Reliability of Graphs Disclosed in Annual Reports of Central and Eastern European Banks

Abstract

This article investigates the changes in the reliability of graphs disclosed in the annual reports of Central and Eastern European (CEE) commercial banks during crisis and noncrisis periods. The analysis covers thirty-three commercial banks from seven CEE countries during 2006–2013. The results reveal that, on average, one-third of the graphs disclosed by a bank violate at least one graph-construction principle and there appears to be a favorable measurement-distortion bias. In line with expectations, there are signs that the impression motivation of managers increases during crisis periods—severe graph distortions are twice as frequent, and distortions of financial indicator graphs increase.

Keywords:

• banks
• crisis
• financial reporting
• impression management

JEL Classification:

• G01
• G21
• M41
• M42

ACKNOWLEDGMENTS

The author is grateful to Nele Tamme and Anette Järve for their contributions in preparing the preliminary version of the dataset. The author also appreciates the useful feedback to the earlier drafts of this article provided by colleagues in the Department of Finance and Economics at Tallinn University of Technology and the constructive criticism provided by five anonymous referees.

FUNDING

This work was supported by Tallinn University of Technology under grant B45 “Economic Crises in Europe: Information, Risks and Government Policy” and grant B57 “Efficiency in Financial Sector in Light of Changing Regulatory Environment.” This article represents the work conducted in connection with these grants. The funding providers had no role in the research process from the study design to submission.

Notes

1. The chairman’s statement, director’s report, and management commentary are usually part of the front-end of the annual report. These contain an overview of the company’s goals, strategy, activities, future outlook, risk factors, governance, and corporate social responsibility issues, or provide comments on the financial results.

2. According to ISA 720, “The Auditor’s Responsibilities Relating to Other Information in Documents Containing Audited Financial Statements” (http://www.ifac.org/system/files/downloads/a040-2010-iaasb-handbook-isa-720.pdf), for annual reports drafted after December 15, 2009, the auditor does not have a responsibility to determine whether other information outside of the financial statements is properly stated. The Auditing and Assurance Standards Board (IASB) has specified that without any special requirement, the auditor reads the other information to detect material inconsistencies between the audited financial statements and other information. Considering the vague definition of “other information” and of auditors’ responsibilities, several steps have been taken since December 2009 taken to revise this standard (http://www.iaasb.org/projects/auditor-s-responsibilities-relating-other-information).

3. In the multivariate models used by Muiño and Trombetta (2009), the dependent variable was the cost of equity, and one of the explanatory variables was an indicator of measurement distortions in graphs.

4. It is also possible to consider the agency problems between shareholders and creditors (Berger, Herring, and Szegö 1995). However, the smaller the shareholdings of managers, the higher are the agency costs arising from information asymmetries between managers and shareholders, and the lower are the agency costs of debt (Jensen and Meckling 1976). As indicated in Bouwens and Verriest (2014), bank managers’ shareholdings tend to be very low, remaining on average at 0.03% of the bank’s share capital, referring to the low impact of agency costs of debt on bank disclosures.

5. Selectivity refers to a situation where a graph is disclosed only if the graphed variable or the company’s overall performance has improved and the graph depicting the same indicator would otherwise be omitted.

6. One could argue that the macroeconomic situation in the home country of the parent company could affect the performance of its subsidiaries in a given host country. However, as the annual reports of subsidiaries usually focus mainly on the performance of their local operations, their disclosure decisions are likely to be affected mainly by local and subsidiary-specific factors.

7. Stricter monitoring of banking activities arises from the interlinkages between banking activities and the real economy. These associations have been thoroughly covered in theoretical models based on a financial accelerator framework (Bernanke, Gertler, and Gilchrist 1999).

8. Regulators’ information demands could be expected to resemble those of professional investors. Professional investors have been shown to rely more on timelier information sources and to concentrate less on annual report front-ends (Cascino et al. 2014).

9. Most empirical articles focusing on discretionary disclosures employ rather restricted samples. This relates mainly to data availability issues and to the use of resource-demanding methodologies. It is usually assumed that larger firms should have higher-quality disclosures. Therefore, by focusing on larger companies it should be possible to determine whether an improvement in disclosure quality is needed.

10. The list of banks includes: Ceska Sporitelna (CZ), Ceskoslovenska Obchodni Banka (CZ), Komercni Banka (CZ), Unicredit Bank Czech Republic (CZ), Ceskoslovenska Obchodna Banka (SK), Prima Banka Slovensko (SK), Tatra Banka (SK), UniCredit Bank Slovakia (SK), Vseobecna Uverova Banka (SK), CIB Bank (HU), Erste Bank Hungary (HU), K&H Bank (HU), MKB Bank (HU), OTP Bank (HU), Raiffeisen Bank (HU), UniCredit Bank Hungary (HU), Abanka Vipa (SI), NLB dd-Nova Ljubljanska Banka (SI), Nova Kreditna Banka Maribor (SI), SKB Banka (SI), UniCredit Banka Slovenija (SI), SEB Pank (EE), Swedbank (EE), DnB Banka (LV), SEB Banka (LV), Parex Banka (LV), Swedbank (LV), Bankas SNORAS (LT), DNB Bankas (LT), SEB Bankas (LT), Ukio Bankas (LT), Swedbank (LT), and Danske Bank (LT). Some annual reports are missing starting from 2011 due to the bankruptcies of the Lithuanian Bankas Snoras and Ukio Bankas. The Slovakian Ceskoslovenska Obchodna Banka has had no annual report available since 2011.

11. In order to check whether graphs disclosed in bank annual reports issued in English differ from those prepared in the local language, one bank was randomly selected from each country and two of its (randomly selected) reports were cross-checked. The results indicated that within the fourteen pairs of reports, graph counts matched and same-looking graphs were disclosed in both languages.

12. The percentage of agreement between the two coders after initial screening was 93.4%. Coding disagreements were thereafter rechecked and adjusted according to the coding rules.

13. The same KFVs were identified for CEE banks in Laidroo and Tamme (2016). However, other previous studies focusing on nonfinancial companies used a rather different set of KFVs, including earnings per share, dividends per share, or profits. One could argue that indicators of risk-taking by banks could be more relevant as bank KFVs. Considering the very low count of such graphs in the reports of CEE banks, such indicators cannot be used for conducting a quantitative study. Therefore, a traditional approach, focusing on the most frequently graphed financial indicators, was preferred.

14. The initial list was significantly longer, including all the items mentioned in Beattie and Jones (1997, 1999). However, only seven violations of graph-construction principles were observed within a given dataset.

15. As in many previous articles, measurement distortions in pie charts are not considered, as this would require a slightly different measurement distortion detection approach (see Beattie and Jones 1994).

16. In this article, the GDI calculation has been corrected accordingly if the graph had a reversed axis.

17. In the case of cost and risk indicators, the calculations take into account that a positive trend would refer to a decrease in a variable and a negative trend to an increase in a variable.

18. A uniform definition of crisis period for all countries is not used in this article because of the quite significant differences in the intertemporal patterns of real GDP growth.

19. The regression models were initially also estimated with bank and year fixed effects; however, the Hausman test indicated that fixed-effects models should not be preferred to a random-effects model. As a random-effects model also enabled the inclusion of country dummies, the latter specification was preferred in this article. The χ² statistics of the Hausman tests are shown underneath the tables presenting the results.

²⁰ The principal component analysis showed that if these six variables were considered, only the eigenvalue of the first component was significantly above 1 and this component explained 40% of the variance. The coefficients for variables in the first component were as follows: E/A 0.495, L/A −0.556, LC/A 0.003, ROA1 0.555, LIQ/A 0.347, and MS 0.132. A Keiser-Meyer-Ohlin test confirmed the suitability of using these six variables for the principal component analysis.

21. The possibility of using year dummies instead of the crisis dummy to control for intertemporal effects was considered. The estimations (available from the authors upon request) showed that the coefficients of year dummies did, in most cases (with the exception of 2009), remain insignificant. Considering that this article focuses on crisis periods lasting longer than one year, the crisis dummies were preferred.

22. Reversed time series axis means that when moving from left to right, the years are listed in descending order instead of ascending order. If the graphed variable has deteriorated and the reader does not notice the ordering of years, it is possible to create the impression of an increasing trend.

Additional information

Funding

This work was supported by Tallinn University of Technology under grant B45 “Economic Crises in Europe: Information, Risks and Government Policy” and grant B57 “Efficiency in Financial Sector in Light of Changing Regulatory Environment.” This article represents the work conducted in connection with these grants. The funding providers had no role in the research process from the study design to submission.

Notes on contributors

Laivi Laidroo

Laivi Laidroo is Associate Professor in the Department of Finance and Economics, Tallinn University of Technology, Estonia.