Friday, February 27, 2015

Dairy 20150227

Problem 1
Program20150226.SAS

Last_BM, BM
(1) negative number
(2) a lot of small number, e.g. 0.01


Tuesday, February 24, 2015

Dairy 20150224

Problem 1:

Some control firms do not sustain in three years.

solution:

I keep those firms that only have three year continuous data.

Tables

SAS Pro\spinoff

Program20150220.SAS
step3的Freq结果。

carveout发生频率

SDC_year Frequency Percent Cumulative
Frequency
Cumulative
Percent
1994 18 9.33 18 9.33
1995 10 5.18 28 14.51
1996 26 13.47 54 27.98
1997 14 7.25 68 35.23
1998 12 6.22 80 41.45
1999 20 10.36 100 51.81
2000 11 5.70 111 57.51
2001 11 5.70 122 63.21
2002 8 4.15 130 67.36
2003 1 0.52 131 67.88
2004 3 1.55 134 69.43
2005 4 2.07 138 71.50
2006 5 2.59 143 74.09
2007 1 0.52 144 74.61
2009 1 0.52 145 75.13
2010 3 1.55 148 76.68
2011 5 2.59 153 79.27
2012 13 6.74 166 86.01
2013 27 13.99 193 100.00
M.B. Slovin et al./Journal oj’Financial Economics 37 (1995)

Eckbo - Handbook of Corporate Finance
Attention: freq is really really high!!! Unbelievable!!!

B.A. Jain et al. / Journal of Financial Economics 100 (2011) 594–615
Vertical spinoffs and carve-outs

FRANK AND HARDEN
Journal of Business Finance & Accounting, 28(3) & (4), April/May 2001, 0306-686X
Sample size large!

J Econ Finan (2009) 33:27–42
Agency problems in tracking stock and minority
carve-out decisions: Explaining the discrepancy
in short- and long-term performances

Wei He & Tarun K. Mukherjee & Peihwang Wei



Friday, February 20, 2015

SDC New Issues Database


31OCT1996
New Holland NV(New Holland Holdings NV/Fiat SpA)

http://www.company-histories.com/New-Holland-NV-Company-History.html






Wednesday, February 18, 2015

Dairy 20150218

昨天基本上,自动手动结束了Carve out subsidiary (SDC New Issues Database IPO)和Edgar的合并,结果match成功率依然低下,虽然已经很多手动添加GVKEY, CIK,最主要的。即使成功merge的对应s-1/s-3 filings中,也几乎没有出现"spin off"的字样。

今天把spinoff parent的merge完成,然后开始和IBES合并在一起,看看MF的变化,以及,前后information asymmetry的变化。以及圈定subset 中,先IPO后spin off的公司。

Friday, February 13, 2015

SAS Tips

(1)

proc reg data = want /rinfluence;

can omit the outlier effect

(2) Winsorize Code

https://communities.sas.com/thread/36963

(3) two groups TTEST

https://www.youtube.com/watch?v=ULq9kQtIDMQ

proc ttest data=want;
 paired next_AT*control_next_AT;
run;