Yes.
FYI - Sales Per Employee is not working.
Not sure if this has been answered here - any update on International Stocks? Thanks.
These formulas are not working with FactSet:
Pr2CashFlTTM < ROI%TTM
Pr2CashFlTTM < ROE%TTM
They had been working with Compustat. Can someone look into this please? Thanks.
Yuval:
Fair enough about COGS when it comes to financials. But that still doesn’t explain why we can’t get the following ratios for financial stocks with Factset:
EV/Sales
ROI%
P.S. Let’s not get hung up on Cash “equivalents”. I can work around it if you’ll at least give us something to work with. Thanks.
We will try to supply EV for all firms, financial or not, but at the moment I’m not sure we’ll be able to. EV is basically market cap plus debt minus cash and equivalents. So, obviously, it doesn’t apply to financial firms since for them cash is a liability and issuing debt is how they operate. As you can read in almost any book on valuation, EV is useless for firms in the financial sector–it’s an irrelevant metric and shouldn’t be considered. In addition, we’re having trouble coming up with cash and equivalents. I’m not sure if we can work this out given FactSet’s data.
As for ROI%, I’m not seeing a problem here. We calculate ROI%TTM in house using numbers that both Compustat and FactSet provide. The inputs are net income, interest expense, total equity, and total debt, and none of those is a problem for financial sector firms.
I would strongly suggest, however, that you use ROA%TTM for financial sector firms instead of ROI%TTM. To get ROI%TTM we add back interest expense to net income, and that’s not appropriate for firms in the financial sector. In addition, the denominator in ROA is total assets while in ROI it’s equity plus debt, so you’re deducting non-debt liabilities from ROI but including them for ROA. Non-debt liabilities are essentially the “moat” for insurance companies, so they should not be excluded.
Thanks for bringing this to our attention. I’ve added it to the Trello board and we’ll get working on a fix.
It seems to be working for me. Can you please elaborate?
Try this:
SalesPerEmpTTM>100000
Works in Compustat, not in Factset. And it’s not a function of the value, it doesn’t work for any amount entered.
I did a test on the production/Compustat site using the screener to return the standard valuation factors for the stocks in the SP500. Then I did the same thing on the beta site to get the FactSet data. I listed some potential issues below.
-
ShowVar(@PEExclXorPTM, PEExclXorPTM) gave error in FactSet. Error was → ERROR: UnhandledExceptionHandler: EXCEPTION_ACCESS_VIOLATION
-
There seems to be an issue with the cashflow factors:
Pr2FrCashFlQ has 214 N/A values in FactSet vs 157 in Compustat.
Pr2CashFlQ, Pr2CashFlTTM, Pr2CashFlPTM - these factors returned N/A for almost all the Sp500 tickers. -
I compared the CS value to the FS value for each factor for each stock. Then I looked to see which factors had a high % where the CS and FS values were different by 100% or more. I picked 100% at random to represent a large difference between CS and FS which seems to indicate that there could be an issue in the ratio calculation.
These factors had many tickers with a big difference between the CS and FS value and a high avg % diff → PEHigh, PERelative, PEG
These were not as far off, but still looked sketchy → EarnYield, PEGY, PEExclXorTTM
You should consider creating a new forum like “FactSet issues” rather than having everybody log them in this one long thread. It would make it a lot easier for you guys to track them.
CSvsFactSet_ValueFactors.xlsx (382 KB)
I don’t know what is going on there on your end. It works fine for me. Obviously not for 100000, but for smaller numbers I get just about every stock just fine, with numbers pretty close to Compustat’s.
Thanks for calling these issues to our attention! We’ll get working on fixing them ASAP.
Hi Yuval I’m not sure what you’re seeing. If the only rule is SalesPerEmpTTM>100000 you get the following results:
Compustat: 4,409 matches
Factset: 6 matches
See attached. And once I’ve already created some useful rules, I get no matches with Factset. Can you look into this again please? Thanks.
For SalesPerEmpTTM, the units on the beta site are different;
Ticker(“T”)
Beta:compustat = 731.4972
Beta:factset = 736.8496
current:compustat=723,062.9451
Walter
Thanks for clearing that up, Walter. I was running SalesPerEmpTTM on the beta site using both Compustat and FactSet and seeing very little difference. I didn’t compare it to the production site, which I should have done. We’ll add that to the issues to be fixed.
Why not? Doesn’t Factset have ~800 line items including many industry specific ones? Why can’t users take the underlying line items and roll our own custom ratios?
I concur that these factors are not in the FactSet data set:
Pr2CashFlTTM
Pr2CashFlQ
Inspired by Walter’s answer to mm’s question involving SalesPerEmpTTM, what other unit changes may have been made during the transition from the current CompuStat w/preliminary data site to the CompuStat w/preliminary data Beta site?
This could be a HUGE cause of inconsistent results between the two sites.
Thank you.
Hugh
Hugh, this issue is strictly isolated to changes in our treatment of the ‘Number of Employees’ line item. Originally, this Compustat item was loaded directly, in thousands. Since FactSet gives us the exact number, the code that dealt with Compustat data was changed to multiply it by 1000. The change broke IncPerEmpQ, IncPerEmpTTM, SalesPerEmpQ, and SalesPerEmpTTM, since their code was not revised accordingly.
Because our contract with FactSet allows us to expose only a certain number of line items. We’re working on exposing as many as we can . . .