XOR - Not What You Think

Research
1

I’m reopening another thread here based on something that came up in a different conversation.

XOR is the p123 acronym for extraordinary items (on the income statement).

Many, perhaps most, investor believe these should be excluded from from analysis, from valuations, from growth rate computations, etc. And they would have powerful support. In this belief tracing back to Graham & Dodd, who devote a substantial portion of their classic text to the analyst’s recasting the reported financials into alternatives that make economic sense. This is also consistent with the idea of the DDM foundation as I’ve been discussing it in the on-line seminar. What’s the point of thinking in terms of E if the number isn’t meaningful as a base for assessing the present relationship top P or a base upon which G can follow.

Problem - there’s one group of people that is not on board with this at all. This consists of those who set the accounting standards that govern the way companies must report their results. They are not investors and often are unaware of and/or insensitive to the needs of investors. they are disclosers whose main priority is to disclose the past accounting period in as complete a way as possible. As disclosers, they get antsy and worse when they think it’s too easy for readers of financial information to miss seeing things they feel should not be swept under the rug. Their belief is that when the financial impact of unusual developments is reported as far “down” the income statement as the extraordinary line, that many who see statements will not notice them or pay less attention to them. Remember, accounting statement are designed to say “this is what happened.” They are not interested in saying “This is what’s meaningful to you, who must invest for and make reasonable assumptions about the future.” Instead, they tell us “Reporting rules are our jurisdiction so keep your noses out of it. If you can’t figure out how ot make our statements work for you, go read Graham and Dodd, etc.”

Traditionally, this wasn’t the end of the world since we could easily recognize and exclude extraordinary items.

But over time, the accounting czars felt companies were looking too good, allowing past problems to be papered over by having the financial consequences at the bottom of the income statements. So they kept tweaking the rules to limit what kinds of things could be reported as extraordinary. And they tweaked some more, and then more, and more and more and more . . .

Well, they got what they wanted: Lots of garbage numbers in the main body of the income statement, where everybody could see it more easily. Compustat puts all that junk in the pretax SpcItems. (Unfortunately, never bothered forcing the companies to separately disclose the text implications of these items so they tend to be done in text commentary or not at all.

It’s been a big success for the accounting profession. The income statements are often pig sties containing lots of garbage up high in the reports.

For us, it’s a mess. It’s much harder for us to get to the numbers we really want or need. We have to go through the trouble of subtracting Spcitems (fortunately, Compustat makes this easier to do than do its rivals). But even Compustat can’t solve the tax issue so we have to content ourselves to use pretax numbers or guesstimate a tax impact.

The biggest mess, however, is in terms of clarity. By pulling so many items that strike the reasonable investor as being extraordinary, while not actually repealing the extraordinary line item, they allow many to inadvertently mis-specify their models. Many believe they are working with clean numbers when they exclude XOR. Nothing could be further from the truth:

In All Fundamentals today, 13% of companies have extraordinary items but 64% have SpclItems.

In PRussell3000, 11% of companies have extraordinary items while 74% have SpcItems.

In the SP 500, 14% of companies have extraordinary items while 84% have SpcItems.

So obviously, you’re not analytically getting good numbers when you work with Net Income or EPS. You may think that excluding extraordinary items is protecting ypu from this. It’s not.

If you use a BfrXOR item, you’re really accomplishing little; you’re just adding a teeny bit to what you didn’t realize you have been using all along, and it’s not clear that the extra difference is meaningful. If you need clean numbers, or it you want good visibility on dirt (which may be valid if you model around corporate changes and/or related noise), you need to be focusing manily on SpcItems and building in some tax assumptions (the easy way to presume a generic 35% rate for earnings ex ScpItems).

2

Thanks for the summary.

  1. Do you have absolute Dollar amounts for the comparisons of extraordinary items vs. SpcItems at hand as well as its source?

  2. So in your opinion, is it correct to imply that the following provides the clearest picture for pre tax earnings?
    For example, for TTM-numbers: “IncBTaxTTM - SpcItemsTTM”

3

Marc,

Thank you!

So I assume EBITDA is also a mess?

Much appreciated.

Regards,

Jim

4

I’m getting ready to go out of townj for a few days. I’ll work this up probably next week.

Yes. Bear in mind, too, that it wopjuld include something we call Other Operating. It, too, can bonce around, but it is not quite as abberrant as SclItems – so you copuld choosde to go either way on it.

Jim,

Not sure when you came on board at p123. When we had Reuters dat, EBITDA was pretty dirty becasuer companies put a lot of the gsarbage up above the operating line. One of the things that made Compustat appealking to us was that they do some conversion work on the statements (as if their database designers had a Granham and Dodd book open on their desks) one result being they pulled that stuff out from operating and put it into the separate SpcItem categopry they created. This is one example of why Compustat is preferred by modelers like us, but does not lend itself to people checking numberd agaisnt 10-Ks. (Expect, too, to see our numbers often dioffer from what you’ll see on other web sites that use different data sources.)

5

What about EPSActual? I haven’t used this myself, but I had assumed that the adjustments that analysts make to reported earnings would smooth out the special items. I’d love to know what EPSActual really takes into account, as it sometimes differs wildly from EPSExclXor.

6

Also, sometimes the special items are so huge. Microsoft has net income of 11.4B and special items of -8.6B. Does that mean their net income should really be 11.4 + 8.6 multiplied by their tax rate? BP reported a net loss of -6.5B and special items of -13.5B, which would mean that we should consider them as having made a nice profit last year. Am I getting this right?

7

These numbers, from the Estimates database as opposed to the Fundamentals database, are meant to be apples-to-apples with analyst forecasts, and hence, cleaner.

8

Yes, you are getting it right. You have to approximate the real number since the tax rates of the unusual items can be very different from the tax rates on regular business items, but yes, you are getting it right.

This can have big impact on which companies make it into the tops (or bottoms, for shorting purposes) of some ranking systems/factors! That’s why my main method of evaluating models is to look at samplings of companies that pass (using the p123 panels) – to see if I’m getting too may of the kinds of companies I don’t really want to be getting (i.e. to check for model mis-specification).

NOTE: This does not necessarily mean you have to reject unusuals – sometimes, you can choose to want them (different from getting them without knowing it). The market is not efficient (many would argue that, but market inefficiency is what lets us do our thing on p123).

9

Are special items included in Operating Income?

10

For us, the answe is “no.”

In 10-Ks and 10-Qs, and for most financial databases, the answer would be “yes.”

But we use Compustat, which was designed based on a level of analytic expertis that made them willing to do the sorts of things Graham and Dodd say analysts should do. So they re-shuffle the income statement in a way that cleans up operating income and collecting unusuals in the Special Items category that is excluded from operating income.

11

Marc, so just to check, if I compare Alcoa’s OpInc in their filings to P123 data that their will be differences? I use Alcoa because it seems to have sizable SpcItems every year. I have not done this yet but it sounds like the answer is yes, there will be differences. (And what is interesting is that Apple has no SpcItems for the last 10 years in P123).

also, in P123, what line item is for extraordinary items (as separate from SpcItems). You separated them out in your first posting above.

This discussion is very illuminating.

12

David,

It’s possible Alcoa’s 10K oper egs will differ. It depends on whether the company chooses to put special items there. Many do nowadays (perhpas most) but I can’t say it’s 100%.

I understand it may be a bit uncomfortalbe for some to knmoe that ojur numbers often don’t mastch up with 10-Ks. But the benefit is better modelling. The differences are the kinks of things we should be doing anyway if we really want to work with numbers that are as analytically valid as feasible. (S&P has another database, Capital IQ, which matches more closely to the filings – that’s the database of choice for those whose needs are referentical, rather than modeling rerlated.)

I explain more of this in the data while paer I wrote (in the Help section) to explain why we switched fdrom Reuters data to Compustat.