“Earnings management initiatives and selected characteristics of an entity: a case study of the Visegrad Four”

This paper responds to the current issue of Earnings Management (hereafter EM) ini- tiatives in the Visegrad Four. The aim is to identify EM practices and determine a statistically significant relationship between EM practices and firm size, country, or business sector. The paper contains a literature review, bibliometric analysis, and a de-scription of the methods used in the practical part. Many publications containing the term “EM” in scientific databases, a keyword science map for “EM” publications, a science map of the most cited contributions dealing with EM, a science map of countries with the highest number of cited publications with the term “EM”, and a science map of sources with the highest number of publications with the term “EM” are included. The analysis covers the period from 1900 to 2019. The study is focused on detecting the presence of EM initiatives in the Visegrad Four using the modified Jones model. Results confirmed the EM initiatives in the Czech Republic, Slovak Republic, Poland, and Hungary. Using the one-way ANOVA test, a statistically significant relationship was confirmed between EM practices and firm size, country, and business sector. The study uses a sample of 250 financial statements of entities from each Visegrad Four country for the year 2018. The sample is adjusted for outliers using the IQR method.


INTRODUCTION
This paper deals with EM initiatives since at least the 1960s. Initial research was focused on earnings management and capital markets. Two types of hypotheses can be distinguished, namely the mechanistic hypothesis and the efficient market hypothesis. Unlike the mechanistic hypothesis, the effective market hypothesis assumes that investors deal with the impact of accounting choices on the profit amount. It can be assumed that stock prices are determined by public available information, including the accounting policy. Later studies were focused on the impact of earnings management on the non-capital market ( Mulford and Comiskey (2002) defined the following motives for Earnings Management practices: • avoiding the risk of decline in stock prices if the earnings achieved are lower than expected; • creating the best picture of profit to maximize the selling price of stocks if the company is considering an IPO; • keeping profit at the level needed to get the maximum possible premium; • reducing profit volatility due to one-off items; • also changes in management give rise to new management efforts to show lower earnings in the first periods after accession, in order to attribute such a situation as a result of old management decisions.
The interest in the EM issue is constantly growing. Recently published studies bring new results in this area. However, most studies are conducted abroad and thus do not reflect the economic conditions of the V4 countries. The aim of the current paper is to identify EM practices and to determine a statistically significant relationship between EM practices and firm size, country, or business sector. The study is carried out in the economic conditions of the Visegrad Four countries.

LITERATURE REVIEW
Since the 1960s, a large number of studies have been created that examine EM from different perspectives. Many definitions of EM can be found. McKee (2005) defines EM as the purposeful and legal management of decision-making and reporting in order to achieve stable and predictable economic results. Davidson et al. (2005) define EM as the process of intentional steps within the limits of Generally Accepted Accounting Principles in order to achieve the required level of reported profit. According to Sankar and Subramanam (2001), EM is the use of flexibility in the selection of accounting methods that influence managers' decisions about future cash flows. Schipper (1989) refers to EM as the "disclosure management." Ronen and Yaari (2008) distinguish three types of EM, namely white earnings management, gray earnings management, and black earnings management. It can also be referred to as beneficialindicates the long-term value, neutral -reveals true short-term performance, and harmfulconceals the long-term or the short-term value.
The composition of profit from two components, cash flow and accrual, leads companies to two forms of EM, namely real earnings management and accrual-based earnings management. Entities are able to manage profits through appropriate operational decisions that have an impact on cash flow from operating activities. Such EM is referred to as "real earnings management" (Rowchowdhury, 2006). The second option is to adjust the accrual in order to achieve the required level of profit. Such EM is referred to as "accrual-based earnings management" or "accounting earnings management". The possibility for such profit management is provided by the accounting standards, principles, or laws themselves (

Bibliometric analysis
The use of statistical methods to analyze articles, publications, or books is known as bibliometrics. Scientometrics, also defined as the sub-field of bibliometrics, deals with the analysis of scientific publications. Bibliometric analysis is widely used in various fields of science. Most often, bibliometrics, or scientometrics, is associated with the evaluation of science at various levels. Historical development in selected science fields, as well as hidden relationships between issues, authors, or disciplines, can be defined based on the results of bibliometric analysis. Bibliometric methods are also used to identify the most current topics of scientific research or the level of their obsolescence.
The following bibliometric analysis presents the topicality of the Earnings Management issue. The bibliometric analysis is divided into two parts based on the selected period (1900-2000; 2000-2019). Table 2 shows the characteristics of conducted bibliometric analysis.    Figure  1 shows the number of publications with the term "Earnings Management" in Social Science Research Network.
Social Science Research Network's eLibrary provides 930,539 research papers from 480,575 researchers in more than 50 disciplines. 1,495 records containing the term "Earnings Management" were posted from 1995 to 2020. 24 records have been posted in 2020 (till May 12, 2020) and nine records do not have the date of posting. The oldest record, Dividend-Based Earnings Management: Empirical Evidence from Finland, was posted on June 12, 1995.
Another database, Web of Science, has 1,875 publications containing the term "Earnings Management" published from 1900-2020 (till May 12, 2020). 44 publications have been published in 2020. The first was published in 1969. Figure 2 shows the development of the number of publications containing the term "Earnings Management" in 1900-2019. Hidden relationships between issues, authors, or disciplines, can be defined based on the science maps. Science maps have been used since the 1970s. Science maps can be defined as representations of scientific fields or organizations in which the elements of the map are associated with topics or themes (Noyons, 2001). The elements in the map can be publications, institutes, authors, instruments, scientific topics, etc. It represents relations among the elements. There are various software for creating science maps. VOSviewer version 1.6.14, released on the 27 th of January, 2020 was used.    2009  2007  2005  2003  2001  1999  1997  1995  1993  1991  1989  1987  1985  1983  1978  1976  1969  Number of records Year It was mentioned above that initial studies were focused on earnings management and capital markets. A keyword science map for Earnings Management publications from 1900 to 2000 shows that publications dealing with earnings management also contain keywords such as ac-cruals, performance, stock returns, market efficiency, stock prices, dividends, decisions, information, etc. The size of the circle represents the weight. The larger it is, the higher weight the keyword has. The color of the circles determines the cluster to which the keyword belongs. Lines repre-

DATA, METHODOLOGY AND HYPOTHESES
This section describes data, methods, and hypothesis development. Data for research were obtained from the Amadeus database providing     Table 1 shows that there are several groups of methods for detecting EM. Mathematical modeling of total discretionary accruals using time series data, specifically the modified Jones model, was used to determine earnings management practices. Based on some studies, the modified Jones model provides the most powerful determination of EM initiatives (Dechow et al., 1995). So-called discretionary accruals (proxy, abnormal accruals) are defined using the modified Jones model. Abnormal accruals are evidence of involved EM practices in firms. The greater the value of abnormal accruals, the lower the quality of the reported economic results. Based on the modified Jones model, discretionary accruals are calculated using the following formulas: ε is a statistical error, NDA means non-discretionary accruals, and DA denotes discretionary accruals.
Basically, two methods are known for calculating total accruals: • the balance sheet approach, and • the statement of cash flows approach (Hoglund, 2012).
According to the balance sheet approach, total accruals are determined as: , where TA is the sum of total accruals, CA ∆ is the change in current assets, CL ∆ is the change in current liabilities, Cash ∆ is the change in cash and cash equivalents, STD ∆ is the change in the current maturities of long-term debt and other short-term debt included in current liabilities, and Dep denotes depreciation and amortization expenses.
To verify the existence of earnings management practices, the fictitious file within which the values of discretionary accruals equal zero was set. It means that EM practices are not expected. It allows to find out whether the level of discretionary accruals of a fictitious file is significantly different from the level of discretionary accruals of the analyzed set of V4 companies. The Mann-Whitney nonparametric test was used. If the calculated p-value is less than the selected significance level 0.05, α = the null hypothesis is rejected and hypothesis H 1 is accepted.
The aim of this paper is to identify EM practices and to determine a statistically significant re-lationship between EM practices and firm size, country, and business sector. These hypotheses were set as follows: There is a statistically significant relationship between EM practices and firm size.
There is a statistically significant relationship between EM practices and a country.
There is a statistically significant relationship between EM practices and a business sector.
The one-way ANOVA test was used to verify the hypotheses. If the calculated p-value is less than the selected significance level 0.05, α = the null hypothesis is rejected and hypothesis H 1 is accepted.
All calculations were performed using statistical analysis software in Excel XLSTAT and are presented in the next part of the contribution.

EMPIRICAL RESULTS AND DISCUSSION
According to Hoglund's balance sheet approach, the sum of total accruals was calculated, as well as variables for the modified Jones model. Table  3 shows the results of the variables calculation for five selected companies. It is not allowed to name the company, so the general designation in the form of country serial number is used.
Some of the calculated values show deviation from the other. These values are called outliers. Their use in further calculations could falsify the results. IQR (interquartile range) method of outlier detection was used in sample preparation as was mentioned in the methodology part of this study.
In the next steps, the modified Jones model was used to estimate discretionary accruals. Within different studies, several criteria are used to estimate the explanatory power of the model in the EM initiatives calculation. Table 4 represents these criteria, their values, and authors of the studies who used these criteria to describe the explanatory power. Table 5 shows the explanatory power of the used model within countries. The countries are ranked from those in which the model used achieves the highest explanatory power to those in which the explanatory power of the model is the lowest according to selected criteria. Note: * Discretionary accruals are too small compared to the value of the total assets. The natural logarithm of total assets is recommended to use to describe firm size. R is the coefficient of determination, is the number of elements in the sample, is the number of independent variables. ** If the calculated p-value is less than the selected significance level 0.05, α = the null hypothesis is rejected and hypothesis H 1 is accepted. All calculations were performed using statistical analysis software in Excel XLSTAT and are presented in the next part of the paper.

Testing hypotheses
Subsequently, the study is focused on testing the set hypotheses.
There is a statistically significant relationship between EM practices and firm size.
There is a statistically significant relationship between EM practices and a country.
There is a statistically significant relationship between EM practices and a business sector.
One-way ANOVA test was used to verify the hypotheses. If the calculated p-value is less than the selected significance level 0.05, α = null hypothesis is rejected and hypothesis H 1 is accepted.
All calculations were performed using statistical analysis software in Excel XLSTAT, and results are presented in Table 6.

CONCLUSION
Interest in the EM problem is proved by the conducted bibliometric analysis. Since 2000, the number of published studies on various aspects of EM has increased significantly. The aim of the paper was to identify EM practices and to determine a statistically significant relationship between EM practices and firm size, country, or business sector. The study was carried out in the economic conditions of the Visegrad Four countries. The modified Jones model was used to verify the existence of earnings management practices in the V4 countries. The results show that there are EM practices in the V4 countries. Also, the explanatory power of the model used was calculated. The modified Jones model is a worldwide model used in EM measurements. It can be seen that the highest explanatory power is in Poland, the lowest explanatory power of this model was observed in Hungary. The study was focused on testing set hypothesis. A statistically significant relationship between EM practices and a country was not proved. As mentioned above, many studies were carried out abroad. It can be seen that the EM issue is published worldwide in many journals and is an important topic at many conferences. There are not many studies focused on the V4 countries. Based on the results of the Jones modified model, the existence of EM initiatives in the V4 countries was proved. By confirming the hypothesis, it can be stated that EM initiatives are used in all the V4 countries. The results of research in this area mainly reflect the situation abroad. Given the above, the EM issue can be considered as current, and it is necessary to deal with it not only

CZ
There is a statistically significant relationship between EM practices and firm size. There is no statistically significant relationship between EM practices and a country. There is a statistically significant relationship between EM practices and a business sector.

SVK
There is a statistically significant relationship between EM practices and firm size.
There is no statistically significant relationship between EM practices and a country. There is a statistically significant relationship between EM practices and a business sector.

HU
There is no statistically significant relationship between EM practices and firm size.
There is no statistically significant relationship between EM practices and a country.
There is no statistically significant relationship between EM practices and a business sector.

PL
There is no statistically significant relationship between EM practices and firm size.
There is no statistically significant relationship between EM practices and a country. There is a statistically significant relationship between EM practices and a business sector.
Note: Significance level 0.05 α = was used within testing all hypotheses.
abroad, but also within the V4 countries. The results of such studies will better reflect the specifics and economies of these countries. A statistically significant relationship between EM practices and firm size in the Czech Republic and Slovak Republic was proved. There are various ways to calculate EM initiatives. Based on some research, the modified Jones model provides the most powerful definition of EM initiatives. It is a variation of the original Jones model proposed by Dechow et al. in 1995. However, neither model takes company size into account in the calculations. In some countries, it has been shown that there is a statistically significant relationship between business size and EM initiatives. The next study will focus on how the explanatory power of a given model is affected in samples of companies of different sizes. A statistically significant relationship between EM practices and a business sector in the Czech Republic, Slovak Republic, and Poland was proved. Based on proven hypotheses, companies from different industries should be taken into account when examining the explanatory power of models.
Further research will focus on finding the explanatory power of other models in the study of EM, as well as identifying other factors influencing EM initiatives.