Letters to the Editor

Updated February 22, 2006.

High Dividends: Half DVY, Half RWR

Mike sent me this link to a January 9, 2006 article by Ben Stein: A Retirement Portfolio With Staying Power. It is in his column, How Not to Ruin Your Life, at Yahoo Finance.

A Retirement Portfolio With Staying Power

Ben Stein looked at a 50%-50% split between RWR, an exchange traded fund that tracks the Wilshire REIT index, and DVY, an exchange traded fund consisting of 50 top dividend paying stocks (high quality, high yield, not simply high yield). Phil DeMuth evaluated the portfolio with a Monte Carlo model. He simulated 10000 30-year sequences.

Although Ben Stein did not report the exact details, he did report enough information for me to interpret his results. Here are four important sentences and my interpretation:

“Note that I also asked Phil to track what would happen if the owner of the portfolio withdrew 4 percent per year and allowed the balance to compound.”

“And it shows that portfolio would have grown so much that a fixed percentage of it would easily keep up with inflation.”

“The news gets even better than that: If you withdraw 5 percent a year..you still have only a less-than-one-in-100 chance of running out of money before you reach 30 years of retirement.”

“The most likely scenario with 5 percent withdrawals a year has you with nine times as much at the end of the 30 years as you started with.”

I interpret this to mean that the simulated income streams were 4.0% and 5.0% of the portfolio’s original balance. There were no adjustments for inflation. There were no adjustments for portfolio growth.

Again, this is my interpretation:

After looking at the results, Ben Stein concluded that the portfolio’s balance increased enough to have permitted using a fixed percentage of the portfolio’s current balance.

I draw this conclusion because a handful of the simulated portfolios failed. If the withdrawals had always been a fixed percentage of the portfolio’s current balance, portfolio failure would have been excluded mathematically. The withdrawal amount could have been very low (similar to what I discovered in the The 4% Shocker).

The 4% Shocker

Another factor reinforcing my conclusion is that the portfolio balance with 5.0% withdrawals would have grown nine times in 30 years. The rule of 72 (a good approximation when interest rates are between 7% and 10%) tells us that a 7.2% overall return would have doubled the balance three times in 30 years. That is, the portfolio would have grown by a factor of eight at 7.2%. With 5.0% withdrawals, the reported growth must be a NOMINAL rate. That is, it does not include any adjustments for inflation.

AREAS OF UNCERTAINTY

Modern day REITS have not been around very long. Our projections might turn out wrong. It takes more than 20 or 30 years of data for us to build up a high degree of confidence in our ability to spot the critical risks. Many investments have looked good for 10 or 20 years, only to end badly.

The biggest issue with REITS is whether they are too popular.

Commercial real estate is noted for booms and busts. History has seen building booms that glut the market again and again, always leading to long periods of hard times.

I think that we have a good understanding of the high quality, high dividend stocks that make up DVY.

MY ASSESSMENT

Ben Stein has come up with a winner.

Yes, I think that we can do better. I think that we have quite a bit to offer. But Ben Stein’s basic approach is a good one.

I think that it would help to set aside at least a portion of funds for future purchases, using a TIPS ladder while waiting for valuations to improve. I prefer drawing down principal temporarily via withdrawals from a TIPS ladder to selling REITS or stocks. Between P/E10 and dividend yields, we have a strong basis for deciding when to add to our holdings. Of course, we must be sure that long-term (smoothed) earnings are sufficient to support dividends and dividend growth. This is less of a concern with quality exchange traded funds.

We do not have to be overly concerned with rebalancing. We are better off making adjustments to the extent that we can assess valuations accurately. We always need to be careful about costs.

Consider Benjamin Graham’s constraint. In this case, you might start with 25% in the RWR and DVY portfolio and 75% in TIPS.

You might make another adjustment similar to Gummy’s Sensible Withdrawal strategy. You might keep the initial withdrawal rate below 4.0% (plus inflation) and cap it at 5.0% (plus inflation) in the first few years, depending upon how the markets behave. But remember that 2% TIPS can provide you with a 4.0% income stream for 35 years before running out of money. Don’t cut back too much.

The Getting Started in the Guidelines section has many additional details.

Do visual aids or other simple tools exist?

I received this letter from Don.

I just stumbled on your site while researching various investing in retirement topics. I am retired with a decent Federal pension and a substantial portfolio for additional expenses. I plan to withdraw at a very low rate so I should be in good shape, but I have long been uncomfortable with Monte Carlo analyses and what you and Robert Bennett call the conventional SWR advice. I figured, "OK, TRPrice's Monte Carlo says that with what I have I can keep an 80/20 portfolio and withdraw $X for 30 years with 99% reliability." But what if I was running the simulation in 2000 and then lost 40% of my nest egg -- I assume I now have new info that tells me I am on one of the fairly lousy paths - things don't really compute at 99% any longer, do they?

Your "value" factor seems to address my concern and feels intuitively right. But it is extremely hard for a newbie like me to come up with numbers. I have browsed around your sites, but I don't see any charts that show how value indexing would perform. I also don't see any guidelines about when in the value cycle to move assets into a different class - you can't just constantly switch stuff around to try to match some P/E10 ratio can you? (I.e. for long term timing when do you make allocation changes?). And finally, I don't see any charts showing the P/E10 against the S&P500 over time. Such a chart would be very helpful.

Do these things exist and I am just not finding them?

Thanks,

Don

PS: By the way, you guys talk about several on-line forums that you follow and/or participate in. Would you mind sharing the addresses?

HERE IS MY RESPONSE

I have difficulty putting graphics directly on this site. I believe that I now know how to do it, but I like my interim approach better.

First for some pictures:

I have posted several graphs and Excel tables for public viewing in my Yahoo Briefcase. They are Microsoft Word documents. That is, if you can read documents created by Microsoft Word, you can view these graphs and tables. Your computer does not need to be able to read spreadsheets. My understanding is that Microsoft supplies FREE Microsoft Word readers for those who need them.

Yahoo Briefcase

Professor Robert Shiller tabulates S&P500 data at his web site. This includes P/E10. He gets the credit. He invented it. I recommend that you download his data as a zip file. Downloading it as an HTML file (i.e., directly as a web page) takes a lot longer. His worksheet includes a graph of P/E10 versus calendar year. It is Figure 1.3. [He labels his plot as the Price-Earnings Ratio. It is really P/E10.]

Professor Shiller's Web Site

Do you have a problem with zip files? Solved. I have put several references to free software at the bottom of the current (January 14, 2006) Notes Section. The title is Problem Downloading Calculators? It includes links to free software to zip and unzip files.

Notes starting from January 14, 2006

I recently addressed some of the practical issues that you mention in the Comments Inspired by "Year 10 Choices" portion of the February 5, 2006 Letters to the Editor. You are correct. "Moving in and out of stocks is a bad idea because of costs." In addition, "our models are never EXACTLY right."

February 5, 2006 Letters to the Editor

I believe that you have already visited our Guidelines Section. I have grouped several links to help newbies under Getting Started.

Guidelines Section

I currently post at the Financial Freedom Blog at Rob Bennett's web site.

Rob Bennett's Web Site

I have posted at NoFeeBoards and the Early Retirement Forum.

NoFeeBoards
Early Retirement Forum

I follow several boards at the Morningstar site. You can find them by clicking the "Discuss" tab at the far right.

Morningstar

Raddr has a discussion forum that you might find of interest.

Raddr’s Web Site

Peter Ponzo is our expert on all matters mathematical. He is a retired mathematics professor. He posts under the name Gummy. Even those who do not have technical training can learn a lot from his tutorials.

He is an outstanding teacher. His tutorials are entertaining.

Professor Ponzo's Web Site

Very cool stuff here

I received this letter from Bud.

I haven't been through half yet but you have done some amazing work.

When I read Shiller's PE10 research, the thing that got me was that a 10-year moving average was really slow to adapt to changing valuations. I know he uses it to cut out some of the noise but there are other kinds of averages that reduce noise but are more sensitive to recent data than the standard one.

Have you ever used any other kind of moving average with your research like weighting recent data higher or using exponential smoothing? It could sharpen up your results.

Bud supplied an interesting link with an example of exponential smoothing. The example is related to dieting.

Example of Exponential Weighting

HERE IS MY RESPONSE

Thank you for your kind remarks.

Regarding Professor Shiller's P/E10, only the earnings are smoothed. E10 is the average of the latest ten years of real (i.e., inflation adjusted) earnings of the S&P500 index. Prices fluctuate instantly. P is the current (real) price or index level, adjusted for inflation. With only a few exceptions, I use the January values of P/E10.

Exponential smoothing is something that I have considered, but which I have not pursued. Gummy (retired mathematics Professor Peter Ponzo) suggested that I investigate exponential smoothing shortly before the closing of the original NoFeeBoards site and its SWR Research Group discussion board.

Gummy's (Professor Ponzo's) Web Site

I have taken several steps toward using exponential smoothing, but I have not pursued them. My biggest difficulty is constructing the series. To construct a series, I need to form a weighted average (multiply real earnings by weights, sum and normalize), preferably using a fixed number of terms. I don’t know how to do this easily. Excel has the AVERAGE function, but I am not aware of a WEIGHTED_AVERAGE function.

Assuming that I am able to construct weighted averages, I have two series that I would like to construct: one with monthly data and the other with annual data. Professor Shiller provides a complete set of monthly data. I have constructed a thinned set of annual data. (NOTE: I have placed a copy of my thinned set of Professor Shiller's data in my Yahoo Briefcase. Select Shiller Data by Month.)

Yahoo Briefcase

Our calculators need annual entries.

Once I have calculated the series, entering the data is easy. Just paste the new entries (that replace P/E10) into row 186 of (a copy of) one of my calculators. Typically, this is involves copying from a column, then using Paste Special to transpose this into a row. Making this easier, the years are listed in row 197.

Suddenly, we have a new calculator with a full set of capabilities.

We can do this with the Gummy series of calculators (i.e., with Gummy's database) and the Unclemick 01 calculator (with the Dow Jones Utilities Average) as well as the standard Deluxe Version V1.1A08a with the S&P500 index.

In my earliest investigations, I looked at the ORDER (or rank) of Historical Surviving Withdrawal Rates. I collected 30-year Historical Surviving Withdrawal Rates for each year using portfolios with 50% and with 80% stocks. Then I ordered them using various approaches.

I looked at single-year P/E, P/E10, dividend yield and Tobin's Q (price to book value in terms of replacement costs). I looked at a few conditions involving changes in single-year P/E and in P/E10. I ranked products of different lists of orders. I ranked products of values. Almost all of that work was by hand.

I made an important breakthrough when I discovered that early data differs from later data. I drew a line at 1920-1921. Gummy has provided an important insight by plotting correlation coefficients versus start years (i.e., the years at the beginning of historical sequences). We can gain additional insights from Ed Easterling's findings about inflation, deflation and the Y-Curve at Crestmont Research. Refer to P/E Ratios & Inflation.

Crestmont Research

In my early investigations, I focused on the Dividend Discount Model and its variants. I found that first-year dividend yields do a good job of predicting Historical Surviving Withdrawal Rates. Then I included the effect of valuations along the lines that John Bogle does when he calculates the speculative return of stocks. I modified his procedure to use changes in P/E10 instead of single-year P/E. Later, I used the percentage earnings yield (100/[P/E10]) and dividends. This improved the numbers, but the combination was messy. After all, dividends come out of earnings and dividend yields are closely related to earnings yield. Finally, I considered earnings yield by itself.

Earnings yield 100E10/P worked well and it made sense. It worked better than first-year dividend yields. It eliminated the problem of first-year dividend yields and dividend cuts. It turned out to be useful for individual market segments. Knowing how well the market performs overall tells us a lot about how the individual market segments perform.

I have seen some data that indicate that single-year earnings growth is becoming more volatile. I don’t know exactly what this means to us. Consider the possibility that the correct adjustment is to dampen the influence of accelerated earnings.

Earlier Letters to the Editor

Retirement Planning Tools Offer Dangerous Advice
February 14, 2006 Letters to the Editor


More Comments: This Time Inspired by "Dollar-Cost Averaging Today" and Comments Inspired by "Year 10 Choices."
February 5, 2006 Letters to the Editor


Comments Inspired by Reading "The Story Behind the Numbers" by Rob Bennett. My response, including Risky Alternatives, Dollar Cost Averaging Today and DCA Today: The Point of Frustration.
January 29, 2006 Letters to the Editor


Unclemick about the Dow Jones Utilities. BillW (with Thanks!) about getting started. Rob Bennett about Dividend Theory versus Dividend Reality. This led to a breakthrough.
January 11, 2006 Letters to the Editor