WHY 4% COULD FAIL
Sep 1, 2015 • Wade Pfau & Wade Dokken
http://www.fa-mag.com/news/why-4--could-fail-22881.htmlOur research shows that Americans retiring in 2015 need to be far more conservative in their withdrawal rates during retirement. The historic 4% annual withdrawal rate is over two times the level that Americans can safely withdraw without expecting to outlive their assets. The real safe withdrawal rate, accounting for fees and today’s stock and bond market levels, is under 2% per year.------------------------------------------------------------------------------
Above is a link to a very interesting article that explores impediments to current implementation of a 4% (+ subsequent inflation adjustments) retirement withdrawal rate.
However, the version of the article in the Financial Advisor Magazine ("FA Mag") Sep 2015 issue is a little confusing for a couple of reasons.
First, the article as rendered (either in print or online) suffers from typos or mis-referenced figures that make it more difficult to follow than it should be.
Second, some key assumptions do not appear in the article but can only be found in the appendix of the related whitepaper. These relate to the mutual fund costs and financial advisor fees.
Specifically, a financial advisor fee of 100 bps, and average mutual fund expense ratios of 67 bps for stocks and 60 bps for bonds.
Below is the whitepaper link which appears at the beginning of the FA Mag article. NOTE: It asks for name and contact information.
Whitepaper contains assumptions in an Appendix and the whitepaper's figures are properly referenced and completely labelled. I suggest you read the Appendix first.
WHITEPAPERhttp://www.fa-mag.com/rethinking-retirement-wealthvest-0815And here - from Professor Pfau's website/blog is page with references and links to earlier papers on related topics :
WADE PFAU RETIREMENT RESEARCHER READING ROOM & BLOGhttp://retirementresearcher.com/reading/http://retirementresearcher.com/blog/-----------------------------------------------------------------
NOTE
Wade Pfau is Professor of Retirement Income at the
American College for Financial Services in Bryn Mawr, PA.
Wade Dokken is Co Founder & Co President of
WealthVest Marketing, a firm that designs, markets, and distributes fixed and fixed index annuities.
Comments
Once you are down to the 2% and lower SWR, is it not the case that you might as well put it all under the mattress and take out what you need till it's gone, given life expectancies? In other words, say you really can live on 20k, 2% of a million, plus some inflation, and are say 62 (trying to be worst-case or stupidest-case here). 20k a year, okay, inflated appropriately. How many years does 20k last divided into a million even if you allow for inflation? Is it less than 35? I suppose it must be. Okay, put it all into bonds.
All of which is an argument that 3.3% should be the floor for any 30 year strategy. Or 2.85% for your 35 years. (Draw 1/30th or 1/35th of principal with the rest getting 0% nominal plus inflation adjustments.)
The problem with this strategy is that there's no possibility of the portfolio lasting longer than 30 (or 35) years. Most investment strategies are designed to last at least N years, and usually longer. So you'd better depart on schedule or before.
I suppose if inflation ever got really bad again a mil drawn down $20k/y would disappear in less time.
I could dive into a calculator and attempt to figure out the decline table.
Right, ice floe at 97, if there are any ice floes then.
As for the 4% or whatever withdrawal rate, doesn't the size of one's nest egg count for something? Meaning, a debt free couple with a $3,000,000 nest egg who lives half way frugally could just live off their principal and not worry about the whims of the stock and bond markets. I realize I live in a low cost/income area of the U.S but in my region a single debt free retiree gets by just fine on $36,000 annually and a couple $42,000.
If one wanted to go pure cash and account for inflation, running simulations would seem to be the easiest(?) way to do that. But no matter what number of years you came up with, there would always be some small probability of inflation being worse than the "worst case" used. That's why I preferred to duck the inflation question entirely.
This question provides yet another example of how the rich get richer. If you've got money to burn, you invest more in stocks. Worst case, you've got enough to live on. On average, the expectation return is higher than the "mattress solution". This is not an option for people who can't take a 30% risk of having a shortfall.
I believe that they key thing here is the importance of keeping your investing costs low.
If you don’t, it is difficult to make things “work”.
As noted in the Appendix of the whitepaper mentioned in the original post, the authors assume investing costs of 160+ bps per year, with 100 bps for a financial advisor and 60+ bps for mutual funds.
It seems to me that these high costs, rather than any fancy math or simulations, accounts for the bulk of the reduction in the sustainable spending rate. This is more important, I think, than the peculiarities of the current market environment.
Note: Respectfully, I find it more than a little (!) disingenuous for the authors to “bury” this high cost assumption at the end of an Appendix to a whitepaper, since I don’t believe they reference the specific cost numbers in the original magazine article.
Or to put it another way, suppose the investor’s investors justify a 400 bps (4%) sustainable withdrawal rate, BUT they then have to pay 100 bps to a F/A and 60+ bps for their mutual funds.
Well, that would leave the investor with 400 - 160 bps to spend (before taxes!) of about 240ish bps per year, which is NOT that different from the figures derived using more complex calculations and assumptions.
-------------------------------------------------------
Too many assumptions to go into there. Monte Carlo and others are like many rule of thumb (e.g. 4% rule) estimators - good for generalities but not good for the specific situations.
Generally, a bottoms up approach is better i.e. budget, net worth, pension, SS etc.
This is my 2015 budget own home, no debt, single person
Basic Living
House
2,117 RE Tax
2,556 HOA
489 Electric
928 Insurance
300 Misc Purchases
133 Mail Box
6,522 Subtotal House
Car
138 AAA
744 Routine Mtc.
1,164 Insurance
82 Registration
1,800 Gas
3,929 Subtotal Car
Personal Expenses
327 Income Taxes
1,200 Cash
360 Medical
340 Cell Phone
3,300 Food
600 Wine
59 Misc
396 Internet Access
300 Dining Out/Entertainment
4,029 Health Ins.
300 Clothes
- Driving Lic
11,211 Subtotal Personal Expenses
21,661 Total Basic Living
Incremental Living - 1
91 Travel Trailer Reg
492 Storage
Good Sam
583
Incremental Living - 2
6,256 Travel/Education/Etc
Misc Hobbies
6,256
6,839 Total Discretionary
28,500 Total Basic + Incremental
Let's assume I don't have any pension or SS, and no inflation for now. What do I need?
$114,000 in near cash for 4 years of expenses - this is ride out market (bond & stock downturns.
$407,143 earning 7% to get to 28,500/year expenses
$100,000 to 150,000 contingency money, if wanted, earning ???
$621,143 to 671,143 total excluding house
Does a person need all that money? Maybe not if the person will collect SS. The closer they are to collecting SS would affect that - e.g. if they are within 2 years they could have less money in near cash.
This is not meant to be a perfect example.
Now let's use Junkster's info on SS $1294 monthly - 15,528/yr
$28,500 Total Basic + Incremental
-$15,528 SS
$12,972 to be funded
$51,888 in near cash for 4 years of expenses - this is ride out market (bond & stock downturns.
$185,143 earning 7% to get to 12,972/year expenses to be funded
$100,000 to 150,000 contingency money, if wanted, earning ???
$337,031 to 357,031 total excluding house
Both of these examples are better than monte carlo and top down rule of thumb.
There are two reasons I can think of that the top down method is the most discussed:
1. Advisors use them to scare people into buying their services
2. Budgeting is boring and most don't people don't have one nor do most know where they spend their money.
Thanks for note.
I understand that the math is not simple.
But I simply can't understand why the authors effectively buried their cost assumption - in the appendix of a whitepaper rather than include in the original article (or it was edited out of the article by the editors of F/A Mag).
IMHO, the authors understate the drag of expenses, in the article as it appears in the magazine. I think this is just as important as all the fancy math.
If you pay 160+ bps per year for investing expenses, things will be tough. To put it another way, assume that you can withdraw 400 bps per year, before expenses.
Well, you then need to hand over 40+% of your gross income (160+/400) to your F/A and the management of the funds that were picked for you.
ARRRGH! Talk about burying the lead....
The recent paper by Professors Pfau & Dokken on the erosion of the 4% retirement drawdown rule allows me to get out my old broken record that extols the virtues of Monte Carlo analysis.
I have done this so any times that MFO members must be tired of my pontifications on the subject. No matter, this is an important topic that should encourage those not mathematically inclined to use the Monte Carlo tool. Mathematical sophisticated is not required.
What is required is the persistence and the thought process to define some plausible what-if scenarios for the long term marketplace. You guys do this all the time when committing money to various financial products. The Monte Carlo tool will do the needed calculations. It is not necessary to look under the hood to comprehend the machinery. That only depends on your interest level.
For the purposes of exploring the various dimensions of this retirement withdrawal issue, I recommend the Monte Carlo code accessible on the Portfolio Visualizer website. Here is a direct Link to their version of a Monte Carlo simulator:
https://www.portfoliovisualizer.com/monte-carlo-simulation
I encourage you to visit this tool and play games with it. With just a little effort you can easily examine numerous scenarios that will permit you to develop a feel for which parameters are important and which are not. Have some fun.
The conclusions that Pfau & Dokken reached were basically preordained by the conservative assumptions that they made. The old 4% drawdown rule was doomed by their pessimistic assumptions. I actually agree with some of them, but not all. You must decide for yourself both the merits and shortcomings of them.
Understand that Pfau & Dokken postulated a 30 to 40 year portfolio survival requirement. They presumed a 95% portfolio survival criteria. You get to choose these based on your own situation. They assumed that the equity market is too high based on the current Shiller’s P/E10 ratio, and simultaneously projected an equity regression-to-the-mean.
Pfau & Dokken also assumed that long-term bond returns would be perturbed modestly from its present low rate of return. But they also hypothesized that annual inflation rates would average 3%. That’s lower than the historical average, but is not consistent with the postulated depressed long-term bond annual rate of return.
Since most retirement analyses make adjustments for the inflation rate, this is a critical paired set of inputs. I recommend you play what-if games with these parameters, especially the inflation rate. The Portfolio Visualizer tool allows this parameter to be easily changed. Just do it.
Pfau & Dokken are smart, experienced researchers. But recognize that their findings were predetermined by a set of conservative assumptions. It’s likely that you agree with some but take issue with others. The Portfolio Visualizer tool will allow you to measure the impact of this plethora of assumptions.
In the Simulation Model box of the code, I recommend you click to the “Parameterized Returns” which will permit you to input your estimate of annual “expected return” and “volatility” (standard deviation) for your portfolio. You can change these to explore how survival outcomes change. Also try different Inflation estimates to test their impact on portfolio survival rates. Enjoy.
I hope you find my post useful, but more importantly, I hope you visit and try the referenced Monte Carlo code.
Best Regards.
http://www.cfiresim.com/
Give it a shot.
Second, 4% or 2% or whatever SWR is not at all necessarily topdown, another puzzling claim. I can't imagine anyone not doing it bottomup, granularly, seeing what's what in detail and whether you can come close; and if not, well, then, you know where to start cutting. How I do it anyway. With regular xfers into checking.
Always glad to see it several thou ahead of upcoming bill totals. But then along come LTC premiums, or unexpected house repairs, and there go those few thou. Need better long-view budgeting and cashflow projections.
The 1.58% inflation rate is the 2013 annual inflation rate (for January - December 2013) as of January 1, 2014. Had the same Jan 1, 2014 date been used as the starting point for bond yields, the methodology would seem to have been okay. That rate was 2.86%. Instead, they chose to start the Treasury rates from a later date, Jan 1, 2015 (1.88%).
The 2.86% figure might have been considered a little high, as it was virtually the peak for 2013 (the actual monthly maximum was 2.90% on Dec 1, 2013). But the figure selected, 1.88%, was the lowest monthly figure since 2012. (Yields in 2013, 2014, and 2015 to date have all been at least as high as the Jan 1, 2014 yield.) It was not representative either.
So I don't believe they used historical average for inflation rates while perturbing (regressing) the current bond yield, which would be wrong, but it does appear that they fudged the starting yield on the bonds a bit.
It is quite possible that we interpreted the Rethinking Retirement document differently or even that we went to a different reference from among those posted by ibartman.
I pulled the 3% number from the “whitepaper” link in ibartman’s original post. That click accessed the 14 page Rethinking Retirement report
On page10 of that document, the authors said: “These initial spending rates are specifically calibrated to include a 3% annual cost-of-living adjustment (COLA), rather than having spending adjust precisely with the realized inflation experienced over retirement.”
In doing any forecasting analyses, I always postulate a positive risk return premium for both stocks and bonds over inflation. Reasonable approximations are 6.5% and 1.5%, respectively. But these are easily adjusted when doing Monte Carlo simulations to suit your preferences.
The primary purpose of my post was to reintroduce Monte Carlo simulators to the MFO population. Any input numbers that I suggest or that Pfau and Dokken actually used are nice as a generic guideline or a departure point, but do not necessarily reflect the specifics that each investor needs for his personal portfolio. That’s why I consistently recommend that each investor become familiar and comfortable enough with the Monte Carlo tool to do his own analyses.
That comfort level comes with practice. With sufficient practice comes confidence in the assembled portfolio and whatever withdrawal rate is being planned. Since plans are never perfectly realized, revisions will be needed as a function of time, so the Monte Carlo tool needs to be revisited.
The specific quoted numbers are not important; the tool and the process are the essential ingredients to exploring the robustness of any portfolio, and an acceptable withdrawal rate that might need adjustments over time.
Best Wishes.