sucking right now but that is another story.  

Save for a rainy day... works for me; especially when that rain may be a from a very stormy market.

 

[SteveR]

Cut:  Green-Fees, Tournament fees, fly-fishing,
(The important things) are non-negotiable.
(The unimportant things), like basic shelter, food, etc. can always be adjusted.

Jarhead, who agrees with ReWahoo that Texas is the best college football team in the country.

I second the thought on Texas. Go Horns!

 

[mb]

Thanks for the comments,

Strongly agree with this:

I think the best approach in these 'bad' times like the 30's and 70's, is to slash spending.

Having flexibility on the "spending side" of the equation is the best insurance.  If you have some flexibility that is where you can exercise the most control.

But that doesn't mean you shouldn't try to optimize the "earnings side."

Not sure I completely agree with this.  I certainly respect his opinion.  I'll have to read the book.  Haven't done that yet.

Buying an asset class that zigs while stocks zag is only possible when done in hindsight. Berstein discusses this in his 4 pillars book when he laments 'Oh if it were only so easy!'

I view this in a probabilistic sense.  I agree that you will never find anything that will go up 10% every time the SP500 goes down 10% but you can certainly smooth out the ride.  That's why we diversify?  Right?  Maybe it is a matter of degree?  The point of the post was to try to get some added insight on how best to do that with respect to SWR.

MB

 

[Eagle43]

Jarhead, who agrees with ReWahoo that Texas is the best college football team in the country.

Did I miss something? Did you lose a wager? With REWahoo?

 

[Jarhead*]

Did I miss something?  Did you lose a wager?  With REWahoo? 

Eagle: How about those Longhorns!


Jarhead, who agrees with ReWahoo, that Texas is the best college football team in the country.



1

 

[REWahoo]

Eagle: How about those Longhorns!