Effects of the audit committee and the fiscal council on earnings quality in Brazil

This study evaluates the effects of the audit committee and the fiscal council with their different characteristics on accounting information quality in Brazil. The proxies of accounting information quality used are: relevance of accounting information, timeliness, and conditional conservatism. The sample consists of Brazilian companies listed on the Brazilian Securities, Commodities, and Futures Exchange (BM&FBOVESPA) with annual liquidity above 0.001 within the period from 2010 to 2013. Data were collected from the database Comdinheiro and the Reference Forms of companies available on the website of the Brazilian Securities and Exchange Commission (CVM) or the BM&FBOVESPA. The samples used in the study totaled 718, 688, and 722 observations for the value relevance, timeliness, and conditional conservatism models, respectively. The results indicate that different arrangements of the fiscal council and the existence of the audit committee differently impact the accounting information properties. The presence of the fiscal council positively impacted the relevance of equity, while the presence of the audit committee, the relevance of earnings. Conditional conservatism was evidenced in the group of companies with a permanent fiscal council, demonstrating that it is significant as a governance mechanism, rather than the installation for temporary operation when asked by shareholders in an ordinary general meeting. The presence of both showed significant earnings for the market, but they were not timely, something which exposes restriction to the relevance found. Lastly, the powerful fiscal council showed a positive association only concerning the relevance of equity.


INTRODUCTION
Th e quality of accounting fi gures is intrinsically linked to the quality of corporate governance (Sloan, 2001).Accounting provides information, among other aspects, concerning executives' performance that are needed by most corporate governance mechanisms to eff ectively operate in addressing agency problems (Bushman & Smith, 2001;Sloan, 2001).
Previous studies have shown that the board of directors and its structure and composition, as well as that of its committees, aff ect the quality of accounting fi gures, indicating that good corporate governance practices benefi t shareholders (Trapp, 2009;Vafeas, 2000;Xie, Davidson & Dadalt, 2003).In addition, the eff orts to grasp the function of the audit committee, an advisory body of the board of directors directly responsible for the supervision of accounting processes (Instituto Brasileiro de Governança Corporativa [IBGC], 2009, 2010; U.S. Government, 2002) have also increased.
Brazil has a characteristic different from other countries due to the potential presence of the fi scal council in companies, according to Law 6,404/1976, the so-called Brazilian Corporate Law (Brasil, 1976).It is a body that can be set up and have its members elected at a shareholders' general meeting as an instrument to inspect the managers' acts.Among its functions, the fi scal council is responsible for examining and commenting on the fi nancial statements audited by an external auditor (Brasil, 1976;IBGC, 2010).
In the Brazilian environment, we may ponder that both the board of directors and its committees and the fi scal council have functions that aim at reducing information asymmetry, something which can improve the quality of information disclosed to the market, in addition to provide more effi cient monitoring of executives.Th ese bodies can contribute to reduce the agency confl icts highlighted by Jensen and Meckling (1976).
Studies such as Ahmed and Henry (2012), Baxter and Cotter (2009), and Wild (1996) have provided evidence that the audit committee improves the quality of accounting fi gures.In Brazil, Trapp (2009) argues that the fi scal council is related to lower levels of earnings management.
In this way, investigating and understanding the relation between functions of the fi scal council and the audit committee and earnings quality has proved to be a research opportunity poorly addressed by literature and convenient to the Brazilian environment.Also, the Securities and Exchange Commission (SEC), as a consequence of the Sarbanes-Oxley Act (SOX) (U.S. Government, 2002), which provided for mandatory implementation of an audit committee, authorized the Brazilian companies holding American Depositary Receipts (ADRs) listed on the New York Stock Exchange (NYSE) to structure corporate governance, considering the fi scal council with an overlap to the audit committee, i.e. it is up to the fi scal council to adapt, also taking the functions of this committee, becoming the so-called 'powered fi scal council.' The audit committee and the fiscal council have common functions related to the generation and reporting of accounting information, but they are diff erent in many ways, emerging discussions about the establishment of one or the other.Th e same applies to the implementation of the powered fi scal council (Furuta & Santos, 2010;PricewaterhouseCoopers [PwC], 2007).
Th e two bodies should not be regarded as mutually excluding, since their functions, responsibilities, hierarchy, and composition are diff erent.First, they occupy diff erent hierarchical positions in the organization: the audit committee is subordinate to the board of directors, preferably formed by independent representatives of this, while the fi scal council is autonomous and not linked to any management body, it consists of shareholders' representatives.In this way, the independence rules are also diff erent from each other; regarding the functions and competences, the audit committee is delegated with activities specifi c to the 'management' function, and the fi scal council is an institutional tool of the shareholders' right to exert control over management (Furuta & Santos, 2010;PwC, 2007).In this way, adopting the two bodies in the same organization could contribute to greater transparency in managers' actions and better corporate governance practices (Santos, 2009).
Thus, considering the corporate governance environment in the Brazilian capital market, this study investigates the following research question: what are the eff ects of the audit committee and the fi scal council on earnings quality of public companies listed on the BM&FBOVESPA?
Th is study aims to evaluate the eff ects of the audit committee and the fi scal council and their diff erent arrangements -types of fi scal council (when it exists, either as permanent or temporary), permanent, temporary, or powered fi scal council -on earnings quality in Brazil.
As proxies to measure earnings quality, the relevance, timeliness, and conditional conservatism models were adopted, according to Almeida (2010), Barth, Landsman and Lang (2008), and Lopes and Walker (2008).
Th e contribution of this research lies on the fact that there is an extensive international literature on earnings quality and its relation with corporate governance mechanisms, unlike the Brazilian reality.Also, the Brazilian literature is incipient to consolidate the understanding of the relation between the audit committee and the fi scal council with fi nancial statement quality.Finally, this research brings as a diff erential feature to the international literature the inclusion of the fi scal council as a mechanism of corporate governance and its interactions with earnings quality.
Th e results indicate diff erent eff ects of the bodies under analysis on earnings quality proxies.Th e fi scal council, in general, either permanent or temporary, has positive association with the book value relevance and the audit committee with net income.At the same time, the results for companies that have permanently installed fi scal councils show that both the relevance of equity and net income is increased, as well as the timely recognition of economic losses (conditional conservatism), demonstrating the importance of permanent installation of the fi scal council to control fi nancial report.Th e fi rms with a powered fi scal council showed relevance only for equity among all the earnings quality metrics used, something which contributes to the existing discussion between scholars and professionals about the loss of focus on the attributions of this body: for instance, PwC (2007).
This study is divided, after this Introduction, as follows: 2. Th eoretical framework and development of hypotheses; 3. Methodology, where models' details, sample development, and estimators are presented; 4. Analysis of results by each model and their variations; 5. Robustness analysis; and 6.Final remarks.Bushman and Smith (2001) argue that corporate governance mechanisms are the means by which controlling agents are disciplined to act in accordance with investors' interests.Accounting is a part of the set of corporate governance mechanisms, as it promotes greater transparency regarding the fi nancial performance of fi rms within a period and the dissemination of this information aims to reduce information asymmetry and agency problems (Bushman, Chen, Engel & Smith, 2004).

THEORETICAL FRAMEWORK AND DEVELOPMENT OF HYPOTHESES
Th ere is no precise and fully reliable metric in the literature to measure earnings quality, as discussed by Dechow, Ge and Schrand (2010), but researchers have developed over the years several models that seek to capture diff erent properties of accounting fi gures.Almeida (2010), Barth et al. (2008), and Lopes and Walker (2008) used similar properties to measure earnings quality, such as relevance, timeliness, and conditional conservatism, the same used in this study.

Audit Committee
The audit committee is an advisory body of the board of directors consisting, preferably, of independent board members.It assures to the board of directors the control over earnings quality and internal controls, and it is regarded as an essential mechanism in corporate governance (IBGC, 2009).
Th e concept of audit committee is not new.Since 1939 and 1940, the NYSE and the SEC, respectively, recommend setting up audit committees as a response to the case McKesson and Robbins in the late 1930s.In subsequent years, the number of companies with established committees signifi cantly increased and the U.S. Congress, the SEC, accountants, and others expressed interest and support for this body (Birkett, 1986).
Th e NYSE and National Association of Securities Dealers Automated Quotations (NASDAQ) established, in 1978 and 1989, respectively, as requirements for listing on their stock exchanges, the establishment of an audit committee consisting of fully independent members.In 1999, they jointly developed the Blue Ribbon Committee on Improving the Eff ectiveness of Corporate Audit Committees, with the purpose of making recommendations on the composition and operation of audit committees, subsequently adopted by the SEC.
Th e audit committee gained prominence worldwide aft er the U.S. scandals, mainly involving Enron and WorldCom, which prompted the SOX enactment by the U.S. Congress, in July 2002.Th e SOX introduced corporate governance rules, among others, aiming to ensure greater transparency to fi nancial reporting by the companies, also instituting mandatory establishment of this committee.
In Brazil, Law 6,404/1976(Brasil, 1976) does not deal with the audit committee, but it regulates the general board of directors.In the country, the installation of a committee is not mandatory, except for fi nancial institutions and insurance companies, according to regulation by the National Monetary Council (CMN), the Central Bank of Brazil (BACEN), and the National Council of Private Insurance (CNSP).
Accordingly, the audit committee has the function of inspecting the fi nancial reporting process (internal process management and integrity and eff ectiveness of internal controls) and internal and external audit, including the resolution of confl icts between executives and auditor.
Th e committee ensures the integrity and timeliness of financial information, as well as the interests of shareholders and other stakeholders (IBGC, 2010; U.S. Government, 2002).Th erefore, its activities are closely related to the fi nancial reporting quality (Klein, 2002;He, Labelle, Piot & Th ornton, 2009).
In Brazil, research addressing the infl uence of an audit committee as a governance mechanism on earnings quality is incipient, we may highlight Cunha, Hillesheim, Faveri and Rodrigues (2014) and Filipin (2012).Cunha et al. (2014) investigated whether characteristics of the audit committee (size, expertise, and independence) refl ect on earnings management by Brazilian public companies listed on the BM&FBOVESPA and they concluded that none of the characteristics under study impacted on earnings management.
Using a portfolio of proxies to measure earnings quality, as in this study, Filipin (2012) sought to verify whether companies with an audit committee listed on the BM&FBOVESPA have better earnings quality; the results for these companies suggested greater relevance of accounting information, but less conservative and persistent net income.
Before the audit committee became mandatory in the United States of America (USA), Dechow, Sloan and Sweeney (1996) found evidence that companies that manage their earnings were less likely to have this committee installed than other fi rms.Wild (1996) noticed the reaction of the U.S. market to the release of fi nancial statements before and after the establishment of an audit committee and found that the market reaction to disclosure of fi nancial statements aft er the establishment of a committee is greater than in previous periods.Ahmed and Henry (2012) and Baxter and Cotter (2009) investigated, in Australian companies, whether voluntary installation of the audit committee impacted the earnings quality.Baxter and Cotter (2009) found evidence of reduced earnings management practices, while Ahmed and Henry (2012) showed a positive impact on unconditional conservatism and a negative impact on conditional conservatism.In this context, the fi rst research hypothesis on the presence of an audit committee in companies listed on the BM&FBOVESPA is presented: H 1 : the audit committee enhances the fi nancial reporting quality.

Fiscal Council
In Brazil, the fi scal council is an organ provided for in Law 6,404/1976(Brasil, 1976) and the company's statute will provide for its operation, which may be permanent or installed by the general meeting at shareholders' request.When installed, it must consist of at least 3 and at most 5 members and substitutes in equal numbers, shareholders or not (Brasil, 1976).
Th e fi scal council is a body independent from the company's board and executives and its competences are set in article 163 of Law 6,404/1976(Brasil, 1976), i.e. it reports directly to the shareholders at the meetings.Th is body inspects the actions taken by the company's board and board of directors, comments on proposals by administration bodies, as well as it examines and comments on the fi nancial statements audited by the external audit, seeking to preserve shareholders' rights.
Both the IBGC, from a professional perspective, and scholars (Carvalhal-da-Silva & Leal, 2005;Tinoco, Escuder & Yoshitake, 2011;Trapp, 2009), regard the fi scal council as a part of the corporate governance system in Brazil.Th e fact that the fi scal council is a direct representative of shareholders inspecting executives' management can have a great infl uence on the reduction of agency confl ict and informational asymmetry, something which makes it a signifi cant control mechanism.Also, Tinoco et al. (2011) claim that the fi scal council collaborates with those who are far from control and power (especially minority shareholders).Trapp (2009) sought to evaluate the direct intersection of a fi scal council as a good corporate governance practice to reduce earnings management.Evidence indicates that these factors, considering the existence and qualifi cation of the members of a fi scal council, are related to lower levels of discretionary accruals.
Finally, Carvalhal-da-Silva and Leal (2005) claim, when constructing the Corporate Governance Index, that the existence of a permanent fi scal council in the organization is more eff ective to monitor and arrange company's management than the council installed only at shareholders' request.It is assumed that a fi scal council installed regardless of shareholders' will and on a continued basis provides the company with greater power from this mechanism in the corporate governance system, mainly by increasing earnings quality.
In this context, the second research hypothesis on the presence of a fi scal council in companies listed on the BM&FBOVESPA is presented: H 2 : the fi scal council enhances earnings quality.
Additionally, to analyze the second hypothesis, we investigate on a methodological basis the various fi scal council arrangements (permanent and temporary).

Powered Fiscal Council
An imposition of the SOX was mandatory installation of the audit committee in companies listed on all the U.S. stock exchanges, something which also aff ected foreign companies that issue ADRs in the U.S. market (Furuta & Santos, 2010).In the Brazilian case, aft er an agreement between the Brazilian Securities and Exchange Commission (CVM) and the SEC, in 2003, mandatory installation of the audit committee was abolished if the company used the fi scal council, adapting it to the functions of a committee, so that the fi scal council incorporates new responsibilities and functions (hereaft er, powered fi scal council).
According to the IBGC (2009), considering the legal attributions of the fi scal council provided for in Law 6,404/1976(Brasil, 1976), the possibility that this body takes any administration activity is excluded.Th e creation of a powered fi scal council does not contradict this understanding, since the fi scal council is careful not to act as an advisor or assessor concerning management decisions.Accordingly, certain activities that cannot be delegated by the board of directors or executives, such as hiring audit and 'non-audit' services, should not be assigned to the fi scal council, since they are provided for by article 142 of Law 6,404/1976(Brasil, 1976) as duties of the board of directors.
PwC ( 2007) discusses the strengths and weaknesses of adopting a powered fi scal council.According to this study, opponents argue that the fi scal council is an institutional tool to exert shareholders' inspection does not serve the purpose of the audit committee to be a management instrument of the board of directors.In addition, the fi scal council has functions and duties diff erent from those predicted for the audit committee, such as providing opinions on proposals to change company's capital, issue debentures, investment plans, or capital budgets, among others, something which can lead to loss of focus on specifi c attributions of the audit committee.
According to Santos (2009), many Brazilian companies that issue ADRs still prefer to strictly comply with the SOX requirements and not install the powered fi scal council; this is most likely due to a market-driven and strategic issue of making their stocks available in the U.S. market, since the acquirer of these stocks is more familiar with the audit committee than with any other type of monitoring body.However, Calazans (2003) highlights that the constitution of an audit committee in Brazilian companies might increase their cost and this additional cost can oft en be higher than the benefi t generated.
Furuta and Santos (2010) investigated the perspective of executives from companies operating in Brazil and issuing ADRs and that of market analysts concerning the formation of an audit committee or a powered fi scal council.Th e results indicated there was no consensus of opinion that the fi scal council is more adaptable than the audit committee to the Brazilian business environment if the functions of these bodies are diff erent and if the costs associated with committee formation are relevant.
It is assumed, then, that the fi scal council improves earnings quality, since it absorbs, in addition to the legal functions of the fi scal council, the functions of the audit committee, expanding its scope of action.Also, companies with a powered fi scal council are listed on the U.S. stock exchanges, something which indicates greater rigor in corporate governance.Th us, the third research hypothesis is presented:

Audit Committee and Fiscal Council
Th e fi scal council and the audit committee occupy diff erent hierarchical positions in the organization: the second is subordinate to the board of directors, preferably consisting of independent members, while the fi rst is set up by the general meeting, a body which represents controlling shareholders and minority shareholders.Unlike the committee, the fi scal council is autonomous and is not linked to any organization's body, and it must report to the shareholders directly at the meetings.
Th e bodies also have diff erences in their functions and duties.Audit committees are delegated with activities inherent to the 'management' function, i.e. it is an advisory mechanism made available to the board of directors to fulfi ll its tasks.Among the legal attributions of the fi scal council, the possibility of exercising any activity inherent to management is excluded.Th e council activities are carried out according to the role of wide and unrestricted inspection of managers' legal and statutory duties, at shareholders' request, as a decision of the general meeting.In this way, it is an institutional mechanism to exert the shareholders' right of control on the executive's decision (IBGC, 2009;PwC, 2007).
Among the diff erences between these bodies that may directly interfere with accounting information, we highlight the fact that the functions of an audit committee are more specifi c and focused on the fi nancial reporting process, unlike the fi scal council, which besides recommending whether the fi nancial statements can be approved also inspects the managers' acts.
Each body has its function in the hypothesis of enhancing fi nancial reporting quality, either advising the board of directors (audit committee) or inspecting and reporting to the shareholders (fi scal council).In this way, the acceptance of both bodies in the same organization refers to the reasoning that they are not mutually excluding.Th is trend might lead to greater transparency of managers' actions and adoption of better corporate governance practices, since a large number of views on control coexisting in the same environment could contribute to ensure continuity in the organization, regardless of individual or group interests (Santos, 2009).
Based on the assumption that better corporate governance practices in the organization improve the quality of fi nancial reporting, it is expected that adopting the two bodies has a positive impact on this relation.Th us, the fourth research hypothesis is: H 4 : the presence of the audit committee and the fi scal council, simultaneously, enhances earnings quality.

METHODOLOGY
The study population comprises active publiclytraded corporations listed on the BM&FBOVESPA that have data available for the period within 2010 and 2013.To start constructing the sample, only companies with annual liquidity above 0.001 were selected.Financial institutions and funds were excluded from the sample, because they have a specifi c accounting standard and regulation.Observations with empty cells, with values equal to 0, and errors in the variables of interest of each model, as well as the companies that did not provide the Reference Form, were also excluded.
Th e accounting and market data were obtained from the database Comdinheiro.Information on the fi scal council and the audit committee was obtained from the company Reference Forms, available at the websites of the BM&FBOVESPA and the CVM.
Earnings quality models were estimated using the ordinary least squares (OLS) method, with robust standard errors clustered by fi rm.Th is estimation method has been widely used in accounting and fi nance studies (Almeida, 2010;Giroud & Mueller, 2010;Lopes & Walker, 2008).
As a method to exclude outliers, observations with extreme values above 3 standard deviations from the standardized mean value of each variable were excluded.Table 1 shows how observations were fi ltered, resulting in the fi nal study sample by model.Since there is no single way to measure earnings quality, this study was based on the metrics used by Lopes and Walker (2008), which capture earnings quality, based on its properties, among them relevance, timeliness, and conditional conservatism.It is believed, therefore, that more relevant accounting fi gures, greater timely loss recognition (bad news), and more timely information correspond to better fi nancial reporting process.
Th e model that tests the value relevance considers the information content of net income and book value as a function of share price.
Th is study evaluates the eff ect of the existence of a fi scal council in Brazilian companies and its variations: general fi scal council (when it exists, either on a permanent or temporary basis), permanent, temporary, or powered fi scal council, audit committee, of both organs simultaneously and one or the other on relevance.All these combinations are controlled by dummy variables in alternate models.
To do this, model 1 was prepared, which has dummy variables and control variables (CV), as follows: where: P i,t is the share price of company i in year t adjusted by dividends and splits (four months aft er the end of the year) and scaled by share price in t-1, NI i,t is earnings per share of company i in year t scaled by share price in t-1, BV i,t is equity per share of company i in year t, scaled by share price in t-1i, C n is the variable that takes value 1 for FC n , AC, ACFC or ACOFC [where: FC n = dummy variable if fi rm i has a general fi scal council installed (FC 1 ), a permanent fi scal council (FC 2 ), or a powered fi scal council (FC 3 ); AC = dummy variable if fi rm i has an audit committee installed; ACFC = dummy variable if fi rm i has both, fi scal council and audit committee installed; ACOFC = dummy variable if fi rm i has a fi scal council or an audit committee installed] in each dummy, otherwise it is considered as 0.
Th e CVs used in this research were fi rm size (S), leverage (LEV), and growth opportunity (GRO) measured by the natural logarithm of total assets, total debt divided by total assets, and percentage growth of net revenues, respectively.Lopes (2009) states that net income can infl uence price over a long period of time.Th us, the model to test timeliness aims at revealing whether net income and its variation can explain stock return.Model 2 is presented below, it has been developed in this research, which sought to identify the eff ect on timeliness.
where: R i,t is return on (the most liquid class) share of company i in year t adjusted by dividends and splits (four months aft er the end of the year) and scaled by share price in t-1, NI i,t is earnings per share of company i in year t and scaled by share price in t-1, and ΔBV i,t are changes in net income per share of company i in year t and scaled by share price in t-1 (C n according to model 1).
Th e conditional conservatism model used is that developed by Basu (1997).Th e model is based on the concept that net income is, in its nature, conservative in a conditional sense (asymmetric recognition between losses and gains) and it anticipates bad news in relation to good news.Th e model relates net income to stock return.Model 3 is presented below: where: NI i,t is the earnings per share of company i in year t, scaled by share price at t-1, D i,t is the dummy for return, i.e. 1 for negative return and 0 for positive return of company i in year t, and R i,t is return on share (the most liquid share class) of company i in year t adjusted by dividends and splits (four months aft er the end of the year) and scaled by share price at t-1 (C n according to model 1).
It is expected that the coeffi cients of variables of interest from interactions with variables related to the existence of the fi scal council, audit committee, both of them, one or the other, permanent fi scal council and powered fi scal council are positive and statistically signifi cant in all models.

ANALYSIS OF RESULTS
Table 2 shows sample composition according to the existence of the bodies under analysis.Th e results are presented for the total sample (annual stock liquidity above 0.001) without excluding outliers.Source: Prepared by the authors.
It is verifi ed that the most common practice in Brazil is adopting a fi scal council, since 457 observations (companies-year) had this body installed within the four years surveyed.The presence of a powered fi scal council is noticed in 38 observations of the sample and it corresponds to companies that issue ADR and did not install the audit committee.
In Brazil, the installation of an audit committee is voluntary, except for financial institutions and insurance companies, so that its existence is found in 254 observations.There are 208 observations from fi rms that have neither installed a fi scal council or an audit committee, something which may suggest weakness in the corporate governance system and lower monitoring of managers in these companies.
Table 3 displays the descriptive statistics of variables in the earnings quality models explained in the methodology.Note: P i,t is the share price of company i in year t adjusted by dividends and splits (four months after the end of the year) and scaled by share price in t-1, NI i,t is earnings per share of company i in year t scaled by share price in t-1, BV i,t is equity per share of company i in year t, scaled by share price in t-1, R is annual return of rm i in period t (four months after the beginning of the year), and ∆NI it is pro t variance per share of rm i in period t.The variables P, LPA, PLA, R, ∆LPA are scaled by price in t-1 (four months after the beginning of the year).

ACFC = dummy variable if rm i has both, scal council and audit committee, installed; AC = dummy variable if rm i has an audit committee installed; FC n = dummy variable if rm i has a general scal council installed (FC 1 ), a permanent scal council (FC 2 ), or a powerful scal council (FC 3 ); D = dummy variable, considering 1 if the return is negative and 0, otherwise; LEV = leverage of rm i in period t; GRO = revenue growth of rm i in period t; ACOFC = dummy variable if rm i has a scal council or an audit committee installed; S = natural logarithm of total assets of rm i in period t.
Source: Prepared by the authors.We observe greater data dispersion between the variables NI and GRO in all models, as well as the variables R and ΔNI in the timeliness model and R in conditional conservatism, since they have a standard deviation above mean value.For the variables body presence and arrangement, extreme values (minimum and maximum value) are always the same, due to the characteristics of these variables, but they have diff erences in dispersion between the various variables.
Table 4 shows the results referring to regressions for the relevance model.In column 1 of Table 4, the result for the eff ect of presence of a general fi scal council has a positive and signifi cant coeffi cient only for the variable BV*FC 1 (0.0281, t = 2.60).Th us, when the company chooses to set up the general fi scal council (permanent or temporary), the results suggest that equity is relevant to the market, unlike net income, which has a negative and non-signifi cant coeffi cient, partly corroborating the results of Trapp (2009), which associates the fi scal council installation in Brazil with lower levels of earnings management.
Column 2 displays the results for the eff ect of an audit committee on relevance.Th e results are contrary to those of the general fi scal council -the variable NI*AC has a positive and signifi cant coeffi cient of 0.568 (t = 4.08) -, suggesting that the presence of this control mechanism has a positive impact on relevance of the reported net income.Interaction with the variable BV is positive, but not signifi cant.Th ese results corroborate those of Filipin (2012) in Brazil and they are convergent with Baxter and Cotter (2009) and Wild (1996).
Th e eff ects of the presence of both (fi scal council and audit committee, simultaneously), represented in column 3, show a positive and signifi cant coeffi cient only for the variable NI*ACFC (0.382, t = 3.08), highlighting that the implementation of the two bodies, simultaneously, only aff ects net income.Also, in column 4, the existence of one or other body has an impact on equity relevance, since the variable BV*ACOFC has a positive and signifi cant coeffi cient of 0.0206 (t = 1.76).
Th e eff ects of a permanent fi scal council are displayed in column 5. Th e results suggest for both the variable NI*FC 2 and BV*FC 2 positive and signifi cant coeffi cients of 0.431 (t = 3.57) and 0.0902 (t = 3.48), respectively.In this way, the presence of the fi scal council on a permanent basis increases the relevance of net income and equity.Such a result confirms that the market recognizes accounting fi gures as more relevant information, thus of better quality, when there is a permanent presence of a fi scal council in the companies.Th is fi nding confi rms the claim of Carvalhal-da-Silva and Leal ( 2005) that permanent fi scal councils may be regarded as stronger governance mechanisms than when they are installed only at shareholders' request.
Th e results for the eff ects of the powered fi scal council (column 6) have evidence similar to that from the general fi scal council: positive and signifi cant coeffi cient only for the variable BV*FC 3 (0.0568, t = 2.51) and negative for net income, although not signifi cant, showing impact only on equity relevance.
Table 5 displays the results for the timeliness model.Th e models of columns 1, 2, 4, and 5 of Table 5 have, for the variables of interest, coeffi cients with no statistical signifi cance, something which indicates that companies with a general fi scal council, an audit committee, and a permanent fi scal council do not raise market expectations in relation to net income and its contemporary variation to stock return.
Contrary to hypotheses 3 and 4, the results for a powered fi scal council and the presence of both have negative and signifi cant coeffi cients for the variables NI*FC 3 (-0.183,t = -0.68)and NI*ACFC (-0.114, t = -1.99),showing that their presence does not improve the timeliness of accounting fi gures, i.e. the market anticipates net income information content.Th is result, as well as the result for relevance, contributes to academic and professional discussions about the overlapping of actions taken by the powered fi scal council.
Table 6 shows the results of regressions for the conditional conservatism model.

ROBUSTNESS ANALYSIS
In order to reinforce the evidence found previously, we sought to change the variable of interest presence of an audit committee and a fi scal council by the independence degree of their members.
Th e fi scal council independence was analyzed in two ways: (i) members representing minority shareholders, elected at the general meeting (thus, independence from the controlling shareholders is expected), and (ii) independence with management, verifying present and past relations between the member and the company or economic group by means of item 12.8 of the company Reference Form.
As for independence of the audit committee, we could only use as a proxy member independence from management, because they were not directly elected by shareholders, making it impossible to analyze independence of the controlling shareholders.Th e number of observations in this analysis changed when compared to the previous one is due to the fact that some companies did not provide the full information, in the company Reference Form, of the members' curricula, both those from the fi scal council and the audit committee, something which resulted in a decreased number of observations.Th e variable DIND 1 is a dummy variable that takes value 1 when the company has at least one member elected by minority shareholders and 0 otherwise, the variable IND 1 measures the percentage of members elected by minority shareholders, while the variable DIND 2 takes value 1 when the company has at least one independent member and 0 otherwise, and the variable IND 2 measures the percentage of independent members.
In the same structure, we analyze relevance, timeliness, and conditional conservatism.Table 8 displays results for the relevance model.Th e results for relevance indicate that fi scal council independence positively impacts equity relevance in the two independence analyses and that of the audit committee only impacts net income relevance, reinforcing the previous results for relevance of analyzing the presence of these bodies.
Table 9 displays the results for the timeliness model.According to the results for timeliness, we cannot attest to the eff ect of member independence on earnings timeliness and its persistence, as well as when analyzing the presence of these bodies.
Table 10 displays the results for the conditional conservatism model.According to the results presented in Table 10, the percentage of independent members in the fi scal council, either elected by minority shareholders or independent members, has an impact on reported profi t conservatism.However, the results do not point out an association between member independence in the audit committee and the practice of conditional conservatism.
Th erefore, we notice that member independence in the fi scal council managed to change the impact of the presence of this body on conditional conservatism.Th e results for audit committee independence reinforce the evidence found for its presence.

FINAL REMARKS
Th is study evaluated the eff ects of the presence of the audit committee and the fi scal council and their various arrangements on earnings quality of Brazilian public companies listed on the BM&FBOVESPA from 2010 to 2013.Th e metrics to assess earnings quality were: relevance, timeliness, and conditional conservatism.Th e motivation for conducting this research focused on the perspective of these bodies functioning as a mechanism of corporate governance, especially in the functions of supervision and inspection of the fi nancial reporting process.
In general, the results indicated that the bodies under study have an eff ect on the quality of accounting fi gures.However, a conclusion must be drawn according to each research hypothesis, body confi gurations, and accounting information properties.
Th e results for the audit committee and the general fi scal council indicated impacts only on the relevance of accounting information: impact on net income, for the audit committee, and on equity, for the general fi scal council, partially confi rming the hypotheses H 1 e H 2 .
In parallel, the results pointed out a high impact of the audit committee on net income relevance (high coeffi cient and signifi cance).Th is evidence may be explained by the specifi city of the functions of the committee in the fi nancial reporting process, coming from internal controls to internal and external audit inspection, while the fi scal council has broader functions/powers, but it has the possibility of accessing information from the auditors and the audit committee when inspecting audited fi nancial statements and the executives' acts.
Th e fact that the general fi scal council has had a positive and signifi cant eff ect only on equity relevance and a negative eff ect on earnings relevance (although not signifi cant) may indicate a greater concern of the fi scal council in protecting the shareholders' equity, while the concern of the audit committee may be greater in ensuring more reliable data on managers' performance (contained in the income statements).
It is worth highlighting that this assumption is that the concerns might be diff erently evidenced in each body (greater focus on one or another goal, based on its functions and hierarchies).Objectively, if net income is relevant to an investor, then the audit committee acts by improving fi nancial reporting quality, since its members, directly or indirectly, were nominated by controlling and minority shareholders, while the fi scal council members are predominantly elected by controlling shareholders or a group of shareholders, detaching small investors and shareholders, and giving greater emphasis to equity, as the fi scal council members are elected and directly report to shareholders at meetings.
It was expected that the presence of both in the same organization might improve the financial reporting process.Th e evidence showed an increase in net income relevance, but the results for timeliness pointed out negative eff ects.Th us, in this scenario, net income is relevant to the market, but not in a timely way, something which exposes a restriction to the relevance found.Also, the existence of one or the other body, in contrast to none, only impacted equity relevance.
Th e results obtained in this research pointed out higher relevance of net income and equity in companies with a permanent fi scal council installed, as well as higher conditional conservatism, something which refl ects the fi nancial reporting function as a contractual mechanism.Th ese results are mutually related, since early recognition of economic losses in earnings (conditional conservatism) tends to make this information more relevant to the market.Such evidence has indicated that fi scal councils permanently installed may be considered as more effi cient corporate governance mechanisms than when installed only at shareholders' request.
We also found a positive impact of a powered fi scal council only on equity relevance.Th is evidence suggests that poor association of a powered fi scal council with earnings quality metrics may be a refl ection of the various functions of this body, and this can aff ect its focus.Such a result may be useful for regulators, market players, and shareholders when discussing better corporate governance models.
Finally, corporate governance is a complex and dynamic system that involves the interaction of a set of components.Th us, evaluating the impact of one or the other component on earnings quality may be a limiting factor, something which gives room for further research in order to deepen and improve a major theme for accounting practitioners, and above all for accounting information users, seeking to relate other governance mechanisms, as well as other accounting information properties

Table 1
Sample selection

Table 2
Global composition of the scal council and the audit committee in Brazil between 2010 and 2013

Table 3
Descriptive statistics of earnings quality models

Table 4
Results of the relevance model= β 0 + β 1 .NI i,t + β 2 .BV i,t + β 3 .C n + β 4 .NI i,t *C n + β 5 .BV i,t *C n + ∑ .βn .Control + ε i,t Note: P i,t is the share price of company i in year t adjusted by dividends and splits (four months after the end of the year) and scaled by share price in t-1, NI i,t is earnings per share of company i in year t scaled by share price in t-1, BV i,t is equity per share of company i in year t, scaled by share price in t-1.The rst line shows the coef cients and the second line, the t statistics.Standard error is robust and clustered by rm.The variables P, LPA, and PLA are scaled by price in t-1 (four months after the n 1 R. Cont.Fin.-USP, São Paulo, v. 28, n. 74, p. 229-248, mai./ago.2017

Table 5
Results of the timeliness modelModel 2 R i,t = β 0 + β 1 .NI i,t + β 2 .∆NIi,t + β 3 .C n + β 4 .NI i,t *C n + β 5 .∆NI i,t *C n + ∑ .βn .Control + ε i,t Note: R i,t is return on (the most liquid class) share of company i in year t adjusted by dividends and splits (four months after the end of the year) and scaled by share price in t-1, NI i,t is earnings per share of company i in year t and scaled by share price in t-1, and ∆BV i,t are changes in net income per share of company i in year t and scaled by share price in t-1.In the rst line there are coef cients and in the second line, the t statistics.Standard error is robust and clustered by rm.The variables R, LPA, and ∆LPA are scaled by price in t-1 (four months after the beginning of the year).ACFC = dummy variable if rm i has both, scal council and audit committee, installed; AC = dummy variable if rm i has an audit committee installed; FC n = dummy variable if rm i has a general scal council installed (FC 1 ), a permanent scal council (FC 1 ), or a powerful scal council (FC 3 ); C n = variable that takes values for FC n , AC, ACFC, or ACOFC; LEV = leverage of rm i in period t; GRO = revenue growth of rm i in period t; ACOFC = dummy variable if rm i has scal council or audit committee installed; S = natural logarithm of total assets of rm i in period t. ***, **, *: signi cant at 1%, 5%, and 10%, respectively.Source: Prepared by the authors.

Table 6
Results of the conditional conservatism model t

Table 7
Summary of results for the research hypotheses

Table 9
Results for the timeliness model (independence) = β 0 + β 1 .NI i,t + β 2 .∆NIi,t + β 3 .I n + β 4 .NI i,t *I n + β 5 .∆NI i,t *I n + ∑ .βn .Control + ε i,t Note: R i,t is return on (the most liquid class) share of company i in year t adjusted by dividends and splits (four months after the end of the year) and scaled by share price in t-1, NI i,t is earnings per share of company i in year t and scaled by share price in t-1, and ∆BV i,t are changes in net income per share of company i in year t and scaled by share price in t-1 (I n according to model 1).The rst line shows the coef cients and the second line, the t statistics.Standard error is robust and clustered by rm.The variables P, LPA, and PLA are scaled by price in t-1 (four months after the beginning of the year).Source: Prepared by the authors.

Table 10
Results for the conditional conservatism model (independence) NIi,t = β 0 + β 1 .D i,t + β 2 .R i,t + β 3 .R i,t *D i,t + β 4 .I n + β 5 .D i,t *I n + β 6 .R i,t *I n + β 7 .R i,t *D i,t *I n + ∑ .βn .Control + ε i,t Note: NI i,t is earnings per share of company i in year t, scaled by share price at t-1 and R i,t is return on share (the most liquid share class) of company i in year t adjusted by dividends and splits (four months after the end of the year) and scaled by share price at t-1 (C n according to model 1).The rst line shows the coef cients and the second line, the t statistics.Standard error is robust and clustered by rm.The variables LPA and R are scaled by price in t-1 (four months after the beginning of the year).D = dummy variable considering 1 if the return is negative and 0, otherwise; DIND n = dummy variable if rm i has at least one independent controller member (DIND 1 ) or administration member (DIND 2 ) in the scal council or the audit committee; LEV = leverage of rm i in period t; I = variable that takes values for member independence; IND n = percentage of independent controller members (IND1) or administration members (IND 2 ) in the scal council or the audit committee in rm i; GRO = revenue growth of rm i in period t; S = natural logarithm of total assets of rm i in period t. ***, **, *: signi cant at 1%, 5%, and 10%, respectively.Source: Prepared by the authors.