expected squared error in mean forecast
Posted: 12 February 2013 10:47 AM   [ Ignore ]
Total Posts:  22
Joined  2013-01-15

Hi, all

some of you probably computed this statistics, if so, could you let me know

if I should winsorize the data

variables I computed

1. (mean analyst forecast - actual EPS ) **2 (i.e. second power)

2. forecast dispersion, which is sample variance of analyst forecast

when I run proc means in sas for the two variables, I go maxmium greater than 200,000.

is this an error on my part OR you have seen such large values, and dealt with it using winsorization. if so, which one to winsorize?


Posted: 12 February 2013 09:38 PM   [ Ignore ]   [ # 1 ]
Total Posts:  901
Joined  2011-09-19

hi Rochelle,

I would winsorize the first variable before taking its power. Did you consider scaling this variable? All things equal, a higher stock price will give a higher EPS and that will boost the metrics (so instead of capturing - I suppose - uncertainty, this gets mixed with firm size as larger firms have higher stock prices).

200,000 seems like a very large number; if you would look at that obs and recalculate it should be clear if there were errors.

Hope this helps (a little),



To reply/post new questions: Please use the group WRDS/SAS on Google Groups! http://groups.google.com/d/forum/wrdssas