January 1, 2007 Letters to the Editor

Updated: January 11, 2007.

CPI Change on P/E 10 Calculations

I received this from CPI Methodology Wonderer.

I have limited understanding of this issue, so forgive me if this question makes no sense. Does Professor Shiller use CPI adjusted earnings to calculate P/E10? The CPI calculation method was changed in the mid 90s. To properly compare apples to apples, would it be necessary to either adjust today's P/E10 downward, or pre mid 90s P/E10 numbers upward to account for the differences in CPI methodology?

Using pre change methodology, CPI would be calculated about 3% higher than current calculations. Alternatively, using today's calculation methodology, past CPI numbers would be about 3% lower. See the chart on this web site that keeps track of the changes:

John Williams' Shadow Government Statistics

If earnings are not CPI adjusted, would high inflation periods distort the statistics? Thank you for being patient with my limited understanding of this subject.

HERE IS MY RESPONSE

Thank you.

Yes, Professor Shiller uses CPI adjusted earnings to calculate P/E10. He includes those CPI numbers as part of his data.

His explanations are in the Online Data section of his main site. His latest updates, however, are included only in his database at his Irrational Exuberance web site.

Professor Shiller’s Web Site
Professor Shiller’s Online Data
Professor Shiller’s Irrational Exuberance Web Site

Professor Shiller’s data appeared first as tables in his 2000 edition of “Irrational Exuberance.”

From the Online Data section of his main site:

“The CPI-U (Consumer Price Index-All Urban Consumers) published by the U.S. Bureau of Labor Statistics begins in 1913..December 1999 and January 2000 values for the CPI-U are extrapolated..For the Plots, I have multiplied the inflation-corrected series by a constant so that their value in January 2000 equals their nominal value, i.e., so that all prices are effectively in January 2000 dollars.”

This dates the earliest versions of Professor Shiller’s CPI-U around 1999, safely beyond the mid 1990s.

The U.S. Bureau of Labor Statistics updates its earlier data when it makes a change. Of course, there can still be disputes about such adjustments. The CPI-U draws special attentions because Social Security, Federal Military and Civilian Retirement Pensions and many contracts in private industry are tied specifically to those numbers. Even a 0.1% change in CPI-U affects billions of dollars.

High inflation periods distort the statistics considerably. Using inflation adjusted (real) dollar amounts greatly enhances the stability of many relationships involving P/E10.

However, both inflation and deflation reduce (real) stock market returns. This was a major finding of Ed Easterling of Crestmont Research. He refers to this as the Y curve. He included it in his book “Unexpected Returns,” but he has an excellent short version in the Stock Market section of his web site. See the slide about “P/E Ratios & Inflation.”

Crestmont Research
Crestmont Research Stock Section

CPI Methodology Wonderer sent me a note of thanks, which I appreciate greatly.

Payout Ratios

I received this from Payout Ratio Commenter.

I saw this article and thought you might find it interesting. Low payout ratios have historically tended to be correlated with low subsequent 10 year earnings growth. Today's payout ratio is very low.

How Fast will S&P500 Earnings Grow?

HERE IS MY RESPONSE

Thank you for an interesting article, Payout Ratio Commenter.

The article reports how well the current payout ratio (the current dividend D1 divided by current earnings E1) predicts the next ten years of earnings growth. The payout ratio fluctuates considerably because, even though dividends tend to be stable, single year earnings fluctuate wildly.

As happens so often, history contradicts what we might expect.

The article reports that higher earnings growth follows higher payout ratios. Apparently, retaining more earnings retards growth. It does not augment growth.

One possible explanation is that corporate managers have to focus on their best ideas when they have limited funds (in the form of retained earnings). Another possible explanation is that a high payout ratio sends a signal that corporations are confident of future earnings. I believe that both explanations are true.

Now, I draw your attention to the time period. The article presents results from the modern era, in this case starting from 1950. Corporate boards have been very careful not to repeat the dividend cuts that were commonplace in earlier decades, especially during the Great Depression. I suspect that the relationship was still present during earlier time periods, but that it was not so strong.

Recently, I have placed several earnings related files into my Yahoo Briefcase. Let me draw special attention to Sheet 7 in my SP500 Returns 01 06 2007 Excel spreadsheet in the Shiller Data by Month folder. It contains information about E1, E3, E4, E5, E6, E7 and E10. I draw attention also to my recent article about the Stock Return Predictor with Earnings Growth Rate Adjustment. This version of the Stock Return Predictor augments P/E10 with historical information about earnings growth. I wrote:

“I generated charts of smoothed real earnings Ex, where x = 1, 3, 4, 5, 6, 7 and 10. I selected E7 for determining changes in the long term earnings growth rate.”

Here is the reference material that I provide to users of the new Stock Return Predictor with Earnings Growth Rate Adjustment:

Reference Earnings Growth Rate (1901-2000): 1.55% per year.
Reference Earnings Growth Rate (1881-1940): 0.72% per year.
Reference Earnings Growth Rate (1941-1970): 3.18% per year.
Reference Earnings Growth Rate (1971-1995): 0.24% per year.
Reference Earnings Growth Rate (1996-2006): 4.45% per year.

You can download a picture of what is going on from my Yahoo Briefcase. Look in the Real E1 and E10 file in the Graphs vs E and D Yields folder.

Earnings grew rapidly from the Great Depression to around 1970. Earnings paused, fluctuating considerably from one year to the next, from 1971 to the mid-1990s. They have resumed their upward course since then, with a couple of down years.

Exhibit 1 in the linked article does not inspire confidence about the cause and effect relationship. The underlying cause is far from obvious. One alternative explanation might be that earnings growth was great up to 1970 at the same time that the payout ratio was high is a coincidence. Instead of a subtle relationship between the payout ratio and earnings growth, there might be an underlying, slowly varying factor that dominates earnings.

The picture in Real E1 and E10 suggests an alternative explanation. There may have been a one-time, persistent increase in the earnings growth rate since World War 2. Such changes have not happened often, less than once per century, but they have happened. Yet, there has been a considerable amount of literature about accelerating business trends since World War 2 and how to cope with the stress that it generates.

What is really interesting is how little the earnings growth rate affects the three versions of the Stock Return Predictor. You can access the basic version using the navigation button at the left at this site or at the PassionSaving site. It is available (but not updated) in my Retirement Trainers as well. The Bear Market version appears in all of my recent versions of Retirement Trainers, including the Simplified Retirement Trainer with Dividends (Simp Ret Tr w Div A2). The third version is the SP500 Stock Return Predictor with Earnings Growth Adjustment (Stock Returns Adjusted).

Entering earnings growth rates of plus and minus 3% per year barely shifts the 10 and 20 year predicted (most likely) returns by 1.5% when starting from today’s P/E10 level of 28. The Bear Market version is the most pessimistic over the next decade. But it is most optimistic at Year 20. All versions report virtually identical returns at Year 30.

Earnings growth rates introduce an important refinement, especially over the next two decades. All calculations easily fall within each other’s confidence limits. If the earnings growth rate has increased permanently, it will gradually show up in Year 30 calculations.

P/E10 continues to dominate all calculations of stock market returns.

Stock Return Predictor with Earnings Growth Rate Adjustment
Yahoo Briefcase

UPON FURTHER REFLECTION, I HAVE ADDED THIS CLARIFICATION

I wrote, “Exhibit 1 in the linked article does not inspire confidence about the cause and effect relationship.”

Look at the Exhibit and think about projections starting between 1950 and 1970. The short-term fluctuations that you see would have been shifted if the predictions were for 8 or 9 years instead of 10. Many of the high-to-high correlations that work well at 10 years would have become high-to-low at 8 or 9 years. Projections that work well at 10 years of earnings growth would not have worked well at 8 or 9 years.

Absent a reason to do otherwise, most analysts would smooth the data to reduce the effects of short-term (two and three year) fluctuations. That is, they would filter out the high frequency noise. They might look at the short-term fluctuations separately. In the form presented, the true relationship might be an artifact of very short-term variations instead of a longer, ten-year relationship.

Payout Ratio Commenter thanked me “for the in depth discussion of the article.” He further stated, “You have certainly given me much to think about.”

There remains a lot for me to think about as well. I doubt that this is my last investigation of earnings, earnings growth and payout ratios. I see similarities with my early research into dividend strategies. It took a variety of small steps for me to get a handle on how to analyze dividend based strategies. I did not know where my dividend research would lead. It turned out to be an outstanding success. I do not know where this new area of research will lead. We will just have to wait and see.

Letters to the Editor in 2007

Letters to the Editor in 2006

Letters to the Editor in 2005