1 of 2
1
Converting fiscal year to calendar date
Posted: 23 May 2015 03:40 AM   [ Ignore ]
Newbie
Rank
Total Posts:  11
Joined  2015-05-23

Hi,

I have a problem converting my fiscal year dates to calendar data.

What I wanted to do is to use the RDQE (report date of quarterly earnings) of 10 different companies as my starting point of the sample period. The professor however, told me to convert the fiscal years to calendar date and then lag the data by 3 months.

Does anybody know what he might mean and could anyone help me on this?

Thanks a lot!

Profile
 
 
Posted: 23 May 2015 07:12 AM   [ Ignore ]   [ # 1 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi,

I am not sure if I understand correctly, but it looks like you want to start measuring data for companies after they release earnings. So, you initially wanted to use RDQE (earnings reporting date).

It sounds like your professor suggests an alternative way to start the measuring period. Instead of using the actual earnings reporting date, you add 3 months to the end of the fiscal year, and assume that by then the earnings are reported.

Hope this helps,

Joost

 Signature 

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

Profile
 
 
Posted: 23 May 2015 07:29 AM   [ Ignore ]   [ # 2 ]
Newbie
Rank
Total Posts:  11
Joined  2015-05-23

Hi Joost,

Thanks for your quick response!

I think you’re right. How should I adress this in practice? Where do I find these end dates of the fiscal year for each company and how do I reconcile them? Since the professor wants me to pick 1 start date I should accept the fact that (that lag) for company A might mean 89 days and for company B 65 days?

Gr.

Anton

Profile
 
 
Posted: 23 May 2015 07:40 AM   [ Ignore ]   [ # 3 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi Anton,

Relevant variables are ‘datadate’ and ‘fyear’ in Compustat Fundamental Annual. Datadata is the last day of the month, closes to the end of the fiscal year. So, if end of fiscal year is January 2, 2015, then datadate is December 31, 2014.

Just for completeness: Date of earnings announcement are in Compustat Fundamental Quarterly, as ‘rdq’, which would be a bit more work to use (as you would need to match annual data with quarterly data). IBES datasets with the actual earnings also have the earnings announcement date and is considered more precise than rdq (but requires matching where a good chunk of observations are lost that are not covered by IBES).

If you are adding 3 months for all firms, you are basically introducing some measurement error if rdq would be more precise. The idea of the lag is usually to exclude the effect of (delayed) information about the previous year that is in the 10-K. The 3 month delay will not work (theoretically) for firms that report after 3 months. In practice, it usually doesn’t matter much. For example, some studies take both the returns over the fiscal year, as well as the returns over a 3 month shifted window, and find very similar results. So it probably won’t matter much to have a 3 month lag or a more precise firm-year specific lag based on the actual earnings announcement; and, a 3 month delay is easier to construct then using a window based on the earnings announcement date.

Best regards,

Joost

 Signature 

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

Profile
 
 
Posted: 23 May 2015 07:48 AM   [ Ignore ]   [ # 4 ]
Newbie
Rank
Total Posts:  11
Joined  2015-05-23

Thanks a lot. I’ll try to get the correct data later today and let you know whether I made progress.

If not, I hope I can ask you where it goes wrong.

Gr.

Anton

Profile
 
 
Posted: 23 May 2015 09:35 AM   [ Ignore ]   [ # 5 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi Anton,

Sure; let me know how it goes.

Best,

Joost

 Signature 

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

Profile
 
 
Posted: 23 May 2015 10:42 AM   [ Ignore ]   [ # 6 ]
Newbie
Rank
Total Posts:  11
Joined  2015-05-23

Hi Joost,

I’ve been working with compustat and found that 9 out of 10 companies have data date 31/12/2004.

I’m thinking about erasing the 10th company which has data date 30/09/2004. So then I can use 1/04/2005 as the start of the sample period with the assumption that a 3 month period is enough for earnings reporting right?

Gr.

Anton

Profile
 
 
Posted: 23 May 2015 01:01 PM   [ Ignore ]   [ # 7 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi Anton,

The typical thing is to add 3 months to the end of fiscal year, regardless when it ends. That way, the ‘benchmark’ of starting the measuring period is relative to the firm’s reporting. That way you don’t lose observations because of differences in fiscal year end.

Best,

Joost

 

 Signature 

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

Profile
 
 
Posted: 23 May 2015 01:28 PM   [ Ignore ]   [ # 8 ]
Newbie
Rank
Total Posts:  11
Joined  2015-05-23

Hi Joost

I understand that. The professor however asked me to choose only 1 starting point of the sample period and not use different starting points for different companies. Knowing that, my solution is the consequence right?

I’m now comparing the returns during my sample period 01/04/2005 - 01/04/2015 and am wondering how I should cope with the dividends paid between that period. Is there a compustat variable that can makes this easier?

Thanks,

Anton

Profile
 
 
Posted: 23 May 2015 03:27 PM   [ Ignore ]   [ # 9 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi,

Ok smile

I would think it isn’t possible to compute stock return using Compustat reliably, as you want to measure it 3 months into the next fiscal year.

Return data in CRSP are most reliable, these are corrected for dividends, stock splits, etc. That requires a match between Compustat and CRSP, for which CCM Linktable is the best way to do it.

Best regards,

Joost

 Signature 

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

Profile
 
 
Posted: 24 May 2015 03:59 AM   [ Ignore ]   [ # 10 ]
Newbie
Rank
Total Posts:  11
Joined  2015-05-23

Hi I’m now searching for monthly stock files in period 03/2005 - 04/2015 in CRSP.  What variables would get me those corrected return data?

I’ve also never heard of matching these two datasets. Could u explain how that works?

Gr. Anton

Profile
 
 
Posted: 24 May 2015 04:22 AM   [ Ignore ]   [ # 11 ]
Newbie
Rank
Total Posts:  11
Joined  2015-05-23

I just found adjusted close prices on yahoo finance so I think I got the data I wanted.

Profile
 
 
Posted: 24 May 2015 06:23 AM   [ Ignore ]   [ # 12 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi Anton,

I would think that adjusted stock prices are not corrected for cash dividends. In CRSP MSF, the variable ‘RET’ is the monthly return. CRSP uses ‘permno’ as the firm identifier’, and Compustat uses ‘gvkey’. CCM linktable will give you the permno for a given gvkey on a given date.

Hope this helps,

Joost

 Signature 

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

Profile
 
 
Posted: 24 May 2015 06:25 AM   [ Ignore ]   [ # 13 ]
Newbie
Rank
Total Posts:  11
Joined  2015-05-23

https://help.yahoo.com/kb/finance/historical-prices-sln2311.html

Hi I read this help page and I think they also use CRSP data?

Profile
 
 
Posted: 24 May 2015 06:30 AM   [ Ignore ]   [ # 14 ]
Newbie
Rank
Total Posts:  11
Joined  2015-05-23

I’ll compare them in order to see if they’re the same.

Profile
 
 
Posted: 24 May 2015 06:35 AM   [ Ignore ]   [ # 15 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi Anton,

Interesting—the ‘adjusted close’ description of yahoo indeed indicates that cash dividends are taken into account.

For completeness (and for other that may stumble across this page)—with CCM and CRSP MSF you probably will get a better quality match (in general, stock ticker symbols are not a good variable to match on as these ticker symbols get ‘reissued’/recycled over time).

Best regards,

Joost

 Signature 

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

Profile
 
 
   
1 of 2
1