The Impact of Corporate Governance on IFRS Adoption Choices

Abstract

We investigate the association between corporate governance strength and EU listed firms' choices with respect to International Financial Reporting Standards (IFRS) adoption in 2005. We measure governance strength by aggregating variables such as board independence, board functioning and audit committee effectiveness. The firms exhibit heterogeneity in both compliance and disclosure quality; some firms do not even meet the minimum disclosure requirements. Regression results show that stronger governance firms disclose more information, comply more fully and use IAS 39's carve-out provision less opportunistically. These findings are germane to accountants, managers and regulators in countries soon to adopt IFRS.

Acknowledgements

We appreciate the helpful comments from two anonymous referees, Ulf Brüggemann, Liesbeth Bruynseels, Salvador Carmona, Laurens Cherchye, Hans Christensen, Peter Easton, Pingyang Gao, Stephan Hollander, Laurence van Lent, Edith Leung, Frank Liu, Don Stokes, Jeroen Suijs, Marleen Willekens and Yachang Zeng. This paper also benefited from being presented at the 2009 EAA Meeting in Tampere, the 2010 IAS Midyear Conference in Palm Springs and the 2010 INTACCT meeting in Varna, and from comments from the workshop participants at Tilburg University and the Katholieke Universiteit Leuven. The authors gratefully acknowledge support from the INTACCT Research Network, the FWO and the Chartered Accountants of Ontario Professorship.

Notes

IAS 1 (Primary Statement of Changes in Equity), IAS 7 (Cash Flow Statement), IAS 14 (Segmental Reporting), IAS 24 (Related Party Transactions), IAS 39 (Financial Assets and Liabilities) and IAS 33 (Earnings per Share).

Previous studies extensively document the heterogeneity in fair value inputs used for financial instruments (Song et al., 2010; Kolev, 2009; Goh et al., 2009) and in the fair value or historical cost choice for investment property (Muller et al., 2009; Avallone and Quagli, 2008; Christensen and Nikolaev, 2009).

We thank the referee for bringing this notion to our attention.

This type of conservatism also encourages awareness of riskiness about financial product investments by company insiders (Ball, 2001; Watts, 2003). While Garcia Lara et al. (2007) find a positive association between the strength of corporate governance and accounting conservatism using the measure of Basu (1997) and Ball and Shivakumar (2005), we investigate the decision to adopt IAS 39 early or late dependent on whether application has a positive or negative outcome as another measure of IFRS compliance.

Corporate governance strength has been measured by means of these Deminor Ratings in prior studies such as Bauer et al. (2004), Bozec (2007), Florou and Galarniotis (2007), Vander Bauwhede and Willekens (2008) and Renders et al. (2010).

Our sample selection criteria limit our analysis to large firms only. Although we are aware of a potential external validity problem, we emphasise that these sample firms represent a substantial fraction of the market value of all European public companies. The fact that our sample involves only large and widely known companies validates our results. Furthermore, corporate governance practices tend to vary less among large firms than among smaller firms. Therefore, finding an association between governance and disclosure and compliance properties for large companies could suggest an even stronger association for smaller companies.

The MSCI Pan-Euro Index contains securities with a free float-adjusted market capitalisation of €4348 billion on 30 June 2006.

Results are robust to using 2004 or 2005 ratings. This is expected since corporate governance tends to be sticky over time. We prefer 2004 figures to mitigate endogeneity. For 10 observations, we had to use the corporate governance rating of 2005 because the 2004 rating was not available. Regression results remain qualitatively the same without those 10 observations.

In further tests, we also take into account the notion that governance choices may be endogenously determined by firm characteristics. We estimate a first-stage model where governance is predicted by firm size, profitability, growth opportunities and leverage in addition to the country controls. Results remain qualitatively the same for all of the models tested below.

In the Restated Items score equity is considered twice because both beginning and end of the year equity should be restated according to IFRS.

It is not possible to split RESTATEMENT up into a mandatory and voluntary part, as there is not a mandatory amount of pages firms must provide to restate financial data. For the other three components of RESTATEMENT, we provide separate information on whether firms comply with mandatory requirements or not in Table 2, Panel A.

To ensure that figures are comparable between financial and other firms, a loss dummy and MTBV are preferred to ROA and sales growth to capture performance and growth, respectively. However, results remain qualitatively the same when the preferred variables are replaced by the alternatives.

Using Nobes' framework (2001), Li (2010) finds IFRS application leads to a larger drop in information asymmetry in countries where the increase in disclosure rules due to IFRS is stronger.

Nobes (2001) considers three additional standards: discontinued operations (IAS 35), inventories (IAS 2) and investment properties (IAS 40). We do not consider these standards because economic transactions related to those standards do not occur for every firm in the sample.

Net income differs from comprehensive income because some gains and losses go directly to stockholders' equity, bypassing the income statement.

All firms have either multiple geographical segments or multiple business segments. Most firms have both; for those we calculate a score with a maximum of 6. For firms only operating in one business or in one geographical segment, we adjust this item to a score out of 3 (i.e. More than 1 Year, score of 1; and Primary Segment 0, 1, 2, score of 2) and double the score to ensure comparability with the other observations.

We cannot test the effect of IAS 39 on net income since we only have the latter information for the early adopters and not for the late adopters.

Including industry effects based on 2-digit SIC codes does not change our results.

For brevity we only show coefficients for BFUNCTIONCC, BINDEPCC and AUDITCC at the bottom of the panel. Control variables are included in these analyses but not shown.

We thank the referee for bringing this alternative notion to our attention.

Additional information

Notes on contributors

Arnt Verriest

Paper accepted by Salvador Carmona.

Ann Gaeremynck

Paper accepted by Salvador Carmona.

Daniel B. Thornton

Paper accepted by Salvador Carmona.