SAS pro/spinoff
Program20140722
SDC_GVKEY9b
Target名称里本身有提示:
Sunbelt Nursery Grp(Pier 1) -- PIER 1 IMPORTS INC/DE
Westmark Intl(Squibb Corp) -- SQUIBB CORP
看时间段:
Alliant Techsystems -- HONEYWELL INC (1950-98)另一个obs则是2011年后才有数据的
Cytec Industries -- AMERICAN CYANAMID CO (spinoff发生于1994,母公司信息截至1993,另一个同名TIC,在08才开始有数据
National Gaming Corp -- HFS INC (94年的spinoff,另一个公司的信息始于2010年)
Pittston Svcs Grp-Pittston Bur -- PITTSTON CO-MINERALS GROUP (PZM) 查看了他们的COMPUSTAT,AT(Assets)规模相当,PZM比PZX成立时间稍微早了几年,所以选了PZM。96年spinoff, 98年PZM,PZX均再无信息更新。
Choice Hotels Holdings Inc -- MANOR CARE INC-OLD(MNR.1)
spinoff发生于1996,MNR.1止于1998,但是98之前,MNR.1比HCR体量要大。
ManorCare Health Services -- MANOR CARE INC-OLD(MNR.1)
同上。
Talbert Medical Management -- PACIFICARE HEALTH SYSTEMS (PHS)spinoff于1997,但 另一个止于1996.
Lakes Gaming Inc -- HILTON WORLDWIDE HOLDINGS (感觉还是HLT靠谱),同理,另一个TIC的截至数据期,发生于spinoff之前。
Key3Media Group Inc -- ZIFF-DAVIS INC(ZD比ZDZ资产大很多)。但他们的信息截止期都早于spinoff时间。
PS:
结果似乎不甚理想。这样merge得到的match有711个。但是之前从cusip=>CRSP=>linktable=>gvkey的路线,得到765个match。
Showing posts with label Diary. Show all posts
Showing posts with label Diary. Show all posts
Tuesday, July 22, 2014
SAS - Control Group
SAS - Control Group
好帖: http://www2.sas.com/proceedings/sugi29/173-29.pdf
1)两个条件下找exactly
2)配对后,随机的找2个
3)条件放宽,比如年龄+1或-1(range)
4)Keep the matches for the low frequency control subjects first.(也是一种法子,然后再随机)
5)Macro
好帖: http://www2.sas.com/proceedings/sugi29/173-29.pdf
1)两个条件下找exactly
2)配对后,随机的找2个
3)条件放宽,比如年龄+1或-1(range)
4)Keep the matches for the low frequency control subjects first.(也是一种法子,然后再随机)
5)Macro
Monday, July 21, 2014
Diary
1)SDC和IBES的merge也不知道对不对。用的是CUSIP。
2)小马说,他用PERMCO的aggregation来merge gvkey,而不是permno。因为很多公司都是major interest,所以这样计算比较正确。
接下来做这几件事:
1)找control group。按照2 digit SIC, size, MB
2)分别在restructuring之前和之后。两种time trend
3)找breakup子公司之后的disclosure情况,以及他们的control group
2)小马说,他用PERMCO的aggregation来merge gvkey,而不是permno。因为很多公司都是major interest,所以这样计算比较正确。
接下来做这几件事:
1)找control group。按照2 digit SIC, size, MB
2)分别在restructuring之前和之后。两种time trend
3)找breakup子公司之后的disclosure情况,以及他们的control group
Friday, July 18, 2014
Diary
I/B/E/S 有些obs有点诡异。
看Avercrombie & Fitch Co.的management forecasts.同一天,差1个半小时,有两次forecasts,而且都是EPS,数字相同,所以I/B/E/S记录了两次。
除了Announcement时间不同以外,Mean_at_date应该是那个时间点Analysts的estimate mean,也不同。好奇怪的是,难道一个半小时内,analyst预期就调整了吗?
还有,从时间顺序上看,management给了一个较高的期望区间,所以analyst forecasts貌似是往上走了。这科学吗?
所以区别在于:
file path: SAS Pro\spinoff\Program20140718
proc sort data=sdc_Ibes2 noduplicate out=try dupout=dups;
by Target__CUSIP parent_SDCCUSIP
__Date__Effective IBES_ticker
Announce_dt Prd_yr Prd_mon Periodicity Measure
Val_1 Val_2;
run;
proc sort data=sdc_Ibes2 noduplicate out=try dupout=dups;
by Target__CUSIP parent_SDCCUSIP
__Date__Effective IBES_ticker
IBES_Announce_dt Prd_yr Prd_mon Periodicity Measure
Val_1 Val_2;
run;
第一种方式中,没有duplicates。
第二种方式中,会有duplicates。
看Avercrombie & Fitch Co.的management forecasts.同一天,差1个半小时,有两次forecasts,而且都是EPS,数字相同,所以I/B/E/S记录了两次。
除了Announcement时间不同以外,Mean_at_date应该是那个时间点Analysts的estimate mean,也不同。好奇怪的是,难道一个半小时内,analyst预期就调整了吗?
还有,从时间顺序上看,management给了一个较高的期望区间,所以analyst forecasts貌似是往上走了。这科学吗?
所以区别在于:
file path: SAS Pro\spinoff\Program20140718
proc sort data=sdc_Ibes2 noduplicate out=try dupout=dups;
by Target__CUSIP parent_SDCCUSIP
__Date__Effective IBES_ticker
Announce_dt Prd_yr Prd_mon Periodicity Measure
Val_1 Val_2;
run;
proc sort data=sdc_Ibes2 noduplicate out=try dupout=dups;
by Target__CUSIP parent_SDCCUSIP
__Date__Effective IBES_ticker
IBES_Announce_dt Prd_yr Prd_mon Periodicity Measure
Val_1 Val_2;
run;
第一种方式中,没有duplicates。
第二种方式中,会有duplicates。
Tuesday, July 15, 2014
Diary
Reference:
Jo and Kim (2007, JFE)
Try to summarize disclosure frequency like below
Panel A: Number of disclosure frequency
Month -24 to -18 M(-24, -18)
Month -18 to -12 M(-18, -12)
Month -12 to -6 M(-12, -6)
Month -6 to -0 M(-6, 0)
Month 0 to +6 M(0,+6)
Month +6 to +12 M(+6, +12)
Month +12 to +18 M(+12, +18)
Month +18 to +24 M(+18, +24)
Panel B: Changes in the number of disclosure frequency
From M(-24, -18) to M(-18, -12)
From M(-18, -12) to M(-12, -6)
From M(-12, -6) to M(-6, 0)
From M(-6, 0) to M(0,+6)
From M(0,+6) to M(+6, +12)
From M(+6, +12) to M(+12, +18)
From M(+12, +18) to M(+18, +24)
Jo and Kim (2007, JFE)
Try to summarize disclosure frequency like below
Panel A: Number of disclosure frequency
Month -24 to -18 M(-24, -18)
Month -18 to -12 M(-18, -12)
Month -12 to -6 M(-12, -6)
Month -6 to -0 M(-6, 0)
Month 0 to +6 M(0,+6)
Month +6 to +12 M(+6, +12)
Month +12 to +18 M(+12, +18)
Month +18 to +24 M(+18, +24)
Panel B: Changes in the number of disclosure frequency
From M(-24, -18) to M(-18, -12)
From M(-18, -12) to M(-12, -6)
From M(-12, -6) to M(-6, 0)
From M(-6, 0) to M(0,+6)
From M(0,+6) to M(+6, +12)
From M(+6, +12) to M(+12, +18)
From M(+12, +18) to M(+18, +24)
Monday, July 14, 2014
Summary
1029 obs. Since two duplicates in 1032 and one error (same target name, same parent name, same announcement date, but one week difference in the effective date. These two should be the same spinoff event.)
902 spinoffs with public parents. 772 obs with complete cusip/permn match
Process:
1. merge_20140529.SAS => Mainly merge Cusip with CRSP (Permnos)
2. Clear v1.SAS => Mainly merge above with Compustat (Gvkey)
3. Final2.SAS => Event study on the spinoff announcement date
4. Final3.SAS => Event study on the management forecast date
5. Compustat_Firm_Characters => ROA, BM, Leverage of parent firms pre & post-spinoffs
6. list_20140613.SAS => I list target firms with public status and chech whether those firms are active before spinoffs. I start with 181 obs. 91 are active. 66 are inactive, which have less than 15 daily CRSP obs in the previous month of spinoffs. The remaining ones are lack of permnos.
7. permno_gvkey macth_20140613.SAS => permno and gvkey should be 100% match. If not, find a reason. So far, 4 out of 506 target_permnos cannot be matached with a gvkey. 7 out of 772 parent_permnos have missing gvkeys.
8. Program20140718 => Merge SDC with IBES by 6-digit CUSIP
9. Program20140722 => Merge SDC with COMPUSTAT by 6-digit CUSIP
9a. Try to find a closest control group
9b. Same 2-digit SIC / 30% Assets / Closest MB ratio
10. manual 20140804 (3).xlsx 是一个重要的文件!我花了500年时间在上面!校对了很多SDC的spin off的母公司和子公司,以及他们的Permno和GVKEY。里面有全系列的a2,b1,b2,c1,c2,d1,d2。和部分a1,如果spin off当日,母公司的RET呈现missing状态。
11. munual_check.SAS => 这个有校对了一下上面的manual 20140804 (3).xlsx有没有漏信息,漏了两个,已经补齐。所以只需看manual 20140804 (3).xlsx即可.
12. SDC_permno_withdup3是从Program20140725来的。Program20140725是一个新的merge。从SDC到CRSP,通过四种不同的方式merge。
第一步,用6-digit CUSIP和time.
第二步,放宽条件,不要求Time也对。
第三步,用Immediate parent cusip,无关time。
第四步,用Ticker.
第五步,手动查找Missing obs。
看这个链接,关于这几步的详细分解。
http://accandfin.blogspot.com/2014/08/data-summary.html
13. AllInOne现在是最牛逼的文件了。1029个SDC过来的spin off都被手动找了一遍。
902 spinoffs with public parents. 772 obs with complete cusip/permn match
Process:
1. merge_20140529.SAS => Mainly merge Cusip with CRSP (Permnos)
2. Clear v1.SAS => Mainly merge above with Compustat (Gvkey)
3. Final2.SAS => Event study on the spinoff announcement date
4. Final3.SAS => Event study on the management forecast date
5. Compustat_Firm_Characters => ROA, BM, Leverage of parent firms pre & post-spinoffs
6. list_20140613.SAS => I list target firms with public status and chech whether those firms are active before spinoffs. I start with 181 obs. 91 are active. 66 are inactive, which have less than 15 daily CRSP obs in the previous month of spinoffs. The remaining ones are lack of permnos.
7. permno_gvkey macth_20140613.SAS => permno and gvkey should be 100% match. If not, find a reason. So far, 4 out of 506 target_permnos cannot be matached with a gvkey. 7 out of 772 parent_permnos have missing gvkeys.
8. Program20140718 => Merge SDC with IBES by 6-digit CUSIP
9. Program20140722 => Merge SDC with COMPUSTAT by 6-digit CUSIP
9a. Try to find a closest control group
9b. Same 2-digit SIC / 30% Assets / Closest MB ratio
10. manual 20140804 (3).xlsx 是一个重要的文件!我花了500年时间在上面!校对了很多SDC的spin off的母公司和子公司,以及他们的Permno和GVKEY。里面有全系列的a2,b1,b2,c1,c2,d1,d2。和部分a1,如果spin off当日,母公司的RET呈现missing状态。
11. munual_check.SAS => 这个有校对了一下上面的manual 20140804 (3).xlsx有没有漏信息,漏了两个,已经补齐。所以只需看manual 20140804 (3).xlsx即可.
12. SDC_permno_withdup3是从Program20140725来的。Program20140725是一个新的merge。从SDC到CRSP,通过四种不同的方式merge。
第一步,用6-digit CUSIP和time.
第二步,放宽条件,不要求Time也对。
第三步,用Immediate parent cusip,无关time。
第四步,用Ticker.
第五步,手动查找Missing obs。
看这个链接,关于这几步的详细分解。
http://accandfin.blogspot.com/2014/08/data-summary.html
13. AllInOne现在是最牛逼的文件了。1029个SDC过来的spin off都被手动找了一遍。
Subscribe to:
Posts (Atom)