
http://online.wsj.com/article/SB1000142 ... =rss_Money
SORRY!
An "unexpected hazard? All of those old studies purported to be allowing for just such a contingency. And they were not "beautiful," they were at the time considered bad news. The conventional wisdom had been amortization calculations, and Peter Lynch claimed 7% for a 100%-stocks portfolio.In recent years, the 4% rule has been thrown into doubt, thanks to an unexpected hazard: the risk of a prolonged market rout the first two, or even three, years of your retirement. In other words, timing is everything. If your nest egg loses 25% of its value just as you start using it, the 4% may no longer hold, and the danger of running out of money increases.
An "unexpected hazard?nisiprius wrote:Yeah, but see, I object to the switcheroo implied by this statement:In recent years, the 4% rule has been thrown into doubt, thanks to an unexpected hazard: the risk of a prolonged market rout the first two, or even three, years of your retirement. In other words, timing is everything. If your nest egg loses 25% of its value just as you start using it, the 4% may no longer hold, and the danger of running out of money increases.
The S&P 500 is about where it was in 1999 in real terms. How hard is it to imagine that 14 years of zero real returns won't stretch to 30 or 40 years?ResNullius wrote:I read this article earlier today, and I thought there was one major flaw. The data on which the 4% SWR is based includes the 30 and 40 year periods following the start of the Great Depression. It's hard to imagine that current or anticipated future events, other than a total financial collapse around the world, would be worse than the 30 to 40 year period following 1929.
You nailed it once again Nisi. I guess they are expecting worse times than we had during the Great Depression.nisiprius wrote: 4% was supposed to allow for the stock market fluctuation--for the full range experienced historically, and/or in Monte Carlo simulations. And it was supposed to have enough of a safety margin to be a reasonable planning guide for thirty years into the future. And less than 15 years later they are saying "oops?"
Stocks have had zero real growth over more than half of the past 25 years. Treasuries are yielding something close to zero in real terms (depending on maturity). How hard is it to believe in 25 years of no growth?EternalOptimist wrote:To over simplify, with the 4% withdrawal, you are saying your assets would last 25 years--ie, 100%/4%= 25 years. This, of course, assumes no growth over 25 years which is hard for me to believe. We are all adults but I prefer to believe I will be just fine. Of course there are those who are more fear-driven who believe they won't
I don't pray at the alter of the 4% SWR, but I also get tired of reading stories by so-called financial experts who sit back and conjure up situations that bear no rational relationship to past or reasonably anticipated future events. It's very difficult to imagine that there will be a 30-year period that is worse than what followed 1929. To be worse, it would require the virtual end of the world as we know it. In order to have enough to be rich throughout such a period of time would require a starting balance of tens of millions of dollars, which is totally out of reach of 99.9% of society.1210sda wrote:You nailed it once again Nisi. I guess they are expecting worse times than we had during the Great Depression.nisiprius wrote: 4% was supposed to allow for the stock market fluctuation--for the full range experienced historically, and/or in Monte Carlo simulations. And it was supposed to have enough of a safety margin to be a reasonable planning guide for thirty years into the future. And less than 15 years later they are saying "oops?"
1210
Another interpretation is that relying on less than 100 years of data to predict 30 years into the future is not a very good technique. We have fewer than four independent 30 year data points. Would you roll a six-sided die four times, get 1, 3, 5 and 6 and declare the odds of rolling a 2 are minimal?1210sda wrote:You nailed it once again Nisi. I guess they are expecting worse times than we had during the Great Depression.nisiprius wrote: 4% was supposed to allow for the stock market fluctuation--for the full range experienced historically, and/or in Monte Carlo simulations. And it was supposed to have enough of a safety margin to be a reasonable planning guide for thirty years into the future. And less than 15 years later they are saying "oops?"
The 30 year period starting in 1966 is typically used for retirement income "Stress Test" purposes. The high inflation in the late 70's makes it a worse historical sequence than starting in 1929.ResNullius wrote: I don't pray at the alter of the 4% SWR, but I also get tired of reading stories by so-called financial experts who sit back and conjure up situations that bear no rational relationship to past or reasonably anticipated future events. It's very difficult to imagine that there will be a 30-year period that is worse than what followed 1929. To be worse, it would require the virtual end of the world as we know it. In order to have enough to be rich throughout such a period of time would require a starting balance of tens of millions of dollars, which is totally out of reach of 99.9% of society.
Indeed. Now throw into the pot the fact that the CRSP--the source of almost any data for which the starting date is 1926--was created in order to provide Merrill Lynch with numbers it could use in an advertisement. As with drug studies funded by drug companies, it would be churlish to suppose that the Chicago School of Business was in the bag; I have no doubt that the researchers saw it as an opportunity to do what they wanted to do anyway, and did it with skill and integrity. But it would also be idealistic to assume that there was no funding bias at all.richard wrote:...Another interpretation is that relying on less than 100 years of data to predict 30 years into the future is not a very good technique...
Except that, it turned out to almost exactly correspond to the returns calculated on the S&P 500 and it's predecessor S&P 90, or whatever it was. Given that there are no meaningful differences between TSM (CRSP 1-10) and the S&P 500, that's exactly what you'd expect. I think we can relegate the conspiracy theories to Roswell and Nessie!nisiprius wrote:Indeed. Now throw into the pot the fact that the CRSP--the source of almost any data for which the starting date is 1926--was created in order to provide Merrill Lynch with numbers it could use in an advertisement. As with drug studies funded by drug companies, it would be churlish to suppose that the Chicago School of Business was in the bag; I have no doubt that the researchers saw it as an opportunity to do what they wanted to do anyway, and did it with skill and integrity. But it would also be idealistic to assume that there was no funding bias at all.richard wrote:...Another interpretation is that relying on less than 100 years of data to predict 30 years into the future is not a very good technique...
This whole sustainable withdrawal hysteria has really reached a point of absurdity. Yes, LG stock returns have been paltry (TSM, TISM) since 2000 (but great from 1990-1999, and completely average from 1990-2012), and bonds are priced to yield 0% or less real returns going forward. But even assuming terrible timing and a retirement starting on January 1st of 2000, a $1M portfolio calling for $40K in year one and adjusting for future years inflation that is split 42% TSM, 18% TISM, 20% TBM, 20% TIPS (Russell, MSCI, Barclays Indexes) ended 2012 at $997,500. Has anyone actually stopped to check these things out, or are we just assuming portfolio implosion because of a visceral reaction to the extended lost decade?allocator wrote:"Say goodbye to the 4% rule". Those of you planning for, or are in, retirement (I'm guessing that's just about all of us) might find this interesting.
http://online.wsj.com/article/SB1000142 ... =rss_Money
SORRY!
And then I realized that, if pressed, I couldn't say for sure exactly what percentile outcome the 2000-2012 results would be in terms of a Monte Carlo simulation using historical risk and return figures. So I checked it out.Has anyone actually stopped to check these things out, or are we just assuming portfolio implosion because of a visceral reaction to the extended lost decade?
richard wrote:Stocks have had zero real growth over more than half of the past 25 years. Treasuries are yielding something close to zero in real terms (depending on maturity). How hard is it to believe in 25 years of no growth?EternalOptimist wrote:To over simplify, with the 4% withdrawal, you are saying your assets would last 25 years--ie, 100%/4%= 25 years. This, of course, assumes no growth over 25 years which is hard for me to believe. We are all adults but I prefer to believe I will be just fine. Of course there are those who are more fear-driven who believe they won't
I read what you said. You're talking in nominal terms, which doesn't seem an appropriate benchmark for inflation adjusted (i.e. real) withdrawals.EternalOptimist wrote:In 1988, S&P was 267 and today 1522...seems like it grew to me. Read what I said.richard wrote:Stocks have had zero real growth over more than half of the past 25 years. Treasuries are yielding something close to zero in real terms (depending on maturity). How hard is it to believe in 25 years of no growth?EternalOptimist wrote:To over simplify, with the 4% withdrawal, you are saying your assets would last 25 years--ie, 100%/4%= 25 years. This, of course, assumes no growth over 25 years which is hard for me to believe. We are all adults but I prefer to believe I will be just fine. Of course there are those who are more fear-driven who believe they won't
If nisi is correct that CRSP is "the source of almost any data for which the starting date is 1926", that would include S&P data, so you're just looking two versions of the same underlying data. It would not be surprising that two versions of the same underlying data agree.EDN wrote:Except that, it turned out to almost exactly correspond to the returns calculated on the S&P 500 and it's predecessor S&P 90, or whatever it was. Given that there are no meaningful differences between TSM (CRSP 1-10) and the S&P 500, that's exactly what you'd expect. I think we can relegate the conspiracy theories to Roswell and Nessie!nisiprius wrote:Indeed. Now throw into the pot the fact that the CRSP--the source of almost any data for which the starting date is 1926--was created in order to provide Merrill Lynch with numbers it could use in an advertisement. As with drug studies funded by drug companies, it would be churlish to suppose that the Chicago School of Business was in the bag; I have no doubt that the researchers saw it as an opportunity to do what they wanted to do anyway, and did it with skill and integrity. But it would also be idealistic to assume that there was no funding bias at all.richard wrote:...Another interpretation is that relying on less than 100 years of data to predict 30 years into the future is not a very good technique...![]()
If you earn 0% real returns and withdraw 4% adjusted for inflation each year, you run out of money after 25 years, which is a bit quicker than 30 years.EDN wrote:But let's go with this and say for the next 17 years (for a full 30 year retirement), the 60/40 portfolio above earns 0% real returns (significantly worse returns than we've seen from 2000-2012). Even with this extreme left tail outcome, and still continuing to pull out annual income, you still don't run out of money.
The problem is that there never was a solid basis for the 4% rule. It's just a general rule of thumb based on historical data that's not sufficient to be a reliable predictor of the future.FinancialDave wrote:Unfortunately, there was really nothing new in the article
Nisi is not correct. The S&P 90 was compiled and tracked weekly starting in the early 1920s, not backfilled. They also tracked another 400 stocks and their returns that weren't included in the index. In the 1950s, the S&P 90 was expanded to the S&P 500. CRSP attempted to compile a measure of the returns of the "total" stock market that included another thousand or two thousand stocks in the 1960s. They found....wait for it....the same returns as the S&P 500! Cap weighting is cap weighting whether you hold 500 or 5000 stocks. The conspiracy theorists who would posit that CRSP was created to make stocks look better than they were have it all wrong...they didn't turn out to do any better/worse than we already knew!richard wrote:If nisi is correct that CRSP is "the source of almost any data for which the starting date is 1926", that would include S&P data, so you're just looking two versions of the same underlying data. It would not be surprising that two versions of the same underlying data agree.EDN wrote:Except that, it turned out to almost exactly correspond to the returns calculated on the S&P 500 and it's predecessor S&P 90, or whatever it was. Given that there are no meaningful differences between TSM (CRSP 1-10) and the S&P 500, that's exactly what you'd expect. I think we can relegate the conspiracy theories to Roswell and Nessie!nisiprius wrote:Indeed. Now throw into the pot the fact that the CRSP--the source of almost any data for which the starting date is 1926--was created in order to provide Merrill Lynch with numbers it could use in an advertisement. As with drug studies funded by drug companies, it would be churlish to suppose that the Chicago School of Business was in the bag; I have no doubt that the researchers saw it as an opportunity to do what they wanted to do anyway, and did it with skill and integrity. But it would also be idealistic to assume that there was no funding bias at all.richard wrote:...Another interpretation is that relying on less than 100 years of data to predict 30 years into the future is not a very good technique...![]()
Well, that has almost nothing to do with my quote. I already accounted for the first 13 years of a 30 year retirement (2000-2012) with the beginning bias of a 2000 start date (not 1998, not 2003, etc.). And I showed that you'd been able to pull over $600K out, and still have about your original portfolio value. So clearly, real portfolio returns haven't even been 0% since 2000, and we began with a period where equity valuations were as high as they've ever been! 0% going forward, that is a left tail 1% probability, not the base for which you plan a retirement around.richard wrote:If you earn 0% real returns and withdraw 4% adjusted for inflation each year, you run out of money after 25 years, which is a bit quicker than 30 years.EDN wrote:But let's go with this and say for the next 17 years (for a full 30 year retirement), the 60/40 portfolio above earns 0% real returns (significantly worse returns than we've seen from 2000-2012). Even with this extreme left tail outcome, and still continuing to pull out annual income, you still don't run out of money.
Nothing's reliable enough for some. 85 years of US data doesn't have enough independence. A global equity index set that is within 1% real of the US result, still not enough. Monte Carlo tests run on thousands of simulations? Worthless.richard wrote:The problem is that there never was a solid basis for the 4% rule. It's just a general rule of thumb based on historical data that's not sufficient to be a reliable predictor of the future.FinancialDave wrote:Unfortunately, there was really nothing new in the article
Believe what you wishrichard wrote:I read what you said. You're talking in nominal terms, which doesn't seem an appropriate benchmark for inflation adjusted (i.e. real) withdrawals.EternalOptimist wrote:In 1988, S&P was 267 and today 1522...seems like it grew to me. Read what I said.richard wrote:Stocks have had zero real growth over more than half of the past 25 years. Treasuries are yielding something close to zero in real terms (depending on maturity). How hard is it to believe in 25 years of no growth?EternalOptimist wrote:To over simplify, with the 4% withdrawal, you are saying your assets would last 25 years--ie, 100%/4%= 25 years. This, of course, assumes no growth over 25 years which is hard for me to believe. We are all adults but I prefer to believe I will be just fine. Of course there are those who are more fear-driven who believe they won't
Read what I said. It hasn't grown in real terms (including dividends) in the past 13 years. 13 is more than half of 25.
Precisely. It showed what worked IN THE PAST and didn't guarantee that history would repeat itself.richard wrote:The problem is that there never was a solid basis for the 4% rule. It's just a general rule of thumb based on historical data that's not sufficient to be a reliable predictor of the future.FinancialDave wrote:Unfortunately, there was really nothing new in the article
Do you understand the objection? The problem is that there are only about three independent 30 year data points in 85 years of data, not that the entire data set isn't independent of something.EDN wrote:Nothing's reliable enough for some. 85 years of US data doesn't have enough independence.
So?EDN wrote: A global equity index set that is within 1% real of the US result, still not enough.
Monte Carlo simulations are entirely dependent on their inputs, no matter how many times they are run. If you knew future returns, distributions, correlations, etc., then MC would help in predicting the future. OTOH, if you had this information, you wouldn't need to run MC models.EDN wrote:Monte Carlo tests run on thousands of simulations? Worthless.
I agree that this is an oversimplification. Because of sequence of returns risk, you can come up with scenarios in which a portfolio has a time-weighted 0% real return over 25 years, but still runs out of money because its dollar-weighted return is significantly below zero.EternalOptimist wrote:To over simplify, with the 4% withdrawal, you are saying your assets would last 25 years--ie, 100%/4%= 25 years. This, of course, assumes no growth over 25 years which is hard for me to believe.
This statement is very interesting as a sort of litmus test. Personally, I agree with it wholeheartedly. But there are also people who would read it and think you're unreasonable for believing such a thing.richard wrote:MC simulations are mainly useful as a teaching tool - how might changing this number (returns, allocations, correlations, whatever) change outcomes. They don't add much to our ability to predict the future.
I would argue that they're pretty good at telling you what strategies are obviously bad, however the remaining ones aren't necessarily good.ObliviousInvestor wrote:This statement is very interesting as a sort of litmus test. Personally, I agree with it wholeheartedly. But there are also people who would read it and think you're unreasonable for believing such a thing.richard wrote:MC simulations are mainly useful as a teaching tool - how might changing this number (returns, allocations, correlations, whatever) change outcomes. They don't add much to our ability to predict the future.
Well, this shouldn't be difficult to check out. Step 1: Let Morningstar show us the growth in VFINX, and along with it, it will show use the growth in the index itself, including dividends. I will get as close as possible to "today" and "13 years" as Morningstar will let me.EternalOptimist wrote:Believe what you wishRichard wrote:[The S&P] hasn't grown in real terms (including dividends) in the past 13 years.
You mean solid basis - as being able to predict the future? That is not really possible - but never the less it has still been studied to death over the last 15 years, and I have done a study or two myself. In the end in my retirement, I use an income stream that does not require I pull money from the principal and is pretty conservative anyway -- so I'm no where near even the 4% rule.richard wrote:The problem is that there never was a solid basis for the 4% rule. It's just a general rule of thumb based on historical data that's not sufficient to be a reliable predictor of the future.FinancialDave wrote:Unfortunately, there was really nothing new in the article
Yes Richard, I understand independent periods, just as I appreciate the ability and usefulness in using this historical data in bootstrap simulations that eliminate many of these concerns (that come to similar conclusions). But I'm guessing you prefer we forget about those too (and their results).richard wrote: Do you understand the objection? The problem is that there are only about three independent 30 year data points in 85 years of data, not that the entire data set isn't independent of something.
==============================
Monte Carlo simulations are entirely dependent on their inputs, no matter how many times they are run. If you knew future returns, distributions, correlations, etc., then MC would help in predicting the future. OTOH, if you had this information, you wouldn't need to run MC models.
MC simulations are mainly useful as a teaching tool - how might changing this number (returns, allocations, correlations, whatever) change outcomes. They don't add much to our ability to predict the future.
nisiprius wrote:EDN: in September, 1995, in a Worth magazine article entitled "Fear of Crashing," Peter Lynch recommended a 100% stocks portfolio--preferably individual stocks chosen from among Moody's Dividend Achievers, but "You could do this the easy way and invest in an S&P 500 index fund, currently yielding about 3 percent." He stated that you could safely withdraw 7% of the portfolio per year, and gave an illustration (reproduced below). He says this is an investment strategy he described in his book, Beating the Street, which I have not read.
(Note that his 7% is different from the Trinity study's "4%," which means 4% of initial value and then COLAed every year, regardless of portfolio value. Lynch is saying 7% of whatever the portfolio value is, and it will hopefully that 7% withdrawal will not fluctuate intolerably and will keep pace with inflation).
Do you agree with Lynch, or would you judge he was being imprudently optimistic? In your opinion, is an annual 7%-of-portfolio withdrawal from a 100% S&P 500 portfolio a safe, sustainable rate or is it imprudently high?
I would like to infer that you do not agree with Peter Lynch's 1995 article, and that you do not think a 7%-of-portfolio annual withdrawal from a 100% S&P 500 portfolio is sustainable... because that would mean that you and I agree.EDN wrote:...Here is my pearl of free advice for the day: "don't read financial porn and don't listen to active managers"...I wrote:Do you agree with Lynch, or would you judge he was being imprudently optimistic? In your opinion, is an annual 7%-of-portfolio withdrawal from a 100% S&P 500 portfolio a safe, sustainable rate or is it imprudently high?
Why wouldn't they take that into account? Nominal bonds had a rather bad negative real return from 1940 to 1980, well within the range of historical data these studies usually claim to include. I don't know what data Bengen used, but the Trinity authors said they used data from 1926 to 1995.baw703916 wrote:...In answer to Nisi's observation about the 4% taking the fluctuations of equities into account: yes it does. What it does not take into account is a negative real return of bonds. If safe bonds have a negative real return, the SWR (in terms of being sustainable indefinitely) is pretty close to zero. I'm guessing that they don't stay this low indefinitely, but what do I know?
Here's Wade Pfau's data on the 2000 retiree. While the situation is dire, it isn't by any means, the worst historic case. It shows that real portfolio value would have fallen by a third while nominal value declined by 12%.If you had retired Jan. 1, 2000, with an initial 4% withdrawal rate and a portfolio of 55% stocks and 45% bonds rebalanced each month, with the first year's withdrawal amount increased by 3% a year for inflation, your portfolio would have fallen by a third through 2010, according to investment firm T. Rowe Price Group.
...a group of powerful speculators with fortunes made in the automobile business and in the grain markets and in the earlier days of the bull market in stocks--men like W. C. Durant and Arthur Cutten and the Fisher Brothers and John J. Raskob--were buying in unparalleled volume.... The big bull operators knew, too, that thousands of speculators had been selling stocks short in the expectation of a collapse in the market, would continue to sell short, and could be forced to repurchase if prices were driven relentlessly up. And finally, they knew their American public. It could not resist the appeal of a surging market. It had an altogether normal desire to get rich quick, and it was ready to believe anything about the golden future of American business. If stocks started upward the public would buy, no matter what the forecasters said, no matter how obscure was the business prospect.