The value relevance of IFRS in the European banking industry

The International Accounting Standards (IAS) originally issued by the International Accounting Standards Committee (IASC) have evolved into the International Financial Reporting Standards (IFRS) issued by its successor body, the International Accounting Standards Board (IASB). IFRS comprise the standards issued by the IASB and those issued by the IASC, some of which have been amended by the IASB. IFRS have emerged as a leading alternative to US GAAP for global reporting. In 2005 the US Securities and Exchange Commission (SEC) laid down a roadmap to permit cross-listing on US exchanges without requiring firms to reconcile IFRS with US GAAP.

Holthausen and Watts (2001) note that accounting values have multiple uses beyond equity investment. However, Barth et al. (2001) reply that the possible contracting uses of financial statements in no way diminish the importance of value relevance research.

On the features of this model, see Lo and Lys (2000).

Quality is defined for operational purposes as the timely accounting entry of amounts relevant to the income statement (particularly losses).

The sample period for Barth et al. (2008) predates the standards issued by IASB. Accordingly, they refer to use of IAS, not IFRS, by their sample of firms.

Other works deals with the effects produced in one country are Abellán and Aledo (2007), Gjerde et al. (2008), or in few countries are Capkun et al. (2008), Sellhorn and Skaife (2008) and Prather-Kinsey et al. (2008). Very recent empirical evidence on value relevance of banks is provided by Anandarajan et al. (2010).

In Arellano’s words, it becomes possible "to control for possibly correlated, time-invariant heterogeneity without observing it" (2003, p. 8). Previous research based on panel models and addressing the market impact of accounting information is scant and does not bear on Europe (see Naceur and Goaied 2004; Negash 2006, Worthington and West 2004).

The fixed effects estimator is used when the unobserved effects are assumed (or found, using the Hausman test) to be correlated with the explanatory variables. Since this correlation is left unrestricted, an OLS method would produce an inconsistent estimator. So the data must be transformed to eliminate the unobserved effects. The transformed data do so by subtracting the time mean (within-panel) for all observations. In the panel data literature, the OLS estimator on data shorn of the time mean is generally referred to as the fixed-effects or least squares dummy variable estimator (LSDV). The latter term reflects the fact that it is numerically the same as the estimator that would be obtained by running an OLS regression including N dummies, one for each individual.

The fixed effects account for all time-invariant cross-sectional effects may then be observed or unobserved. The internal data transformation eliminates any time-invariant regressors, such as country-fixed effects. Similarly, when a LSDV estimator is used, the bank-fixed effects subsume all other time-invariant effects.

Leuz (2003) deals with Germany only, so he can hold that his work is not affected by these problems. Barth et al. (2006) regress the share price on country- and industry-fixed effects and then regress the residuals from this regression on book value of equity per share and net earnings per share.

Barth et al. (2006) use the same method to mitigate the potential influence of outliers. When, as an alternative way of coping with the presence of extreme values, we drop the observations lying in the first and last percentile of the distribution, results remain substantially unaltered (see Sect. 5.4).

When the unbalanced panel is converted into the balanced one, the number of banks falls from 221 to 81 and observations from 1,201 to 567.

A discrepancy between individual and joint significance, which will recur in other cases, is not unusual in multiplicative interaction models and can in fact be interpreted as a signal of multicollinearity (see Wooldridge 2003; Brambor et al. 2006) induced by the inclusion of interaction terms. As Brambor et al. (2006) point out, however, “even if there really is high multicollinearity and this leads to large standard errors on the model parameters, it is important to remember that these standard errors are never in any sense ‘too’ large—they are always the ‘correct’ standard errors. High multicollinearity simply means that there is not enough information in the data to estimate the model parameters accurately and the standard errors rightfully reflect this”.

According to Bernard et al. (1995) banks in Denmark report accounting information based on market values to a higher degree than other countries in Europe. Therefore, the different impact of mandatory IFRS may be due to this reason.

Actually, the figure used in determining the quartiles was the average market capitalization of each bank over the sample period.

This segmentation differs from the former because some cooperative banks are large (e.g. Crédit-Agricole).

In 2000 Bankscope provides the Thomson BankWatch ratings as well. In the same database, information is missing on Moody and Standard and Poor ratings for the year 2003.

To economize on space, we omit all the results so far commented, making them accessible on demand.

Besides, when we limited our estimation sample to the 33 banks (175 observations year-observations) that used IFRS before 2005, once again, we found an increasing value relevance of the earnings per share and a decreasing value relevance of book value per share, only the marginal impact of earnings being statistically significant after 2005.

We are indebted to an anonymous reviewer for this suggestion.