22
countries where voluntary adoption of
IFRS
was not allowed prior to 2005 (Capkun et al.
2011a). My sample period spans from 2000 to 2009 (i.e., five years in the pre-
IFRS
period
and five years in the post-
IFRS
period). I delete firm-years of non-
IFRS
post
2005.
22
I also delete financial firms and firm-years with no industry affiliation. Finally, I
require a constant set of firms across the sample period to capture the effect of changes in
accounting choices among the same firms in the pre-
IFRS
period and in the post-
IFRS
period. After deleting all firms with
missing data, my total sample is comprised of 2,568
firm-years representing 321 unique firms.
23
Table C2 (Panel A) presents the distribution of firm-years by industry and
country. While there is a reasonable distribution of industries across my sample, the bulk
of my sample comes from the United Kingdom and France. This
country distribution is
quite consistent with prior studies on
IFRS
adoption in the EU (e.g., Chen et al. 2010a;
Capkun et al. 2011b).
24
Table C2 (Panel B) presents the descriptive statistics for my total sample. The
differences between the lower and upper quartile values show there is considerable cross-
sectional dispersion for most variables in my analysis, including my dependent
variable
(
CAPEX
). The mean of
CAPEX
(in log format) is -3.35 corresponding to an annual
average of roughly 3.5% of total assets.
25
Also, the descriptive statistics reveal that, on
average, the firms’ gross value of
PPE
is roughly 50% of the book value of their total
assets. In addition, the firms typically are profitable (81.19% of firm-years), are paying
dividends (79.48% of firm-years), and have a
BIG4_5
auditor (81.31% of firm-years)
.
22
I define firm-years as
adopting
IFRS
if WorldScope reports code 23 in the field of accounting standards
followed after January 1, 2005.
23
Because I scale most variables by lagged total assets, the year 2000 is dropped from my sample in the
pre-
IFRS
period and the year 2005 is dropped from my sample in the post-
IFRS
period.
24
My results are qualitatively similar after dropping UK and French firms from my total sample.
25
I find that my dependent variable, capital expenditures, is skewed. Hence, I normalize this variable by
including it in log format.