While we're at it, will someone please show the Yale professors how to handle negative earnings so that they don't exclude these along with so many others that are not in the standard and poor earnings series he used, which used to report yearly six months after the end of the year if at all, perhaps using earnings price ratio?

Phil McDonnell writes: 

The use of PE ratios is very bad for historical studies. A simple mental experiment shows why. Suppose earnings result in a current PE ratio of 10. If earnings are cut in half the new PE will be 20, other things being equal. As earnings get smaller and smaller the PE rises to near infinity as earnngs approach zero.

When earnings go negative the PE suddenly flips from very large positive to negative. This causes the variable to go from a monotonic increasing to sudden negative.

At this point the Yale professor had no choice but to eliminate any such data.Using a slightly different variable of E/P eliminates the problem. It becomes a monotonic variable. When the E factor gets very small the E/P variable goes smaller. When E starts to go negative the variable goes negative with no loss of continuity. With this change there is no need to throw data out.





Speak your mind


Resources & Links