Notes
1. The article uses the word ‘‘IFRS’’ to indicate International Financial Reporting Standards, which
includes also the previous definition of the same principle, namely, International Accounting
Standards (IAS).
2. The only partial exception is Germany where, however, private companies are allowed to use
IFRS in their separate financial statements, but only in addition to the local (German) generally
accepted accounting principles (GAAP). So this could not be considered as a properly full IFRS
transition.
3. Notice also that, especially after Basel accords (the most relevant for this study being the 2004
one), the stability of banking and financial system has been found to critically depend on client
company financial reporting transparency (Bushman & Landsman, 2010), making earnings attributes
of crucial importance.
4. The database used for sampling is AIDA, the Italian version of Amadeus provided by Bureau
Van Dijk that contains comprehensive information for private companies in Italy.
5. In our analysis, we do not consider the year of the transition because previous research finds particular
pervasive earnings management practices in that particular year (Capkun, Cavezan-Jeny,
Jeanjean, & Weiss, 2011). Data will then refer to the years 2006-2009.
6. The use of a logistic model is the most prevailing approach for estimating propensity-scores
(Lawrence, Minutti-Meza, & Zhang, 2011).
7. The number of companies included in the group of non-IFRS adopters is the results of a preliminary
selection from AIDA where companies met the following criteria: (a) availability of data
from 2005 to 2009; (b) annual report prepared under Italian GAAP; (c) firms not involved in a
liquidation process; (d) limited liability companies; (e) total asset, leverage, and profitability 30% lower than the minimum and 30% higher than the maximum of the same variables for the
group of IFRS adopters.
8. It includes 85 adopters in 2005 (the only firms that could potentially cover all the time-period
investigated: 2006-2009), 125 adopters in 2006, 34 adopters in 2007, and 26 adopters in 2008.
All missing values have been deleted together with the correspondents of the pairs. There are
cases where the full-time series (which normally should go from the year after IFRS adoption
until 2009) was not available. In the case a firm-year observation was missing, also the correspondent
observation referred to its pair was excluded. This further decreased the number of
observations.
9. To remove potential biases due to the presence of outliers, abnormal accruals are winsorized
with a ‘‘top 98% winsorization.’’
10. Basu (1997) model cannot disentangle the role of random errors in accruals and of other types of
earnings management (like reverting excess provisions) and can only identify the existence of
transitory components, and not whether their recognition is timely or untimely (Ball &
Shivakumar, 2005).
11. In all our analyses, we test for potential multicollinearity issues using the Variance Inflation
Factor (VIF) test. All the tests show a maximum VIF factor lower than 5.6, well below the
threshold of 10 suggested by Kennedy (2008).
12. We note that the R2 is different across the two sub-samples, suggesting that the Ball and
Shivakumar (2005) model for timely loss recognition better fits the sub-sample of non-IFRS
adopters.
13. As information on listed ownership was hand-collected, it was impossible to include this variable
as one of the a priori discriminant in the definition of the matched sample. Therefore, the distinction
between the companies that are part of a group where the parent is listed and others was
first made on IFRS adopters only to maintain matched pairs, and therefore allow the analysis to
be made on IFRS adopters and their individual pairs. Then, we rerun the analysis classifying
both IFRS adopters and non-IFRS adopters on the basis of listed ownership (but of course losing
the matches). Results are consistent with those shown in Table 5. Coefficient of variable IFRS
is still positive and significant in the sub-sample of companies controlled by a listed company
(b = .162; p = .000) while it is not significant in the sub-sample of companies not controlled by
a listed company (b = .029; p = .458). The difference between the two coefficients is significant
at the 1% level (p = .001). We similarly repeated the analysis reported in Table 6 and results are
qualitatively the same. The difference between the coefficients g3 between the two groups is negative
and highly significant for firms controlled by a listed company (g3IFRS ADOPTERS 2 g3NONIFRS
ADOPTERS = 21.156; p = .000), indicating that losses are less timely recognized among
IFRS adopters. The same results are obtained for firms not controlled by a listed companies
(g3IFRS ADOPTERS 2 g3NON-IFRS ADOPTERS = 21.034; p = .000). The difference of the differences
between the coefficients of these two sub-samples is not significant (p = .940).
14. Levene’s test is chosen to test differences in standard deviation because it does not assume the
data to be normally distributed (Gastwirth, Yulia, & Miao, 2009).
15. In this particular case, we are considering negative values of abnormal accruals. More negative
values mean more income-decreasing earnings management. So a negative coefficient should be
interpreted as an increase in earnings management and, consequently a decrease in reporting
quality.
16. We also tested the potential impact of the financial crisis on the relation between IFRS adoption
and earnings quality by testing differences in means of abnormal working capital accruals
(AWCA) between IFRS adopters and non-IFRS adopters before and after 2007. Results show
that the level of AWCA is significantly higher for IFRS adopters compared with non-IFRS adopters
at 1% level both before 2007 and after 2007. In particular, before 2007, the mean of AWCA
for non-IFRS adopters is 0.069 while for IFRS adopters is 0.167. After 2007, the mean of
AWCA for non-IFRS adopters is 0.084 while for IFRS adopters is 0.153.
17. We use the larger group of 355 IFRS adopters as in this case we do not lose observation in the
matching procedure.
18. Relevance and comparability are mentioned in the ‘‘joined conceptual framework among the fundamental
and enhancing characteristics of financial information’’ (Phase A of the Joint
Conceptual Framework), available at www.ifrs.org
19. Micro-firms are defined as those that do not exceed the limits of two of the three following criteria:
(a) balance sheet total: EUR 350,000; (b) net turnover: EUR 700,000; (c) average number
of employees during the financial year: 10.