Message Font: Serif | Sans-Serif
 
No. of Recommendations: 12
Check it out folks:

Anyone contemplating actually using the MFI site to make investing decisions really ought to spend twenty minutes at the SEC attempting to verify its earnings yield figures. If you're doing it not too long after now (1/6/2006), I suggest you start with the following, all of which are on today's MFI Top 100 (mkt cap $100m): DHB, RAI, IVII, KFI, HNR, CALL, INTL, KG, ULCM, JOR, INTX, BSX, INSP, HSII, or TOD. By my calculations (about which, more in a moment), these all have earnings yields at least 500 basis points less than the MFI site indicates. The first four are off by over 1000bps.

I got the Little Book for xmas and sat down to it with great expectation. I got up from it, shortly thereafter, with the same. People grouse about backtesting, but that which Greenblatt did sounded nice and rigorous. Past performance is no guarantee, I know, but history definitely rhymes, if you ask me, so: I went straight to Stock Investor Pro (a stock screening program from the American Association of Individual Investors) and set about trying to duplicate the MFI site's results....

The site says to see the book for the formulas it uses; the book says it uses EBIT/EV for earnings yield, and ROC, which it defines and discusses in such a way as to suggest that all they're backing out of the denominator is goodwill and intangibles, not cash.

I used the universe of 3332 stocks SIP dragged up as having mkt cap of $100m or better and not being ADRs or in the financial or utilities sectors. SIP has lots of figures, including ROE and ROA, but it does not have as a regular data point EV, ROC, or ROIC: you have to construct them from other fields they do have. So I tried all sorts of things. EBIT/EV is reasonably straight-forward, but I computed ROC in various ways and tried various forms of ROIC. I tried using ROA (as the book suggests) and ROE. I tried constructing fwd ROIC using fwd consensus earnings estimates, extrapolating from mrq op and net margins, and using a 35% tax rate. Once I'd added the extra fields to the database, I exported it to an excel spreadsheet and ranked, added, and ranked agin', as the Little Book says. I even tried ranking the sum of the rankings of various approaches. In each case, I looked at the percentage of the site's Top 100 that landed in my Top 100 and in my top decile. The best I could manage was about 50% by the former measure and 75% by the latter: good enough for an active investor like myself to proceed on, but still awfully far from the site's results. This bothered me because the metrics involved aren't exactly complicated. I thought I ought to be able to achieve at least 85% overlap, and it bothered me that I couldn't.

So I looked at the site more carefully and noticed, finally, that it posts a specific earnings yield figure for the stocks it selects. I compiled a portfolio of the site's top 100 in SIP and then compared SIP's figure for EBIT/EV to the site's. Yikes! The list at the beginning of this note arose from this comparison, and I don't vouch for it: SIP's data (which come from Thomson/FirstCall) are sometimes off, and some of the co.s above may well be properly represented on the MFI site. But I really don't think they all are because yesterday I completely failed in a careful effort to verify their figure for Reynolds American (RAI), which they say has an earnings yield of 28%.

EBIT/EV is not a complicated metric, nor is it one about the computation of which folks much disagree. I went to the SEC and looked at the past four quarters, and as far as I can tell, RAI has an earnings yield of 8-10%, depending in part on whether you look at the last four quarters or annualize the most recent quarter. This is no small discrepancy, 2000 basis points, so I googled up some reviews to see if anyone else had any problems with the site. I came up with a review at Amazon (from 12/12/2005) by one “Boston Quant” that I've copied below. The situation he describes with Callwave is essentially the same as I found with Reynolds yesterday: there's just NO WAY to back into the site's figure for EBIT/EV, which one ought to be able to do drunk and blindfolded. I've copied the review because his initial comments are very well-taken.

I think a great hue and cry should be raised about this. I believe in what the book says and I found the studies impressive, but the site's opacity and apparent stark contradiction of SEC filings are disturbing, especially insofar as it presents itself as a value-investing service for the Everyman that is to be followed by rote.

<<To make such claims of outperformance without verifiable proof is irresponsible. I was looking to backtest this strategy, but after reading the book and checking the website I am unable to back into either EBIT/EV (which should be very easy) or ROC (also very simple). There is simply not enough detail to match the calculations on the site (which is based on the book). They claim to be using Compustat's point in time database which I have access to. Are they using twelve month trailing numbers, projected numbers, most recent quarter, or something else? Perhaps there is a transform applied to the ratios, we just do not know. There are not answers to these questions in the book anywhere.

Example:
Take Callwave (CALL) today (12/12/05) which is on the sites list as being recommended today. CALL has a Earnings Yield of 16% as calculated by Greenblatt's site. The little book describes EV = Market Value of Equity Including preferred + Net interest bearing debt. Though most in the industry also subtract off cash, it turns out that I am unable to reproduce their numbers either way.

Since CALL does not have debt or preferred stock or minority interest then EV should = Market Value = 104.64. To get an Earnings Yield of 16% EBIT must be 16.74.(If you subtract off cash EV = 104.64-61.9 = 42.74, so for a 16% EY EBIT must = 6.84)

According to COMPUSTAT (period ending 9/05):
Trailing 12 months EBIT = 7.0 (does the author adjust for non-recurring?)
Most recent Quarterly EBIT = 1.8
Most recent year-end EBIT = 9.5

Added Notes: The book's web site just mentioned this review. Instead of responding to this note why not just provide specific step by step calculations. Why not start with CALL and show what numbers are used to make you calculations? I would think this is in the interest of all, especially if this 'magic' formula is so easy. Then others like myself with equally high quality data can validate your results.

Skip this one and read Benjamin Graham.>>
Print the post  

Announcements

What was Your Dumbest Investment?
Share it with us -- and learn from others' stories of flubs.
When Life Gives You Lemons
We all have had hardships and made poor decisions. The important thing is how we respond and grow. Read the story of a Fool who started from nothing, and looks to gain everything.
Contact Us
Contact Customer Service and other Fool departments here.
Work for Fools?
Winner of the Washingtonian great places to work, and Glassdoor #1 Company to Work For 2015! Have access to all of TMF's online and email products for FREE, and be paid for your contributions to TMF! Click the link and start your Fool career.