Corporate Behavior: An Exploratory Study of the Brazilian Tax Management from a Corporate Social Responsibility Perspective

: A look into the literature on corporate social responsibility (CSR) reveals few studies focusing on the relationship between ethical concerns and corporate behavior of companies that perform tax evasion management. This study links tax management with ethics and CSR reporting. The purpose of this article is to analyze financial and social responsibility information disclosed by the five main Brazilian construction companies that are being investigated in Brazil’s Operation Car Wash (Operação Lava-Jato—in Portuguese) because of inappropriate behavior. Based on the theoretical concepts of organizational façades and organized hypocrisy, we used content-analysis methodology and lexical search approach to analyze the consistency between the practices of tax management and CSR reporting. The results reveal evidence of aggressive tax management. To meet its tax management objectives, a company usually manages and plans taxes accordingly, delaying the payment of tax debt and not reporting all tax risks, thus being fined for violations of the law. We found evidence of organized hypocrisy and organizational façades, since there are contradictions between the tax behavior of the investigated companies and their CSR and ethical discourse.


Introduction
Tax management is used to maximize companies' corporate income [1,2]. Their practices play a critical role when we discuss corporate social responsibility. Especially in the case of tax evasion and avoidance, as well as moral hazard problems, tax management practices imply serious social issues associated with corruption in the areas where companies develop their activities.
Enderle [3] claimed that business ethics helps to improve the quality of decision-making and actions at all business levels. Ethics should lead to the moral behavior of their managers and employees in all situations, including tax management, since tax collection and tax revenues are important to assure the provision of well-being to society. That is, tax revenues enable adequate infrastructure, health, education, culture, employment, social income distribution, and public safety. Furthermore, reaching a certain ethical and responsible commitment implies well-being, that is, everyone wins: the company, the workers, and society [4].
The purpose of this article is to analyze the financial and social responsibility information disclosed by the five main Brazilian construction companies that are being investigated in Brazil's Operation Car Wash (OCW) because of inappropriate behavior. In this way, we verify whether these companies manage tax issues in accordance with the principles of CSR and ethics, which they claim to adopt. Having as background the OCW, we seek to understand whether the discourse on CSR of the construction companies under investigation is compatible with their tax behavior.
We analyzed financial and non-financial data of companies Andrade Gutierrez, Camargo Correa, OAS, Odebrecht, and Queiroz Galvão. Based on the theoretical lenses of organized hypocrisy [5] and organizational façades [6][7][8], we used the content-analysis methodology and lexical survey to understand whether companies that are being investigated because of corruption practices pay their tax debts and have the ethical behavior they claim in their CSR reports.
This paper contributes to the CSR accountability literature by voicing the concerns about ethical and corporate behavior of companies that perform tax evasion management using moral hazard behaviors to maximize shareholder interest to the detriment of the social welfare of the community. Since only a few previous studies linked tax management with corporate social reporting and ethics, this paper fills a gap in the literature. To the best of our knowledge, this is the first paper to explore the level of consistency between corporate discourse and tax management in the case of companies that were caught by the OCW.
After this Introduction, the paper proceeds as follows. In Section 2, we present the literature review, providing an intersection regarding our theoretical framework and the previous literature on the relationship between CSR disclosure and tax management. In Section 3, we present the main characteristics of the Brazilian tax system and the tax aspects of Brazil's OCW. In Section 4, we present the research method. Then, we proceed with the interpretation and critical discussion of our empirical results in Section 5. Section 6 offers some concluding thoughts and suggestions for future research.

Literature Review
This section provides a short review of the main concepts addressed in this research: organized hypocrisy, organizational façades, corporate social responsibility, ethics, and tax management. Notwithstanding the few research studies connecting tax management with ethics and social responsibility, there is a narrow intersection between these areas.

Organized Hypocrisy and Organizational Façades as Mechanisms to Manage Legitimacy
The term organized hypocrisy was used by Brunsson [5] to describe organizations that have multiple norm systems. This author argues that organizational decisions sometimes play several roles implying, different designs of decision processes [9]. This prompts managers to argue different things to different stakeholders and lead to inconsistencies between speech and actions [10]. Some organizations are more homogeneous than others, but most, if not all, have divergent identities. This happens because some organizational members also have loyalties to other interests and need to maintain bases of legitimacy in the eyes of such interests. Organizational action requires keeping a balance between the norms and expectations of competing groups [11]. Hypocrisy "is a fundamental type of behavior in the political organization: talking in a way that satisfies one demand, deciding in a way that satisfies another one, and supplying products in a way that satisfies a third one" [5]. To achieve a coordinated and collective action is the fundamental organizational problem [9,12,13].
Bird and Soundararajan [14] argued that sustainability speeches are often seen as empty statements for public consumption. The existence of inconsistent information and the consequent perception of suspicious reasons associated with CSR are closely associated with the concept of organized hypocrisy. A company can be perceived as hypocritical when there is a discrepancy between its statements and its behaviors [15,16].
Lipson [17] explained that organized hypocrisy is characterized by dissociated discourses, decisions, and products. It appears when there is a decoupling between discourse and corporate behavior, reflecting the inconsistencies of the organizational environment [18]. In these cases, organizations deal with inconsistent norms and conflicting pressures, establishing separate structures and processes that are "uncoupled" to satisfy each of the different interests. Nickell and Roberts [19] explained that the solution of hypocrisy represents a less devastating choice to the parties involved than the one that would result from a "heavier" declaration.
More than a possible problem, hypocrisy is seen above all as a solution for modern organizations that are subject to inconsistent demands and conflicting pressures from their stakeholders. Thus, actors in organizational hypocrisies shift within different espoused theories and discourses as they interact with different actors having divergent norm systems [10]. Companies usually build various façades to impress different stakeholders. For example, they can have a "child-friendly" profile of reporting social action for the education of children and adolescents and for the eradication of child labor in CSR reports but they avoid paying taxes that would be applied to early childhood education, health, sports, and school meals.
There is a loss of credibility if the company hides behind make-up, a mask, a kind of symbolic front or a façade, which, as Abrahamson and Baumard [6] argued, if revealed, can make stakeholders withdraw their support-they would divest, dismiss, sue, and generally discredit the true organization revealed behind the façade.
Cherian et al. [20] argued that the CSR is viewed as a strategic approach for companies' reputation and competitiveness. Many companies disclose socially correct speeches in their CSR reports, but, in practice, they do not act according to the principles they claim to follow. Cho et al. [8] considered that these behaviors fit the theoretical models of organized hypocrisy and organizational façades. In this sense, they argued that organized hypocrisy is a way to manage stakeholders' conflicting interests through a false speech that will satisfy all their demands. The organizational façades, on the other hand, would be a symbolic front, erected and conceived by the organization to reassure and mislead its stakeholders about the legitimacy of the organization and its management.

Theoretical and Practical Issues on Social Responsibility, Ethics, and Tax Management
Corporate taxes represent the state's compensation for supporting companies so that they can perform their activities (for example, providing health and education systems to present and future labor work). The state, represented by society, expects retribution for the impacts the company has caused and for the support it provides companies so that profits are achieved. Gribnau and Jallai [21] stated that responsible companies should be willing to pay a fair share of tax, as it is not coherent to benefit from public goods and services without adequate compensation.
Studies such as those developed by Nadiah et al. [22], Van Renselaar [23], and Lanis and Richardson [24] report aggressive behavior of large companies through both elusive and tax evasion planning. In this perspective, Fourati, Affes, and Trigui [25] explained that strong disclosure of economic, environmental, social, and corporate governance activities is associated with a higher level of tax evasion, indicating that CSR and tax evasion are complementary strategies. Hoi, Wu, and Zhang [26] argued that companies with poor CSR performance engage in aggressive tax evasion more than those with better CSR performance. However, Christensen [27] found that companies that disclose CSR information are more ethical since they are less likely to engage in inappropriate actions such as tax evasion, corruption, and bribery. Dietsch [28] argued stronger state intervention and stricter rules are needed to ensure socially responsible corporate conduct.
Joo, Eom, and Shin [29] explained that firms attempt to create social and improved economic value by implementing CSR. Brieger et al. [30] identified CSR as a source of legitimacy, of preservation of capitalism, and as an ideological movement. They argued that CSR consolidate and legitimize the power of large multinational corporations, which, despite issuing well-structured social and environmental reports, are eventually involved in cases of aggressive tax management. Branco [31] mentioned cases of dubious tax behavior by Apple, Google, Amazon, and Starbucks. Christensen and Murphy [32] also identified a series of corporate scandals involving large companies such as Enron, WorldCom, Tyco, Yukos, Parmalat, and the Big-Four global accounting firms. They drew public attention to the growth of tax evasion mechanisms such as transfer pricing, refactoring, special-purpose enterprises, corporate investments, dubious charitable funds, and other modalities for tax abuse. Gomes and Abreu [33] stated that tax planning is characterized by the pursuit of three basic objectives: (i) to prevent tax obligation from materializing, avoiding the occurrence of the chargeable event; (ii) to reduce tax liability by lowering its taxable amount or tax rate; and (iii) to postpone tax payment.
Kim and Zhang [34] found that politically connected companies using a range of corporate political activities, namely political campaign contributions and lobbying, are more aggressive than unconnected companies, due to better tax law information and lower market pressure for transparency. Nadiah et al. [22] sustained that, when an adequate and sound governance structure is embedded in organizations, they become less aggressive in managing tax payments.
In line with the view of those who do not believe that the principles of CSR are real and adequately addressed in management practices, Preuß and Björn [35] found that CSR is negatively associated with tax payments and they suggested that corporate tax payments and CSR from European public companies act as substitutes. These authors concluded that corporate tax payments may not be viewed as part of CSR. Mgbame et al. [36] argued that, since the main objective is to maximize shareholder value, companies have monetary reasons to adopt tax policies which allow them to reduce their tax obligations. What is noticeable is that, in some cases, although the company states that it supports ethics and CSR principles, there is some evidence that its tax management does not seem to be consistent with its discourse. According to Sikka [37], companies engage in tax evasion practices to increase shareholder wealth.
CSR regulators are now aware of the need to report tax and payments to governments. In 2017, the Global Sustainability Standards Board (GSSB), GRI's independent standard-setting body, initiated a project to develop new disclosures related to tax, to help promote greater transparency on a reporting organization's approach to taxes. It is important that tax data are more widely accessible since that will help to build stakeholder trust. The standard was approved in 2019 and is known as GRI 207: Tax 2019. Although earlier adoption is encouraged, the effective date is 1 January 2021.

The Brazilian Tax System Scenario and the Anti-corruption Movement
The Brazilian tax collection framework is one of the most complexes in the world [38] because there are three levels of tax authority: federal, state, and municipal. This is also a characteristic of other huge countries, such as the USA. All three levels of government (federal union, states, and municipalities) have the right to create their specific categories of taxes, as defined in the Brazilian Constitution [39].
Municipalities are the smallest fraction of the political-administrative division of Brazil. Municipalities are part of a state, which is part of a federal unit, also called the union, represented by the federal power.
Tax payment and the accessory obligation are managed by government bodies at three different levels [40]. Table 1 highlights the main current taxes in the Brazilian tax system. The high complexity of the taxation system generates large costs for companies to manage taxes [39]. The taxes incurred differ in business transactions both on the receipt and on delivery of goods and services [40], and are embedded in the prices and passed on to the buyer directly or indirectly, that is, some taxes are paid by the buyers. To calculate the tax obligation, it is necessary to consider, among other aspects, the site of origin and destination, special taxation regimes, nature of the transaction (sale, purchase, remittance, and donation), product and service code, and whether the commercial relations are between Brazilian companies or Brazilian companies and other tax jurisdictions.
The company that sells the goods or services calculates and collects taxes that should be passed on to the government within a time limit set by law. Nevertheless, this does not always happen. Some companies collect the tax and never pass it on to the government; others take a long time to make this transfer, using a tax installment program. This is a legal and legitimate procedure in the case the company cannot afford to pay the tax. According to the Brazilian Court of Auditors [41], the large number of installments that arose in Brazil spread the culture of non-payment of tax debt, since it creates the expectation that new programs will be provided by the state, permitting the delay of tax payment. Nevertheless, there is a low rate of installment payments, due to: default; breach of rules; or because taxpayer chooses to include the installment debt in another supervening program.
The creation of installment programs has not achieved its objectives, namely the increase in the collection of tax debts [42]. Faber and Silva [43] stated that there are no other tax administrations in the world that allow for installments as those granted in the Brazilian federal level, which vary from 60 to 180 months, and there may also be installments with no defined deadline.
Despite these circumstances, which place Brazil at a significant disadvantage when compared to other countries, the Brazilian tax authorities (especially at the federal level) have implemented a strong digital program for tax control named SPED (a Portuguese acronym for the Public System of Digital Bookkeeping). Conversely, Deloitte [40] argued that the Brazilian tax system computerization has rendered tax compliance much more dynamic: electronic monitoring makes the inspection process to require increased attention by companies so as to ensure that no information is omitted or incorrect.
Started in the middle of the past decade, at the moment this digital system comprises a large framework of detailed tax, commercial, and operating data, organized in standardized formats, which taxpayers must prepare and submit electronically to the tax authorities [38]. SPED is a tax framework organized in different projects: Digital Accounting Bookkeeping (SPED Accounting), Digital Tax Bookkeeping (SPED Fiscal), and Electronic Tax Invoice (NF-e), in Portuguese, "nota fiscal eletrônica", functioning in an integrated way within the three governmental supervisory levels (federal, state, and municipal).
With the SPED filing obligations coming into force, irrespective of any related tax inspection, the SPED submissions are mandatory on a regular basis. Depending on the file, electronic filing could be made monthly, annually, or even instantly, on a transaction-by-transaction basis (which is the case of the electronic tax invoice). According to PwC [38], SPED is already giving rise to positive changes regarding compliance, particularly in relation to tax inspection and tax collection. Additionally, it is important to reduce space for illegitimate tax procedures and to combat tax evasion, a typical problem of informal-economy situations in which the exposure to corruption (bribery) is higher. PwC [38] pointed out that SPED also increased transparency on the part of taxpayers to tax authorities via digital reporting.
Brazil's OCW investigates acts involving crimes of corruption, money laundering, and tax evasion under the criminal law. It has also included crimes such as fraudulent financial movements, implying tax crimes arising from relevant tax evasions. Hence, tax offences are being discovered by unraveling fraudulent schemes, thereby allowing for the identification of several taxes and social contributions evasion and the consequent application of pecuniary penalties. Initiated in 2014, OCW is the largest corruption-combating and money-laundering investigation ever performed in Brazil. The participation of the construction companies in this fraud scheme is quite relevant. They were organized into a type of cartel to commit irregularities. This anticompetitive behavior meant fixing the value of the work in advance as well as determining who would win the contract with stateowned company Petrobras. As a result, the companies were able to overprice public works for their own benefit, enticing Petrobras employees and other civil servants and bribing political agents. In turn, the winning company shared illicit gains with other companies and participants in the fraud.
As a result, they are accused of corruption, cartel formation, bidding fraud, money laundering, and criminal organization. When convicted, in addition to paying taxes, they must pay a fine or provide services. The companies involved have already signed or may sign Leniency Agreements (Mercy Programs) with national and international authorities, to whom they confess violations, promising to collaborate with the investigation and in return have the benefits of penalty reduction.

Research Method
This section presents the methodological procedures applied in our investigation to analyze our research question: "How compatible is the discourse on CSR of the five main Brazilian construction companies with their tax behavior?"

Research Design and Data Collection
To conduct the research, we used descriptive and exploratory analysis to understand how the main Brazilian construction companies managed tax issues in accordance with the principles of CSR and ethics, which they claimed to adopt. Using a multi-case-study approach [44], we investigated the phenomenon in its natural environment [45], adopting multiple sources of evidence about five entities. Yin [44] stated that, in general, a case study represents an appropriate strategy when questions such as "how" and "why" are posed.
Companies were selected based on the criterion of "inappropriate practices that were made public" by any mass media communication vehicle, in the period 2009-2018. This technique was also used by Lanis and Richardson [24], who conducted an electronic survey on tax-aggression cases by public corporations, using terms such as "tax aggression", "tax evasion", and "amended tax review". We used similar expressions, but we included terms related to scandals, bribery, and corruption. Thus, data were collected from the Web in April 2019 in national-media online publications, seeking news about companies that allegedly behaved inappropriately.
This search procedure helped to identify the initial population sample of this study, and companies from various business activities appeared in Web search results. Several companies emerged such as Apple, Siemens, Volkswagen, Uber, Cisco, Daslu, Dolly, H. Stern, JBS, and Schincariol (this firm changed its name to Brasil Kirin), which were excluded because they belong to different business segments, which would make comparability difficult. In addition, in some cases, neither their annual report nor their CSR report was found. However, the construction sector was the most cited.
Thus, Andrade Gutierrez, Camargo Correa, OAS, Odebrecht, and Queiroz Galvão were selected to be investigated in depth through extensive content analysis [46]. Their financial and non-financial reports were found on the company website or in newspapers. In the appendix we provide information on the reports that were analyzed in this study.
After an initial organization of documents, we conducted the content analysis process of the data that were thoroughly examined and contrasted, by analyzing written communication [47] to provide a condensed description of the patterns revealed. We did not predefine the categories of similarities, differences, and relationship types because there is a gap in the explicit definition of this type of matter in the reviewed literature. Therefore, it was an exploratory process of knowledge-building throughout the content analysis with a subsequent triangulation among the researchers of this article [48], aiming to validate the results and to ensure robustness and comprehensiveness of conclusions.
During the period analyzed, various news was disclosed about the business scandal arising from OCW results. Table 2 shows the inappropriate behaviors that were published in the news until 2018.

Disclosure of Financial Data
We examined companies' annual reports, seeking to identify tax management practices, i.e. finding the tools used by companies to reduce, postpone, or avoid paying tax. We did not analyze the effects of these tools on the companies' financial situation. By using the financial reports, the objective was to verify which tax management practices were adopted before and after the OCW.
On the selected companies' homepages or in regional newspapers, we could find many financial statements of the analyzed companies but not all. Despite our intense efforts, we were unable to obtain same-period information for all companies. Thus, the analyses were performed in different years, but preserving the premise of belonging to the period before and after 2014, the year in which OCW began ("Stage 1-before 2014 (exclusive)" and "Stage 2-after 2014 (inclusive)"). As shown in Table 3, we collected 40 financial statements (80% of the total expected for the five companies).

Years
Annual reports Note: N represents "No" and is equal "10"; Y represents "Yes" and is equal "40".
Reading the financial reports, we understood that companies used similar tools to reduce, postpone, or avoid the payment of taxes. We were able to map the management tools that characterized the companies' tax behavior.

Disclosure of Non-Financial Data
To analyze the content of the companies' speech, a search measured the frequency with which some keywords (attributes) appeared in the text of their CSR reports, homepages, and codes of conduct. The more a word is quoted, the greater its relevance to the company discourse. In this sense, Bardin [46] explained that the importance of a word in the context of the content analysis methodology increases with the frequency it appears in the text. To count the frequency of appearance of words, we used a demo version of the software MAXQDA 2018.2, a product of VERBI [49], which is suitable for qualitative data analysis and enables the systematic evaluation of qualitative data and interpretation of textual data through content analysis [50]. This tool has been used in different research areas, such as sociology, economics, law, and business administration [49]. The result of the lexical search enables the classification of terms according to their meaning, with pairing of synonyms and close meanings. The MAXQDA returns a table with an excerpt of the sentence where the word is inserted, which allows us to understand if the meaning in which it was used meets the objectives of the analysis.
We found different types of documents that showed companies' official discourse on CSR, as shown in Table 4. The results show that companies disclosed 23 standalone reports using the GRI guidelines, and 11 companies provided combined reports (the CSR information plus annual reports). When CSR information was not found in standalone or combined reports, we used the companies' code of ethics and conduct and the corporate social information found on their website (ANG, CCO and OAS). We observed that, after OCW, some companies created or updated their codes of ethics and conduct and disclosed CSR information on their websites as good governance initiatives. Preuss [51] also used companies' code of ethics; Everett and Tremblay [52] used information from websites to analyze narrative cases about ethics and Joo, Eom, and Shin [29] to analyze the impact of CSR on the business ecosystem; and Packer et al. [53] used companies' websites to analyze corporate social responsibility (CSR) in the seafood industry. Table 4. Types of CSR Information analyzed.

Years
Types of CSR Information  ANG  CCO  OAS  ODE  QGA  2009  GM  GM  N  CR  N  2010  GM  GM  N  CR  CR  2011  GM  GM  GM  CR  GM  2012  GM  CR  GM  CR  GM  2013  GM  CR  GM  CR  GM  2014  GM  CR  N  CR  N  2015  GM  CR  N  GM  N  2016  GM  N  N  GM  N  2017  N  N  N  GM  GM  2018 WEC* WEC* WEC* GM GM Note: N is "No data"; CR is "Combined Reports" (annual reports combined with non-financial information); GM is the "GRI model"; WEC* is the website for code of ethics and conduct.
The ISO 26000:2010 provides guidance to all types of organizations, regardless of their size or location. It is recognized as a reference document aimed at motivating social responsibility initiatives and it is intended to assist organizations in contributing to sustainable development. This international standard, reviewed and released in 2017, was reproduced in full in Brazil by ABNT (the Brazilian Association of Technical Standards). The document explains that socially responsible organizations need to adopt ethical behavior which is in accordance with the accepted principles of moral and correct conduct. In the context of community development, the ABNT Standard [54] highlights that compliance with tax obligations is essential to help governments to generate revenue that is addressed to crucial development issues, including the health and welfare of society.
Thus, we used some terms that were extracted from this Standard to perform the lexical survey. The words searched appear in Figure 1, representing meanings that are similar to Ethical Behavior (ABNT Items 2.7, 2.18, and 4.4), Tax Revenue (ABNT Items 6.8.7.1 and 6.8.7.2), Government or public administration (ABNT Items 4.6 and 6.8.7.1), and Social Welfare (ABNT Items 6.8.1 and 6.8.7.1). In the annual CSR Reports, companies are expected to use these terms, which are significant attributes for proper conduct (ABNT Items 3.3 and 4.7), since, if companies behave ethically, taxes will be responsibly passed on to the government, which will provide more goods and services to society, increasing social welfare. Lee and Yu-Lan [55] also used a computational linguistic analysis to quantify the importance of each special term on CSR reports.

Results and Discussions
In this section, we describe the empirical findings of the data analysis. Firstly, we analyzed the financial statements and its explanatory notes to identify how they managed, accounted for, and highlighted tax issues. Our interest was to know about payments, installments, and other transactions to postpone, omit, or avoid tax. These issues provided information about the management practices that characterized the tax behavior of each company in the two-stage survey-before and after 2014. Secondly, to provide the main insights into the obtained dataset, we identified unusual and atypical transactions that could directly or indirectly influence illegitimate reduction of tax. The independent audit reports were also considered as important. A much larger tax liability value was not recognized as provision of contingent liability because the likelihood of having to pay these taxes and consequently the possibility of an outflow of resources was judged as remote by CCO. The provisions for taxes liabilities correspond to lawsuits and administrative proceedings related to all these taxes. In these years, the provisions for risks arising from differences in labor rights were recorded (INSS and FGTS). In Stage 1, there were no records on installments or tax arrears, but there were many contingencies related to the risks of tax payments. Some expenses with donations were recorded, but there is no clarification about this. The independent auditor's reports were issued without qualifications in Stage 1.  OAS The company Construtora OAS was founded in 1976. In Stage 1, the company distributed part of the profit to its shareholders, except 2011 and 2012 when it paid interest on equity. In 2009, the spun-off portion was merged into OAS Engenharia, which took the Tax Recovery Program-REFIS related to PIS, COFINS, IRRF, INSS, IRPJ, and CSLL. The OAS Group had offshore companies in several countries. The contingent liabilities for tax claims were significant, but the company recorded only a small amount as a provision for tax risks to cover tax claims, such as ISS, IPTU, ICMS, IRPJ, CSLL, PIS, and COFINS, as well as Social Security. The notes to the financial statements did not include information on tax-payable balances recorded in non-current liabilities, which may represent overdue taxes. The independent auditor's reports for 2009, 2012, and 2013 were issued without qualifications. The independent auditor's reports for 2010 and 2011 expressed a qualified opinion because OAS sold a subsidiary, resulting in a gain recorded in the Profit or Loss for the year when it should have been recorded directly in equity.  Odebrecht-ODE The company Construtora Norberto Odebrecht was created in 1944. We were not able to find the annual reports for 2009 and 2011. The company distributed profits from 2010 to 2013 and reduced its share capital several times. From 2010 to 2013, contingent liabilities were significant, but the company recorded only a small amount as a provision for risks to cover tax or labor compensation claims; nonetheless, the notes did not detail the types of taxes involved in the claims. Since 2009, the company jointed to installment programs such as REFIS or PAEX, aiming to regularize its tax and social security liabilities. The independent auditor's reports for 2010, 2012, and 2013 were issued without qualifications.  Queiroz Galvão-QGA The company Construtora Queiroz Galvão was founded in 1953. We did not find the annual reports for 2012. The company distributed profits and paid interest on equity in Stage 1. In 2009, 2010, 2011, and 2013, the board of QGA considered that the appropriate measures for any tax, social security, and labor contingencies claims had already been considered. The board reassured that there was no need to recognize any provisions for taxes in the annual reports. QGA has not published evidence of legal discussion on tax issues, tax arrears, or installments. The independent auditors' reports for 2009, 2010, 2011, and 2013 were issued without qualifications.

Companies' Tax Behavior after 2014 (inclusive): Stage 2
We present Stage 2, which refers to the analysis of the reported financial information in the dataset from 2014 to 2018.  Andrade Gutierrez-ANG In 2014, on media news about OCW investigations, the ANG board reported in note no. 1 that it was unaware of irregularities in contracts and that there was no filing of any lawsuit against the company resulting from these investigations. In the same year, ANG jointed installment payments of INSS, IPI, and IRPJ. This year the company recorded provisions for tax risks arising from ongoing lawsuits but did not specify which taxes. ANG explained that it also had contingent liabilities related to ISS, ICMS, IRPJ, and CSLL, whose likelihood of loss was only considered possible. The financial statements for 2015 and 2016 revealed that there was a reasonable increase in the installment payment and tax provisions recorded, given that the ANG had infraction notices because of the calculation of the IRPJ and CSLL taxes. In 2015, ANG had considered several expenses as deductible, but in accordance to the tax law they should not have been deducted. In 2016, ANG reported in note no. 24 that it had to regularize assets and revenues which were realized abroad without being recognized in Brazil. These revenues were subject to taxes and penalties in Brazil. The independent auditors also emphasized the resources repatriation event. In 2015 and 2016, there were dividend payments to shareholders based on the annual profits and reserves. In 2014 and 2017, the company made a loss, but dividends were paid from the profit reserve. In 2015, 2017, and 2018, according to the notes, ANG reported that it signed the Leniency Agreement with MPF, CADE, CGU, and AGU. In 2017, ANG opted to join the Tax Regularization Program (PRT), including debts already recognized in its financial statements, as well as debts in the process of administrative and judicial trials. It also joined the Special Tax Regularization Program (PERT). In 2017 and 2018, contingent liabilities representing about 50% of the company's share capital were recorded. ANG recognized some inappropriate practices in public administration contracts and paid a fine. The Brazilian Court of Auditors (TCU) blocked ANG's non-financial assets to guarantee possible damages caused in other contracts, with other public company, under investigation. The independent auditors' reports for 2014, 2015, and 2016 expressed a qualified opinion because of: (1) uncertainties in accounts receivable; (2) lack of information on the Leniency Agreement with CADE and other public administration bodies; and (3) difficulties in determining whether there would be any need to make adjustments to the balance of investments and to the equity (that is, they could not identify if the equity was over or underestimated). In 2018, its parent company Andrade Gutierrez S.A. (AGSA) reported that it operated with three offshore companies. In 2017 and 2018, ANG published the annual reports without the notes, but it reported that the notes could be consulted at its headquarters. We used some ANG information disclosed by AGSA to overcome this obstacle. The independent Auditors' reports for 2017 and 2018 were not found in the newspaper or on its homepage.  Camargo Corrêa-CCO The company made a profit in 2014 and a loss in 2015, but, in these years, it used the profit reserve to pay dividends. Throughout this research timeline, provisions for tax risks were recorded for tax claims that correspond to lawsuits related to the payment of taxes such as PIS, COFINS, IRPJ, CSLL, INSS, ISS, ICMS, among others. The identified contingent liabilities refer to ICMS, COFINS, and IRRF from 2014 to 2018. Besides, CCO recorded a provision for labor claims and related taxes and contributions. In 2014, the company reported in note no. 29 that it was aware, through the press, of OCW investigations and that it had taken several measures to clarify the facts. Despite mentioning a very significant possible loss, it recorded only a small amount as a probable loss. Significant amounts of liabilities with MPF and CADE were recorded in 2015 and 2016, arising from terms of commitments and Leniency Agreement. According to note no. 29 to the financial statements for 2015, this agreement reduced uncertainties about legal sanctions and made cash flow more predictable. In 2016, CCO reported in note no. 15 that it committed to return funds as compensation for damages caused to Brazilian society. According to the independent auditors, differences between the amounts of irregularities that will be agreed upon at OCW and the amounts of provisions and contingent liabilities disclosed in 2017 and 2018 can be significant. CCO paid part of the tax debt, including amounts determined during the investigation. This company adhered to the PERT to pay IRRF and other taxes, which were not provisioned. The company was fined in 2017 and had to reverse expenses that were considered previously as deductible for tax purposes.
In 2014, the independent auditor's report expressed a qualified opinion because the outcome of OCW could not be determined and could affect the provisions amounts and the contingency liabilities amounts. In 2015, the independent auditor's opinion was issued without qualifications. In 2016, CCO published the annual reports without the independent audit report, but in the following year the independent auditor informed that there was an audit report for 2016 with a caveat because of the possible effects of in-progress collaboration agreements and internal investigations. The independent auditor's reports for 2017 and 2018 expressed a qualified opinion because the supervisory board had identified irregularities in others contracts not covered in the agreements previously signed in the scope of OCW.  OAS OAS has been in the process of Judicial Recovery since April 2015, as a pre-bankruptcy. The last annual reports of OAS made available on its website referred to 2014 and 2015. The company made losses exceeding its share capital in 2014 and 2015, but, despite this, in 2014, it paid dividends and interests on equity to its shareholders. In 2015, the company presented negative equity. OAS had infraction notifications and it had to respond to some tax, civil, and labor lawsuits. In 2014 and 2015, the annual reports indicated a provision for tax risks with irrelevant value; however, tax contingencies for possible losses comprised large amounts. No provisions were recorded to cover any fines and taxes due to irregularities found under the OCW. The notes to financial statements do not clarify the nature of taxes payable, whether they were overdue or not, nor if they would be paid in installments. The company reported on its website that it had signed a Leniency Agreement in November 2019 with CGU and AGU, agreeing to pay high amounts of fines. In 2014, the independent auditor's report expressed a qualified opinion because: (a) there was uncertainty about receivables from subsidiaries; and (b) possible impacts of OCW. In the emphasis of matter paragraph, the auditor stressed that the Judicial Recovery plan indicates the existence of significant uncertainty that may cast doubt on the company's ability to continue as a going concern. The external auditor issued an adverse opinion in 2015, because: (a) No evidence was found to confirm the realization of accounts receivable from clients and of related parties and, therefore, impairment losses should have been recorded. (b) The board adjusted the balance of its liabilities based on the new conditions set forth in the Judicial Recovery Plan. In doing so, the board calculated gains that were credited in the income account for the year. Thus, a liability should have been recorded and only when the conditions are met a gain can be recorded. (c) Considering the content of the investigations, it seems likely that other facts could affect the financial statements, but the company has not recorded any provision to address this matter.  Odebrecht-ODE As in 2014, the ODE made profits in 2015, but, in these years, it did not pay dividends to shareholders. In 2015, the company reported that the annual reports did not address the impacts of OCW investigations or legal proceedings arising from them. Therefore, only other probable losses on labor, tax, and civil claims against ODE were accounted for, but the notes did not clarify what these claims were. In 2017 and 2018, the company recorded tax provisions to cover claims related to taxes debts, but the ODE did not disclose the amount related to each type of tax. From 2016 to 2018, ODE and other companies from the group jointed to Leniency Agreements under OCW and began to cooperate with the investigation. In 2015 and 2018, ODE recognized the existence of contingent liabilities not recorded, which may occur in the long term, whose values are very significant due to OCW. In 2017 and 2018, the company suffered a very significant loss, partly absorbed by profit reserves. In 2017 and 2018, ODE adhered to the PERT to regularize the tax liabilities, including those from the OCW, not accounted for at the appropriate time.
In 2015, the independent auditor's report expressed a qualified opinion because no evidence was provided to clarify matters related to OCW and its impacts on financial statements. The independent auditor's report for 2018 expressed a qualified opinion because it was not possible to obtain appropriate evidence related to the amount of the provisions recorded and there was uncertainty about the need for additional provisions to deal with the agreements that were in progress in Brazil and abroad. ODE has been in the process of Judicial Recovery since 2019, as a pre-bankruptcy.  Queiroz Galvão-QGA Despite the news in the media and lawsuits against QGA, the company claimed that it had no involvement with OCW. We did not find information about the QGA Leniency Agreement. In 2014 and 2016, QGA made losses. In these years, there was payment of dividends using unrealized profit reserves. In 2014 and 2015, there were records of short-and long-term tax provisions, but the notes do not clarify the type of tax. In the financial statements from 2014 to 2016, there were provisions for labor risks, with insignificant value, but, in 2017, the value increased. In 2017 and 2018, QGA disclosed some estimated contingent liabilities related to civil lawsuits involving contractual disputes and third parties' indemnity claims, labor and tax contingent liabilities, whose likelihood of loss was considered possible, but the notes do not clarify the details. QGA did not disclose any note about installments or late payment of taxes. QGA argued that there is no need to record a provision for events that may result from OCW because it has carried out internal and external investigations to identify any non-compliance and disobedience to the law, but no irregularities have been pointed out yet. However, in a trial related to OCW, in 2017 and 2018, TCU stated that OCW was unable to enter into new contracts with the federal public administration because of irregularities in the contracts.
The independent auditors' reports for 2014 and 2015 were issued without qualifications, but, in the emphasis of matter paragraph, they reported that financial statements do not include any effects that might come from OCW. In 2016, 2017, and 2018, the independent auditor's reports were issued without qualifications. However, the auditors reported some uncertainties regarding the results of the investigation of the OCW.
The results from the previous analysis of the companies' financial statements provided important information on the characteristics of financial and tax planning in the different periods analyzed, enabling us to obtain some evidence on what they do in practice. The discussion is presented in the following section.

Aggressive Tax Management
In the period after 2014, transparency was improved since off-balance sheet amounts were recorded and taxed. From 2014 onwards, accountability increased since companies started to disclose better-quality accounting information, a fundamental requirement for the compliance program that they have committed to implement. A search was conducted to identify the main types of tax management that appear in the companies' accounting reports and their position in a ranking of the level of aggressive tax management.
The survey was performed using the dummy variable method. In this study, these values represented the absence (0) or presence (1) of a certain aggressive tax management attribute, found in their financial statements. We also included the audit opinion because, as shown in the previous section, in many cases, qualifications are related to OCW. Table 5 presents the 11 types of evidence found in the companies' annual reports and that evince inappropriate or risky behavior.  From 2009 to 2018  ANG  CCO  OAS  ODE  QGA  Sum  1  Administrative tax collection process 1  1  1  1  1  1  5  2  Adverse audit report  0  0  1  0  0  1  3  Audit report with qualifications  1  1  1  1  0  4  4 Fines for tax and labor law violations 1 1 1 1 1 5 5

Evidence Types
High value of possible unrecognized tax risks 1 1 1 1 1 5 6 Late payment of taxes (tax debt) Lawsuit against some taxes 2  Note. 1 Prior to filing a collection lawsuit, the government proposes a friendly settlement through an administrative tax collection process. The existence of an administrative tax collection process indicates that the tax agent found evidence of non-compliance with the tax rule. 2 Normally when a company goes to the court against a tax, the payment of the tax is suspended. Companies have complained about almost all federal, state, and municipal taxes. 3 Tax infringement notification is the document issued by tax inspectors when they encounter irregularities in tax calculation. 4 In accordance with applicable laws, taxes are within the prices of services and products sold. The seller should pass the amount of tax to the government monthly and not use it for other purposes.
Nine of these pieces of evidence refer to typical tax management events and another two are also related to reliability of accounting information. As shown in Table 5, ODE is the company with the most types of evidence of tax aggressiveness, with nine occurrences out of 11. ANG have eight occurrences, followed by OAS and QGA, with seven and five, respectively. However, OAS had an adverse opinion expressed by the independent auditor in his report.
The most common types of evidence identified in the five companies are: (a) administrative tax collection process; (b) fines for tax and labor law violations; (c) high value of possible unrecognized tax risks; and (d) tax infringement notification.
Venard and Hanafi [56] argued that, if companies from an industry adopt a large degree of unfair behavior (not only corrupt behavior), a specific enterprise in the same industry will certainly imitate this behavior. They defined unfair behavior as the likelihood of companies to be dishonest. A company using unfair behavior is one that does not respect local regulations, such as paying government taxes, or which fails to observe or follow trade regulation. These five companies appear to have used tax aggressiveness. Studies by Gimenez, Hayashi Júnior, and Grave [57] indicate that a company's strategic position is associated with the management perception of a more frequent modal strategy perceived among its competitors. This mimetic behavior explains why all these major construction companies were involved in the OCW scandal and similar tax frauds. Although they did not pay the due tax debts, evidence reveals that they paid dividends to shareholders, even when companies made losses and probably had overdue taxes.

Organizational Façade and Hypocrisy in Business Discourse
Business commitment with ethics and with stakeholders is known to be formalized in the Codes of Conduct, annual CSR reports, the combined annual reports and on companies' website communications. In this section, we analyze if the companies' discourse on CSR, ethics, and sustainability is consistent with aggressive tax management.
All investigated companies voluntarily adopted the CSR guidelines in conducting their business. Before 2014, their CSR reports stated that they were ethical, socially responsible, and always concerned with social development, as shown in the excerpts in Table 6. Table 6. Excerpts of CSR discourse before 2014.

Company
Excerpts of CSR Discourse -Before 2014

ANG
In the CSR report for 2008, ANG argued that business growth is only possible from a serious, ethical, and responsible business attitude that leads to the creation of economic, social, and environmental value. In 2012, it confirmed that social and environmental responsibility and safety at work are pillars of a sustainable company.

CCO
In 2009, CCO stated the commitment with the sustainable development of Brazil and all countries in which the Group operates. In the CSR report for 2011, it pledged to always act fairly and correctly towards shareholders, professionals, customers, suppliers, governments, local communities, and society at large. CCO argued to comply with legal norms.

OAS
In 2012, OAS reported that its governance values were aligned with ethics in decision-making processes, and that all practices are based on the internal Declaration of Conduct. The company stated that, by paying taxes and assuming a zero-contingency policy, OAS generates value at various levels contributing to wealth generation.

ODE
In the CSR report for 2009, ODE explained that as part of its CSR policies, the company was helping in building a prosperous, fair, and environmentally sustainable society. It explained that it had a clear and permanent commitment to sustainability.

QGA
In the CSR report for 2011, QGA argued that the organization sought a transparent dialogue with the communities in its areas of influence. In 2013, the board added that social responsibility and care for the environment are integrated into the strategic planning.
These discourses are not consistent with the tax practice found in the companies' financial reports. Santos [15] claimed that corporate hypocrisy exists when there is a discrepancy between statements and behaviors. Organizational hypocrisy is promoted by boilerplates discourses that are a façade for providing an impression of ethical behavior and corporate social responsibility. As can be observed, there is a tension between sustainability discourse and practice [8].
Wagner, Lutz, and Weitz [16] observed that, when there is a lack of sincerity about the true motivations of companies to adopt CSR or when the motivations are companies' self-interest, stakeholders can perceive corporate hypocrisy. Hypocritical organizations frequently claim to be something that they are not. When there is inconsistency between CSR initiatives disclosed by a company and its actions, this may have an adverse impact. A company (for example, CCO) cannot claim to be committed with the development of Brazil and at the same time avoiding paying taxes that are necessary to improve health systems and education.
The OCW scandal influenced their CSR reporting after 2014. Some companies reported their mistakes and how they are getting over them; that they are devoted to creating or improving compliance and integrity programs and implementing code of ethics and conduct. In Table 7, we show some excerpts from the investigated companies' speeches. Table 7. Excerpts of CSR Discourse after 2014.

Company Excerpts of CSR Discourse -after 2014
ANG ANG decided in 2015 to initiate talks with the MPF, recognizing and clarifying wrongdoing, and to compensate companies and public agencies that were harmed. In addition, ANG publicly portrayed itself through the national press, recognizing the illegal practice. ANG stated that it had already paid taxes and fines that were overdue (or it was paying in installments).
CCOa CCO explained that, in 2014, it was striving to collaborate with the authorities. In the CSR report for 2015, it made the decision to recognize and correct misconduct found in the OCW investigations. CCO argued that it corrected deviations and reinforced its internal control system with modern and effective mechanisms; and that it also reimbursed society. Additionally, in 2015, it took fundamental steps to consolidate its commitment to integrity. This commitment was intensified with the updating of their code of ethics and conduct.

OAS
With the implementation of its code of conduct in 2014, OAS reaffirmed its commitment to comply with all laws and ethical values applicable to its economic activities. OAS argued that it invested in professional governance, as well as in strict guidelines to reduce risks in conducting business. In 2017, the CSR reported that OAS signed the Business Pact for Integrity and anti-Corruption Practices that aims to unite companies to promote a fairer and ethical market and to eradicate bribery and corruption. Regarding the leniency agreement signed in 2019, with CGU and AGU, OAS argued that the correction of inappropriate actions of the past opened the door to the future that it wanted.

ODE
During 2014, ODE implemented its code of conduct. In the CSR report for 2015, ODE stated that it acknowledged its mistakes and that it was learning a lot from them. It also recognized the need to apologize to all who may have been harmed by its organization's involvement in such episodes: its members and their families, customers, shareholders, suppliers, and also to the whole Brazilian society. In 2016, it approved its compliance policy. It recognized that it has been fined and it received penalties for leniency agreements. It publicly apologized to society. In the CSR report for 2018, ODE stated that it was committed to act ethically and transparently and that the most immediate challenge was the financial stabilization of the group by approving and executing the Recovery Plan, approved in June 2019.

QGA
In the CSR report for 2014, QGA launched a new code of ethics to guide employee actions in its relationships with stakeholders. QGA reinforced the public commitment to ethics and integrity in its business environment in 2016. In 2017, QGA implemented a new governance regulation and profound changes were made to its compliance program. QGA also decided to implement a sustainability policy which defines the guidelines of an economically viable, environmentally sound, and socially fair company. In the CSR report for 2018, QGA argued that will promote the risk analysis, seeking continuous improvement of its Compliance Program.
We analyzed the relevance of certain terms in companies' CSR reporting before and after the company was investigated by OCW. The result of the MAXQDA lexical survey shows the frequency of appearance of terms in CSR discourse for each one of them. Before 2014, the most cited attribute was "government-public administration". The frequency of this attribute remains relevant after 2014, standing out as the second in number of citations. The government remains the main client because of the major works it can contract. The relevance of this attribute in business discourse suggests that there was a preference for the economic dimension over the social and environmental dimensions of the sustainability triple bottom line.
Undoubtedly, in the post-2014 period, the companies started to further value "ethical behavior" in their speeches. The results show that "ethics" was the most used attribute in the reports analyzed (except CCO). This emphasis on ethics seems to contemplate reflections on the right, wrong, and moral conduct of management.
The low use of the terms such as "tax" and "welfare" seems to result from the low importance attributed to these concepts by the companies under analysis. All companies speak of social aspects, which include society's expectations for better public services and quality of life. However, this can only be achieved if tax collection is enough. The low frequency of the term "tax" and higher frequency of the term "welfare" in CSR discourses appear to indicate that the companies still do not understand very well the social function of tax and its importance for the social welfare.
Thus, we inferred that, before 2014, when the term "government" was emphasized in CSR discourses, management was riskier; taxes were probably not paid as required by law. In the next phase, after the scandals of OCW, companies' CSR discourses started to highlight "ethical behavior", thus companies are showing signs that they want to correct previous mistakes. However, it is still too early to say that they are fully complying with the ethical principles and CSR guidelines they argue to adopt.
Urquiza and Marques [58] also used corporate reports to analyze the discourse of a financial company. They found evidence of contradictions, concluding that although there is the discourse of sustainability and ethics, in organizational practice only profits really matter. Cho, Guidry, and Hageman [59] found evidence that the Dow Jones Sustainability Index (DJSI), one of the most visible indicators of excellence in corporate sustainability, is more influenced by what companies disclose than by what they perform.
In our cases, we identified that, after 2014, companies' speeches began to present terms that indicated changes in behavior and a greater commitment to society and ethics. Some companies recognized their mistakes and promised to correct them (ANG, CCO, OAS, and ODE), some apologized (ANG and ODE), and all started procedures to contain inappropriate conduct (ANG, CCO, OAS, ODE, and QGA). They used terms such as "apology", "compliance", and "integrity" more often. However, it is unclear whether they understand the important role of taxes in society. Thus, the new GRI standard on taxes is very welcome. It is expected that it can contribute to emphasize that there is no sustainability or social welfare without the payment of the correct amount of taxes by companies.

Conclusions
This paper shows how main Brazilian construction companies appear to have practiced aggressive tax management, even though in their CSR discourse they claim to be committed with the development of Brazil. The study revealed that the companies investigated in this research did not consistently manage tax issues with the principles of CSR and ethics. Based on our results, we show that the most frequent pieces of evidence of aggressive tax management were: (a) administrative tax collection processes; (b) fines for tax and labor law violations; (c) high value of possible unrecognized tax risks; and (d) tax infringement notifications. Although the construction companies reported they had adopted CSR principles and ethical and fair behavior, accounting evidence showed that there were unrecognized taxes, unrecognized provisions, high amounts of installments, many tax disputes, tax evasion, and unrecognized foreign revenues. Thus, we show how different the CSR discourse can be from the corporate actions. We agree with Rodrigue et al. [60] that CSR in these companies only can be symbolic since substantive practices would bring changes in organizations, most notably in terms of tax payment.
Our analysis reveals that, before 2014, when the CSR discourse emphasized the "government" attribute, tax management was very risky. After 2014, when CSR discourse came to highlight "ethical behavior" as the most important attribute, tax management showed signs of falling into a more moderate profile.
None of the five companies wrote in their reports the importance of collecting tax promptly to preserve or improve social welfare. They do not see tax payment as an important action for the benefit of society yet, as ABNT CSR guidelines or GRI 207 on tax suggest. Companies' practices of questioning, delaying, not paying, or paying taxes in installments reveal that they do not adequately understand the social function of tax.
Our study showed that the analyzed companies acted in a similar way. Venard and Hanafi [56] also concluded that a company's decision to adopt corrupt behavior could therefore be linked to its competitor's behavior. Thus, we agree with Cho et al. [8] that, when considering the broader societal context, an individual organization may have little choice but to engage in organized hypocrisy.
Our study also found that four companies had followed GRI guidelines before 2014; however, all of them had practices incompatible with the social responsibility perspective. Thus, the GRI standard on tax is necessary to make companies aware of the need to pay taxes on time.
We believe that the analyzed companies have learned a good lesson because a new style of management seems to be developing. Our study revealed that more tax provisions have been made, part of the taxes paid, financial resources that were abroad (off-balance sheet) were recorded in Brazil (repatriated revenues), and fines have been paid; these attitudes seem to show that companies wish to improve their compliance with legal and social issues and fulfill their commitments to government agencies.
Our paper makes some contributions to the literature by focusing on tax management and studying the interaction between CSR and taxes. We analyze for the first time the incoherence between Brazilian construction companies' speeches and their practice. These companies are privately held. Studies using these companies are very rare in Brazil. Thus, our study has theoretical and practical implications. In terms of theoretical implications, our study reveals that CSR reports or discourses do not prevent companies from behaving inappropriately in terms of social welfare. This reinforces the idea that CSR disclosures should go behind the boilerplate disclosures they are providing to avoid organizational hypocrisy. In terms of practical implications, following ISO and GRI standards, CSR regulators should emphasize in their standards the role of tax in social welfare. Additionally, accountants and managers need to reinforce ethics education and training to avoid thinking that tax evasion is an acceptable practice.
As with any other research, this study has limitations. Nevertheless, these do not undermine the validity of the results. The main limitations are related to the reduced number of companies that are analyzed and the unavailability of documents for analysis. Future studies could go on this line of research by analyzing the CSR discourse of companies that are having unethical behavior. Further research should also investigate the individual and institutional barriers that discourage unethical behavior in tax management in different international contexts. This study could also be extended to other industries in Brazil or other countries. Another future line of research could be to explore how tax avoidance actions affect companies' market value in short and long term. Funding: This paper was financed by National Funds of the FCT-Portuguese Foundation for Science and Technology within the project "UIDB/03182/2020".