SWR race to the bottom

[Focus]

Here are the latest and most depressing findings from Wade Pfau and friends, just out:

Asset Valuations and Safe Portfolio Withdrawal Rates

In a nutshell, they propose that from here on it's going to be different (and worse than you ever imagined). Accordingly, they write:

The issue is that bond yields today are well below and stock market valuations are well above their historical average. Importantly, there are no historical periods in the United States where comparable low bond yields and high equity valuations have occurred simultaneously.

The fun really starts on page 11 of the linked paper (most of which is over my head), with a table showing the probability of success for a 4% initial withdrawal rate over 30 years for different equity allocations, initial bond yields, and initial cyclically adjusted price-earnings ratio (CAPE) values. And it gets even more interesting on page 13, with a table estimating the safe initial withdrawal rates for various equity allocations and retirement periods.

Brace yourself, because the SWR numbers shown are very low (e.g., 1.4% SWR for 30 years, 99% probability of success with 60% equity allocation). Ouch.

 

[pb4uski]

This will give those with OMY syndrome comfort.

 

[brewer12345]

So the evil Dr. Pfau is at it again, eh?



If we really believed this, we would all be piling into annuities or perhaps 30 year TIPS. Anyone ready to do so? Bueller?

Didn't think so (aside from obgyn who would do so even if Dr. Pfau predicted a 10% SWR from an all equity portfolio). We are clearly in unusual times, but the US population has never been so old, so wealthy, or so open to international investment.

 

[bjorn2bwild]

Well, 15 pages is a lot to wade phrou, but I don't mind a little drama. Thanks for the link, I'll give it a look.

 

[Meadbh]

We are all doomed!

The most remarkable thing about the video was Wade Pfau's boyish appearance and squeaky voice. I'm sure he is highly intelligent......

 

[MooreBonds]

Brace yourself, because the SWR numbers shown are very low (e.g., 1.4% SWR for 30 years, 99% probability of success with 60% equity allocation). Ouch.

Hmm....current dividend yield on the S&P 500 ETF SPY (according to Yahoo!) is 2.03%.

Using his assumptions, I wonder if he would suggest an 80%-100% equity portion and living off of just the dividends to increase your projected SWR?

 

[brewer12345]

Hmm....current dividend yield on the S&P 500 ETF SPY (according to Yahoo!) is 2.03%.

Using his assumptions, I wonder if he would suggest an 80%-100% equity portion and living off of just the dividends to increase your projected SWR?

Offer him some grant money and he will look into it for you.

 

[FIREd]

I didn't read the whole paper, but scanning through it, my thoughts are:

1) 1.4% SWR for 30 years would require some seriously negative average real returns over that time span.

2) In Table 3, lower equity allocations yield higher withdrawal rates than higher equity allocations. For example, for a 30 year retirement and 95% success, a 20% equity allocation allows for a 2.8% WR while a 80% equity allocation allows for only a 2% WR. Seems counter-intuitive.

 

[kramer]

I didn't browse the paper yet, but I just bought a nice block of 30 Years TIPs at the last auction a couple of weeks ago and they went for 1.42% real. Theoretically, that gives you a 1.42% SWR for 30 years without touching inflation-adjusted principal.

The theoretical 30 year SWR for this bond, which would be quite difficult to do in practice for logistical reasons like reinvestment risk etc, would be 4.06%.

 

[Symbiotic]

It seems to me that there's really a diminishing return to all this model input refinement that Wade and others are going through.

It's just my opinion, but I believe that the "success percentage" quantification makes it too easy to draw the wrong conclusions from these kinds of analyses. This problem is closely coupled to the oversimplified spending models that most simulations provide.

As an alternative (or complement, perhaps) to speculating further about modeling the earning side, I think a more meaningful step forward would be to build a new tool that

(a) allows for parameterization of spending in a variety of more realistic ways,
(b) produces a broader set of outcome data, in both numeric and visual form, that are more tractable and intuitive, and
(c) automatically simulates a nice variety of A's and provides interesting ways to explore the differences in the results.

Maybe when I retire I'll get to work on this.

 

[ejman]

Let's face it. The end conclusion from all this tweaking will be that the only safe SWR is 0%. And then there is Rewahoo's asteroid.

 

[photoguy]

Here are the latest and most depressing findings from Wade Pfau and friends, just out:

I only skimmed through this paper but my first impression is that it is telling us a whole lot about what the author's think is a good "forecasting" model and maybe not so much about SWR rates.

 

[daylatedollarshort]

The thing is all these SWR forecasts from various "experts" change year to year based on the state of the current economy, when they are supposed to be predicting 30 years in advance.

If they had any degree of accuracy, a 30 year forecast for a SWR rate wouldn't change year to year based on one additional year's worth of economic data. How many financial experts predicted current interest rates of today 30 years ago accurately? I'm guessing close to zero or at most no more than would occur from random chance guessing.

As other have said anyone could just buy 30 year TIPS this week and lock in ~1.20% real, which would provide a 1.2% SWR without even dipping into any principal.

 

[48Fire]

Thanks for posting. I will take a better look. Sure glad I did not read this before I turned in my notice yesterday, and waivered on the decision. Whatever % I withdraw, I am looking forward to it.

 

[clifp]

Well if you decrease the expected returns of the US stock market by 2% to make it more in line with rest of the world and then further use CAPE to make the results even more dire you can end up with really low withdrawal rates.

If you have a 50/50 portfolio and Firecalc uses historical equity returns of X% and Dr. Pfau uses returns of X-2% logically you'll end up with a withdrawal rate roughly 1% lower than FIRECalc. I posted comment on his blog explaining my objection to his arbitrary decreasing of US equities returns.

But I think much of the results defy common sense. If you simply buy a 30 year ladder of TIPs bonds you'll get a real yield of ~.4% This means you have 100% SWR rate of 3.3% (spending principal)+.4%= 3.7%. The bad news is you have a 100% chance of being broke in year 31. But if you spend less than than 3.7% say 3.0% you can stretch your retirement.

When I look at my own portfolio its current yield was 2.7% at the beginning of the year, which is as low it has ever been. This is also the same as 90% SWR rate. However, my portfolio is designed to be an evergreen portfolio meaning I'll die with my principal intact. However for the typical retiree it is perfectly ok to spend a portion of their principal. Meaning for an ~80% equity portfolio 2.7% is darn near the floor of a SWR not the ceiling.

 

[marko]

So, for years and years this guy was talking about 3.5-4% SWR. "Don't worry, be happy...years-of-analysis, 100 years of historical data, yada yada yada".

Now, "ooops!"?

 

[Brett_Cameron]

So, they performed Monte Carlo simulations using current bond yields and projected equity yields derived from CAPE correlation. I am not surprised at the results.

Their ability to predict the future is not enhanced by this method.

 

[MichaelB]

So, they performed Monte Carlo simulations using current bond yields and projected equity yields derived from CAPE correlation. I am not surprised at the results.

Their ability to predict the future is not enhanced by this method.

+1

 

[REWahoo]

Their ability to predict the future is not enhanced by this method.

But their ability to gain media attention and sell books certainly is. You can be the judge on which is the primary motivating factor.

 

[obgyn65]

I tend to be very cautious by nature. But a 1.4% SWR does sound very low indeed. If these predictions come true, I guess some here may need to reconsider the use of SPIAs

Brace yourself, because the SWR numbers shown are very low (e.g., 1.4% SWR for 30 years, 99% probability of success with 60% equity allocation). Ouch.

 

[Which Roger]

A couple of thoughts (although I'll admit I didn't read the paper):
- If stocks will have lower expected/average returns than in the past, this probably means they are less risky than in the past. Less risk means less variability of returns from year-to-year. Less variability correlates to higher SWR since you are less likely to get a really bad sequence of returns for the first few years of ER.
- Part of the reason that bonds have very low yields is that inflation (and inflation expectations) are currently low. Lower inflation means lower increase in spending each each year for someone using the "initial withdrawal of X% adjusted for inflation each year" method.

 

[Brett_Cameron]

But their ability to gain media attention and sell books certainly is. You can be the judge on which is the primary motivating factor.

This is true.

 

[ERD50]

...

But I think much of the results defy common sense. If you simply buy a 30 year ladder of TIPs bonds you'll get a real yield of ~.4% This means you have 100% SWR rate of 3.3% (spending principal)+.4%= 3.7%. The bad news is you have a 100% chance of being broke in year 31. But if you spend less than than 3.7% say 3.0% you can stretch your retirement.

When I look at my own portfolio its current yield was 2.7% at the beginning of the year, which is as low it has ever been. ... Meaning for an ~80% equity portfolio 2.7% is darn near the floor of a SWR not the ceiling.

But that isn't a 2.7% real return, is it? I'm not sure your comparison is apples-apples to TIPS?

I tend to be very cautious by nature. But a 1.4% SWER does sound very low indeed. If these predictions come true, I guess some here may need to reconsider the use of SPIAs

Maybe my perception is wrong, but it seems to me that most of the posters here are very willing to consider/reconsider SPIAs. But most of us (again, my perception) seem to want to reduce the number of years we hold them, as most of are very cautious about predicting the health of the insurer 40 years out. And from what I've read, SPIAs are relatively expensive right now.

If the future economy is bleak, I doubt the SPIAs are going to look amazingly attractive in comparison, where will that money come from? They can't give away candy and remain solvent for the long haul. That is a balancing act (plus expenses). They don't have any magic mojo.

-ERD50

 

[youbet]

I didn't browse the paper yet, but I just bought a nice block of 30 Years TIPs at the last auction a couple of weeks ago and they went for 1.42% real. Theoretically, that gives you a 1.42% SWR for 30 years without touching inflation-adjusted principal.

The 1.42% SWR would only be possible if the gov't measured CPI reflected your personal rate of inflation. Or, in general, the personal rate of inflation for aging, retired folks. At our house, it wouldn't work. Our personal rate of inflation has been higher than the gov't published CPI figures the past several years. Food, travel, medical....... all bringing up our costs beyond the gov't figures.

 

[Gotadimple]

The thing is all these SWR forecasts from various "experts" change year to year based on the state of the current economy, when they are supposed to be predicting 30 years in advance..

+1
Exactly my thoughts - recency bias.

-- Rita