The JWR retirement calculator

[sgeeeee]

Has anyone else used any of the JWR calculators? I downloaded the equity/I-bond switching calculator and ran some simulations. For the 30 year retirement case, switching between the three cases

1) 100% I-bond (for high PE periods)
2) 50% equity and 50% I-bond (for mid PE periods)
3) 100% equity (for low PE periods)

The simulator indicated that historically you could have increased your SWR from about 4.2% to almost 4.4%. That seemed like a pretty minimal advantage for a technique that required such massive movements of allocations. Plus the advantage went away almost entirely when I looked at 40 year peiods. Maybe I'm missing something.

The other point I'll make is that based on my reads of *****' posts, I was expecting a calculator that provided me with a modified SWR using standard fixed allocations and based on today's valuations. I was not expecting a calculator that simulated a result based on utilizing a massive re-allocation scheme.

It's interesting and I appreciate the work JWR1945. Thanks. But I'm certainly not ready to employ this kind of investment strategy to gain a few tenths of a percent SWR based on historical optimization. I'll try to take a look at the other programs when I get a chance. Someone let me know if I'm missing something in these programs.

 

[Michael]

Switching with I-Bonds doesn't seem very practical to me. There are such low limits on annual purchases that it would take many years to put a reasonable sized portfolio into them. It would be time to switch back out of them before you could even get your entire portfolio into them.

I ran simulations with commercial paper, and switching to 100% equity below P/E 14 significantly improved 40 plus year survival rates. 60% equity from P/E 14 to 24, and no equity above P/E 24 led to 100% survival at even 60 plus year retirements at 4.2 % withdrawals. It also significantly improved average terminal portfolio values at 30 years. Of course, this is no guarantee that it will work in the future. It has not worked very well since the early 1990s, and there is no telling whether things will return to their historic pattern.

 

[wabmester]

It has not worked very well since the early 1990s

Probably because the market P/E hasn't been below 14 since the late 80's.

So, what happened in the 80's that might have made the P/E go up, and will we ever see that level again?

* the IRA and 401(k) were introduced in the 80's and are hugely popular today

* discount brokerages were introduced in the 80's (well, 1977) and are hugely popular today

* the PC went mainstream and had a huge productivity impact

* NASDAQ (started in 1971) had a huge influx of offerings and became wildly popular

And now we have the internet, instant access to information, etc. Basically, the markets became much more popular and efficient in the 80's.

Higher efficiency should mean less risk = lower risk premium = higher average P/E. I think it may be safe to expect the average to stay higher than it was, but 14 sure would be a nice entry point .

 

[hocus]

I'm certainly not ready to employ this kind of investment strategy

Thanks for looking at the tool, SalaryGuru.

I think it is awfully early in the discussions to expect too many people to be changing their investment approach based on what the tool reveals. There are so far three posters who have said publicly that they have made changes in their portfolio mix as a result of the findings brought to light during The Great SWR Debate: *****; JWR1945; and BenSolar. There are over 100 others who have expressed a desire to explore the various questions brought to the table, but who have not yet heard what they need to hear to take the step of changing their investment class allocations.

It's not reasonable to think that people are going to change their allocations first and then come to an understanding of what the tool reveals about the realities of SWRs. People need first to come to an understanding of why changes in valuation levels is a critical factor in the determination of SWRs. When people get that, they will be able to make use of the numbers generated by the tool with a lot more confidence.

I hope we can all agree that JWR1945 has been completely upfront in letting us know how he is making use of the historical data in coming to the conclusions he has come to. Given the extent to which his claims differ from the claims that have been dominant in most discussions of investment strategy held over the past 20 years, it is understandable that we would want to be absolutely sure that he is shooting straight with us. It seems unlikely to me that he would have permitted a downloading of the material if it did not in fact back up his claims.

My hope is that we will now be able to move forward to a more fruitful stage of the debate. We know that there is data out there that has been examined by a serious researcher who was led by that data to the conclusion that changes in valuation levels do indeed affect long-term stock returns (and, thus, SWRs). Now we need to put together the pieces, to figure out why it is that JWR1945's findings are so at odds with the findings generated by the conventional methodology findings.

It's that sort of discussion that I refer to as "the good stuff." There's lots of juicy stuff to talk about, and, once we are in agreement that the data does indeed say what JWR1945 says it says, no grounds for any board friction. We are now at a stage in the proceedings where we are primed to move together in a mutual "learning together" experience. I believe that the product of our work in this next stage will make clear to all why I have spent so much time and energy over the past two years trying to obtain a place at the table for discussions of the data-based SWR concept.

 

[unclemick]

Wab

I tend to agree with your points -soooo- grit my teeth and hold my balanced index and soldier on. And poke around a little at the margin with my hobby stocks but not willing to go as far as hankjoy in search of current yield to improve current cash flow.

I think my trusty Ben Graham - The Intelligent Investor (1972) has a 25-75% range for stocks depending on valuations -BUT he was picking them one at a time.

 

[hocus]

this is no guarantee that it will work in the future.

This is an important caveat. It is an important caveat to all SWR analyses. SWR analysis looks to data from the past to make reasonable assessments of what take-out number will work in the future. It is always possible that stocks will perform in the future in ways different from how they have always performed in the past. If that happens, SWR analysis will turn out not to be a perfect guide to the formation of Retire Early plans.

I am not sure that we have any other good options, however. If we assume that stocks will perform in the future in ways entirely at odds with how they have performed in the past, we have nothing to go on but hunches in putting together our plans. It's a good idea to remind people that SWR analysts are not fortune tellers. All they are looking at to generate their findings is hitorical data. That said, the historical data provides powerful insights.

SWR analysis is not perfect. But it is valuable.

 

[unclemick]

Also - JWR's post on div minus 1% didn't thrill me at all - but numbers are numbers. I plan to keep reading but no dramatic changes for now. Variable SWR for me - yield plus 1%. Research on!

 

[hocus]

Higher efficiency should mean less risk = lower risk premium = higher average P/E.   I think it may be safe to expect the average to stay higher than it was

There are a number of serious analysts who agree with this statement.

It's not a factor properly included in an SWR analysis, however. SWR analysis presumes that stocks will perform in the future as they have in the past. To presume that the average PE will be higher in the future than it has been in the past is to assume that stocks will not perform in the future as they have in the past.

The thing to do if you believe that the average PE will be higher in the future than it has been in the past is to add something to the SWR in the determination of your personal withdrawal rate (PWR). You could take SWR Plus One as your PWR.

There is nothing whatsoever wrong with people saying that they personally believe a 4 percent withdrawal to be safe. Everyone is of course entitled to his or her personal opinion on this question. It is a problem, however, when people say that the historical data shows this to be so. The historical data reveals the average PE that applied in the past, not the average PE that may or may not apply in the future.

When reporting the SWR, you need to make reference to what happened in the past. When determining your PWR, it is perfectly appropriate to make adjustments in line with your personal expectations re ways in which stocks may perform in the future that are at variance from how they have performed in the past.

 

[hocus]

Re: The JWR retirement calculator[b]UncleMick[/b]

JWR's post on div minus 1% didn't thrill me at all - but numbers are numbers.

UncleMick is making reference to a thread that was put to the SWR board yesterday.

http://www.nofeeboards.com/boards/viewtopic.php?t=2554

I am confident that JWR1945 is right about the numbers. However, I think it might be going too far to read his posts here as a demonstration that the “Dividends Plus One Percent” rule of thumb does not provide meaningful insights.

Here is a PDF link to a recent article by Rob Arnott, editor of the Financial Analyst’s Journal, titled “Is Our Industry Intellectually Lazy?”

http://www.aimrpubs.org/faj/issues/v60n1/pdf/f0600006a.pdf

Please scroll down to the section titled “Pension Accounting 4.”

Arnott: “Returns are for the most part a function of simple arithmetic. For almost any investment, the total return consists of yield, growth, and multiple expansion or yield change....For stocks, based on very long term history, growth tends to be around 1 percent above inflation....So why expect 7 percent in the future? The U.S. equity yield is currently well under 2 percent. And we probably should not count on resumed multiple expansion because the market is not cheap by any conventional definition. Much of our industry seems fearful of simple arithmethic of this sort.”

Arnott is making an argument in favor of the data-based SWR tool here. It seems to me that he is also making an argument supportive of the “Dividends Plus One Percent” rule of thumb. It could be that I am entirely wrong about this. I make it a practice to stay out of discussions of numbers because I am just not good at that sort of stuff.

My reading of what Arnott is saying, however, is that the “Dividends Plus One Percent” rule of thumb makes at least some sense as a rule of thumb. I am sure that JWR1945 is right about what happened in 1966. I think that it is important, however, that we distinguish between mathematical constructs and rules of thumb. A rule of thumb is advertised as a rough guess as to what will work and nothing more. A mathematical construct purports to be based on a calculation of numbers. A higher level of precision is required for a mathematical construct.

I am not able to say whether the “Dividends Plus One Percent” rule of thumb is a good rule of thumb or not. I lack the expertise with numbers required to make an informed statement on the question. But I am not personally convinced that the “Dividends Plus One Percent” rule does not provide any meaningful insight into what withdrawal rate makes sense for an aspiring early retiree. The rule of thumb certainly lacks the precision of an analytically valid SWR analysis. That does not necessarily mean that it does not provide valuable insights of its own, in my view.

It's important to keep in mind that the SWR number is the number that works in the worst-case scenario. It seems possible to me that the "Dividends Plus One Percent" rule of thumb is giving you a number that will often but not always work. My guess is that many investors will not see a need to go with a number as conservative as the SWR in determining their personal withdrawal rate (PWR). It could be that the "Dividends Plus One Percent" rule of thumb provides a not-great indication of the SWR that applies but a not-too-bad indication of the PWR that some investors would want to make use of in their plans.

 

[JWR1945]

unclemick

Also - JWR's post on div minus 1% didn't thrill me at all - but numbers are numbers. I plan to keep reading but no dramatic changes for now. Variable SWR for me - yield plus 1%. Research on!

I followed that up with some numbers about dividends. I have been wrong about the behavior of dividends in terms of real dollars. They do not always grow enough to match inflation except over very long time periods (e.g., 20 or 25 years). They consistently grow in terms of nominal dollars. The 1966 sequence ran right into the stagflation days of the 1970s.

You can still count on taking out all dividend amounts plus 1%. You cannot count on taking out the initial dividend and increasing withdrawals each year to match inflation.

Have fun.

John R.

 

[JWR1945]

I will be examining some of the reported calculator conditions.

Keep in mind that the behavior of switching before the early 1920s and afterward differ. I can point to several qualitative reasons as to why this might be so, but not why it must be so.

I usually report the results for 30-year retirements begun in 1921-1980. [I can separate these years very easily by entering 1921 into cell M1 and 1980 into cell M2.]

As an example: in the boom times of the late 1800s, prices declined steadily because productivity increased and the dollar was still on the gold standard. A side benefit was that a portfolio 100% in commercial paper would have allowed a 30-year withdrawal rate of 6% or higher. Don't count on doing that today.

Take many of the comments from others with a grain of salt. Those who use the Fama/French data (which is great for looking at slices of the market) have sequences that begin in 1927. They do not use data from the earlier period.

Have fun.

John R.

 

[JWR1945]

Be careful when looking at partially completed sequences.

Keep in mind that all of the calculator's data after 2002 is dummy data with poor returns. That is why I added a data reduction column with balances in the year 2000. [I could have used 2002, but 2000 is a round number.]

Have fun.

John R.

 

[John Galt]

Besides "partially completed sequences", be careful with
partially completed seances. Like planning to ER, it's
easy to get spooked

John Galt

 

[JWR1945]

Look at middle thresholds around 11 or 12 or 13.

I have made a few quick, spot checks. The earlier reports with withdrawal rates around 4.4% or so appear to be with a middle threshold around 14 or 15.

Remembering some results from my earlier sensitivity studies, I looked at lower thresholds around 12 and an upper threshold of either 21 or 24.

I stuck with the 100%-50%-0% stock allocations (for lowest P/E10 values, middle values and highest values, respectively). [Allocations of 100%-30%-0% and/or 100%-40%-0% may do better.]

I used the default value of 2% for the real interest rate of ibonds. I also looked at commercial paper. I did not look at TIPS.

With commercial paper and with thresholds of 12 and 21, portfolios with a withdrawal rate of 5.0% would have survived all 30-year sequences starting in 1921-1980.

With 2% ibonds and with thresholds of 11 and 24, portfolios with a withdrawal rate of 5.0% would have survived all 30-year sequences starting in 1921-1980. [Portfolios with a withdrawal rate of 4.8% would have survived all 30-year sequences starting in 1871-1980.]

The corresponding withdrawal rates for 100% survival for 30-year sequences beginning in 1921-1980 with thresholds of 12 and 21 and with 12 and 24 were 4.8% and 4.9%, respectively.

The earliest failures in the 1921-1980 timeframe are found in sequences beginning in the 1960s. [In one case, it was 1964. In another case they were 1965, 1966 and 1968.]

The earliest failures in the 1871-1980 timeframe were in 1907 and 1910.

Have fun.

John R.

 

[Michael]

So, what happened in the 80's that might have made the P/E go up, and will we ever see that level again?

Everything you have written is true. The nascent economic recovery, following a period of super low interest rates, speak well for the next few years. If President Bush gets his wish to put part of Social Security contributions in private equity accounts, the multiples could expand further.

That said, I am given pause by the fact that people in the past have said the words "this time is different", and "we have entered a permanently higher plateau of valuations" before. They were always wrong. I can think of a number of good reasons for multiples to contract in the future, but most of the reasons allow for the possibility of several more years of upside first. Since none of us mortals can know enough to predict the future, the best that I can think of is to stay partly in equity, but adopt risk reduction strategies in case things go wrong. High valuation eras have historically tended to be high risk times, and it is a long way down this time around.

 

[hocus]

Since none of us mortals can know enough to predict the future, the best that I can think of is to stay partly in equity, but adopt risk reduction strategies in case things go wrong.

This is a good expression of my viewpoint on the strategy question.

If an investor gets out of equities altogether because of what she learns from an SWR analysis, there is a chance that prices will go up in the short term and the investor will feel regret over the decision to exit from stocks. She then may lose confidence in the SWR tool and get heavily back into stocks at just the wrong time.

I prefer a more measured response to what the historical data says. I see a good bit of sense in the rule of thumb from Benjamin Graham that was posted by UncleMick earlier in this thread. That was to go as low as 25 percent stocks at times of high valuation and to go as high as 75 percent stocks at times of low valuation, and to have a stock allocation somewhere between those two at times of moderate valuation.

I personally have a zero allocation to stocks today. But that is because of the particular circumstances that apply with my particular Retire Early plan. I tend to think that most investors should try to keep some skin in the stock game even at times of extremely high valuations.

I of course do not think that the average investor should be 75 percent invested in stocks at the sorts of valuation levels that apply today. That is a high-risk valuation level according to the historical data and we should be forthright in letting people who use this board as an information resource know that.

 

[unclemick]

Beware the 'average' investor

My nephew stopped by yesterday on his last day on leave. Age 30, with a 30 yr 'savings span' target age 60. 70-80% stocks - half way between Bogle and Bernstein, over half way to a military pension, with civilian wife - saving 40% of taxable income. I'm glad I gave him the 94 Bogle book and De Gaul lecture back then. But he's weakened since reading 'Four Pillars'. They will do much better - even at high stock valuations than I did starting in 1966 with 'no guidance'. Of course I'm prejudiced.

 

[ShokWaveRider]

Where is the JWR retirement calculator

 

[JWR1945]

http://www.retireearlyhomepage.com/jwr1945.html

Have fun.

John R.