Buckets and Bears

[clifp]

I had the same reaction last time this was paper was posted. Let see over long periods of time stock have higher returns than bonds.

So it doesn't really take an excel spreadsheet much less a PHd in finance to conclude that if you withdraw your bond portion first on average you'll end up with more money. Nor is very surprising that in almost all case your chances of 30 years survival are greater with higher stock allocation.

The trick is getting a real good crystal ball that will predict that the next 30 or 40 years will be a lot like the last 100....

 

[Nords]

The trick is getting a real good crystal ball that will predict that the next 30 or 40 years will be a lot like the last 100....

Yikes, I sure hope not.

 

[audreyh1]

Retiring into a bear market with a high allocation to stocks is a potential retirement killer

I handled it by taking almost 3 years to average in a lump sum into my asset allocation. Initial plan was 2 years, but as the bear market stretched out, so did my averaging in plan!

On top of that upon retiring I already had 2 years of expenses available in cash in a separate account so that I didn't need to withdraw from the portfolio for a while.

This was all a good thing, because I retired in late 1999!!!

It was tough to do that last investment in late 2002, but by late 2003, I was well ahead of the game!

Audrey

 

[twaddle]

The period beginning in 1966 makes the 2000-2003 bust look like a cake walk.

From 1966-1986, the S&P 500 was flat on an inflation-adjusted basis. 20 years with no real capital gains! The first part of the sequence was fairly boring and flat, punctuated by a crash in 1973-1974, followed by years of high inflation.

It hurts just thinking about it.

 

[ats5g]

Ah, nothing like bringing up the Retirement Calculator from Hell for the 1965-1995 period. It seems that the real determinant of portfolio success was the withdrawal rate, not necessarily the allocation to stocks/bonds. Even scarier [sp], is when stock and bond returns aren't that far apart, as in Part II.

- Alec

 

[cute fuzzy bunny]

BTW, I think Bunny has a great plan. Just retire twice as rich as you need to be, so you can live off 1/2 your stash without selling securities.

Oh Ha, come on. Thats not what I said and you're smarter than that.

Retirement is a balance of certainties and uncertainties. I've simply chosen a different balance of those. I dont have twice as much money, just less costs and less uncertainty.

I havent created a bunch of constructions that may produce a future uncertain higher rate of return and then made a bunch of certain constructions with poor rates of return to offset the future uncertain volatility and uncertain inflation levels.

Its just a matter of ascertaining ones spending needs going forward, minimizing unnecessary costs and risks and putting together a mix that meets that need.

Theres two ways to run a business.

1) Think up all the cool products and services you can conceive, then try to figure out how to sell those to people, spend a bunch of money in the process and hope your income meets the expenses and produces a profit. Cut expenses, headcount and projects if you dont. Do lots of projections, make lots of guesses, and build all sorts of layers of failsafes into the company.

2) Determine what kind of company you want to run, what sorts of needs customers have that you can solve, determine what people will pay for those products and services, produce them, sell them, keep the spending and company size in line with the expected income, and have a fairly predictable business.

The sort of planning most ER's engage in is #1..and thats how many businesses are run as well. Its not needs based and planned, its ad-hoc and improvised. Which is what most people are comfortable with. Until they start bitching about having to deal with all the surprises and shortfalls, or they fail because they built in too many failsafes that drained off too many resources.

ER Planning Process: How can I build an asset allocation that produces the most money with the best expected risk to reward factor, coupled with a bunch of personal implications like ones willingness to experience volatility and loss. Then when I get that worked out, how much money can I take out without it imploding. Now, can I live within that spending limit.

Problems come up when any expectations or uncertainties in the process crop up. A bear market. Lower future returns than historic. Unexpected inflation.

I did #2. Figured out how much I needed to spend, eliminated unnecessary expenses, broke my portfolio in half, crafted one half to meet the reduced current expenses with expected future cost increases over the next 20 years, and deployed the other half for maximum growth to service the second half+ of my retirement.

Most of those uncertainties no longer matter. Granted a severe long term bear market with extensive defaults and bankruptcies would screw me. But it'll screw everyone equally.

Hence not needing to outrun the bear, just the other participants...

I'd rather used fixed income to produce a steady predictable stream of funding than as a brake and parachute against all the other frightening uncertainties and failsafes I've invested in to try and maximize future earnings so as to meet my costs and debt payments.

 

[twaddle]

Ah, nothing like bringing up the Retirement Calculator from Hell for the 1965-1995 period. It seems that the real determinant of portfolio success was the withdrawal rate, not necessarily the allocation to stocks/bonds.

Actually, playing with Bob's simulator has caused me to reexamine my asset allocation. I was surprised by the historical robustness that a high exposure to small stocks provided -- not only in the 1965-1995 period, but in every 30-year period examined.

My current allocation is basically 50/50 stocks/bonds, with an 80/20 large/small mix and a strong value tilt.

If I move from 80/20 large/small to 50/50 large/small, the historic volatility stays very similar (it goes from a Std Dev of 9.2% to 9.4%), and FIREcalc gives a 4.59% SWR with 100% success.

I've been turned off by Small Value due to the recent historically high run-up for that asset class, but I think I may average into a higher exposure to it over the next few years....

I also like the idea of the bonds-first withdrawal strategy, but mostly for psychological reasons. By withdrawing from cash/bonds/cds in the first 10 years, it's simply easier to ignore my stock portfolio. I could figuratively bury it in the backyard for 10 years and completely ignore the volatility. Maybe.

 

[Nords]

Ah, nothing like bringing up the Retirement Calculator from Hell for the 1965-1995 period. It seems that the real determinant of portfolio success was the withdrawal rate, not necessarily the allocation to stocks/bonds. Even scarier [sp], is when stock and bond returns aren't that far apart, as in Part II.

- Alec

FAQ archives: http://www.early-retirement.org/forums/f47/william-bernstein-s-books-website-info-30505.html

 

[cute fuzzy bunny]

Hey Twad...make those changes you're talking about and you're pretty much at the same AA and strategy that i'm at.

Now what are the odds of that...?

 

[Shabber2]

Well twaddle I thought that was very interesting and will definitely take note as I was a big fan of rebalancing before looking at that!

 

[twaddle]

Hey Twad...make those changes you're talking about and you're pretty much at the same AA and strategy that i'm at.

Now what are the odds of that...?

Great. See you in the soup line if this strategy fails us.

 

[cute fuzzy bunny]

I'll give you a little extra. My promise to you.

 

[bpp3]

It'd be interesting to me to see how this strategy would have worked in Japan-1990. I'm not sure 50% bonds would have been enough to make it through that bear.

You rang?

50/50 Japanese stocks/Japanese bonds, starting 1 Jan 1990.
Expense ratio 0.5%
Yearly withdrawals: 4% in first year, indexed to inflation thereafter.
Starting portfolio: 1,000,000 yen.

[Removed bad numbers. See next post for fixed numbers]

But nowadays, we all know better than to go 100% domestic, right? So, same as above, except:

25/25/25/25 Japanese stocks/foreign stocks/Japanese bonds/foreign bonds, starting 1 Jan. 1990.

[Removed bad numbers. See next post for fixed numbers.]

Summary:
[Removed bad numbers.]

Bonds first seems a bit riskier overall, in this historical case.

 

[twaddle]

Bonds first: Hit 100% stocks by 1993, portfolio goes bust in 2004.

Can that be right? How can a 4% WR deplete the 50% allocation to bonds in just 3 years?

In any case, an ER bust after just 14 years would definitely be a record I wouldn't want to beat.

 

[bpp3]

Can that be right? How can a 4% WR deplete the 50% allocation to bonds in just 3 years?

Oops, found a mistake. Thanks for the sanity check.

If I have not made a mistake this time (no guarantee), then:

100% Japanese portfolio:
Rebalanced: 101,000 yen by end of 2006.
Bonds First: 12,000 yen by end of 2006, hit 100% stocks in 1999.

50/50 Japanese/foreign portfolio:
Rebalanced: 608,000 yen by end of 2006, minimum 560,000 yen hit in 2002.
Bonds First: 729,000 yen by end of 2006, minimum 487,000 yen hit in 2002, hit 100% stocks in 2002.

I think the conclusion remains, that Bonds First looks riskier.
(Another conclusion is that there are bigger issues to worry about, like diversification.)

 

[FUEGO]

Interesting article. At first I thought it was breaking major ground, maybe it still is. However I think the strength of the authors' conclusion is overstated.

The article makes it sound like the "take out bonds first" strategy is solely responsible for higher portfolio survivability over 30 years. For example, bonds first gives a 27% failure rate at 50% stocks and 5% withdrawal rate. Rebalancing gives 37% failure rate. But I would suggest this is an apples to oranges approach to analyzing failure. The portfolio under the bonds first strategy would be 80-100% stock for the great majority of the 30 year withdrawal period. I would say an apples to apples comparison should be the 40% or 50% or 60% stock allocation "bonds first" withdrawal strategy versus the 80% stock rebalancing strategy. In this manner, you get a more similar average % stocks over the course of the 30 years. When comparing the two withdrawal strategies on this basis, the bonds first withdrawal strategy looses its luster. At 50% stock, bonds first withdrawal at 5%, the failure rate is 27%, versus a failure rate of 22% with 80% stock, rebalance, 5% WR. (See attached diagram).

I also have questions about how adding in multiple asset classes (SCV, international, commodities, intl bonds, etc) would affect the survival percentages using bonds first versus annual rebalancing harvesting strategies.

The takeaway from the paper seems to be "higher allocation to higher returning asset classes = better portfolio survivability". And the data used in the paper (US stock/bond returns from 1926-2003) is fairly limited and maybe optimistic, given widely varying market returns seen in other countries throughout the same time period.

 

[twaddle]

As others have said, there isn't really any new ground being broken here. It's just another way to look at bonds and volatility.

Why do people want bonds in their portfolio? To reduce volatility, right? Why don't they like volatility? Because they don't want to experience drawdowns, right?

So, you have three choices as I see it:

1) Go with a high stock allocation and learn to live with the drawdowns.

2) Stick with your bond allocation, sleep well, but suffer in terms of both long-term portfolio survival and terminal value.

3) Start with a high allocation to bonds, consume them while you let your stocks simmer for 10 years or so in the hope that they grow, and then live with the same volatility in (1) but with a potentially much larger nest egg.

The simulator doesn't explore this, but once you hit that 100% stock allocation, maybe your portfolio has grown enough that you can buy another slug of bonds to help you sleep better. (Think of it as a 10-year rebalancing if you like.)

Also, the simulator does include small cap, and you're right -- other asset classes have historically improved things a bunch.

 

[ats5g]

Fuego,

IMHO, there doesn't appear to be that great a difference, or that much of an improvement, with rebalancing vs. bond first strategy. For example, a couple of quotes:

From Table 1:

The amount of information gathered is too voluminous to present, but for the particular (and arguably “most practical”) cases of (1) 4 percent withdrawals, with 50/50 stock/bonds and (2) 4 percent withdrawals with 60/40 stock/bonds, the shortfall frequency pattern by year for the Rebalance model and the Bonds First model were trivially different. Shortfalls did not occur earlier when there was no rebalancing. In both cases, the timing of the shortfalls was quite similar; the earliest shortfalls occurred at year 12, with the frequency of shortfalls increasing over time. The risks appear to be very much the same for these cases.

From Conclusions:

3. The probability of shortfall with either Rebalance or Bonds First is relatively low (10 percent or less) as long as withdrawals do not exceed 4 percent of the starting portfolio value and stock exposure is less than 70 percent (λ < 0.7.) Shortfall risk increases as the withdrawal rate, γ, increases. This is in line with findings of Cooley et al. (1999, 2003), Bengen (2004), and others.​

and

The conclusions on the efficacy of depleting bonds first (and hence not rebalancing) may seem risky. Suppose that an investor has a 50/50 stock/bond portfolio at the onset of retirement. If the bonds are withdrawn first, over a period of, say, 12 years, the stock component of the portfolio has had 12 years to grow undisturbed by withdrawals. Even though stocks are more volatile, this 12-year hiatus could have allowed a significant increase in the stock part of the portfolio. Both of the shortfall analyses demonstrate that one is no more likely to run out of money using this strategy than if one rebalances. While one can argue that the retiree’s portfolio will get more and more volatile over time as fewer and fewer bonds remain, the evidence does not suggest that shortfall is less likely with rebalancing. It then becomes a matter of how risk is viewed—as portfolio volatility (return variance) or as shortfall risk.

The PIBR analysis suggests that the Bonds First withdrawal strategy is likely to leave a larger portfolio at the end of 30 years than rebalancing would. A reasonable recommendation to a retiree would be to first harvest bonds and then harvest stocks. This strategy provides outcomes at least equivalent to rebalancing with respect to shortfalls and has the added benefit of a potentially larger estate. An advisor, however, still needs to be cognizant of the behavioral aspects associated with portfolio volatility.

I think the bootstrap analysis/simulation, which considers many more stock/bond return series than the temporal analysis/simulation, may be a little better when comparing rebalancing vs bond first. This is mainly because the temporal analysis/simulation could be labelled as data mining.

Overall, if you're 50/50 or 60/40 stocks/bonds and withdrawing 3-4%, this article is probably much ado about nothing.


- Alec

 

[unclemick]

current yield 3.99%(that's close to 4%) on Yahoo at 1 pm local time.

Pssst - Wellesley!



heh heh heh - And my sister tells me the Pats are a good football team.

 

[cute fuzzy bunny]

Yep, a few more years of seasoning and the pats might make the playoffs.

Wellesley adm shares...4.44% yield....shoot...forgot the 'pssst'!

 

[Nords]

Overall, if you're 50/50 or 60/40 stocks/bonds and withdrawing 3-4%, this article is probably much ado about nothing.

current yield 3.99%(that's close to 4%) on Yahoo at 1 pm local time.

Wellesley adm shares...4.44% yield...

So it's even less fuss if the principal remains intact.

 

[twaddle]

I thought the idea of a 4% SWR was to increase the withdrawal each year for inflation.

Yahoo tells me I couldn't do that with Wellesley. The dividend in 2006 was about the same as it was in 1989, for example.

Also, everybody knows you're not supposed to consume all of your income, right? The returns on stocks only look good if you reinvest dividends (and that's what this study assumes).

 

[cute fuzzy bunny]

Yahoo tells me I couldn't do that with Wellesley. The dividend in 2006 was about the same as it was in 1989, for example.

VWINX 1, 3, 5, 10 and 38 year (since inception) annualized returns are all more than 4% plus the CPI for those terms.

But there are plenty of reasons why it may not repeat that performance. Or why it will. Or better. I guess the future might throw some curveballs at us that the last ~40 years didnt.

 

[twaddle]

I have no idea what the future performance will be. I'm just saying that the fact that the current yield is 4% is meaningless in terms of SWR. Yahoo data only goes back to 1987, but the dividend history doesn't show any increase with inflation since that time. It goes up and down, and right now it's about the same as the 18-year average.

It would be interesting to see a SWR simulation based on consuming dividends, but so far every simulation I've seen assumes that dividends are reinvested. And everybody knows that half of the returns from stocks over the last 100 years or so have come from dividends, so reinvestment is a huge assumption to ignore.

 

[cute fuzzy bunny]

Now if you'd be so kind as to explain why I'd get a 4% yield on something, pay the taxes on that, reinvest it, then sell shares, and pay taxes on that?

Isnt it kinda nice to just get your 4% paid out, let the principal (and therefore the payout) adjust somewhat with inflation, and throw in a few lumps of faster moving stuff to assure maintaining principal levels against cost of living?

Isnt that the "black swan" idea you were so fascinated with a month or so ago? A big lump of slow moving assets with a few choice juicier picks made with smaller amounts?