Hi All...   I have asked similar questions but cannot get the right answer yet. I thought I got the answer... but realized it is not.   CRSP Daily data only has trading dates( no holidays, no Saturdays, No Sundays). I have other data(let me call it OTHER) that has just random calendar dates.    If I want to merge these two, I inevitably lose some non-trading days that are in OTHER. Rather losing OTHER's date, I want to keep them by matching the closest trading dates.   For example, 2019/07/14 (it is Sunday) will be matched with 2019/07/15. In other words, I do not want to lose any data in OTHER because of not matching dates.   Can anyone know this kind of code that matched the closet available date?   Tons of Thanks!   Jerry ... View more

data bmt5r; set bmt5; format Filing_Date yymmddn8.; run; data bmt5rr; /*Making connecting identifiers*/ set bmt5r; d_10 = intnx('day',Filing_Date,-10); d_9 = intnx('day',Filing_Date,-9); d_8 = intnx('day',Filing_Date,-8); d_7 = intnx('day',Filing_Date,-7); d_6 = intnx('day',Filing_Date,-6); d_5 = intnx('day',Filing_Date,-5); d_4 = intnx('day',Filing_Date,-4); d_3 = intnx('day',Filing_Date,-3); d_2 = intnx('day',Filing_Date,-2); d_1 = intnx('day',Filing_Date,-1); d_0 = intnx('day',Filing_Date,0); d1 = intnx('day',Filing_Date,1); d2 = intnx('day',Filing_Date,2); d3 = intnx('day',Filing_Date,3); d4 = intnx('day',Filing_Date,4); d5 = intnx('day',Filing_Date,5); d6 = intnx('day',Filing_Date,6); run; proc sql; /*Merging by connecting identifiersto obtain PRICE from CRSP*/ create table d_10 as select a.* , b.* from CRSP1 a inner join bmt5rr b on a.crspcusip6 = b.cusip6 and a.DATE - b.d_10 >=0 having (a.DATE - b.d_10)=min(a.DATE - b.d_10); quit; proc sql; /*Merging by connecting identifiers to obtain PRICE from CRSP*/ create table d_9 as select a.* , b.* from CRSP1 a inner join bmt5rr b on a.crspcusip6 = b.cusip6 and a.DATE - b.d_9 >=0 having (a.DATE - b.d_9)=min(a.DATE - b.d_9); quit; Hi All,   I need -10 to +60 interval from specific dates(Filling_Date), and try to obtain daily stock price returns' standard deviations using different intervals. However, my SAS code is so inefficient.   1) Is there other codes to have '-10 to +60 day's interval at once other than specifying -10, -9 to +60 each? What I did is to need to create more than 70 separate data files and these are... hassles.   2)If I have price information from -10days of Filing_Date to +60 days, How can I get variances or standard deviations? I know SAS code calculating these but it only works on a single variable but not across variables.   So long as I follow the lengthy code above, I will have D-10 to D60 and these match with Price_10 to Price60, and finally, Return_9 to Return_60 from the simple calculation price_10/Price_9 and so on.  SO, I can get returns anyway. BUT I do not have any clue to calculate Var or STD across returns.    These are quite tough... I may need other codes.  Please share your better ideas!! Thank you!!   Ah...there should be some missing values as d_# cannot be matched with DATE(CRSP trading dates). I did not put the code that solves this issue and this issue is not my worry at all.  ... View more

Hi All,   Yes. It is easy to merge the Compustat and the CRSP. What if I want something different?   I now have been trying to match these two in a more precise way. For example, if a fiscal year-end of a firm in the Compustat is 2019/11/30(Saturday), I cannot match with the exact same date from the CRSP as the CRSP only contains trading days! (no holidays, no Saturdays, no Sundays!!!)  I want to match this fiscal year-end with the closed next trading day. In the case of the above example, it should be 2019/12/02(Monday).   To achieve this matching, I referred this link: https://wrds-www.wharton.upenn.edu/pages/support/applications/event-studies/event-study-research-application/   Specifically in Step3. and this code:    proc sql; create table temp as select a.permno, b.* from input a left join caldates b on b.evtdate-a.edate>=0 group by a.edate having (b.evtdate-a.edate)=min(b.evtdate-a.edate); The last row shows that it minimizes the days between trading day and fiscal year-end(calendar day).  But... **bleep**:( That last code did not work and match quite weirdly... This codes matched fiscal year-end of 2010/10/30 to 2009/03/31 (WHAT???) and many like this. So far... my conclusion is having(~~~~) code does not work and need another. I have tried revising on b.~~~ part as well but could not get meaningful one. Since I set fiscal year-ends to be ahead of trading days, the code looks logical(yes. it should be trading-fiscal >= 0).    Cannot guess what's wrong with my code... looks all fine but wired outcomes. Please have a look and share your thoughts!   Thank you,   Jerry ... View more

  proc sql; select HAVE.FRIMID, COMP_T, CRSP_T, ASSET, RETURN from HAVE, CRSP where HAVE.date =CRSP.(date+1year or +1month) and HAVE.ID =CRSP.ID quit; This code just represents what I want...     you can see  HAVE.date =CRSP.(date+1year) What I am doing now is to calculate annual return starting from a specific date. e.g. 2019/10/31~2020/10/31. To do this, I need to have prices now and a year later. However, I do not know how to match. Since it is not just a number but year... +1 does not work.    Please help me how can I add day/month/year on my identifiers when they are calendar dates... I first made code to transform that YYMMDD to YYMM and then +1. but it sometimes does not make sense as if the case is 2013/12 then +1 is 2013/13... or 2013/10/31 became 2013/10/32... instead 2013/11/01.   Because I need to match same to same+@ instead same to same, I am struggling with this.   Please help me. I searched for other posts but haven't found similar.         ... View more

HI!!!   I found a very close answer here:  https://wrds-www.wharton.upenn.edu/pages/support/applications/event-studies/event-study-research-application/   But still... I wonder how can I get a monthly return from a specific date. e.g. from July 13th to Aug 13. If either or both dates are nontrading days, then select the next trading days. Since I have to have a month interval, the code used in the link that can only sets intervals instead of a certain period of time(like month and year) is not the right one.   I wonder how other CRSP users can get stock returns too. If you use CRSP and know how to get month-long returns please share your knowledge.   Thank you,   ... View more