Calculator Fright

Research on Safe Withdrawal Rates

Moderator: hocus2004

JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Calculator Fright

Post by JWR1945 »

It was a long time before I became comfortable with Safe Withdrawal Rate calculators. It was made worse because I had had very little experience in using Excel spreadsheets. I had had enough experience with Excel to know what it was capable of doing. But I had never used Excel long enough to remember many of the standard tricks and procedures. That was true in every respect. I knew that HELP contained the information. I did not know where it was in HELP. A good bit of time is lost while trying to ask the right question.

My first experience was not with Excel, but with FIRECalc, which has an outstanding user interface. I was able to learn a lot from it right from the start. I discovered almost immediately that Historical Database Rates have come in three clusters. That was immediately obvious from looking at the display. Failures are colored red. If you put in withdrawal rates with some failures, but not many, you will see that the bad years are clustered together.

Later on, I discovered how to use it best. The best procedure is to enter a withdrawal rate and then determine when failures occur. My original procedure was to vary withdrawal rates until the percentage of failures equaled a predetermined amount. That required repeated entries. It was aggravated because of the problem of partially completed sequences. They influence the denominator. The best choice is to count the number of failures within a specified range of years. I have chosen to exclude sequences that begin after 1980 but to include all sequences from 1970 through 1980. That is because frequently there are failures from 1970 through 1974, which means that one should include at least some partial sequences. I put the upper limit at 1980 because it is a round number.

It was much later that I realized the significance of data overlap and how it favors the clustering of failures. It was much later that I found out that the calculator outputs are all in nominal dollars (that is, without adjustments for inflation) although withdrawals are applied in real dollars (unless you specify otherwise).

I had had difficulty when I first tried to download the Retire Early Safe Withdrawal [Rate] Calculator. It is in a zip file. I had never succeeded in downloading the related software needed to open it.

Much, much later, I reached the point that I felt that I absolutely had to download it along with other zip files. I was within a week of paying cash when I received a Kim Komando computer tip of the day that linked me to Coffee Cup's free software (for nonprofessional users). I downloaded Coffee Cup's software and it works beautifully.

As I have mentioned, I was new to Excel in terms of being an actual user. As such, I took me quite a while to learn how to use it efficiently. Even now, I have important gaps in my knowledge. One thing that I regret is that nobody helped me when I asked for assistance. That makes me especially willing to help others. It also motivates me to be especially clear when telling others how to modify calculators.

Overcoming my fear of calculators and persisting long enough to learn Excel has paid off handsomely. We have now learned that switching portfolio allocations according to P/E10 levels would have increased Historical Database Rates substantially (first from 3.9% to 4.8% and now to 5.1%). That is a dramatic improvement. We had been led to expect next to nothing. I have discovered that the original calculator had not actually implemented its intended two-threshold, three-allocation capability. I have now written modifications to do this.

I understand fully why people are reluctant to use calculators. I am willing to provide any assistance that I can to help. I have found using them to have been worth the effort many times over. Both FIRECalc and the Retire Early Safe Withdrawal Calculator are tremendously powerful and useful.

Have fun.

John R.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

...the calculator outputs are all in nominal dollars (that is, without adjustments for inflation) although withdrawals are applied in real dollars (unless you specify otherwise).
How does this affect the results?
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

There are strategies that call for you to take some money out of stocks at peaks. If you are planning on a 40-year retirement and your portfolio has grown enough to meet all of your needs at a 2.5% withdrawal rate, why not convert everything to TIPS and/or ibonds? Even if they produce no real growth, they can meet all of your needs at zero risk. (I am excluding taxes in this discussion.) There would be no need to accept the risk of a market crash. You would have total safety available for the rest of your retirement.

In a real life you would identify more than one income level. You would demand that the lowest level be fully funded regardless. You would not take all of the money out of the market. You would take enough of it out of the market to meet those minimal needs at zero risk.

Have fun.

John R.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

You would not take all of the money out of the market.
One might to decide to leave the S&P during high P/E years regardless, but I see your point.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

Keep in mind that our results with allocation switching are brand new. Prior to my downloading the Retire Early Safe Withdrawal Calculator, it was thought that switching provided minimal improvement at best. Results had been presented to that effect. Almost immediately after I downloaded the calculator, I was able to get a substantial improvement with the calculator as it was. That is, with only one threshold and two allocations. Later, we have been able to make even greater improvements by using two thresholds and three allocations.

In the early days of the NoFeeBoards, I made several studies with FIRECalc that showed that taking the money off the table increases portfolio survivability significantly. Only recently have I found out that those studies are flawed because FIRECalc lists balances in nominal dollars instead of real dollars. It is not necessary to rerun those studies since switching produces all of the gains that were hoped for and even more.

I have run some tests with our best switching algorithm for stocks and commercial paper. It switches the stock allocations to 100%-40%-0% with P/E10 thresholds at 11 and 21. All portfolios beginning in 1921-1980 survive for 30 years at a withdrawal rate of 5.1%, but not at 5.2%. I have examined this portfolio at withdrawal rates of 5.2%, 5.3%, 5.4% and 5.5% with the idea of taking all of the money out of stocks and putting it into (0% real interest) cash equivalents when the balance was sufficient to guarantee reaching 30 years successfully. In none of the cases did it help to make this change. Switching already incorporates any improvement to be gained from using such a strategy.

Before our research on this board, the maximum withdrawal rate that insured 30-year survivability for portfolios that began in 1921-1980 was close to 3.9%. It is now 5.1%.

In terms of what to do at this moment, we have shown that 0% stocks is the best S&P500 allocation at today's valuations. I have found the evidence so compelling that I have sold some stock with the intention of keeping it out of stocks and without a requirement for interest on the cash (equivalents). That is radically new behavior for me. This is the first time that I have moved out of a 100% stock allocation (while monitoring the portfolio, but not selling shares unless absolutely necessary).

My pension is sufficient to meet all of my retirement needs and even more. My withdrawal rate is zero. My criterion for selling is whether the stock would be a buy at normal valuations if earnings were to double (or something else that would be dramatic, yet plausible).

[I have picked up some Merck shares because I considered its price and dividend to be compelling even if the drug pipeline remains empty for a few years. But I have added to my planned cash allocation and I have placed limit orders to make additional sales.]

Have fun.

John R.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

My pension is somewhat less generous than yours, but it took less than 1% of my portfolio to supplement it during the first 2 years of my retirement. I am currently about 60% equities overall, but not S&P. As a play on favorable demographics plus the nascent economic recovery, in my main IRA I have begun experimenting with small cap and foreign mutual funds chosen by short term momentum. The results so far have been promising, with this IRA (currently 95% equity) now larger than it was in March 2000. Historically this method has also provided a measure of downside protection.

In deference to valuations, my taxable account has more bond/treasury type of investments than equity. I also overweight certain sectors such as health care that I think have favorable prospects. This is the part of my portfolio that I plan to draw down first, so it is positioned more conservatively than my IRAs are. If S&P valuations were below 11, I would probably just load up on them and sit tight. I expect the current valuations to end badly eventually. However, the current trend is up, so thus my 60% equity exposure. As further data becomes available, I may modify this.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

I think that you are doing the right thing.

I believe that we are in the beginning of a prolonged bear market. The market seldom moves gradually. There are always some great rallies in a bear market and some dramatic setbacks in a bull market.

I keep in mind that even great stocks drop to single digit P/E ratios at the bottom of a bear market. At least, they did in the early 1970s.

Have fun.

John R.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

I believe that we are in the beginning of a prolonged bear market.
I see the recent bear mainly as a reaction to profit drop in the wake of post Y2K corporate spending on tech taking a 3 year pause. I watched the lagging S&P P/E ratio in Barrons stay around 30 or so the entire time. It is still there in the latest issue. With demographics still favorable, the tax code still ushering boomers plus the pension plans that serve them towards equities, and S&P profits starting to recover, I suspect that the near future will see new S&P highs.

However, I don't see most of these trends lasting for more than a few years. In other words, the 2000-2002 bear market may turn out to be a mere shadow of the one to come. Corporate profit decline, coupled with P/E multiple contraction could be a double whammy. Of course, this is just a wild guess since there are far too many variables to keep track of. I will have to see how things play out as we get closer.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

I looked at using the lower of P/E and P/E10 before I had the ability to investigate switching. My thinking was that sharply rising earnings would show up immediately in the P/E ratio, but much later in P/E10. It turned out that using P/E10 alone was better.

Indicators and Retirement Portfolios from Saturday, Sep 27, 2003 at 10:28 am CDT.
http://nofeeboards.com/boards/viewtopic.php?t=1461

The nature of that study was similar to buy-and-hold. There was only one allocation. Taking advantage of rallies in today's market is essentially a trading strategy. The greatest single concern about trading is the cost. Costs are certain. Rewards are uncertain.

That having been said, I do believe that there is room for selecting some stocks (including index funds) that differ from the S&P500 in critical respects. Even though the S&P500 index may be dominated with expensive stocks with sky-high P/E ratios, some individual companies have stock that sell at reasonable prices, sometimes with attractive dividends as well.

Still, it is worth keeping in mind that a prolonged bear market can bring valuations into the single digits. With a good company, you may end up with prices remaining flat or increasing slightly as earnings grow enough to overcome the downward pressure of the bear on the market in general.

With rare exception, IMHO, selling short is a fool's game. You can be right about the company but wrong about the time element.

Demographics is something that I don't know how to handle separately. I remember hearing a coworker say that she would get out of the market just before everybody else did. There may be lots of baby boomers planning on dumping there shares just a few days before everybody else does.

Have fun.

John R.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

Demographics is something that I don't know how to handle separately.
I suppose one could simplify it by substituting the number of live births shifted 46 years or so for P/E 10 in the models. In years when the number of 46 year olds were increasing, hold equity, and when they are decreasing, hold bonds. Perhaps an average of a few years to smooth out transient drops. I may undertake this project when I have developed more skill with the spreadsheet. If I do, I will attempt to post any interesting results, but it will probably be quite a while. I don't know how accurate the live birth data is for 1871 minus 46. I don't expect earth shaking results, this is just one of the things that interest me.

To date I have been going by the charts showing the number of middle aged people growing rapidly. The rise in births actually started in mid depression years (1936 plus 46 =1982), and grew from 3 million per year to 5 million at the height of the baby boom. So far, so good because a rising population means an expanding economy, which equates to rising corporate profits. Then births dropped off to around 4 million per year during the baby bust. That is what could cause problems, since it means a declining number of people of prime spending age each year for more than a decade. In addition to declining profits, P/E multiples could contract as the boomers attempt to switch from equity to bonds in accordance with standard advice for aging investors (e.g., 100 minus age = % of equity). This is just one factor of many though, events such as Y2K can cause a big bear even when the prime spending population is growing.
I remember hearing a coworker say that she would get out of the market just before everybody else did.
Ask her to let me know when she is about to sell so I can sell just before she does.

Mike
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

Mike wrote:Ask her to let me know when she is about to sell so I can sell just before she does.
I'm retired now. But when I heard her remark, it struck me that I had exactly the same thing in mind.

I recommend using the FirstGov website for demographics material. Everything in every census is there along with lots of studies. I would start by clicking the Data&Statistics link on the left hand side of the Home page.
www.FirstGov.gov
Note: be sure to use the dot gov extension.

You are right about the old data. Economic data were sketchy before the Great Depression. When the Federal Government started intentionally and actively influencing the economy on a grand scale, they needed data.

There have been books and articles about demographics and investing. The main difficulty is in separating separate fact and fiction in terms of what is happening today. Back in the early 1980s I read an article about how the 1990s and later would bring in a new Golden Era. And it has. But if you listen to what is being reported, you would think that we are in dire straights. In fact, we have had two of the mildest recessions ever known, productivity is sky high, opportunities are incredibly good and unemployment is virtually nonexistent in historical terms. For example, it was not long ago that an unemployment rate of 6% was considered full employment. In fact, it was considered dangerous to try for anything lower.

Have fun.

John R.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

Thank you for the link. I have bookmarked it.
In fact, we have had two of the mildest recessions ever known, productivity is sky high, opportunities are incredibly good and unemployment is virtually nonexistent in historical terms.
This prosperity is exactly the phenomenon that leads me to consider demographics as one of the possible contributing factors.
User avatar
BenSolar
*** Veteran
Posts: 242
Joined: Mon Nov 25, 2002 5:46 am
Location: Western NC

Post by BenSolar »

JWR1945 wrote:I have discovered that the original calculator had not actually implemented its intended two-threshold, three-allocation capability. I have now written modifications to do this.
Hi John :)

I was wondering if Greaney ever published your modified calculator to his web site? I remember there being an offer to at one point ...

Hope you are well. I've been enjoying the posts you and Mike have been putting up over here. :great:
"Do not spoil what you have by desiring what you have not; remember that what you now have was once among the things only hoped for." - Epicurus
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

I tried to email my code. I requested a response as to whether the code got through. I did not receive any response whatsoever. I don't think that my attempt was successful.

It is a big program (since the code includes the calculator itself). I have no experience in sending messages with megabytes of data.

I have allowed by subscription to the Motley Fool to expire. I will not return before they start enforcing their own rules.

I registered at the Retire Early discussion board and looked over some posts. A couple of days later, my username was no longer recognized. I received no notice. Apparently, I am a non-person at that site.

I am still waiting for more activity on this board. That is what I appreciated about having hocus. He was always interested in my research and he always had a ton of good questions.

Have fun.

John R.
User avatar
ElSupremo
Admin Board Member
Posts: 343
Joined: Thu Nov 21, 2002 12:53 pm
Location: Cincinnati, Ohio

Post by ElSupremo »

Greetings John :)
I tried to email my code. I requested a response as to whether the code got through. I did not receive any response whatsoever. I don't think that my attempt was successful.
If you'd like to set something up at NFB I'd be happy to work with you on that. We could do something similar to what raddr has or whatever you think would be best. If your interested send me a PM or email. I'll respond! ;)

FWIW I for one greatly appreciate your efforts on this board. I know there are others. Hang in there! :great:
"The best things in life are FREE!"

www.nofeeboards.com
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

If you'd like to set something up at NFB I'd be happy to work with you on that.

I don't know that there is a demand for my calculators.

Only Mike and BenSolar have indicated that they have made their own copies. JanSz showed interest in how well his particular algorithm worked, but not in running calculators himself. (His primary interest seemed to be making a spreadsheet for tracking how well he followed his own algorithm.)

There is also the matter of acknowledging John Greaney's role as the creator of the basic calculator. I would prefer to have his blessing if we were to distribute it ourselves.

Have fun.

John R.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

Mike wrote:I suppose one could simplify it by substituting the number of live births shifted 46 years or so for P/E 10 in the models. In years when the number of 46 year olds were increasing, hold equity, and when they are decreasing, hold bonds. Perhaps an average of a few years to smooth out transient drops. I may undertake this project when I have developed more skill with the spreadsheet. If I do, I will attempt to post any interesting results, but it will probably be quite a while. I don't know how accurate the live birth data is for 1871 minus 46. I don't expect earth shaking results, this is just one of the things that interest me.
If you are able to collect the relevant data and if you have some ideas about how to use them, go ahead and post what you can. I will supply such assistance with the spreadsheets as I am able to provide.

Right now, you have suggested a simple algorithm which is easy to implement. Another easy to implement algorithm would be to adjust P/E10 thresholds in accordance with the demographics.

Algorithms that are fixed relative to individual years are generally easy to implement. I only have to change a line or two. Even an algorithm with memory associated with previous years is easy to implement. Again, I only have to write code for a single set of lines.

Algorithms that depend upon current balances are more difficult because I have to enter them separately for each start year. The JanSz algorithm is an example. It is possible that Excel has a better way for me to enter the code and that I just don't know what it is. It would be a form of relative addressing, which Excel has with the fill handle, but which would skip rows (specifically, 18 rows) as it shifts to the right.

Have fun.

John R.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

Thank you. I am wrestling valiantly with my tax forms now, but will keep this in mind for the future.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

I have found a partial list, the existence of which encourages me that the data is available if I can figure out how to find it.

http://www.infoplease.com/ipa/A0005067.html

There is no adjustment for immigration, and earlier dates appear to be less precise than later ones, but the data may be close enough for my purposes. I will let you know if I find a complete list.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

I have made a preliminary review of the demographic data. I had originally thought of ordering Historical Database Rates in accordance with time-shifted demographic weightings. It turns out to be a little bit more difficult than that.

I made some graphs. The number of live births and (real) E10 are similar, with the number of live births occurring about 10 years before anything shows up in E10.

E10 is the earnings component of P/E10. It is the average of the past ten years of earnings in real (i.e., inflation adjusted) dollars.

Tailor your algorithms to take advantage of this observation. The number of live births could be part of a weighting of P/E10 or its inverse, E10/P, the earnings yield. Or you might look at the number of live births/price (as measured by the index in real dollars). (The relationship between earnings yield as measured by E10/P and Historical Database Rates is very close.)

The birth rate declined gradually throughout the last century. It was depressed a little bit more than it would have been during the Great Depression and during the late 1970s. Described differently, the baby boom was a higher than normal swell upward in a steadily downward trend.

The birth rate seems to contain two types of information. It can be depressed during bad times economically, which should correspond to current business conditions. It is also a forerunner of the increase in E10 that will occur about a decade later.

Have fun.

John R.
Post Reply