Accounting choices under ifrs and their effect on over-investment in capital expenditures

3.3 Sample Selection 
3.3.1 Total Sample 
My total sample consists of all publicly-listed firms available on WorldScope 
database in nine EU countries that mandated 
IFRS
adoption in 2005 (see Panel A of 
Table C2). Prior literature (e.g., Christensen et al. 2008) shows that capital market and 
economic benefits resulting from 
IFRS
adoption are highly dependent on managers’ 
incentives to voluntarily adopt 
IFRS
(i.e., early adoption). Therefore, in order to 
minimize the self-selection bias related to managers’ incentives to voluntarily adopt 
IFRS
before 
IFRS
became mandatory in 2005, I include firms only from those nine EU 
20
WorldScope defines closely-held shares as shares held by insiders. This variable includes shares held by 
officers, directors, and their immediate families. 
21
WorldScope reports the auditor for each firm only for year 2009, the last year of data currently available 
on WorldScope. I assume that this control variable is fairly sticky and does not vary significantly over time. 
I also assume that firms that had Arthur Andersen as their auditor in 2002 moved to another 
Big4_5
auditor 
upon Arthur Andersen’s closure. 

---

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. 

---

23