65
traditional definitions in survival analysis setting because the event is not necessarily
bad events that people want to avoid completely. As mentioned in Section 3.1, we
define an event as an insurance company being acquired in M&A deal(s), a depen-
dent censoring as an insurance company filed bankruptcy, and an independent
censoring as otherwise. Also, we are more interested in knowing whether there are
any negative impacts of the GLB Act. Thus, in this research for M&A deal(s), we
focus on “being acquired”, which is dependent with going bankrupt.
Given different definitions of events and censoring, the methodology developed in
Chapter 3 is applicable to both scenarios under the competing-risks setting.
4.2 Data source and data structure
Given our research interest, we used the combined database of the University of
Chicago’s Center for Research on Security Prices (CRSP) and Standard & Poor’s
COMPUSTAT, which is created by a division of Standard & Poor’s called COM-
PUSTAT. Those data sets are available through Wharton Research Data Services
(WRDS). The major insurance companies doing business in U.S., with Standard In-
dustrial Classification (SIC) Codes 6311, 6321, 6331, and 6351, were extracted from
the database. This choice of insurance companies is consistent with Marlett et al.
(2003).
The industrial annual data with CRSP and COMPUSTAT combined is used to get
the covariate information. Daily stocks data from CRSP is used to get the detailed