Mylan (MYL) free cash flow is greater than its operating cash flow for TTM

Since the difference between the two is essentially CapEx, how can a company have negative CapEx?
This seems to be the only stock in the S&P500 with this situation.
Any ideas of a cause?
OperCashFlttm = 1,680
FCFttm = 2,089

Something funky with change in working capital?

In the PRussell 1000, these other companies have a similar situation: UTHR, ARW, JLL and NBIX.

I have seen this a couple of times in the past as well but could never figure out what was happening.

Anyone have an idea? thanks.

P123 does FCF as;

Cash flow from Operations - CapEx - DivPaid

Which is neither FCF to Firm nor FCF to Equity. Just sayin’.

Hi Walter, yes, I think I knew that distinction about to the firm or equity.
It still begs the question because neither CapEx nor Dividends can be negative numbers.
I have a feeling it is just a short term anomaly of how the data is reported into the P123 database and then updated. Just a guess.

Check out the ANN numbers for MYL. They add correctly. There’s something different about the TTM data, I guess.

It looks like there was a recent quarterly update which caused a drop in OperCashFlTTM and no change in DivPaidTTM, CapExTTM, or FCFTTM. Maybe this is related to the cashflow items issue discussed in https://www.portfolio123.com/mvnforum/viewthread_thread,11721

Dunno.

Ah, thanks. That might be it.

Here is the Cash Flow statement as it shows up on in “Panels”.

Apparently, FCF is falling back to the 2018 annual figure. Why? Because “dividends paid” ttm is N/A. And dividends is part of the FCF calculation. (FCF = “Operating Cash Flow” - “CapEx” - “Dividends Paid”)


yes, I saw that as well. thanks.

I wonder if use of TTM is ultimately less accurate than the annual data (ANN). I don’t know under what conditions P123 will default to ANN when it has insufficient data for TTM so I can’t anticipate this for each rule and write an EVAL statement or whatever. Maybe this is just an extreme corner case.

I wouldn’t make that assumption.

I can guarantee that dividends can be negative in the CompuStat database. We’ve seen it before. It comes into play when a company is receiving payments from a unconsolidated subsidiary. CompuStat normalizes that by putting the receipts into dividends paid/(received).

We’ve also absolutely seen negative revenues (!!!) when a company is paying more above the gross than it’s keeping. (I would think that the CXOs would recall from their MBA days that this is a cause for immediate shutdown of operations, but I’m no CXO.)

Can it happen with CapEx? The S&P documentation says that they’ve encountered capex lines in the cash flow statements that are net of the sale of property, plant and equipment, and are therefore negative. It notes that they will then report it in PPE, but…

A quick check of CapExQ<0 and EPSExclXorQ>0 yields 46 companies. The largest of these, W.P. Carey (WPC) is a REIT. It has a negative capex of -0.388. That figure does not appear in the latest Q, so they’re definitely combining some of the other investing cash flow figures in the capex line.

I am no expert in REIT accounting, but it seems clear to me that they’re at least including “Return of capital from equity method investments” and “Capital contributions to equity method investments” in capex. CapEx then becomes slightly positive, but still not equal to the reported CapEx number in our system. I’m not diving into the footnotes. :slight_smile:

As I will tell people, 95% of the time financial data does exactly what you expect. We don’t spend our time with that 95%.

Anyway, for MYL, capex is nice and negative. DivPaid is NA when tested with the KEEPNA parameter.

Paul, ok. I appreciate the background. I learned something today.
I exclude REITs and Financial sector companies in my universe and just focus on operating companies.
But it looks like other corner case issues (the 5%) can pop up.

I have been learning that building a model on free cash flow is interesting.

More to do!
thanks.

For MYL, both CapEx(0,TTM,KeepNA) and CapEx(1,TTM,KeepNA) are positive.

What’s happening is since DivPaid(0,TTM,KeepNA) is NA, the FCF expression is falling back to the prior period.

Given FCF=OperCashFl-CapEx-DivPaid

Without Fallback,
OperCashFl(0,TTM,KeepNA)-CapEx(0,TTM,KeepNA)-DivPaid(0,TTM,KeepNA) = 1680.2 - 274.5 - NA = NA and FCFTTM(0,TTM,KeepNA)=NA // match!
OperCashFl(1,TTM,KeepNA)-CapEx(1,TTM,KeepNA)-DivPaid(1,TTM,KeepNA) = 2341.7 - 252.1 - 0 = 2089.6 and FCFTTM(1,TTM,KeepNA)=2089.6 // match!

With Fallback,
OperCashFl(0,TTM)-CapEx(0,TTM)-DivPaid(0,TTM) = 1680.2 - 274.5 - 0 = 1405.7 and FCFTTM(0,TTM)=2089.6 // mismatch!

I think I’m going to start using my own FCF expression. Perhaps something like;

isNA(OperCashFl(0,TTM,KeepNA)-CapEx(0,TTM,KeepNA)-isNA(DivPaid(0,TTM,KeepNA),isNA(DivPaid(1,TTM,KeepNA),NA)),OperCashFl(1,TTM,KeepNA)-CapEx(1,TTM,KeepNA)-isNA(DivPaid(1,TTM,KeepNA)))

If the current period DivPaid is NA, it will use the prior period value. If that’s NA or if the current period OpercashFl or CapEx are NA, then it will fallback all the component values.

Nice work. thank you.
DivPaid tends to either stay the same or go up, decreasing FCF. But, it is not a big change usually from period to period so I think what you are doing is reasonable.
The other thing I was looking at was using an SP900 (SP400+SP500) universe instead of PRussel1000. Although S&P does some interesting things to build these two universes, they seem to be a slightly higher quality, in terms of data availability, than the straight top 1000 by marketcap.
Although Mylan is an SP500 stock, I think it is a rare case.
Thanks for the examples, though, I learned something from your code. May come in handy for some other things.