negative EPS growth rates

I ran into an interesting problem at work. I’m trying to rank companies in the S&P500 by relative EPS growth rates. The question is how to deal w the companies that have negative EPS. The problem here is that in some cases, the classic EPS growth rate calculation, i.e. (EPSt - EPSt-1) / EPSt-1 produces an absurd result. For example, if EPSt-1 = -10, and EPSt = -15, we get a positive growth rate: (-15 - (-10)) / (-10) = +.5! which, of course, is not right. I can think of several possible fixes. 1. In cases where there is a decrease in EPS from an already negative EPS number, I’d take the ratio using the absolute value of prior period’s EPS, in which case my calculation becomes : -15 - (-10) / abs (10) = -.5, which works. 2. Another fix might be simply to delegate ALL companies w. losses to their own class, and rank them separately from the companies that don’t have losses. 3. Another suggestion I heard is to exclude companies w. losses in both periods entirely from the rankings. Thoughts, anyone?

PEG Ratio

i don’t think that answer corresponds to the question i posted…

I found this problem at work too. I used the first method i.e. look at the sign of both current and year-ago Eps , and used a case statement for eps growth rate. current -ve , year-ago negative = just do usual for diff but take absolute value of year-ago in denominator current -ve, year-ago positive = no problem, usual works current +ve , year-ago negative = do usual but take abs value of year-ago in denominator current +ve , year-ago positive = no problem , usual works

I had read this problem online somewhere a long time back and that article has used the second option. However, I would say that you could use both the first and second option.

Thanks janakisri & idreesz. This occurs to me now, and I’m not sure but… why not just ALWAYS use ABS(EPSt-1) in the denominator? I can’t think of a case where that WOULDN’T work… and you don’t even need if / thens! maybe im missing something, but I just cant think of a situation where that would produce the wrong result… thanks again!

zoya , you are correct.

thanks, janakisri! i tested using bloomberg adjusted EPS for the S&P500 as of 12/31/09 and as of 12/31/10 and it worked.