A Multi-Issue/Multi-Period Analysis of the Geographic Diversity of IASB Comment Letter Participation

Abstract

The International Accounting Standards Board (IASB) establishes accounting standards now used in some form in over 100 countries. Diverse geographical participation in International Financial Reporting Standards (IFRS) standard-setting is seen as desirable as it may improve the consistency of IFRS applications, reduce criticism of regional over-influence, and promote the legitimacy of the IASB. This study investigates country participation and the regional and institutional factors that influence the geographic diversity of comment letters (CLs) in the IASB's standard-setting process. Using CLs regarding 57 IASB issues from 2001 through 2008, we find that countries with EU membership, G4+1 membership, donations to the IASB, and larger equity market development are associated with larger numbers of CLs and CL writers. Analysis of a subsample of more developed countries finds some evidence that countries with more historic divergence in accounting standards from IFRS also have more CL writers. In most countries, one of several major stakeholder interest groups, such as professional accountancy bodies, accounting standard-setters, and public accounting firms, send at least half of the CLs. While response levels for most countries vary greatly depending upon the nature or topic of an IASB issue, overall response levels remain low at just over 100 responses per issue and did not increase over time. While geographic diversity and response rates are greater than its predecessor the International Accounting Standards Committee, they are lower than those of many national standard-setters, possibly raising due process and legitimacy issues for the IASB.

Acknowledgments

The authors gratefully acknowledge the helpful comments and suggestions provided by Professor Lisa Evans (the journal editor), the two anonymous reviewers, Gary Braun, Sebastian Hoffmann, Mark Myring, Charles Page, Ann Tarca, and the other participants of the 2010 Mid-Year Meeting of the International Accounting Section of the American Accounting Association (AAA) and the 2011 Annual AAA Meeting for their helpful comments. The authors also thank Shelby Elking, Michael Little, and Amy Schimmoeller for their research assistance. Any remaining errors are our own.

Notes

Richardson and Eberlein (2011) often calls exchange legitimacy either input legitimacy or participatory legitimacy.

Publicly available summaries of IASB issues and any related CLs are now regularly prepared by the IASB staff and the large international public accounting firms.

Jorrisen et al. (2012) assign each letter to the country they felt is best associated with the first author, even if jointly written letters had authors from several different countries. Thus, the IASB's and Jorrisen et al.'s number of letters for a particular issue and their totals by country are often different than the response totals in this paper due to these different approaches. For example, the IASB and Jorrisen et al. report 123 CLs for the 2005 Proposed Amendments to IAS 37. We report 130 responses because while we start with 123, two letters have two authors (we add 2 more responses), two letters have three authors (we add 4 more), and one letter is from a company domiciled in two countries (we add 1 more to our country count as Royal Dutch Shell was domiciled in both the UK and the Netherlands). There also might be some confusion about dates as we assigned the year based on the release of the issue, while Jorrisen et al. (2012) assign the year based on the due date of comments.

The IFRSIC was first known as the IFRIC.

Katselas et al. (2011) conclusions are based on their investigation of 27 corporate CL writers using an EU dummy variable which includes five German, one Dutch, and one French corporation. One Australian, 6 Swiss, and 13 UK firms are not in their EU dummy variable.

While not an official part of the IASB due process, the IASB issued one ED in 2007 and six EDs in 2008 in French.

All joint FASB/IASB projects where identical documents are issued simultaneously are excluded. A large number of those projects' CLs are not clearly identified as belonging to either the IASB or the FASB, as many are addressed to both organizations or no organization at all. The projects excluded are Consolidations (2005), Conceptual Framework DP Phase A and D (2006), Conceptual Framework ED Phase A (2008), Revenue Recognition (2008), and Financial Statement Presentation DP (2008).

Given suggestions, Equity Market Development was operationalized two ways, with a second part focusing on the number of companies listed on a country's equity markets. However, multicollinearity issues prevented both from simultaneously being in the models.

The vast majority of Pan-European writers are trade associations for businesses, financial services organizations, and cooperatives.

The current goal of IASB funding is that each country should contribute an amount roughly in line with their GDP, which is used here to operationalize Economic Development Level. Larson and Kenny (2011) find that IASB donations and GDP are significantly correlated with a coefficient of .89. In the larger sample GDP per capita and IASB donors are significantly correlated at .49, but regressions for Models 1 through 6 with both variables did not note any multicollinearity problems. However, it is still possible that inclusion of both affects the results.

While there is no generally agreed upon VIF cutoff value (4, 5, or 10 are not unusual), particularly with smaller samples some suggest cutoff values as low as 2 (Craney and Surles, 2002), 2.5 (Allison, 1999), or 3 (Minkov, 2013) in order to minimize multicollinearity issues. If a VIF cutoff of 2 is used in Models 7 to 12 and the overall least significant variable (economic development) is removed, then all VIFs are below 2, most adjusted R2s increase slightly, and in all models IFRS Differences is significant (from .030 to .063). Given the existence of many significant correlations of .5 and above among the main and sensitivity analysis variables, this paper uses 3 as a general VIF cutoff point.

While a .8 correlation is often thought to indicate a high level of multicollinearity, Berry and Feldman (1985) suggest .7 in very small samples, and Gujarati (2003) notes that it may exist even with a .5 pair-wise correlation. Multicollinearity may also occur when an independent variable is correlated with a combination of several independent variables while not being highly correlated with any single independent variable. That is a concern here because several variables have correlations of .5 or above with two or more other variables. We try to follow Jaccard et al.’s (1990, 41) view that ‘the inclusion or exclusion of a variable is first and foremost a theory problem and requires a theory-based solution’.