國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
13
firm-year observation of cross-listers based on matching year, industry, and sales growth. Because the U.S. matched firms are not experiencing the mandatory IFRS adoption but will be affected by any other factors in the U.S. capital market, they serve as a direct control group for those identified foreign firms. Moreover, to control for other factors that U.S.-domiciled firms may not able to capture but could still influence cross-listers, I also perform regression-based tests separately on a sample consisting of foreign firms cross-listed in the U.S. from countries that have not mandated IFRS.
5. Sample selection and descriptive statistics
I collect all accounting data and market-based data on Compustat and CRSP from year 2003 to year 2008. To test the impact of IFRS adoption on the mispricing of discretionary accruals, I define the pre-IFRS period as 2003-2004 and the post-period as 2005-2008. Following the existing literatures (e.g., Beaulieu and Bellemare 2000;
Hail and Leuz 2008), I exclude Canadian firms because the capital market of Canada and the U.S. are highly integrated and the cross-listing motives of Canadian companies are different from those of companies in other countries. In addition to addressing the similarity of capital market issue, following the existing literatures (e.g., Doidge et al. 2009; Sarkissian and Schill 2009), I exclude firms from tax havens such as Bermuda and the Virgin Islands since many of these firms are actually U.S.
firms that have acquired a foreign status to experience tax advantages. Excluding firms that had mandated IFRS but had not done by the fiscal year-end on December 31, 2005, I can compare the extent of accruals anomaly firms experienced before and after the adoption of IFRS with any mispricing of discretionary accruals experienced by their comparable U.S. firms or by foreign firms from non-IFRS adoption countries from the same fiscal year-end. Finally, I exclude firms with missing any of the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
14
accounting data or market-based data required for portfolio and regression tests and exclude foreign companies for which no comparable U.S.-matched firms can be identified.
The final sample contains 1705 firm-year observations of foreign firms cross-listed in the U.S., including 859 firm-year observations from countries that mandated IFRS adoption and 846 firm-year observations from countries that did not adopt IFRS over the sample period. Table 1 presents the country distribution of foreign firms in the sample. Among 23 countries in panel A, U.K. represents the biggest portion, contributing 177 firm-year observations, about 21% in the sample of foreign cross-listed firms from mandatory IFRS countries. On the other hand, Israel accounts for the largest proportion in panel B, amounting 319 firm-year observations, about 38% in the sample of non-IFRS adoption countries.
‧
Table 1 Country distribution of foreign firms cross-listed in the U.S.
Panel A: Mandatory IFRS adoption countries
Country Number of Observations Number of Firms
Australia 35 13
Country Number of Observations Number of firms
Argentina 40 10
‧
This table presents distribution of countries included in the analyses. Panel A reports a distribution of cross-listers from countries that mandate IFRS, and Panel B provides information of cross-listed firms from non-IFRS adoption countries. The test period covers from fiscal year ending in December 2003 to 2008.
Table 2 exhibits descriptive statistics of the main variables for the treatment group, namely, cross-listers from mandatory IFRS countries, and for the two control groups, i.e., cross-listers from non-IFRS adoption countries, and matched comparable U.S. firms respectively. From pre- to post-IFRS period, there is no statistically significant change in the mean value of size-adjusted return (SARET), for either the cross-listers from mandatory IFRS countries, or their comparable U.S. firms. As for control variables, discretionary accruals (DACC), log market value (MV), and residual variance (RVAR) show significantly differences between two periods for the cross-listers from countries mandating IFRS. For the cross-listed firms from non-IFRS adoption countries, discretionary accruals (DACC), book to market value (BM) and residual variance (RVAR) differ between two periods. As for the comparable U.S.-domiciled firms, only discretionary accruals (DACC), book to market value (BM) show difference from pre- to post-IFRS period. Besides the comparison between pre- and post-IFRS periods for each of three groups, I compare the difference in the mean values of the main variables between sub-samples. For the comparison between cross-listers from IFRS and non-IFRS countries, there are
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
17
significant differences in the mean values of DACC, MV, BM, and RVAR in both the pre- and the post-IFRS periods. However, for cross-listers from IFRS countries and their matched U.S. firms, there are significant differences in the average levels of OCF, MV, BM, and RVAR.
Table 3 exhibits the correlation of variables for cross-listers from IFRS and non-IFRS countries. In particular, operating cash flows (OCF) and log market value (MV) are significantly, negatively correlated with discretionary accruals (DACC), and residual variance is positively correlated with discretionary accruals (DACC).
Size-adjusted return (SARET) is negatively correlated with log market value (MV) and is positively correlated with book to market value (BM) at 1 percent level of significance.
‧
Panel A: Cross-listed foreign firms from mandatory IFRS adoption countries (859 observations)
Panel B: Cross-listed foreign firms from non-IFRS adoption countries
(846 observations)
‧
control groups (i.e., cross-listers from non-IFRS adoption countries and comparable U.S.-domiciled firms). The sample period covers from fiscal year ending in December 2003 to 2008.a Indicates5 percent level of significance in mean difference between pre- and post-IFRS periods.
b Indicates 5 percent level of significance in mean difference between cross-listers from mandatory IFRS countries and from non-IFRS adoption countries.
c Indicates 5 percent level of significance in mean difference between cross-listed firms from mandatory IFRS countries and their matched comparable U.S.-domiciled firms.
Variable definitions
DACC = Discretionary accruals estimated from the modified Jones (1991) model following Kothari et al. (2005)
OCF = Operating cash flow
SARET = Annual size-adjusted return calculated as the difference between the raw buy-and-hold return of each firm and the return of matched portfolio to which each firm is assigned
MV = Log of market value BM = Book-to-market ratio CP = Cash flow-to-price ratio
RVAR = Residual variance estimated from the Fama and French (1996) three-factor model
‧
DACC = Discretionary accruals estimated from the modified Jones (1991) model following Kothari et al.
(2005)
OCF = Operating cash flow divided by total assets
SARET = Annual size-adjusted return calculated as the difference between the raw buy-and-hold return of each firm and the return of matched portfolio to which each firm is assigned
MV = Log of market value BM = Book-to-market ratio CP = Cash flow-to-price ratio
RVAR = Residual variance estimated from the Fama and French (1996) three-factor model
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
21