Pondering the Preferred Procedure

  

The SSG’s Preferred Procedure is one method to estimate future EPS growth for the next 5 years.  Cy Lynch, a NAIC/BI and Manifest Infesting guru, has written a series of articles in Better Investing magazine that explain the PP using Oracle Corp as an illustration.  Since Oracle has no Preferred Dividends, only four estimates are needed to calculate its PP: Sales Growth (less) Pre-Tax Profit Margin (less) Tax Rate (divided by) Shares Outstanding (equals) Earnings Per Share. 

 

I have long held that the Preferred Procedure involves too many estimates and way too much guesswork.  My earlier assessment is at:  Considering Kohl’s (KSS) and Reconsidering NAIC’s Preferred Procedure, September 14, 2006.  Oracle (ORCL) is yet another illustration that the PP is full of weak speculation.

 

(1) Final PP Result for Oracle: 11.30% Estimated EPS

 

Cy’s PP resulted in an estimated 11.30% EPS growth rate which, he says in the 3-09 issue of BI magazine, is reasonable compared to Value Line’s 11.60% estimate and the 13.40% estimate from Yahoo Finance.

 

(A) VL’s EPS estimate was 17.00% (not 11.60%) and is now 15.00% as of 2-20-09, VL’s latest report on ORCL.  By comparison, Cy’s 11.30% EPS estimate is way low.

 

(B) Cy did not compare his 11.30% against other EPS estimates: 15.00% @ FactSet CallStreet via CNNMoney; 15.00% @ S&P via BI.org; 13.82% @ Zacks.com; and 13.73% @ Reuters.com.  Again, Cy’s 11.30% estimate is low by comparison.

 

Note that these EPS estimates are from six different data sources, not just different websites, and I ALWAYS check all six for every SSG I do.  Here’s a link to how I estimate EPS; see: Estimating EPS.  Checking only two sources, as Cy did, is not sufficient to learn which are in the ball park and which may be out-of-whack.

 

(C) My method would be to stop here and use one of the following: the lowest of the six EPS estimates (13.43% currently by Thomson Reuters via YahooFinance), or the average of all six (14.33%), or the average less 1 Standard Deviation (13.65% and easily calculated by Excel).

 

I think these three choices are reasonably conservative and each of them “wrings-out” what many consider to be analyst over-optimism.  If you were still worried, the average less 2 Standard Deviations is 12.97% which I consider a useful benchmark as anything less (like Cy’s 11.3% estimate) seems unreasonably conservative to me.

 

We need some criteria to tell us what’s too high as well as what’s too low!!

 

(2) The First PP Element: 12.00% Estimated Sales Growth for ORCL

 

We can see why Cy got such a low EPS estimate by examining the separate parts of his Preferred Procedure.  The first element of Cy’s PP began with a 12.00% estimate of Sales growth for the next 5 years.

 

(A) Cy used Value Line data for his SSG and found that ORCL’s Sales growth was 9.4% over the past 10 years which, he says, was flat for the first five and increased to 22% over the following four years due to mergers.  

 

(B) Using S&P data from BetterInvesting.org, I see that ORCL’s sales have grown 9.2% over the last 10 years, 16.0% during the last 7, 22.2% during the last 5, and 24.9% over the last 3.  Mergers in 2005 and 2006 complicate using ORCL’s recent historical record to project future growth. 

 

One approach might be to use its 16.00% sales growth over the last 7 years which is the same with S&P or VL data.  Estimating 16.00% instead of Cy’s 12.00% sales growth, with no other changes in the PP, results in an estimated EPS of 15.40% instead of 11.30%.  That difference is no “small potatoes.”

 

(C) Another approach would be to check what sales growth the analysts were estimating.  Cy says his 12.00% sales estimate is supported by Morningstar’s 13.00% estimate and Value Line’s 11.5% estimate.

 

** However, Cy did not check the only other analyst that makes a Sales estimate for the next 5 years, Zacks.com which was estimating 21.92% on 11-1-08.  Three months later, on 3-18-09, its Sales estimate was down slightly to 22.19%.

 

** Morningstar’s Sales estimate for the next 5 years was way down from 13.00% to 8.00% on 3-18.

 

** Value Line does not make an explicit estimate of Sales growth and Cy implied VL’s 11.5% from its 8-22-08 dollar estimate which has dropped in VL’s next two quarterly reports to, using Cy’s method, 9.9% and 6.9%. 

 

** So, should the Sales estimate for the next 5 years be 22% (from Zacks), 16% (ORCL’s 7 year history), 12% (by Cy), 11.5% or 9.9% or 6.9% (all implied from VL), or 8% (from Morningstar)?  As you might guess, there is no “right” or “preferred” answer and your guess is as good as mine.

 

(3) The Second PP Element: 40.8% estimated PTPM for ORCL

 

The second element of the Preferred Procedure is estimating the Pre-Tax Profit Margin (PTPM) for the next 5 years.  Cy estimated 40.8% PTPM which was ORCL’s average for the past 5 years based on VL data.

 

However, S&P data shows a much different PTPM: 36.6% average for the last 5 years and 35.2% for the last 3 years.  Hemscott data from StockCentral.com shows 36.2% which is similar to S&P but again much different than VL.

 

Sooooooo, what is the best PTPM to use: a three year (35.2%) or five year average (36.6%) from S&P data, or 36.2% from Hemscott data, or 40.8% from VL data?  Again, your guess is as good as mine as there is no “right” or “preferred” answer.

 

(4) The Third PP Element: 29.00% estimated Tax Rate for Oracle

 

Here, Cy estimated 29.00% as ORCL’s Tax Rate for the next 5 years which again was based on VL’s estimate.  Cy says this seems reasonable, but does not explain that no other data source makes such an estimate and that we are pretty much stuck with relying on VL.

 

However, Value Line only makes estimates for 1500 stocks while over 7000 are included in the S&P data files and 7700 in the Hemscott data files.  Still another problem with the PP is that, for the thousands of companies not covered by VL, there is no reality check to test the reasonableness of our estimates, no benchmarks to compare against.

 

(5) The Fourth PP Element: 5,230.8 estimated Shares Outstanding

 

For the fourth and last component of the Preferred Procedure, Cy makes a big deal of not relying on VL’s estimate of 4500 M shares and instead uses 5230.8 M estimated shares outstanding in the next 5 years.  VL’s estimate is for basic shares while Cy’s estimate is for diluted shares which, he says, are what we should use for our SSGs.

 

Cy first converts VL’s basic shares to diluted and finds that to be unreasonable given ORCL’s share history and stated intent.  He appears to suggest that VL doesn’t adequately know or appreciate those factors.  Cy then decides to rely on 5230.8 M shares, some 18% more than VL, without telling us how he derived his final estimate.

 

Conclusions:

 

– By working through Oracle as an example, I’ve tried to demonstrate how the Preferred Procedure involves too many estimates and too much guesswork;

 

– The PP’s weakest link is its first component, an estimate of Sales growth for the next 5 years.  Start low and you are almost certain to wind up low;

 

– Instead of making separate estimates of the PP’s four components to derive one EPS estimate, I prefer to more directly check long-term EPS estimates from six different data sources. 

 

** I can determine what is out-of-whack and which are in the ball park.

 

** I don’t have to wonder about which of several possible Sales estimates to use, or weigh how best to estimate future PTPM and Taxes, or wrestle with converting basic shares to diluted shares, or worry about the differences between VL and S&P data.

 

– Even if you continue to use the Preferred Procedure, it’s still a good idea to make some comparisons to check the reasonableness of your estimates and, most importantly, your final result.

 

Armin

– please help me improve this SSG Blog by leaving a comment and/or taking this short and easy survey: