Profiting from Forward Earnings Estimates

There are many elements of conventional Street wisdom that are completely wrong.  One of my missions at "A Dash" is to challenge what many find to be obvious.

As we embark on a new earnings season, a consideration of how we think about earnings may be especially helpful.

As background, here are two elements of Wall Street Truthiness about earnings:

1. Companies and analysts are widly optimistic about forward earnings.
2. Companies lowball earnings expectations so that they can deliver an earnings "beat."

The astute readers of "A Dash" will immediately see that these widely-held beliefs seem to be inconsistent.  Many of those in the financial punditry happily advance both arguments -- but not in the same post.

There is an obvious explanation.  At some point there is a "crossover," a point where the forward estimate is accurate.  If both statements are true, we know that estimates are accurate at some point, so when does the crossover occur?

Background:  My Viewpoint

My contention is that most people are making a big mistake by dismissing forward earnings.  If people form a negative opinion about a source of information, they just tune it out.  My approach is to embrace many sources, but be careful to define the happy zone for each.

In the case of stock analysts, I have some fairly strong opinions.  When I began in the business in 1987, making the transition from the academic world, I enjoyed reading widely.  Since my company had access to research from many sell-side firms, I would often read twenty analyst reports on a single company.  Most of the reports sounded the same, like something that might have been written by one of my old students.  As I learned more about how analysts were recruited, I realized that my perception was accurate!

The reason for reading many reports on the same company was to discover who had a different idea - -something unique that others had missed.  It was a valuable exercise.

In my most successful investment program one of the factors is an excessive reaction by analysts.  I have studied thousands of reports, always with a skeptical eye.  I like to be on the opposite side of the upgrade/downgrade cycle.

For these reasons I find it amusing that some readers -- people who have never actually read a report, relying soley on someone telling them about an "upgrade," insist that my use of analyst data is "naive."  I mention this only to show the grip that preconceptions have on our thinking.

My position is that we should look at each source for possible useful information.  The army of stock analysts is very good at doing one thing:

Analysts' forward earnings estimates are very good for a one-year time frame.

 

The Common Mistakes

A few weeks ago I wrote a piece arguing that forward earnings estimates were better at forecasting future earnings than were the popular backward looking methods.  I was quite surprised by the reaction.  You would think that I was writing about religion or politics!

I am always delighted to have comments, but none of the commenters engaged me on the proposition raised.  Not a single one showed that Shiller or Hussman or anyone else who specializes in a 20-20 view of the past was better at forecasting future earnings.  They all skipped ahead to a discussion of stock valuation.

To emphasize, in the forward earnings series I am taking this one step at a time.

1. Earnings are important.  Eventually, stock prices reflect the fundamentals.  If you do not understand this, you should reach some of the work from Chuck Carnevale, whom I have cited on several occasions.
2. If you have a better estimate of next year's earnings, you will have an investment edge.
3. The consensus forward estimates are the best method of looking one year ahead.  I do not know of any better source.  Suggestions are most welcome, but use evidence.

These three points are the current agenda.  In a future article I'll discuss more about ways of using the information, but there is only so much you can do in a single piece.  This article, and the related research, took about 25 hours.  It would be easier (and generate more page views) to post a breezy opinion piece every day.  My approach is more valuable for investors who will take the time to read, and to keep an open mind.

So please -- be fair!  If you want to be a perma-bear who is waiting for the Shiller method to signal another era of single-digit P/E ratios, that is your privlege.  You are not an investor.  You are a spectator.  Unless interest rates go to 16%, you'll never see it.  You are not trying to forecast earnings, and you basically have no winning investment options.

Why People are Going Wrong

One of the reasons so many people are mistaken about forward earnings is the widely-cited study from the McKinsey Quarterly.  Scores of blogs have shown the key chart from this.  Since it confirms everyone's pre-existing viewpoint, no one bothered to consider the dilemma I posed at the start of this article.

I want to be completely clear that McKinsey is focused on unrealistic long-term expectations for companies that include their clients.  The article talks about 3-5 year growth forecasts.  They do not address forecasts in the one-year time frame.  The authors responded to a preliminary inquiry in my research for this article, but they did not respond to questions about the "crossover" or the one-year time frame.  Most of the people citing the article are taking the McKinsey conclusion and incorrectly applying it to a one-year time frame.

Here is the McKinsey chart:

 

McKinsey does not explain the scale on the vertical axis, so we do not know the meaning of EPS.  Their main point seems to be the general downward path of estimates.  They do not attempt to explain the fact that 2/3 of the companies beat estimates.  We have no way of checking this research, since unlike academic researchers, they are not sharing the data.

I can do better.  I myself have some of the Thomson Reuters data and a friend has documented other years.  Here is my own chart for the time period that I can verify.

 

 

While my data do not go back as far in history as McKinsey's, I think the information is more accurate in showing the final earnings result (months after the calendar year ends) and showing recent years.

There has also been a change in behavior in the post SOX era.  Companies and their accountants have become more accurate.

Since the squiggles are difficult to interpret, let me focus on forecasts for the twelve-month time horizon.  This table shows the calendar year forecast made at the start of each  year, comapred to the actual result (known about sixteen months later).

These are impressive results.  With the exception of the year of the Lehman failure and the concomittant economic collapse, the estimates have been too pessimistic.

Conclusion

The criticism of investment analysts typically gets a round of cheers from the investment world.  It lets us all feel smug and superior.

It is more profitable to ask if there is anything analysts actually contribute.  This article shows that the consensus is quite useful in forecasting earnings for the upcoming twelve months.  Since most people do not believe this, the knowledge gives you a significant investment advantage.

Acknowledgements

I very much appreciate the advice and cooperation of Tom Brakke, the leading expert on analysts and earnings estimates.  The conclusions here, and any mistakes, are mine, but I appreciate Tom's advice and help.  He has many excellent ideas on how to use forward earnings.  I hope he chooses to join in.

I also appreciate the continuing discussion of forward earnings by Brian Gilmartin of Trinity Asset Management.  He always has a finger on the pulse of estimate changes.

I have requested more data from Thomson Reuters.  To my surprise, they have not responded.  One would think that they would want to cooperate in demonstrating the value of their research.  So far, no luck.

Meanwhile, my own future articles in this series will discuss issues in interpreting earnings estimates.

UPDATE:  Chart corrected 10/15/10, 1:30 CDT to reflect a few missing data points.  Thanks to reader "Angel" for spotting this.  (JM:  My data source is excellent, but the record switched from weekly to monthly reporting, as you can see from the lines.  This was the only year affected.)

 

Comments

[jason george]

Jeff ~

Thanks so much for your efforts.

This is one of my favorite blogs on the Internet.

Good luck this week!

Jason

[oldprof]

Thanks to everyone for the encouraging comments. Most bloggers do not do any original research where they have to collect data, verify, analyze and interpret -- all before writing.

It is nice to know that people recognize and appreciate the effort.

Jeff

[oldprof]

Brett -- The one-year estimates on the individual stocks that I follow constitute a mixed bag. I really do not have a good statistical measure. I am frequently focused on finding analyst errors in my universe, so I am not sure it is a good sample.

Good question -- I'll give it more thought.

Jeff

[oldprof]

Tom -- Thanks for following up with a great article. I hope that everyone will take a look: http://researchpuzzle.com/blog/2010/10/18/taking-another-look/

Meanwhile, there is much more to add to the forward earnings discussion. I know that you have a number of themes and I also have a couple of ideas. I look forward to further discussions.

Thanks,

Jeff

[tom brakke]

I followed Jeff's lead and am also taking another look at the McKinsey study and its ripple effects in the blogosphere.

[Brett Alexander]

Jeff,

Excellent work. One question, do you find that one year earnings forecast for individual companies to also be quite accurate?

[Joe]

Jeff,

Just wanted to comment how valuable your research has been to me and how much I appreciate the effort you put into your blog. Great stuff!

Regards,

Joe

[Paul Nunes]

This was very helpful Jeff.

[Chris Tinker]

spot on that should have said...

[Chris Tinker]

Hi Jeff,

You are spon on here. I have built up an awful lot of analytical work on this over the years and totally agree with your conclusions. 1 year forward earnings are a very important information set - especially when exanmined in relation to changes in expectations andcombined with the year two growth estimate of earnings followeing on from that. I would be happy to share data and resources offline.

kind regards Chris

(note that I've changed email address as I now run my own business )

[Leo]

Thank you for this insightful article.

I have discovered your blog about two years ago, and it's always a pleasure to read. I regard it as one of my most valuable sources.

Leo

[Mike C]

Appreciate the level of work that went into this note! Good stuff