Cultural Impact on Earnings Management


Term Paper (Advanced seminar), 2014

27 Pages, Grade: 1,0


Excerpt

Table of Content

List of Tables

List of Figures

List of Variables

1. Introduction and Background
1.1 Hofstede’s cultural dimensions
1.2 Hypotheses

2. Research Design
2.1 Measurement of earnings management
2.2 Model specification

3. Sample Data and univariate analysis
3.1 Sample selection
3.2 Descriptive statistics

4. Results of the multivariate regression
4.1 Pooled sample
4.2 Results for each year

5. Conclusion

Appendix

References

Sources of Law

Source materials

List of Tables

Table 1: Distributions of Variables

Table 2: Correlation Matrix

Table 3: Regression results for the pooled sample

Table 4: Regression results for each year of the sample

Table 5: Hofstede's cultural values for each country of the sample

Table 6: Composition of the sample

Table 7: Variance Inflation Factors with Uncertainty Avoidance

Table 8: Variance Inflation Factors without Uncertainty Avoidance

List of Figures

Figure 1: Residual Plot

List of Variables

illustration not visible in this excerpt

Abstract

This paper analyses the effect of Hofstede’s cultural dimensions on earnings management on 433 firms in 18 european countries, all reporting under IFRS. The results of multivariate regression reveal that only one cultural variable, namely Power Distance, has a significant effect on earnings management measured by discretionary accruals. In contrast, the scores for Individualism and Masculinity have no significant influence. These findings suggest that culture still has an influence on earnings management as suggested by previous studies, but is diminishing due to international accounting harmonization.

1. Introduction and Background

The International Financial Reporting Standards (IFRS) became mandatory for every listed company in the European Union on January 1st, 2005.[1] This regulation was a result of a longlasting accounting harmonization process with the US-GAAP, with the intention to make annual financial reports at first sight more comparable for investors on an international level. However, this increased comparability is threatened by various factors out of which culture and earnings management are two.

Previous research shows that accounting choices differ systematically depending on the nationality of the accountant.[2] Thus, one may assume that IFRS application is heavily influenced by national GAAP, which emerged from various cultural and legal factors. As a consequence international comparability might be biased by different cultural values. In addition, earnings management is a significant threat for the comparability and fair presentation of companies’ financial statements, which became an important tool for managers to fulfill investors’ expectations.[3] Research shows that there are significant differences across nations in the use of earnings management.[4] However, there are many factors influencing this variation ranging from economic incentives to the institutional and legal environment. Since earnings management offers the broadest sphere of an accountant’s influence in creating the financial statements, it is also the domain where culture could reveal itself throughout the judgemental choices of the accountant. That is why this paper tries to link these two threats for comparability and examines the impact national culture has on earnings management and is closely related to a study by Guan and Pourjalali.[5]

The limited empirical research on the cultural influence on earnings management to date indicates that culture has a significant impact.[6] However, there is no study known to the author investigating the cultural effect on earnings management within the IFRS or any comparable coherent accounting system, which is crucial for uniform circumstances. It is questioned, whether accounting harmonization due to IFRS adoption diminishes cultural effects on earnings management or not.[7] Therefore, the results of this paper have to be compared with the results of Guan and Pourjalali , who tested the cultural influence over several different accounting systems.[8] The model to test the hypotheses is taken over from Guan and Pourjalali but with the addition of the Heritage index as a control variable.[9] Furthermore, this study is based on 433 firms in 18 european countries part of the EURO STOXX 600 all reporting under IFRS.

The remainder of this paper is organized as follows: the next part describes Hofstede’s cultural dimensions used as a proxy for culture, consequential hypotheses are formulated thereafter. The second chapter contains the model specification and research design, before sample selection and univariate analysis are presented in chapter three. Results from the regression and the tested hypotheses are analysed in chapter four. The last section summarizes and gives a short outlook for future research.

1.1 Hofstede’s cultural dimensions

The social psychologist Geert Hofstede defines culture as „the collective programming of the mind which distinguishes the members of one human group from another“[10]. In an extensive IBM employee survey across 72 countries and 116.000 questionnaires, he was able to identify four cultural variables that he labeled: Uncertainty Avoidance, Power Distance, Masculinity and Individualism.[11]

- Uncertainty Avoidance. Hofstede defines this value as „the extent to which the members of a culture feel threatened by uncertain or unknown situations“[12]. High uncertainty avoidance correlates with more regulation and the fear of changes in society.
- Power Distance. This dimension measures how countries handle social inequality and how minorities or less powerful citizens accept it. The higher the respective power distance index, the more inequality and hierarchy is accepted.[13]
- Masculinity. This cultural variable stands for the distinction of gender roles in society. Cultures with high masculinity emphasize gender differences and are driven by competition and achievement.[14]
- Individualism. This dimension describes the degree of interdependence of a society. In individualistic countries people are more self-sufficient, self-oriented and taking care of their immediate environment and family only.[15]

The above-mentioned measurable indices offer a useful tool to implement culture in empirical research, explaining its widely use. Hofstede’s work has also been closely linked to accounting values, trying to explain cross-country differences in accounting systems, making it applicable for empirical accounting research.[16] However, it has to be considered that Hofstede’s work is criticized for its methodology and one-company approach, so the results of this paper have to be evaluated cautiously.[17] Especially, one has to criticize that culture is restricted by national borders, suggesting that a country stands for one culture.[18]

For the purpose of this paper it is concentrated on the Power Distance, Masculinity and Individualism index in order to formulate hypotheses to explain the cross-country variation in earnings management.

1.2 Hypotheses

Countries that score high on Masculinity, focus more on financial data than those who score lower. In this context, earnings make economic achievement of a company visible and are a key figure for investors in a society that is more assertive and driven by success.[19] As a consequence, the first hypothesis is suggested:

H1: The higher a country ranks in terms of Masculinity, the higher the magnitude of earnings management.

In societies with large Power Distance, individuals accept their place in society. In a wider sense companies have less incentives in large Power Distance countries to manage their earnings because they tolerate their respective status quo and are more obedient to authority, meaning that they conscientiously obey the law. Giving this relationship, the second hypothesis is as follows:

H2: The higher a country ranks in terms of Power Distance, the lower the magnitude of earnings management.

In a broader sense, Individualism could represent a general proxy for a share- or stakeholder orientation of a company. Organizations in countries that score low in Individualism are expected to look after their stakeholders as an extended family in order to defend their interests. In that respect, earnings management serves as an instrument to protect stakeholder’s welfare threatened by missed debt convenants or analysts’ forecasts.[20] A positive correlation between collectivism and the magnitude of earnings management is expected. Consequently, the last hypothesis reads as follows:

H3: The higher a country ranks in terms of Individualism, the lower the magnitude of earnings management.

For the last cultural dimension, Uncertainty Avoidance, no hypothesis is formulated.

2. Research Design

2.1 Measurement of earnings management

In order to measure the dependend variable an accruals-based model, namely the modified Jones - Model[21], is used because it is seen as the most powerful model in detecting earnings management to date.[22]

The maximum extent to which a company can manage its earnings can be estimated through total accruals of a period, since they contain all non-cash revenues and expenses. For the purpose of detecting earnings management these total accruals are seperated into a nondiscretionary (NDA) and a discretionary part (DA). The nondiscretionary part embodies all accruals demanded by accounting standards or statutory law, whereas the discretionary accruals go beyond and capture all accruals set up at companies’ free judgement. Before this discretionary part can be calculated, a model is needed to derive the normal part of total accruals from equation (1).

(1) illustration not visible in this excerpt

where:

TAit = total accruals measured as net income before extraordinary items minus operating cash flow for firm i in year t.

NDAit = nondiscretionary accruals for firm i in year t.

DAit = discretionary accruals for firm i in year t.

Consequently, discretionary accruals are the proxy for earnings management and DAit from equation (1) is defined as the dependent variable of this study.[23] To assess nondiscretionary accruals, Jones assumes that these “normal” accruals change with the economic development of a company.[24] Therefore, nondiscretionary accruals for firm i in year t are estimated first before the discretionary part can be derived.

(2)illustration not visible in this excerpt

where:

Ait-1 = total assets for firm i at the end of year t-1.

∆REVit = change in revenue for firm i in year t.

∆RECit = change in net receivables for firm i in year t.

PPEit = gross property, plant and equipment for firm i at the end of year t .

αit, β1it , β2it = firm specific parameters for firm i in year t.

et = error term with normal OLS properties.

The estimation results from equation (2) lead to firm-specific parameters which form the basis for calculating the discretionary accruals. Nondiscretionary or “normal” accruals are estimated for every year and industry, afterwards fitted values are computed for every firm.[25] Consequently, equation (1) is used to obtain discretionary accruals for every firm i in year t.

2.2 Model specification

To test how culture influences the magnitude of earnings management, the unsigned discretionary accruals are regressed on Hofstede’s cultural indices in addition to control variables based on Guan and Pourjalali.[26]

illustration not visible in this excerpt

(3)

where:

DEit-1 = book debt to book equity ratio for firm i at the end of year t-1.

ASSETit-1 = total assets for firm i at the end of year t-1.

GROWTHit= one-year growth rate of a firm’s sales revenues.

POWERi = Power Distance index of firm i.

UNCERi = Uncertainty avoidance index of firm i.

INDIVi = Individualism index of firm i.

MASCi = Masculinity index of firm i.

HERit = Heritage index of firm i in year t.

b0,b1,… = model variables.

eit = error term with normal OLS properties.

The control variables ASSET and DE are taken over from Guan and Pourjalali to control for company size and debt convenants, respectively .[27] The magnitude of discretionary accruals is expected to rise with an associated increase in lagged assets and debt to equity ratio, because larger companies have more ability to manage their earnings and companies with a higher indebtedness have more incentives to do so. The one-year growth rate of a firm’s sales revenues is added as a control variable because accruals are expected to rise as sales increase.[28] Finally, the Heritage index is included, as suggested by Guan and Pourjalali , to control for investor protection and institutional factors for each country.[29] The country-based Heritage index takes into account 10 other indices concerning economic and financial freedom and is provided on a yearly basis.[30]

3. Sample data and univariate analysis

3.1 Sample selection

Data was obtained for the 7-year period from 2005 to 2012 for all 600 EURO STOXX companies from the Thomson Reuters Datastream.[31] The sample period begins in 2006 because IFRS became mandatory for all listed EU-companies in 2005.[32] 130 financial institutions and insurance companies are excluded, as well as 37 companies because of missing data, leading to a full sample of 433 firms. The EURO STOXX 600 companies are headquartered in 18 different countries, for which all Hofestede and Heritage indices are available. Table 1 shows all cultural values for each country in the sample. As already indicated, these values are time-invariant, so no further distinction between years is possible. Hofestede’s cultural indices are provided irregularly. For the purpose of this paper cultural values from 2001 are used.

illustration not visible in this excerpt

Table 2 shows the number of firms per country and their respective share of total observations. About 58% of all observations are based on the United Kingdom, France and Germany, making it a slightly more balanced sample compared to the study of Guan and Pourjalali.[33] Their study was to about 50% based on two countries. However, the results of this paper still have to be evaluated cautiously due to the high concentration on only three countries.

Firm-data in foreign currency are converted into EUR with historical closing rates for balance sheet items and historical average rates for income statement items.[34] Currency translation is necessary even though a percentage measure for discretionary accruals is applied, because the unstandardized lagged assets are used in the regression model.

illustration not visible in this excerpt

3.2 Descriptive statistics

Table 3 shows the descriptive statistics for all used variables. The average magnitude of unsigned discretionary accruals for the sample firms is about 4% of lagged assets whereas the median value is about 2.6%, suggesting that the DA distribution is skewed. Concerning the cultural values it is worth noting that there is a remarkable variation in the sample. For example the Masculinity index ranges from 5 (Sweden) to a maximum of 79 (Austria). This variation should facilitate to find a possible relationship between the discretionary accruals and the cultural values.

Unfortunately, it is not possible to compare the distributions of variables with the study of Guan and Pourjalali, because a corresponding descriptive statistic was not provided in their paper.[35]

illustration not visible in this excerpt

Table 4 reports the pearson-correlation matrix for the variables included in equation (3). For the dependent cultural variables, it is worth noting the following: The dimensions POWER and UNCER are highly significant correlated with the dependent variable whereas INDIV is only significant correlated on a 5%-level. In contrast, MASC shows no statistically significant relation to DA. It should also be noted, that the Heritage index correlates with the cultural values Individualism, Uncertainty Avoidance and Power Distance. Since the Heritage index represents economic freedom, it can be assumed that Power Distance and Uncertainty Avoidance rather inhibit economic freedom in contrast to Individualism.

With respect to the structure of correlation among the dependent variables it becomes apparent that multicollinearity might be a problem for the multivariate regression because UNCER is highly positive correlated with POWER (p = 0.8162). To further examine this problem the variance inflation factors (VIF) are calculated (see appendix, table 7) and confirm multicollinearity.

illustration not visible in this excerpt

To address this issue, UNCER the variable with the highest VIF is excluded from equation (3). After removing this variable, VIF values for the other independent variables are all below 2, so results of the multivariate regression are unlikely to suffer from multicollinearity (see appendix, table 8).

Furthermore the residual assumptions, namely homoscedasticity and no autocorrelation, are tested before carrying out the regression to assure that the Ordinary Least Square-estimator is the best linear-unbiased one. In Figure 1 in the appendix, the residuals are plotted against the fitted values, suggesting a certain structure and therefore autocorrelation. Breusch-Godfrey and Durbin’s alternative test for autocorrelation are carried out to gain more insight. The p-values of both tests are at 1%-level, so autocorrelation can be assumed. Residuals are not independent from each other, thus one of the Gauss-Markov assumptions is not satisfied. In addition, Breusch-Pagan and Szroeter’s rank tests were carried out, suggesting heteroscedasticity in the data, so residuals do not have the same variance. These problems might mainly arise from the time-invariant culture variables leading to correlation amongst the residuals from previous periods. In addition, missing variables in the model could cause these problems, because a number of different parameters have an influence on earnings management. To be able to apply the Ordinary Least Square-Regression, Newey-West standard errors corrected for heteroscedasticity and autocorrelation are used to adress these problems, making p-values useable while not changing the respective coefficients.

4. Results of the multivariate regression

To test the hypotheses the Ordinary Least Square method is used. First, results for the regression on the pooled sample are presented. In order to see if the results differ over time, results for each year are also displayed.

4.1 Pooled sample

Table 3 shows the multivariate regression results for the pooled sample. It can be seen

that only the Power Distance variable has a highly significant effect on the magnitude of discretionary accruals with the predicted sign. This indicates that the higher a country scores on Power Distance, the lower the magnitude of earnings management for resident firms – which is consistent with the second hypothesis. However, the other cultural values MASC and INDIV show no significant relationship with the dependent variable, which runs contrary to the results of Guan and Pourjalali .[36] This indicates that culture still has an impact on earnings management but it is diminishing due to IFRS application.

In addition, GROWTH and ASSET are highly significant at 1%-level, in contrast to DE and HER. The higher the growth of sales in a year, the larger the magnitude of earnings management. Interestingly, lagged assets have a slightly negative magnitude effect on earnings management in contrast to the study of Guan and Pourjalali, suggesting that the lower the assets of the previous year, the lower the magnitude of earnings management.[37] These results raise questions whether IFRS adoption could actually alter the relationship between DE and ASSET to discretionary accruals. Furthermore, it is apparent that the Heritage index is not related to the magnitude of earnings management.

The model only explains about 4% of total variance making it slightly worse than the study of Guan and Pourjalali, resulting mainly from the omitted tax rate and uncertainty avoidance variable.[38] However, the p-value of the F-Test is at 1%-level suggesting that the overall model is valid.

[...]


[1] (EG) No. 1606/2002.

[2] Cf. eg. Tsakumis (2007).

[3] Cf. Mulford/Comiskey (2002) p. 57-59.

[4] Cf. e.g. Leuz et al. (2003).

[5] Cf. Guan/Pourjalali (2010).

[6] Cf. e.g. Doupnik (2008), Han et al. (2008).

[7] Akman (2011) finds that culture still has an impact on disclosure practices within IFRS.

[8] Cf. Guan/Pourjalali (2010).

[9] Ib.

[10] Hofstede (2001) p. 9.

[11] Ib. p. 41.

[12] Ib. p. 161.

[13] Ib. p. 98.

[14] Ib. p. 297 ff.

[15] Ib. p. 225 ff.

[16] C.f. Gray (1988).

[17] e.g. Olie (1995) p.124-143 and McSweeney (2002).

[18] C.f. Baskerville (2003).

[19] C.f. Hofstede (2001) p. 383.

[20] I.b. p. 323.

[21] C.f. Jones (1991).

[22] C.f. Chen (2010).

[23] C.f. Wagenhofer/Ewert (2003), p.212-214.

[24] C.f. Jones (1991).

[25] Mnemonic WC06010 from Thomson Reuters Datastream is used for industry classification. After excluding financials, three industries are left in the sample: industrial, utility and transport.

[26] 26 C.f. Guan and Pourjalali (2010). The earnings and disclosure score are not taken over from Guan and Pourjalali because in their study, they had no significant results. In addition, the effective tax rate was not included because of missing data. The sample would have been reduced to 376 firms.

[27] Ib.

[28] C.f. Ernstberger et. al (2012)

[29] C.f. Guan and Pourjalali (2010).

[30] The Heritage Foundation http://www.heritage.org/index/explore?view=by-region-country-year [access: 22.01.2014]

[31] EURO STOXX 600 composition: 13.01.2014

[32] The sample does not begin in 2005 because lagged variables are used in the regression and to guarantee homoge nous data.

[33] C.f. Guan and Pourjalali (2010).

[34] Source: OANDA http://www.oanda.com/lang/de/currency/historical-rates/ [access: 22.01.2014].

[35] C.f. Guan and Pourjalali (2010).

[36] C.f. Guan and Pourjalali (2010).

[37] Ib.

[38] Ib.

Excerpt out of 27 pages

Details

Title
Cultural Impact on Earnings Management
College
Catholic University Eichstätt-Ingolstadt  (WFI)
Grade
1,0
Author
Year
2014
Pages
27
Catalog Number
V276790
ISBN (eBook)
9783656725787
ISBN (Book)
9783656725718
File size
651 KB
Language
English
Tags
Bilanzpolitik, Earning, Management, Culture, Kultur, Hofstede, Jones, EURO STOXX, Europa, IFRS, Accounting, GLOBE, Earnings Management, Accruals, Periodenabgrenzung
Quote paper
Stephan Küster (Author), 2014, Cultural Impact on Earnings Management, Munich, GRIN Verlag, https://www.grin.com/document/276790

Comments

  • No comments yet.
Read the ebook
Title: Cultural Impact on Earnings Management



Upload papers

Your term paper / thesis:

- Publication as eBook and book
- High royalties for the sales
- Completely free - with ISBN
- It only takes five minutes
- Every paper finds readers

Publish now - it's free