Introduction of International Accounting Standards, Disclosure Quality and Accuracy of Analysts' Earnings Forecasts

Abstract

We examine whether the introduction of international accounting standards by German companies has improved the accuracy of analysts' forecasts, and what role changes in the quality of disclosures have played in this process. We develop a structural equation model that allows us to separate the effects of changes in disclosure quality from other effects of the introduction of international accounting standards on forecast errors. Our sample comprises 1,908 firm-years covering the period from 1997 to 2005. We measure disclosure quality with data from a yearly annual-report competition. We find that the introduction of international accounting standards has been associated with a significant improvement in forecast accuracy. Increases in the quality of companies' disclosures appear to have contributed to this improvement. However, the disclosure effect, while significant, explains only a small portion of the overall improvement in forecast accuracy. Further analyses show that differences in disclosure quality are more relevant for German GAAP companies than for IFRS/US GAAP companies. Moreover, only the quality of notes to companies' financial statements appears to matter to analysts; the quality of management reports appears to make no difference. Our results are robust to a variety of tests concerning the sample composition, the operationalisation of variables and the estimation procedure.

Acknowledgements

We are grateful to Salvador Carmona, the editor, two anonymous reviewers as well as to Vasiliki Athanasakou, Paquita Davis-Friday, Allan Hodgson, Bill Rees, Alfred Wagenhofer and, especially, Peter Schmidt, for helpful comments. We also appreciate comments from workshop participants at the University of Glasgow, the University of Graz, the Vienna University of Economics and Business, the London School of Economics, the Center for Quantitative Economics, University of Muenster, and from participants at the IAAER Conference in Istanbul (2006), the EAA Conference in Lisbon (2007), Verein für Socialpolitik, Ausschuss für Unternehmensrechnung (2009).

Notes

Standards issued by the International Accounting Standards Board (IASB) are called International Financial Reporting Standards (IFRS). Standards issued until 2000 by the IASB's predecessor, the International Accounting Standards Committee, are called International Accounting Standards (IAS). In this paper we use the term IFRS to refer to both.

In order to strengthen the database for our empirical investigation, we consider IFRS and US GAAP annual reports to be equivalent and place them together in a single group. IFRS and US GAAP are based on very similar fundamental concepts, and the standards are also very similar in their detail (and as a result of the convergence endeavours of IASB and FASB they are becoming even more alike in the course of time), whereas the differences between IFRS and US GAAP on the one hand and German GAAP on the other hand are very marked (see Leuz and Verrecchia, 2000; Cuijpers and Buijink, 2005; Daske, 2006; Daske and Gebhardt, 2006).

Focusing on accounting changes in a single country has the advantage that results are not affected by different economic, legal, cultural or other factors that can affect the quality of financial statements (Leuz, 2003; Ruland et al., 2007). We focus on German companies for two reasons. First, given the large number of companies that were already publishing IFRS or US GAAP annual reports in the 1990s, the consequences of the introduction of international accounting standards can be investigated comparatively well on the German capital market. Second, the differences between traditional German domestic accounting and IFRS or US GAAP are very pronounced, so that tests on the consequences of the transition should be relatively powerful (Bartov et al., 2005).

It should be noted that improvements in analysts' forecast errors per se do not provide conclusive evidence of improvements in the quality of companies' financial statements and that reducing forecast errors is not an objective of standard setters. First, improvements in the accuracy of analysts' forecasts may be brought about by causes other than changes in financial reporting (e.g. by changes in investor relations or changes in analyst behaviour). Second, improvements in forecast accuracy could also be effected by giving company management more scope to make earnings smoother and thus easier to predict. It is an important innovation of the present paper that we isolate the indirect effect of the introduction of international standards on forecast errors that is brought about by changes in the quality of disclosure. However, even observing a rise in the quality of disclosure that mediates improvements in forecast accuracy does not allow us to conclude that the introduction of IFRS was successful. First, the standards change could have other, adverse effects on addressees of financial reporting other than financial analysts. Second, in our analyses we consider neither the overall cost of the introduction of IFRS to the companies nor, more specifically, the cost of the expanded disclosures.

Also see Cuijpers and Buijink (2005), who work with a sample of European companies and find that the dispersion of analysts' forecasts is higher for companies that use either IFRS or US GAAP than for companies using local GAAP.

Two standards developed by the German Accounting Standards Board (GASB) also deal with the management report, GAS 5 ‘Risk Reporting’ and GAS 15 ‘Management Report’. However, German financial reporting is regulated by law and companies are not legally obligated to observe the standards of the private-sector GASB. Empirical studies document that non-compliance with GASB standards is widespread (Kajüter and Winkler, 2003; see also Gebhardt and Heilmann, 2004).

See also the results of Hodgdon et al. (2008), who in agreement with Barron et al. (1999) show for an international sample of companies that the degree of compliance with IFRS accounting standards is negatively correlated with analysts' forecasting errors.

Beretta and Bozzolan (2008) develop an index for the richness of disclosure and find for a sample of Italian companies that disclosure quality has a stronger positive (negative) association with analysts' forecast accuracy (dispersion) than mere disclosure quantity.

Mediation explains through which transformation processes independent variables cause effects on dependent variables. Following the seminal work of Baron and Kenny (1986), in a three-variable system where X is the impulse variable and Y is the outcome variable, a third variable M is said to function as a mediator if (i) X is correlated with Y, (ii) X is also correlated with M, (iii) M is correlated with Y, and (iv) when M is controlled, the association between X and Y is significantly reduced.

While so far SEM has not been used extensively in financial accounting literature, its application is more common in other areas of business research such as marketing, strategic management research and managerial accounting; see, for instance, Shook et al. (2004) and Smith and Langfield-Smith (2004).

We also conduct a test of the endogeneity of the choice between German GAAP and IFRS/US GAAP within a simple OLS framework. Following the example of Lapointe-Antunes et al. (2006), we use the residual-variable test, a variant of the Hausman test (1978). We regress the potentially endogenous variable INT on variables that have been used in pertinent studies (Leuz and Verrecchia, 2000; Hail, 2002; Gassen and Sellhorn, 2006; Lapointe-Antunes et al., 2006). In the next step the residuals of this regression are placed in equation (2) as an additional explanatory variable. Equation (2) is then estimated using OLS. We find that the regression coefficient of the first-stage residuals in the second-stage equation does not differ significantly from zero. This analysis (Wooldridge, 2003) indicates that the choice between German GAAP and IFRS or US GAAP cannot be regarded as endogenous for our model constellation. A similar result has been obtained by Lapointe-Antunes et al. (2006).

Wagenhofer and Dücker (2007) point out, however, that the interpretation of the measures commonly used in this literature from a theoretical point of view is not always clear (also see Dechow et al., 2010).

This checklist has been compiled since the middle of the 1990s on the basis of several empirical surveys involving fund managers, financial analysts, auditors and private investors (see Krumbholz, 1994; Armeloh, 1996; Heumann, 2005; Prigge, 2006).

Examples of items on the checklist for the notes are information on the scope and method of consolidation, information on methods of recognition and valuation; information on business combinations; descriptions and explanations on depreciation of plant and equipment, amortisation of intangibles, impairment of assets, the valuation of finished and half-finished goods and other inventories, provisions, deferred taxes, financial instruments (the foregoing items refer to notes on the balance sheet); information on the breakdown of operating income and expenses as well as on other income and cost categories (from notes on the income statement); and, finally, information on notes pertaining to the cash flow statement and segment reporting.

Examples of items on the checklist for the management report are descriptions and explanations on the legal and organisational structure of the company, its products and markets; the past performance including explanations for possible deviations from previously predicted results; the company's current strategy; its control system; the management incentive system; research and development; the company's financial position; individual economic and operational risks and the overall risk situation; business opportunities and the business outlook with regard to markets; expected trends in revenues, costs and results; and expectations regarding future investments, financing and liquidity.

The first adjustment to the management-report criteria was made in 1999 in response to an amendment of the Commercial Code that for the first time obliged companies to comment on ‘the risks of future business developments’. The second change was in 2005 after requirements for the management report were extended and defined in more concrete terms by another reform of the German GAAP and by the publication of German Accounting Standard (GAS) 15 on the management report (Prigge, 2006).

The difference between MANREP _{t −2}, the quality score of the management report in the penultimate German GAAP-year, and MANREP _{t +1}, the quality score in the second IFRS-year, is also not statistically different from zero (z = −0.511; p = 0.6095).

A disadvantage of rank regression is that the regression coefficients cannot be interpreted economically (Iman and Conover, 1979). We return in Section 5.5 to the issue of the economic interpretation and the economic magnitudes of the main coefficients of our model.

Apart from auditors, until 2005 there were no institutions in Germany – such as the Securities and Exchange Commission in the USA or the Financial Reporting Review Panel in the UK – responsible for ensuring compliance with and enforcement of accounting rules by publicly traded companies. The legal foundations for an explicit German enforcement system were first established through an amendment of the Commercial Code in December 2004. The new German enforcement system is a two-stage system, with the private Financial Reporting Enforcement Panel (Deutsche Prüfstelle für Rechnungslegung) as the first and the Federal Financial Services Authority (Bundesanstalt für Finanzdienstleistungsaufsicht – BaFin) as the second stage. For further details of changes to the German enforcement system see Ernstberger et al. (2010).

Using market capitalisation (stock price times number of shares outstanding) or the book-to-market ratio as independent variables in our model may be problematic because we deflate forecast errors with stock prices. Using variables on both sides of a regression equation may lead to biased results.

Additional information

Notes on contributors

Martin Glaum

Paper accepted by Salvador Carmona.

Jörg Baetge

Paper accepted by Salvador Carmona.

Alexander Grothe

Paper accepted by Salvador Carmona.

Tatjana Oberdörster

Paper accepted by Salvador Carmona.