Thanks a lot for the quick reply! To answer your questions, 1), yes, indeed. My data is in a SAS data set. 2), My date values are SAS data values with a format such as YYMMDDn8. (Datadate in Compustat North America Quarterly dataset) 3), The variables I have on hand is for quarter value already. Therefore, I do not need to compute the sales or costs for each quarter again. All I need to do is to use these quarterly values to computer firm-level cost stickiness measure following the Weiss (2010). 4), unfortunately, it is not. What I observed is that for some firm, some year, there are differences between how quarter end is recorded. Also, some firms in certain years may have quarterly data missing (only contain three quarters) for example, the first a couple of rows I have looks like the following: GVKEY DataDate Calendar Data Year and Quarter Fiscal Date Year and Quarter 001003 19840331 1984Q1 1984Q1 001003 19840630 1984Q2 1984Q2 001003 19840930 1984Q3 1984Q3 001003 19860131 1985Q4 1985Q4 001003 19860430 1986Q1 1986Q1 001003 19860731 1986Q2 1986Q2 001003 19861031 1986Q3 1986Q3 ------ ----- ----- --------- As you can see, even within the same year, there are situations like missing quarters and different datadate for the same quarter (1984 vs 1986). However. I would say that at least 80% of my observations follows regular pattern as 0331 for first calendar quarter, 0630 for the second calendar quarter and so on. The rest of obs seems to follow the pattern like 0430, 0731, 1031 and 0131 for most of times. (I have 361,688 obs in total). Again, thanks a lot for your time and help and hopefully there is a way to resolve this problem here. Thanks!
... View more