The Impact of the International Accounting Convergence on Income Smoothing in Brazillian Companies

In Brazil, the convergence to IASB standards began in 2008 with the entry into force of the Law n. 11.638/07. This study has as objective to verify the influence of the convergence process to the international accounting standards in income smoothing level in Brazilian companies. The study has a quantitative approach, based on the model of Barth, Landsman and Lang (2008), was carried out from the financial statements of 133 companies in the period of 2005-2007 (pre-IFRS) totaling 344 observations, as well as 106 companies in the period of 2010-2012 (post-IFRS), totaling 289 observations. The financial statements were extracted from Thomson ONE Banker website. In the data analysis, were applied statistical techniques of multivariate linear regression, F test and Pearson correlation. The results show a rising in income smoothing levels after convergence. It is concluded that the effectiveness of the process of convergence of accounting standards of the country to international standards, in order to get better quality of accounting information, does not depend only on the convergence of accounting standards.


INTRODUCTION
ccounting, as the main provider of information of companies, plays a key role in trade and financial transactions.The differences between accounting standards of the various countries represented an obstacle to this process as it hindered business language by the lack of uniformity in the assessment of the economic events of companies.
The accounting standards convergence of each country to international accounting standards is important to mitigate these barriers.In this sense, the International Accounting Standard Board (IASB) played a key role, to develop a set of high quality accounting standards, which could serve as a global standard reference for the preparation of financial statements, called International Accounting Standards (IAS) and International Financial Reporting Standards (IFRS).
In Brazil, this process of convergence to the IASB standards has as key milestones the creation of the Comitê de Pronunciamentos Contábeis (CPC) in 2005, the Instruction N. 457/07 of the Brazilian Securities Commission (Comissão de Valores Mobiliários -CVM), which led to public companies the preparation of consolidated financial statements under IFRS from 2010, in addition to the Law N. 11.638/07, which amended several provisions of Law N. 6.404/76.The convergence of local accounting standards to the IASB standards aims to reduce the distortion of information reported by companies from an international trade scenario and search for developing foreign capital flows.It is expected, therefore, a reduction in information asymmetry between the various users of financial information, such as investors, analysts, business partners and other stakeholders in different countries.
The convergence of accounting standards at the international level motivated the research studies on the impact of this process in the financial statements.Some of these studies address the impact of IFRS on the capital market (ASHBAUGH; OLSSON, 2002;BENEISH;YOHN, 2008), others investigate its effects on the cost of capital injections (ERNSTBERGER;VOGLER, 2008), the protection to investors (RENDERS; GAEREMYNCK, 2007).Other investigated the adequacy of local levels of GAAP to IFRS (DING; SU, 2008).There is still others that examined the economic consequences of the adoption of IFRS (CHRISTENSEN;LEE;WALKER, 2007;DASKE et al., 2008).

A
The Impact of the International Accounting Convergence on Income Smoothing in Brazilian Companies 3 BBR, Braz.Bus.Rev. (Engl.ed., Online), Vitória, v. 12, n. 2, Art. 1, p. 1 -24, mar.-apr. 2015 www.bbronline.com.brConvergence to IFRS also raised queries related to earnings management (GR), but the results of international studies disagree on the effects of convergence on GR levels.Several studies show that certain contextual characteristics of countries can influence the effects of convergence on the GR, as investor protection mechanisms ( DYCK; ZINGALES, 2004;RENDERS;GAEREMYNCK, 2007), legal system based on common law or code law (LEUZ; NANDA; WYSOCKI, 2003), local accounting standards based on rules (VAN Tendeloo;Vanstraelen, 2005;GONCHAROV;ZIMMERMANN, 2006), access conditions to capital market (JEANJEAN; STOLOWY, 2008), more effective system of mandatory adoption of IASB standards and greater regulation of corporate governance (CHEN et al., 2010).
In and the effects of inflation on management practices.They point out that not all the earnings management models used in international studies are suitable to Brazilian reality.
Studies on the subject that analyze companies from various countries, such as Yoon (2007), Barth, Landsman and Lang (2008) and Chen et al. (2010) do not show a comparison of the effects of convergence between countries' GR, separating groups in these countries, e.g., in accordance with their contextual characteristics.To investigate the influence of the international accounting convergence process in Brazilian companies, in relation to income smoothing, a form of GR, can contribute to a better understanding of the benefits of convergence, in terms of quality of accounting information.is the influence of convergence with international accounting standards in income smoothing level in Brazilian companies?The objective of the study is to assess the influence of convergence with the IASB standards on the level of income smoothing in Brazilian companies.
Despite expectations about the effects of the international accounting convergence process, are still incipient studies on the subject in Brazil.As this convergence process in Brazil is recent, it is justified to study that seeks to identify its influence in the GR of the companies through the income smoothing approach.The improvement over the research of Barth, Landsman and Lang (2008) stems from the fact of been considered a distinct field of study, with different contextual characteristics and the outstanding economic importance of Brazil, besides considering a longitudinal period ex-ante and ex -post to convergence.

EARNINGS MANAGEMENT
In addition to term earnings management (gerenciamento de resultados -GR), Paul (2007) also cites the income smoothing, conservative accounting for reduction of current profits for the sake of future profits or big bath accounting and make-up of financial statements or window dressing.
Two of the most found definitions of GR in academic, international or national literature, are presented by Schipper (1989) and Healy and Wahlen (1999).Schipper (1989, p. 92) considers the GR as a "purposeful intervention in the foreign disclosure process, with the intention of obtaining some private gain."With this definition, Schipper limits the discussion of GR to external disclosures, excluding managerial accounting reports.
For Healy and Wahlen (1999, p. 368), the GR occurs when administrators use as judgment in the reported financial information and decisions relating to operating activities, in order to "change financial information, or mislead some investors on economic performance the company or to influence contractual outcomes that depend on reported accounting numbers." In these perspectives, the GR implies modification of the reported financial information, in order to gain some individual manager, or to allow the company to present a more favorable situation to its stakeholders.Santos and Paul (2006, p. 16) point out that "the possibility of distortion of accounting information usually comes from the existence of alternative criteria for recognition, measurement and/or disclosure within the accounting system."It gives opportunity to administrators choose one of many meaningful alternatives to portray the information in the most convenient form and may distort the analysis of business performance.
The motivations or incentives for GR practice have been the subject of several studies.Watts and Zimmerman (1986), from the Positive Theory approach, developed three hypotheses to guide studies on the explanation of the adoption of certain accounting practices by companies: incentive plan, indebtedness level and political costs.Healy and Wahlen (1999) reported three incentives should be based on empirical research in GR: capital markets, contractual incentives and regulation.Martinez (2001) also adds the political costs in regulatory reasons for the existence of GR.Martinez (2001) states that there are different types of GR, depending on the motivations involved in the process, but highlights three main types.The modality of the target earnings consists in managing to achieve certain predetermined profit targets, which may be higher or lower than earnings.In income smoothing, the objective is to reduce the variability of results, remaining stable results in a certain level.In conservative accounting (big bath accounting), the company worsening its current income in order to improve future incomes.
Several models or methodologies for investigation or detection of GR by the companies are found in literature, in qualitative or quantitative models.According to Martinez (2001), the quantitative approach is based on three approaches: distribution of frequency analysis of specific accrual and analysis of aggregate or total accrual.
The uniformity in the frequency distribution of income was associated with GR by Burgstahler and Dichev (1997).It consists of the empirical evidence of a small number of companies that have small losses and a large number of companies with small profits.The main advantage of this methodology is not having to estimate accruals, however, does not analyze how the GR was done (TRAPP, 2009).
The analysis of specific accruals, according to Trapp (2009), aims to analyze the behavior of certain classes of companies or specific accounting accounts.It requires large number of observations.The methodology of the aggregate or total accruals was first used by Healy (1985).The objective is to estimate the discretionary component of the accruals.
Total accruals are divided into: Discretionary and non-discretionary.The discretionary part of the accruals is manageable, or rely on the judgment of the manager.

MANAGEMENT RESULTS AND CONVERGENCE TO IFRS
The relationship between GR and the convergence of accounting standards at the international level is discussed in several studies, but the results are not similar.Some found no significant differences between the reports on IAS/IFRS and reports in local GAAP (VAN TENDELOO;VANSTRAELEN, 2005;JEANJEAN;STOLOWY, 2008;ELBANNAN, 2011).Others point increase in GR after the adoption of IASB standards (GONCHAROV; ZIMMERMANN, 2006;PAANANEN;LIN, 2009).
Jeanjean and Stolowy ( 2008) observed that the GR levels of UK and Australian companies did not decrease after the adoption of IFRS.In French companies, there was an increase in GR after the convergence to IFRS, which can increase the information asymmetry between managers and external users of financial statements of these companies.
By studying the adoption of IFRS by Egyptian companies, Elbannan (2011) found no significant evidence of reduced GR after the process of convergence.He attributes these results to the lack of conformity between the preparers of financial statements, inappropriate regulatory mechanisms, poor infrastructure accounting in that country and inadequate training of professionals.
According to Van Tendeloo and Vanstraelen (2005), Goncharov and Zimmermann (2006) and Jeanjean and Stolowy (2008), the process of convergence to the IASB standards, more flexible, because they are based on principles rather than rules, may result in increased GR levels, particularly in countries with accounting standards based on rules, with weak stock market and little protection to investors.This combination of factors can lead to a GR setting higher under the IASB standards, in relation to local accounting standards.In this sense, it was elaborated the research hypothesis as follows: H 1 -the process of convergence with international accounting standards increased the earnings management in Brazilian companies.
On the other hand, there are studies that point to improvement in the quality of reports based on IAS/IFRS (ASHBAUGH; PINCUS, 2001;YOON, 2007;BARTH;LANDSMAN;LANG, 2008;CHRISTENSEN;LEE;WALKER, 2008;CHEN et al., 2010;IATRIDIS, 2010;IATRIDIS;ROUVOLIS, 2010).Yoon (2007), when analyzing the accounting convergence process in 41 countries, found some evidence of a positive association between the convergence of accounting standards and the improvement of accounting quality.One of the proxies used was the quality of accruals related to GR. Christensen, Lee and Walker (2008) observed in their research a quality improvement with the adoption of IFRS only for companies with incentives to adopt IASB standards.
Barth, Landsman and Lang (2008), researching companies from 21 countries, found that those that adopt IASB international standards have less evidence of GR, more timely loss recognition and higher value relevance to the financial statements.Iatridis (2010), in search of the adoption of IFRS in British companies, obtained the same results as Barth, Landsman and Lang (2008).
However, the results of study and the Iatridis and Rouvolis (2010) in Greece companies show a small increase in the GR soon after convergence and reduction in the following periods.In general, there noted improvements in the quality of accounting statements with the adoption of IFRS.
When researching companies from 15 countries of the European Union, Chen et al. ( 2010) found a reduction of GR levels after the adoption of IFRS.These authors add that the improvement of accounting quality is attributable to IFRS, rather than changes in the managerial incentives, institutional characteristics of the capital market and general business environment.
Furthermore, based on the results of research Barth, Landsman and Lang (2008), Iatridis (2010), and Iatridis and Rouvolis (2010), the high requirements of disclosure and the quality of financial reports required by the IASB standards lead to reduction in GR levels, especially in countries with strong stock market and robust mechanisms for the protection of investors.In this perspective, we elaborated the following research hypothesis: H 2 -the process of convergence with international accounting standards decreased earnings management in Brazilian companies.
Barth, Landsman and Lang (2008) weigh in their study that the results of the companies may be influenced by other factors not attributable to the financial reporting system of the entity, reflecting the GR measures.The authors used factors such as size, sales growth, percentage change in the preferred shares, debt ratios, cash flow and auditing, as control variables to ensure the reliability of the metric used to GR.
It appears that the convergence of accounting standards at the international level, with the adoption or adaptation to the accounting standards of the IASB, may result in greater or lesser flexibility in comparison with the local rules of the IASB.This process can provide, according to some studies presented here, the GR, considering the opportunistic behavior of the agent (preparer of financial statements).

RESEARCH METHOD AND PROCEDURES
This section presents the population and sample of the research, as well as procedures for collecting and analyzing data.

PROCEDURES FOR DATA COLLECTING AND ANALYZING
The study used as proxies of GR three measures related to income smoothing, based on the model adopted in Barth, Landsman and Lang (2008).The metrics used to measure GR levels of Brazilian companies in the pre-convergence period (period 1) and post-convergence period (period 2) are described below.
Metric 1: Variability of changes in net income (ΔNI) Dependent variable: Where: LO n = Net operating income for the n period (includes sales minus all operating expenses, but does not include the financial results nor other non-operating income and expenses); AT n = Total Assets in the n period.Equation 1: Independent variables (of control): The variability of ΔNI is its variance of residuals of Equation 1, applied in the preand post-convergence.The smaller the variation, the greater the evidence of income smoothing (earnings management).
The Where: CFOn = Operating cash flow in the n period; ATn = Total Assets in the n period.
Equation 2: Independent variables (of control): The same metric as 1.
Equation 4: Independent variables: The same as metric 1, excluding the CF variable.

ACC _ CF n  (ACC,CF)
Were compared correlations of the residuals of Equations 3 and 4, instead of correlations between CF and ACC, directly.A negative correlation is an indicator of GR, since companies tend to influence their accruals up when the cash flows appear smaller.

DATA DESCRIPTION AND ANALYSIS
First, it presents the descriptive statistics for the test variables and control variables used in the regression equations, as shown in Table 1. of changes in net income; ΔCFO = variability of changes in Operating Cash Flow; ACC = accruals; CF = cash flow; SIZE = natural logarithm of total assets; GROWTH = percentage of change in sales; LEV = total liabilities divided by total equity at the end of the year; DISSUE = percentage change in total liabilities; TURN = sales divided by total assets at the end of the year; CF = annual net cash flow from operating activities divided by total assets at the end of the year; AUD = dummy variable, with 1 being to audited company by PwC, KPMG, E&Y or D&T, and zero for others; XLIST = trading of shares on the stock exchange in the United States of America, 1 for yes and zero for no; CLOSE = Percentage of shares held by individuals or entities related to the company, as directors and their families, foundations, among others.Source: research data.
The test variables shown in Table 1 demonstrate greater standard deviation in the preconvergence period, which may indicate that the results of the companies showed less variability in the post-convergence period, with income smoothing evidence.
However it is emphasized that this analysis of test variables should be treated with caution as they may be influenced by other external factors beyond the international accounting convergence process, which the model by Barth, Landsman and Lang (2008) seeks to mitigate with the use of control variables.
Regarding the control variables, firm size, measured by the natural logarithm of the total assets of the companies, showed an increase in the post-IFRS period.There was also an increase in average sales growth from the period 1 to period 2. It is also noted an increase in debt (LEV) in period 2. However, as the change in liabilities (DISSUE) shows a reduction in period 2, it appears that the increase in debt (LEV) may have been more due to the decrease in the value of net assets of businesses than by rising debts properly.Based on these variables, we calculated the three measures of GR in the model of Barth, Landsman and Lang (2008), from the four multiple linear regression equations already presented.Before the calculation of these measures, were tested the assumptions of linear regression of residual autocorrelation, the homoscedasticity, multicollinearity and normal distribution.The results show that the autocorrelation of the assumptions of residual (Durbin-Watson test) and multicollinearity (VIF test) were met, except for the assumption of homoscedasticity and normal distribution.
The problem of heteroscedasticity data was identified by the Test of White in Stata 11.2 software.To remedy it, we used robust regression in the same software.For normality of the data, given the sample size of 357 and 335 observations for periods 1 and 2, respectively, this assumption is relaxed in this analysis.Gujarati (2006) states that the assumption of normality is restricted to small samples (less than 100 observations) and may assume the  It is observed in Table 2, that the model is significant in both periods, with explanatory power of 5.8% and 9.3%, respectively.The residual value of the model used in the calculation of smoothing shows values of 0.004 (period 1) and 0.001 (period 2).These values were obtained from the variance of the set of observations residues.It was found the company by company residue and, as a result, we calculated the variance of this set of residue.
The control variables considered significant for explaining the variability of profits (ΔNI) in period 1 were of the size (SIZE) and cash flow (CF), 10%; auditing (AUD) 5%; and the change in liabilities (DISSUE) and sales growth (GROWTH) 1%.In period 2, the significant variables were sales by total assets (TURN), 10%; the percentage of shares held by individuals or entities related to the company (CLOSE), 5%; and cash flow (CF) 1%.The calculation of the second measure of GR takes into consideration the profit rate variability on the variability of cash flows, and ΔNI the variance of residuals of equation 1 and ΔCFO the variance of residuals of the equation 2.
Barth, Landsman and Lang (2008) argue that firms with higher volatility in cash flow typically have higher volatility in net income.This second measure attempts to control this factor.If companies use accruals to manage results, the variability of changes in net income BBR, Braz. Bus. Rev. (Engl. ed., Online), Vitória, v. 12, n. 2, Art. 1, p. 1 -24, mar.-apr. 2015 www.bbronline.com.brThe Impact of the International Accounting Convergence on Income Smoothing in Brazilian Companies 15 may be less than the variability of changes in operating cash flow.The results of the equation of the regression 2 are shown in Table 3. Table 3 shows that the model is significant for two periods, with 14% of explaining power for one period and 61% for the period 2. The variance of results used for calculation of the smoothed model, is 0.014 (period 1) and 0.006 (period 2).
The control variables considered significant variability explanation of operating cash flows (ΔCFO) were cash flow of business (CF) in both periods, the change in liabilities (DISSUE) in period 1 and the size (SIZE) 2 in the period.
The second GR metric is the division of the variability of changes in operating profit on the variability of changes in cash flow, with the following value for the two periods: Pre-IFRS: Post-IFRS The third measure of GR is based on the Pearson correlation between accruals (ACC) and cash flows (CF).As in the previous two steps, are compared the residues of equations 3  and Lang (2008), is calculated as the difference between operating income and operating cash flow divided by total assets.
According to Iatridis and Rouvolis (2010), a negative correlation is an indicator of GR, since companies tend to influence their accruals up when the cash flows appear smaller.The results of equation 3 are shown in Table 4. Equation 3 uses as the dependent variable cash flow (CF).Table 4 shows the model of equation 3 as mean in the two study periods, the clarification of 8.6% and 7.2% in each period, respectively.It is observed that in the period 1, the variable that helps explain the CF companies, significantly, is the size of the companies (SIZE) and sales divided by assets (TURN).In the period 2, the significant variables were size (SIZE), sales growth (GROWTH) and sales of assets (TURN).Table 5 presents the results of the regression of the equation 4. Equation 4 considers as the dependent variable accruals.The model presented in Table 5 is not significant in both periods, with explanatory power of 1% and 2%, respectively.No tested variable is significant to explain the value of accruals (ACC) in period 1.The ACC is explained significantly in period 2 by the variable accounting firm (AUD), but only 10%.To reach the third measure of the GR model of Barth, Landsman and Lang (2008), we calculate the Pearson correlation between the residuals of the equations 3 and 4. If the correction coefficient is negative and significant, there are signs of GR through the use of accruals in order to align the result to the company's cash flow.It is shown in Table 6 that the Pearson correlation coefficient between the cash flow (CF) and accruals (ACC) was negative in both study periods, indicating that there GR evidence in both periods.However, considering that the equation 4 is not significantly, we denote their residues (ResAcct) not only represent the effect of the adoption of IFRS, but also other variances in accruals not explained by the model.Thus, the results of measurement of the third GR cannot be considered to address the problem of search.Table 7 presents a Brazil, Santos et al. (2011) investigated whether the enactment of Law N. 11.638/07 changed the degree of accounting conservatism, measured by Basu's model, of nonfinancial Brazilian companies, between the first quarter of 2005 and the third quarter of 2009.These authors did not found significant effects, but point out that their results are preliminary, given the limitation of the sample.Grecco (2013) using the Modified Jones Model, found no changes in levels of earnings management of the 361 nonfinancial Brazilian companies from 2005 to 2011.Rosa and Tiras (2013) investigated aspects of the economic environment in which were drawn the financial statements and the potential for earnings management in Brazil.These authors suggest to Brazilian researchers examine aspects of the Brazilian economic environment, as the recent attractiveness of foreign investments, corporate structure

4. 1
POPULATION AND SAMPLEThe population of the research is made up of Brazilian companies listed in the Thomson ONE Banker (http://banker.thomson.com)database, totaling 364 companies.The criteria for sample selection was that the company present values for all the variables studied in both study periods(2005-2007 and 2010-2012).Based on this criterion, it was identified 123 companies, totaling 369 observations in each period.Finally, during the analysis, 12 observations were excluded, classified as outliers in the pre-convergence period, resulting in a final sample of 357 observations during this period.In the post-convergence period, were excluded 34 observations classified as outliers, resulting in a final sample of 335 cases.Were classified as outliers observations above three standard deviations, obtained directly by SPSS 20.0 software.
SIZE = value of total assets; GROWTH = percentage change in sales; LEV = total liabilities divided by total equity at the end of the year; DISSUE = percentage change in total liabilities; TURN = sales divided by total assets at the end of the year; CF = annual net cash flow from operating activities divided by total assets at the end of the year; AUD = dummy variable, with 1 being audited company by PwC, KPMG, E&Y or D&T, and zero for others; XLIST = trading of shares on the stock exchange in the United States of America, 1 for yes and zero for no; CLOSE = Percentage of shares held by individuals or entities related to the company, as directors and their families, foundations, among others.

CFO
residual of equation 1 in the n period;ΔCFO n = residual of equation 2 in the n period.The variability ΔCFO is the variance of residuals from Equation 2 groups, the first group pre-convergence and the second, post-convergence.The smaller the variation, the greater the likelihood of GR.Measure 3: Pearson correlation between accruals and cash flow Dependent variables: .Bus. Rev. (Engl.ed., Online),Vitória, v. 12, n. 2, Art. 1, p. 1 -24, mar.-apr.2015  www.bbronline.com.brThe Impact of the International Accounting Convergence on Income Smoothing in Brazilian Companies 11 normality assumption for larger samples based on the Central Limit Theorem (Teorema do Limite Central -TLC).According toStevenson (2001, p. 181), by central limit theorem "[...]    even in the case of a non-normal distribution, the distribution of the sample means will be approximately normal, since the sample is large."These results, according toStevenson (2001), are known as Central Limit Theorem and represent one of the most important concepts related to statistical inference.The only restriction for its use is that the sample is greater than 30 observations.The first measure of GR refers to the variability of income divided by total assets of each company, from one year to the next.According to Barth, Landsman and Lang (2008), significant variations in changes in income may be income smoothing earnings evidence.As the profit can be varied influenced by external factors and directly related to the accounting standards, the variable ΔNIn was used as the dependent variable in the equation of regression 1, which includes some control variables.The calculation of the GR level occurs from the variance of residuals of equation 1, operating profit variability place directly in order to mitigate external effects on variability of operating income.The results from the regression from equation 1 are shown in Barth, Landsman and Lang (2008)changes in preferred shares), present in the model applied byBarth, Landsman and Lang (2008), was not used in this study.This variable is capital market in recent years, long-term debt have yet as the main source to financial institutions, especially government, such as BNDES.So to capture the effects of the financing structure in this study, we used the LEV and DISSUE variables.Measure 2: Mean rate BBR, Braz.Bus.Rev.(Engl.ed., Online),Vitória, v. 12, n. 2, Art. 1, p. 1 -24, mar.-apr.2015  www.bbronline.com.br results of various companies.In the year 2008, was not considered in the 2009 analysis.First, because ΔNI and ΔCFO variables for 2009 are influenced by the 2008 values (range compared to the previous year).In addition, the year 2009 can still be considered as a transition period to IFRS.
Data collection was carried out from the website of Thomson ONE Banker (http://banker.thomson.com), in Thomson Financial and Worldscope databases.The pronouncements.On the second aspect, the economic, remember that the year 2008 was marked by strong global economic crisis, which started in the United States with the real estate crisis, known as the subprime crisis, which led to the bankruptcy institutions financial and insurance companies known worldwide.This crisis spread worldwide, impacting the BBR, Braz.Bus.Rev. (Engl.ed., Online), Vitória, v. 12, n. 2, Art. 1, p. 1 -24, mar.-apr.2015 www.bbronline.com.br

Table 3 -Results of the Regression of Equation 2
and 4, instead of correlations between ACC and CF directly.The control variables are the same used in equations 1 and 2, except for CF variable, which is treated as a dependent variable in the equation 3. The value of accruals, following the model of Barth, Landsman 014  BBR, Braz.Bus.Rev. (Engl.ed., Online), Vitória, v. 12, n. 2, Art. 1, p. 1 -24, mar.-apr.2015 www.bbronline.com.br