emerging markets review - fucape business school · determinants of asymmetric loss recognition...

Emerging Markets Review xxx (2017) xxx–xxx

EMEMAR-00494; No of Pages 15

Contents lists available at ScienceDirect

Emerging Markets Review

j ourna l homepage: www.e lsev ie r .com/ locate /emr

Determinants of asymmetric loss recognition timeliness in public and privatefirms in Brazil

Antonio Carlos Coelho a,1, Fernando Caio Galdi b,⁎,2, Alexsandro Broedel Lopes c

a Universidade Federal do Ceará, Avenida da Universidade, 2431, Benfica, Fortaleza, CE 60020-180, Brazilb Fucape Business School, Av. Fernando Ferrari, 1358, Vitória, ES 29075-505, Brazilc Universidade de São Paulo, Av. Professor Luciano Gualberto, 908, FEA 3, São Paulo, SP 05508-900, Brazil

a r t i c l e i n f o

⁎ Corresponding author.E-mail addresses: [email protected] (A.C. Coel

1 Antonio acknowledges financial support from CNPq2 Fernando acknowledges financial support from FAP

http://dx.doi.org/10.1016/j.ememar.2017.02.0021566-0141/© 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Coelho, A.C., et al.Brazil..., Emerg. Mark. Rev. (2017), http://d

a b s t r a c t

Article history:Received 8 January 2016Received in revised form 27 January 2017Accepted 17 February 2017Available online xxxx

We investigate the determinants of asymmetric loss recognition timeliness (ALRT) for publicand private firms in Brazil. We complement Ball and Shivakumar (2005) by investigatingALRT in Brazil where, unlike in the UK setting, equity markets do not provide the adequate in-centives for high quality financial reporting. In Brazil, even listed public companies do not facethe same institutional environment of the British firms. Using a unique database of Brazilianpublic and private firms we document that Brazilian public and private firms present similarALRT, different from the evidence that Ball and Shivakumar (2005) reported for the UK.

© 2017 Elsevier B.V. All rights reserved.

Keywords:Earnings qualityLoss recognitionPrivate firmsEmerging marketsBrazil

JEL classification:M41K22N26

1. Introduction

Prior research (Ball and Shivakumar, 2005; Burgstahler et al., 2006) indicates that public firms have higher earnings qualitythan private firms due to the lower market demand for high-quality earnings for private firms. In this paper, we consider thatincentives of managers and auditors play a central role to determine the quality of financial statements and we investigate asym-metric loss recognition timeliness in a situation different from previous research. In our setting preparers and auditors of both pri-vate and public firms face similar incentives, thus we expect no difference in earnings quality from private and public companies.

According to Ball et al. (2003), we consider the hypothesis that reporting quality is determined by the underlying economicand political factors influencing managers' and auditors' incentives and is the main driver of public and private companies' earn-ings quality. Specifically, we investigate the determinants of asymmetric loss recognition timeliness (ALRT) for public and privateBrazilian firms conditional on their reporting incentives prior to the adoption of IFRS because we want to reduce the potential ofambiguous effects in earnings management due to convergence to IFRS as evidenced by Zhang et al. (2013) and Pelucio-Greccoet al. (2014).

ho), [email protected] (F.C. Galdi), [email protected] (A.B. Lopes)..ES.

, Determinants of asymmetric loss recognition timeliness in public and private firms inx.doi.org/10.1016/j.ememar.2017.02.002

http://dx.doi.org/10.1016/j.ememar.2017.02.002

mailto:[email protected]




Imprint logo

http://www.sciencedirect.com/science/journal/15660141

www.elsevier.com/locate/emr


2 A.C. Coelho et al. / Emerging Markets Review xxx (2017) xxx–xxx

Ball and Shivakumar (2005) argue that asymmetric loss recognition timeliness make financial statements more useful in cor-porate governance and lending agreements. The governance effect of asymmetric loss recognition timeliness is that managershave asymmetric incentives to report good and bad news. Without conservatism managers can defer loss recognition to futureperiods passing to subsequent managers the results of their improper investment decisions. Timely loss recognition creates incen-tives for managers to cut economic losses quickly. Conservatism is also useful in loan agreements by making financial statementsreflect quickly bad news and allowing creditors to impose contractual restrictions arising from covenants. Conservatism transfersdecision right from managers to creditors more quickly. In Brazil, these functions of conservatism are of diminished importancebecause shareholders can control managers' actions directly and loan agreements based on accounting numbers are very rare re-ducing the demand for timely recognition of economic losses. Brito and Martins (2013) argue that the Brazilian institutional en-vironment of weak legal protection and the low demand for high-quality earnings restrict the potential benefits generated byconservatism.

We believe that by addressing this problem we extend prior research by shedding additional light of the determinants of theactual properties of accounting reports. Brazil provides a unique opportunity to extend Ball and Shivakumar (2005) work becauselike the British firms Brazilian private and public firms face the same accounting standards. However, unlike the British firms,Brazilian public and private firms also face similar incentives to provide conservative accounting reports. Thus we expect theALRT in Brazilian public and private firms to be the similar and lower than the level of ALRT presented by British firms showingthe importance of contractual incentives to shape the actual properties of financial statements.

To implement our analysis we use a unique database with nearly 60,000 firm-years observations being 2.4% public firmsand 97.6% private firms over the 1995–2006 period. Our results show that there is no significant difference in ALRT forBrazilian public and private firms unlike Ball and Shivakumar (2005) reported for the UK. We also show that ALRT forBrazilian firms is not significantly affected by firm's issuing activity in capital markets. Additionally we detected strong sig-nals of opportunistic behavior for both samples of firms.

Our paper contribute to literature by shedding light on the determinants of the quality of accounting reports for publicand private firms immersed in a low quality accounting environment. We contribute to a strand of the literature (Ballet al., 2000a,b, 2003; Ball and Shivakumar, 2005; Bushman and Piotroski, 2006; Ali and Hwang, 2000) by providing addition-al evidence that accounting quality as measured by asymmetric loss recognition timeliness is not solely determined by thetype of firm, accounting standards, regulations, taxes or auditing and listing requirements (Brazilian public and private firmsface the same environment regarding these factors) but is mainly influenced by the demand and incentives for high qualityreports.

In this sense we extend and complement Ball and Shivakumar (2005) by investigating ALRT for private and public firmsin Brazil where, unlike in the UK, equity markets do not provide the adequate incentives for high quality financial reporting.In Brazil even the public companies listed on the São Paulo Stock Exchange (BOVESPA) do not face the same incentives toreport high quality earnings as the British public firms do. Most of them do not rely on capital markets to raise funds (ratherthey use special private channels with private and state owned banks) and the ownership concentration is large.3 Brazilcomplies negatively with all the five criteria pointed by Ali and Hwang (2000) as being related to the informativeness of ac-counting reports: (i) code law origin, (ii) influence of tax on accounting, (iii) government standard-setting, (iv) credit-basedinsider model of financing and (v) small amounts spend on auditing. Third, there is pervasive evidence that investor protec-tion in Brazil is very low (Anderson, 1999; Studart, 2000) and the law enforcement in the country is very low – according toDurnev and Kim (2005) only Colombia ranks below Brazil in terms of law enforcement. These factors are likely to drivedown the quality of accounting reports for public (and private) firms. In these circumstances, equity investors anddebtholders do not base their decisions primary on publicly available financial reports, because they have private accessto other sources of information.

Finally, we consider this study is also relevant due to the lack of evidence about private firms' earnings quality in emergingeconomies and because of the importance of private companies in the real economy. Private companies (i.e. firms that do nothave public traded securities) are predominant in economies around the world. The 2016 Forbes's list of largest private companiesin US4 includes 223 firms with revenue greater than $2 billion. Ball and Shivakumar (2005) comment that over 90% of UK regis-tered companies are private. In emerging markets the prominence of private companies is even more exacerbated. According toS&P Capital IQ,5 private companies have higher aggregate market value than public companies in Brazil. However, there is fewevidence on the quality of private companies' financial reporting. According to Chen et al. (2011) accounting research on privatefirms from emerging markets is practically nonexistent.

The rest of the paper is organized as follows. Section 2 presents the background, the Brazilian institutional environment forpublic and private firms, and our main hypothesis. Section 3 presents the data. Section 4 discusses the methodology and results.Finally, Section 5 concludes the paper.

3 Brazilian equities market has recently improved regarding size, volume and sophistication. From 2004 to June 2008, 102 firms listed at Bovespa (São Paulo StockExchange). From all this listings, 78 were voluntarily made on the highest corporate governance level, called “Novo Mercado”. At this corporate governance level,the main feature is that listed firms can only issue common stocks (what has been historically minority in Brazil). However, our sample comprises the period from1995–2006, where most firms have a controlling shareholder (controlling group) and financed themselves primary by using private debt.

4 Available at: http://www.forbes.com/sites/andreamurphy/2016/07/20/americas-largest-private-companies-2016/#7d795fee14175 http://exame.abril.com.br/mercados/empresas-de-capital-fechado-valem-mais-do-que-as-listadas/.

Please cite this article as: Coelho, A.C., et al., Determinants of asymmetric loss recognition timeliness in public and private firms inBrazil..., Emerg. Mark. Rev. (2017), http://dx.doi.org/10.1016/j.ememar.2017.02.002

http://www.forbes.com/sites/andreamurphy/2016/07/20/americas-largest-private-companies-2016/#7d795fee1417

http://exame.abril.com.br/mercados/empresas-de-capital-fechado-valem-mais-do-que-as-listadas/


3A.C. Coelho et al. / Emerging Markets Review xxx (2017) xxx–xxx

2. Background and hypotheses development

Young et al. (2008) recognize that there is not a unique agency model that describes all corporate governance designs aroundthe globe. The authors argue that in emerging economies the institutional context enhance the costs and problems to enforceagency contracts, resulting in the prevalence of concentrated ownership as a response to the principal-principal conflict (insteadof the traditional principal-agent conflicts in developed markets).

Corporate governance literature (La Porta et al., 1999; Bebchuk and Hamdani, 2009) suggests that the presence of controlling share-holders changes the fundamental governance problem. According to Bebchuk and Weisbach (2010) the presence of a controlling share-holder, shifts the fundamental governance problem from opportunism by executives and directors at the expense of public shareholdersto opportunism by the controlling shareholder at the expense of the minority shareholders. This is the typical case of Brazilian publiccompanies where the controlling shareholder (controlling group) hold a mean (median) 76.1% (79.5%) of the voting shares(Okimura et al., 2007). This exacerbated concentration probably exerts some (negative) impact on the properties of accounting numbersfor Brazilian public firms.

The demand for high quality in public companies' financial statements is less likely to exist for public firms in emerging mar-kets with developing capital markets and concentrated ownership (e.g. Brazil). Parallel, and consistent to the argument of Ball andShivakumar (2005, p.84) that private companies are more likely to resort to the “insider access” model to resolve informationasymmetry, we argue that in developing markets the incentives faced by public and private firms to recognize losses on a timelybasis do not differ significantly.

It is an established fact that institutional factors may impact the properties of accounting earnings. Previous literature (Ballet al., 2000a, 2000b; Hung, 2000; Ball et al., 2003) shows evidence that the demand and usefulness of accounting numbers arerelated to general characteristics of each country (e.g., Code-law or Common-law). Supplemental studies (Ball and Shivakumar,2005; Ball et al., 2008 and Pae et al., 2005) present evidences that asymmetric loss recognition timeliness (ALRT – hereafter) isrelated to specific features regarding the firm, the level of activities of capital markets and demand for public information withineach country (region). In a comprehensive survey about earnings quality Dechow et al. (2010, p. 364) suggest that timely lossrecognition has an endogenous component related to firms' reporting incentives, primarily equity incentives. In the same direc-tion, Gopalan and Jayaraman (2012) show that insider-controlled firms are associated with more earnings management thannoninsider-controlled firms in weak investor protection countries.

Ball and Shivakumar (2005) found that ALRT is substantially less prevalent in private than in public companies in the UK. Theyinterpret this result as a direct consequence of the lower demand for high quality information in the environment in which pri-vate firms are immersed. However their results impound economic and institutional features that may not prevail in other coun-tries, such as the level of UK capital market development and the dispersed ownership structure of most public British firms.

Thus, we conjecture our hypothesis for this study as:

H0. Brazilian public and private firms exhibit the same level of timely recognition of economic losses because they face the sameaccounting, auditing and tax rules and regulations, like the British firms, but their preparers also face similar incentives to producelow quality financial reports – unlike the British.

In this sense we consider that reporting quality is ultimately determined by the underlying economic and political factorsinfluencing managers' and auditors' incentives and is the main driver of public and private earnings quality in Brazil.

To better understand the environment in Brazil, the next section discuss in detail the accounting, auditing and tax norms forBrazilian private and public firms and the market forces shaping corporate financial reporting in Brazil.

2.1. Accounting, auditing and tax rules for Brazilian public and private firms

In Brazil most commercial and industrial firms are incorporated under two forms: (i) sociedades limitadas (CL) and (ii)sociedades por ações (SA), both of them characterized as limited liability companies. CLs are always registered as private firms,while SAs may assume the form of public company or private company (depending whether or not they offer securities to themarket). CLs are not required to publish financial statements or any other form of public financial information irrespective oftheir size. The other group (SA) can be subdivided on companhias abertas (SAca) – the group of public companies – andcompanhias fechadas (SAcf) – the group of private companies. SAs on both groups are regulated by the Brazilian company law(Lei 6404/1976) and have to publish financial reports according to Brazilian GAAP. However, only SAca have shares listed onthe São Paulo Stock Exchange (BOVESPA). Additionally, the Brazilian Revenue Service (RFB) does not discriminate between thetwo groups of SAs who face the same set of tax rules.

Brazilian SAs present the same setting reported by Ball and Shivakumar (2005) for public and private British firms. SAca arethe ones that can raise money from the public by issuing shares and debt. SAcf are firms that decided to adopt SA status forother reasons rather than to access public equity and credit markets.6 The only significant difference between both types ofSAs is their listing status and access to public equity and debt markets. In the next section we show that despite SAca being listedon the BOVESPA and SAcf are not, both types of SAs end up having the same market for financial reporting environment. TheBrazilian GAAP also applies equally for both sets of firms.

6 One of the possible reasons is to have access to funds from the Brazilian National Development Bank (BNDES).




2.2. The market for corporate financial reporting in Brazil

Our main hypothesis in this paper is that Brazilian private (SAcf) and public companies (SAca) face similar incentives to provide in-formative accounting reports. Based on this assumption we do not expect to find different levels of timely loss recognition betweenthese two groups. We believe that Brazilian private and public firms have no incentives to provide conservative accounting reportsdue to four reasons: (i) poor investor protection and legal enforcement, (ii) insider model of financing, (iii) ownership concentration,(iv) strong links between tax and accounting reports under the so-called Continental European accounting model (Nobes, 1998). Inthis section we present these features in detail.

2.2.1. Poor investor protection and legal enforcement7

There is pervasive evidence that investor protection and legal enforcement in Brazil are at world's lowest levels (Anderson, 1999).Durnev and Kim (2005) show that only Colombia ranks below Brazil in terms of legal enforcement and obedience to the rule of law.According to Black (2001) these features are reflected also in securities market regulation. Recently, Chong and Lopez-de-Silanes (2007)documented that poor investor protection is the rule in Latin America causing capital markets in the region to be anemic and not workto finance firm's activities. Firms in the region have to form alliances with foreign partners or look for finance abroad – mainly by is-suing American Depositary Receipts. Recent data8 show that Brazilian firms are one of the biggest ADR's issuers in the world.

Recent literature on corporate governance (La Porta and Lopes-de-Silanez, 2000; La Porta et al., 2002) has shown the pervasive in-fluence that poor investor protection and legal enforcement have on the development of capital markets. To have access to externalsources of funds firms have to credibly commit not to expropriate investors and to maintain a reasonable flow of information inorder to facilitate monitoring by boards, auditors and regulators. Financial statements play an essential role on this process becausethey help to reduce information asymmetry and are inputs to contracts between managers and shareholders (Sunder, 1997). Account-ing earnings are especially important because they provide a summary measure of firm performance which can be used to evaluatemanagers' behavior, for example. Timely loss recognition of economic losses play a special role in this monitoring process because ithelps to constrain managers opportunistic behavior and to provide more reliable measures of firm performance for other contractslike lending arrangements. However, timely loss recognition must be verified and deviations must be punished in order for conserva-tism to be reliable ex ante. Timely loss recognition depend on managers' discretionary judgment and consequently on their incentives tobehave properly. In Brazil we have a situation where the market does not demand conservative financial statements and managers donot face clear impediments to behave improperly.

2.2.2. Insider model of financingThe economic environment that shaped Brazilian capital markets during the 1990s took place on early 1980s. Institutional problems

were coupled with high levels of inflation9 and macroeconomic instability in the eighties – factors that reduced significantly economicactivity in Brazil and generated what Brazilians call the ‘lost decade’. The eighties in Brazil constitute a text-book case of the so-calledcrowding out phenomenon. To finance a very generous fiscal policy the Brazilian government had to issue growing volumes of publicnotes and bonds paying increasing interest rates. This phenomenon was accompanied by high inflation levels. The Brazilian governmentissued very short-term notes paying floating interest rates to ensure its own funding. This process became so exacerbated that in thelate eighties almost the entire stock of financial assets in the country were allocated to treasury notes. This process continued duringthe first half of 1990s and reduced dramatically the demand for stocks in Brazilian capital markets because investors were not interestedin buying shares once treasury bills were available yielding high returns.

Beginning in 1994, deep macroeconomic reforms took place in Brazil. Initially a currency stabilization plan (Plano Real) wasimplemented. However, according to Nassif (2007, p.22) despite the fact that the successful stabilization program had reducedannual inflation rates from 916% to 1.65%, the combination of high real interest rates with significant real overvaluation of theBrazilian currency against the principal trade partners' currencies produced increasing twin deficits. In 1999 Brazil adopted a float-ing exchange rate regime, but still maintained very high interest rates, continuing to attract investors to government bonds.

A natural consequence of the crowding out process implemented in Brazil was that firms could not count on capital markets to fi-nance their growth opportunities. The Brazilian Bank for National Development (BNDES – a state controlled bank) was responsible foralmost all long term finance available in Brazil. The private banks provided almost no funds for firms because their assets were allocatedto public notes. Most of the trading volume on the São Paulo Stock Exchange during this period was made of shares of state-controlledenterprises. Under these macroeconomic conditions firms in Brazil could not rely on public capital markets to finance their develop-ment. Finance was obtained through private channels and government connections.

Chong and Lopez-de-Silanes (2007) provide a detailed discussion of this scenario for Latin America as whole and found similarpatterns. Macroeconomic instability and poor investor protection and legal enforcement caused investors to be very unlikely toprovide long term financing for firms in the region. For capital markets to flourish, investors must be confident that they arenot going to be expropriated or have their returns reduced by high inflation levels. In an inimical environment like theBrazilian during this period investors and firms reached their agreements through private deals. These deals are based on relation-ships and long term influence and not on public financial statements. Corporate financial reporting is not a serious input into

7 This analysis is based on the timeframe covered in this study and do not reflect the recent initiatives undertaken in to improve investor protection and corporategovernance in Brazil.

8 www.adr.com.9 The Brazilian consumer inflation rate reached 99.7% in 1982 (http://www.bcb.gov.br).


http://www.adr.com

http://www.bcb.gov.br



contracts under such circumstances. In this setting there is no demand for informative accounting reports and timely recognitionof economic losses serves no need for public firms.

2.2.3. Ownership concentrationThe corporate governance literature has shown that ownership concentration can function as a response to poor investor protection

(Shleifer and Vishny, 1997). Knowing that they cannot rely on courts to solve conflicts, shareholders decide to act ex ante and to buylarge blocks of shares in order to improve monitoring over managers. This is exactly the case in Brazil where ownership concentration ishigh and during our sample period not a single company had its control floated in the São Paulo Stock Exchange. We believe that thishigh level of ownership concentration is a response to the low investor protection level investors face in the country and a consequenceof the fact that firms in Brazil do not rely on public markets to finance their activities.

The typical agency conflict between managers and shareholders is dominated in Brazil by the conflict between minority and control-ling shareholders. The financial accounting literature has shown (Watts, 2003a, 2003b) that timely loss recognition of losses plays a cen-tral role in mitigating managers' opportunistic behavior which increases earnings usefulness in contracts and other arrangements. InBrazil, the conflict between controlling and minority shareholders changes the importance given to conservatism because most of thedisputes occur because of structured transactions designed to dilute minority participations (and related problems) and not onperformance-related matters on which earnings are important inputs. We conjecture that the high level of ownership concentrationin Brazil reduces dramatically the role that asymmetric recognition of losses has on contracts between managers and shareholders.

This aspect turn the Brazilian firms apart from the British because UK private and public firms have distinct ownership struc-tures – public firms are more dispersed than private ones. In Brazil the level of ownership concentration is similarly high causingshareholders to have special access to information and to not depend on public financial statements. In this situation, there is nodemand for informative accounting reports. Investigating this question, Fan and Wong (2002) exam the relation between infor-mativeness and the ownership structure of firms based on seven East Asian countries. They found a negative relation betweenownership concentration and timeliness which is consistent with two alternative explanations. First, concentrated ownership cre-ate agency problems between controlling shareholders and outside investors causing controlling shareholders to report earningsfor self-interested purposes. Second, concentrated ownership may prevent leakage of proprietary information and earnings con-sequently lose information content.

Additionally Ahmed and Duellman (2007) find evidences that the percentage of outside directors' shareholdings is positivelyrelated to conservatism, while the percentage of inside directors is negatively related to conservatism. LaFond and Watts (2008)document that firms characterized by greater information asymmetry present greater timeliness in loss recognition of bad newswhen compared to good news.

Private and public companies in Brazil operate in a similar environment regarding agency conflict between controlling shareholdersand management. Large shareholder concentration in public firms is a key-feature of Brazilian capital markets. In a recent research,KMPG (2010) documented only 13.4% of independent board members for the 50 most liquid companies traded on the traditional seg-ment at BM&FBovespa. We can conjecture that in Brazil the average public and private firms present similar corporate governanceschemes. In such scenario, we posit that information asymmetry between controlling shareholders and managers are similar in privateand public firms in Brazil. We believe that the situation in Brazil is an interesting setting where we can test ALRT between private andpublic firms facing similar information asymmetry features between controlling shareholders and management. Our paper contributesto the findings and arguments of Fan and Wong (2002) and LaFond and Watts (2008).

2.2.4. Strong links between tax and accounting reports – Continental European modelAccording to the Nobes (1998) classification, Brazil can be classified as a Class B country. It's accounting and legal systems are based

on its Iberian colonizers – code law system. All accounting norms for public firms are stated on the Company Law and professional ac-counting bodies have no de facto power to influence accounting standards.10 We observe a strong link between tax and financial reportsin the country – financial statements are prepared following the Company Law but adherence to tax regulations is the rule not the ex-ception. Additionally, dividends must be paid at a minimum fraction (25%) of corporate profits according to the Company Law for bothSAcf (private firms) and SAca (public firms). In Brazil financial reporting is strongly influenced by taxation and dividend policies. Ac-cording to Ball and Shivakumar (2005) these influences are likely to cause a lower demand for earnings which reflect underlying eco-nomic performance.

Taken together these factors lead us to hypothesize that private and public firms in Brazil face the same set of market incentives –the market for financial reporting is identical for these two sets of firms. Most of the listed firms were state-controlled and others listedonly to have special access to government funds. Thus our main hypothesis in this paper is that the incentives to incorporate asymmet-ric loss recognition into will be similar between public and private firms in Brazil.

Watts (2003a, 2003b) adopt the contractual explanation for conservatism. According to him conservatism arises as an equilibriumfeature of accounting numbers which are used to form contracts between parties with asymmetric information and conflicts of interest.In the typical principal-agent conflict conservatism arises to mitigate managers' opportunistic behavior – to inflate profits to increasebonus payments, for example. We conjecture that Brazilian private and public firms face the same contractual environments unlikethe British firms reported by Ball and Shivakumar (2005).

10 The Brazilian Company Lawwas reformed in the end of 2007 and from 2008 on Brazilian accounting standards started to be settled by an independent body (seek-ing convergence to IFRS). However, the standards have to comply with the requisites of the Company Law.




3. Data

3.1. Sample selection

Our database is composed by public and private Brazilian corporations that released financial statements from 1995 to 2006. Weconsider a period of analysis prior to IFRS adoption in Brazil because we want to reduce the potential of ambiguous effects in earningsmanagement due to convergence to IFRS. Recent studies about the adoption of international financial reporting standards (IFRS) indicatethat this event has impact on earnings management conditional on other institutional factors. Zhang et al. (2013) find that Chineselisted companies that experienced mandatory adoption of IFRS significantly increased earnings management. By the other hand,Pelucio-Grecco et al. (2014) find that the adoption of IFRS by Brazilian firms had a restrictive effect on earnings management.

First, we collected data from all private firms registered on SERASA, which is the largest Latin American financial database onindividuals and corporations. SERASA plays an active role in most credit and business-related decisions made in Brazil, respondingto 4 million inquiries per day made by over 400 thousand direct or indirect customers.11 Secondly, we use ECONOMATICA data-base to collect data from public firms. Then we merge the data from SERASA and ECONOMATICA and our sample results in 5364firms, from which 5000 are private corporations and 364 are public-held corporations. We lose one year of observation becausewe use changes in income and other variables to calculate operating cash flows. Thus, our final sample contains 5364 firms during11 years. It is also important to note that according to Brazilian corporate law, all financial statements from corporations (“SAs”)must present auditor's report, what enforces the application of GAAP to Brazilian corporations.

3.2. Descriptive statistics

We provide detailed income statement (Table 1) and balance sheet (Table 2) descriptive statistics for public and private firms.Panels A, B and C present statistics for public, private and all sample of firms, respectively, with t-statistics for a two tailed test ofdifferences in means between private and public. As stated by Ball and Shivakumar (2005) the t-statistics must be interpretedwith caution, since its control for neither cross-sectional nor serial correlation.

3.2.1. Income statement itemsTable 1 presents descriptive statistics for the income statement variables. The mean (median) annual sales for public-held

companies are R$1299 million (R$89 million), compared to R$22 million (R$3.6 million) for private firms. This difference is sta-tistically significant at 1% and represents the difference in size for firms pertaining to each group. However sales scaled by totalassets at beginning of year are not significantly different from public-held to private firms in Brazil. The mean (median) net returnon assets averages −63% (2%) for publicly listed firms and 19% (5%) for private firms. Other profitability measures also point outto positive and statistically significant differences on accounting returns between private and public-held companies. These fea-tures can be compared to the presented by Ball and Shivakumar (2005) on Table 1 where private firms also present higherreturn-metrics than public-listed firms. The analysis of the payout ratio shows different behaviors when comparing private andpublic firms. Apparently, private firms are more aggressive in paying dividends. A possible explanation for higher dividends in pri-vate firms is that many of them are subsidiaries from international firms and payback large dividends to the parent company.

3.2.2. Balance sheet itemsTable 2 presents summary statistics for balance sheet items. Opposite to income statements variables, substantial differences

between private and public companies are evident. The average (median) public-held firm has total assets of R$2267 million12

(R$147 million), while the average (median) private firm has total assets of R$25 million (R$ 2.1 million). Equivalently mean (me-dian) equity from public-held firms respond to R$927 million (R$47 million), while the average (median) private firm has equityof R$13 million (R$0.9 million).

Consistent to the idea that public-held companies make more intense use of financial markets when compared to privatefirms, leverage statistics on Table 2 (scaled total debt, scaled net debt and scaled long term debt) indicate that the public-heldfirms' group presents significantly higher mean (median) debt variables.

4. Methodology and results

4.1. Methodology

Our basic model considers the transitory nature of economic income and is based on Basu (1997) and Ball and Shivakumar(2005). Basically the model considers that increases and decreases in accounting income tend to revert, what indicates the tran-sitory components of earnings. To measure the difference in income decrease (increase), the basic model (1) is applied consid-ered:

11 ww12 The

PleasBraz

ΔIt ¼ α0 þ α1DΔIt−1 þα2ΔIt−1 þ α3DΔIt−1 � ΔIt−1 þ εt ð1Þ

w.serasa.com.br.average exchange rate between the Brazilian Real and the US dollar in the first quarter of 2010 was 1.8BRL:1USD.

e cite this article as: Coelho, A.C., et al., Determinants of asymmetric loss recognition timeliness in public and private firms inil..., Emerg. Mark. Rev. (2017), http://dx.doi.org/10.1016/j.ememar.2017.02.002

http://www.serasa.com.br


Table 1Descriptive statistics - income statement items.

Sales(R$ mn)

Sales/Assets TCosts/Sales Deprec/LTAssets

NonOI/Assets NROA OROA ROABT Accruals/Assets OCF/Assets Payoutratio

Panel A -public-heldcorporationsNumber 4368 4084 2853 2162 4061 2723 4041 4041 4019 4014 2799Mean 1299.18 0.87 9.59 0.09 0.00 −0.63 −0.37 −0.37 −0.31 0.21 0.18Median 89.16 0.47 0.98 0.05 0.00 0.02 0.00 0.00 0.00 0.01 0.18Standarddeviation

6322.57 12.27 260.93 1.17 0.37 25.71 21.32 21.41 12.60 12.97 3.31

Skewness 20.02 63.42 36.07 45.22 −53.67 −50.74 −59.39 −58.97 −52.40 45.58 −20.72Panel B -privatecorporationsNumber 57,540 52,856 56,784 32,952 54,708 52,664 52,779 52,667 52,707 52,707 14,188Mean 21.58 8.36 1.24 0.07 0.00 0.19 0.23 0.23 1.45 −1.26 0.92Median 3.62 2.06 0.97 0.05 0.00 0.05 0.06 0.07 0.00 0.05 0.53Standarddeviation

133.53 1042.18 43.67 0.09 2.20 11.86 14.99 13.21 279.66 273.17 13.74

Skewness 34.99 223.63 80.06 28.92 −206.91 132.55 136.12 131.17 223.66 −225.26 −8.82t-stata 13.35⁎⁎⁎ −0.46 1.71⁎ 0.97 0.04 −1.66⁎ −2.38⁎⁎ −2.63⁎⁎⁎ −0.40 0.34 −5.68⁎⁎⁎

Panel C - totalsampleNumber 61,908 56,940 59,637 35,114 58,769 55,387 56,820 56,708 56,726 56,721 16,987Mean 111.72 7.82 1.64 0.07 0.00 0.15 0.19 0.19 1.32 −1.15 0.80Median 3.80 1.92 0.97 0.05 0.00 0.05 0.05 0.06 0.00 0.05 0.45Standarddeviation

1715.66 1004.12 71.24 0.30 2.12 12.89 15.53 13.96 269.59 263.35 12.63

Skewness 72.67 232.10 102.47 161.97 −213.80 78.27 102.84 88.18 231.98 −233.62 −9.51

Variable definitions: Sales, total annual sales revenue in R$ millions; Sales / Assets, Sales divided by total assets at beginning of year; TCosts / Sales, calculated as(sales − operating income) / sales; Deprec / Fixed Assets, depreciation and amortization expense divided by long-term assets (fixed assets + deferred assets) atend of year; NonOI / Assets, non-operating income scaled by beginning total assets; NROA, Net return on assets calculated as net income scaled by total assets atbeginning of year; OROA, Operating return on assets calculated as operating income scaled by total assets at beginning of year; ROABT, Return on assets beforetaxes calculated as net profit before income taxes scaled by total assets at beginning of year; Accruals / Assets, accruals scaled by beginning total assets, whereaccruals are change in inventory + change in debtors + change in other current assets − change in creditors − change in other current liabilities − depreciation;OCF / assets, operating cash flow scaled by beginning total assets, where operating cash flow is calculated as operating income + depreciation andamortization + direct taxes − change in working capital − non operating income; Payout ratio is defined as dividend scaled by net income.

a t-statistic for two-tailed test of difference between private and public firmmeans.⁎ Represents significance at 10%.⁎⁎ Represents significance at 5%.⁎⁎⁎ Represents significance at 1%.


where ΔIt represents changes in net income at year t scaled by total assets at the end of the year t − 1, DΔIt − 1 equals 1 ifprevious year's changes in net income is negative or 0 otherwise, ΔIt − 1 represents changes on net income at year t − 1,DΔIt − 1 ∗ ΔIt − 1 is the interaction variable of negative changes of net income at year t − 1.

To consider differences between private and public companies, our empirical analyses is derivate from (1) and we use pooledordinary least squares (POLS) to enable us to estimate the models considering a variety of dummies variables that changes acrossspecifications. Ball and Shivakumar (2005) recognize there is a potential endogeneity problem in this specification. If firms selectlisting status based on expected changes in earnings, the dummy variable DPR is endogenous. However, the authors argue(p. 117) that this problem “is highly unlikely in the income changes regression, because any relation between choice of listingstatus and income is likely to involve the average level of income in the years following the listing decision, not the change ina single future year”.

To select between OLS and random effects regression we run the Breusch-Pagan Lagrange multiplier test (LM). The null hy-pothesis in the LM test is that variances across entities is zero. Results show that we cannot reject the null for the model(2) and we can reject the null for the model (3). Hence, for the specifications presented on Table 4 we also run random effectsregressions, but our main results do not change. For the sake of simplicity, we tabulate and present our OLS results.

4.2. Timeliness in loss recognition between public and private firms

To measure earnings ALRT in Brazilian public and private firms we use the approach proposed by Ball and Shivakumar (2005).Initially we apply the model to evaluate transitory gain or loss components on earnings among public and private firms as



Table 2Descriptive statistics - balance sheet items.

Total assets (R$mn)

Equity (R$mn)

Scaled totaldebt (%)

Scaled netdebt (%)

Scaled LTdebt (%)

Bonds/TotalDebt

Deferred/Assets FixedAssets/Assets

Panel A - public-heldcorporationsNumber 4368 4368 3081 3081 3081 2872 3081 3081Mean 2267.04 927.43 0.43 0.34 0.16 0.09 0.02 0.60Median 146.88 46.67 0.24 0.17 0.10 0.00 0.00 0.61Standard deviation 10,539.44 4771.79 2.00 2.02 0.86 0.20 0.07 0.59Skewness 11.50 12.51 18.17 17.95 51.05 2.81 4.99 3.78

Panel B - privatecorporationsNumber 57,540 57,540 57,540 57,540 57,540 na 57,540 57,540Mean 25.00 13.26 0.28 0.16 0.21 na 0.01 0.55Median 2.08 0.88 0.11 0.04 0.03 na 0.00 0.42Standard deviation 395.79 252.67 4.90 4.91 4.90 na 0.04 0.70Skewness 44.98 46.36 107.49 107.18 107.75 na 11.45 26.72t-stata 14.06⁎⁎⁎ 12.66⁎⁎⁎ 3.61⁎⁎⁎ 4.169⁎⁎⁎ 1.77⁎⁎ na 13.82⁎⁎⁎ 4.51⁎⁎⁎

Panel C - total sampleNumber 61,908 61,908 60,621 60,621 60,621 2872 60,621 60,621Mean 183.19 77.76 0.28 0.17 0.20 0.09 0.01 0.55Median 2.23 0.92 0.11 0.04 0.03 0.00 0.00 0.42Standard deviation 2882.87 1311.63 4.80 4.81 4.78 0.20 0.04 0.69Skewness 41.80 44.62 108.86 108.54 110.24 2.81 10.47 26.01

Variable definitions: Total Assets, total assets at end of year in R$ millions; Equity, equity at end of year in R$ millions; Scaled total debt, total debt divided by totalassets at end of year; Scaled net debt, net debt divided by total assets at end of year; Scaled LT debt, long-term debt divided by total assets at end of year; De-ferred/Assets, deferred assets divided by total assets at end of year; Fixed Assets/Assets, fixed assets at the end of the year divided by total assets at the end ofthe year.

a t-statistic for two-tailed test of difference between private and public firmmeans.⁎⁎ Represents significance at 5%.⁎⁎⁎ Represents significance at 1%.


follows:

PleasBraz

ΔIt ¼ α0 þα1DΔIt−1 þα2ΔIt−1 þα3DΔIt−1 � ΔIt−1 þα4DPRþα5DPR � DΔIt−1þα6DPR � ΔIt−1 þ α7DPR � DΔIt−1 � ΔIt−1 þ εt

ð2Þ

where ΔIt represents changes in net income at year t scaled by total assets at the end of the year t − 1, DΔIt − 1 equals 1 if pre-vious year's changes in net income is negative or 0 otherwise, ΔIt − 1 represents changes on net income at year t − 1, DΔIt −

1 ∗ ΔIt − 1 is the interaction variable of negative changes of net income at year t − 1, DPR equals 1 for private firms or 0 for publicfirms, DPR ∗ DΔIt − 1 is the interaction variable for private firms with negative changes of income at year t − 1, DPR ∗ ΔIt − 1 rep-resents the interaction variable for private firms with changes of net income at year t − 1 and DPR ∗ DΔIt − 1 ∗ ΔIt − 1 representsthe interaction variable between private firms with negative changes of net income at year t − 1. These definitions are used inour first regression, presented on Table 3, column 1. Alternative specifications are used (Table 3, columns 2 to 9), where DPR def-inition varies. Additionally, we investigate the sample of public firms in more detail. We investigate three additional features ofpublic firms and their impact on ALRT: (i) firms' activity in public debt markets, (ii) degree of leverage using public debt and(iii) presence of covenants. In this sense we also investigate whether firms financing activities with public debt act as an incentiveregarding ALRT.

DPR assumes different meanings across our regressions. On Table 3, column 2 DPR equals 1 for firms with no public debt and 0otherwise. On Table 3, columns 3, 4, 5 and 6 DPR is not defined considering our regressions specifications on subsamples. OnTable 3, column 7 DPR equals 1 for firms without covenants and 0 otherwise. Finally on Table 3, columns 8 and 9 DPR is definedas firms' leverage ratio.

Eq. (2) adopts a time-series test of timeliness in loss recognition. That approach allows one to analyze if lagged income mea-sures convey incremental explanatory power to current changes on income, by negative (positive) income and by private (public)firms. Ball and Shivakumar (2005) hypothesis suggest less reversal of income decreases in private companies than in public com-panies, reflecting a lower frequency of timely loss recognition due to lower demand for financial reporting quality. Table 3, col-umn 1 presents the results of regression (2) considering a pooled OLS model.

We focus on the coefficients α6 and α7 because we expect Brazilian public and private firms to exhibit earnings of thesame quality and consequently we do not expect these coefficients to be significant. Table 3, column 1 shows that besidesthe significantly reversion of public firms' results when there are negative changes on profits (α3 = −0.27), one cannot dif-fer public and private firms behavior considering that α7 is not statistically significant. This result in coherent with our hy-pothesis and shows no significant difference between public and private companies in Brazil unlike Ball and Shivakumar



Table 3Pooled OLS regression of change in earnings on lagged change in earnings for all firm-years. Dependent variable measured as change in net income. DPR variable variesacross columns.

ΔIt = α0 + α1DΔIt − 1 + α2ΔIt − 1 + α3DΔIt − 1 ∗ ΔIt − 1 + α4DPR + α5DPR ∗ DΔIt − 1 + α6DPR ∗ ΔIt − 1 + α7DPR ∗ DΔIt − 1 ∗ ΔIt − 1 + εt

1 2 3 4 5 6 7 8 9All firms(DPR = 1for privatefirms; 0for publicfirms)

Public firms(DPR = 1for firmswith nopublicdebt; 0otherwise)

Mostleveragedfirms(publicdebt/totaldebtaverage)

Lessleveragedfirms(publicdebt/totaldebtaverage)

Mostleveragedfirms(publicdebt/equityaverage)

Lessleveragedfirms(publicdebt/equityaverage)

Firms withpublic debt(DPR = 1for firmswithoutcovenants;0otherwise)

Moreleveraged(publicdebt/totaldebt) thanthe average(DPR is theleverageratio)

Moreleveraged(publicdebt/equity)than theaverage(DPR is theleverageratio)

Predict Coef. Coef. Coef. Coef. Coef. Coef. Coef. Coef. Coef.

Intercept (α0) ? −0.004 0.0141⁎ −0.008 −0.003 0.000 −0.004 0.012 −0.006 −0.004DΔIt − 1 (α1) ? 0.012 −0.0351⁎⁎⁎ 0.024⁎ 0.006 −0.015 0.0136⁎ −0.039⁎⁎ 0.008 0.012ΔIt − 1 (α2) 0 −0.005 −0.073 −0.002 −0.006 −0.135⁎⁎ 0.003 −0.156⁎ 0.003 −0.003DΔIt − 1 ∗ ΔIt − 1 (α3) − −0.267⁎⁎⁎ −0.735⁎⁎⁎ −0.178 −0.319⁎⁎⁎ −0.487 −0.258⁎⁎ −0.735⁎⁎⁎ −0.372⁎⁎⁎ −0.270⁎⁎⁎

DPR (α4) ? 0.014⁎⁎⁎ −0.026⁎⁎⁎ 0.003 0.045 0.001DPR ∗ DΔIt − 1 (α5) ? −0.010 0.056⁎⁎⁎ 0.006 −0.048 −0.002DPR ∗ ΔIt − 1 (α6) ? −0.103⁎ 0.091*** 0.091 −0.200 −0.023DPR ∗ DΔIt − 1 ∗ ΔIt − 1

(α7)+ −0.019 0.490⁎⁎ 0.042 0.357 0.001

α2 + α3 −0.272 −0.808 −0.179 −0.325 −0.621 −0.256 −0.892 −0.369 −0.273α2 + α6 −0.108 0.019 −0.065 −0.197 −0.026α2 + α3 + α6 + α7 −0.395 −0.226 −0.758 −0.212 −0.295

Adj. R2 0.0777 0.067 0.0333 0.0641 0.1309 0.0473 0.1371 0.0726 0.0515F-Stat. 214.68⁎⁎⁎ 13.95⁎⁎⁎ 3.88⁎⁎⁎ 11.07⁎⁎⁎ 10.41⁎⁎⁎ 10.71⁎⁎⁎ 15.6⁎⁎⁎ 7.15⁎⁎⁎ 6.08⁎⁎⁎

Cross-section obs. 5069 281 163 277 113 303 78 272 281Years 10 10 10 10 10 10 10 10 10No. of obs 49,693 2351 624 1727 420 1931 655 2244 2349

Variables: ΔIt, changes in income at year t scaled by total assets at the end of the year t − 1; DΔIt − 1, equals 1 if previous year's changes in income is negative or 0otherwise; ΔIt − 1, changes on income at year t-1 scaled by total assets at the end of the year t − 2; DΔIt − 1 ∗ ΔIt − 1 is the interaction variable of negative changesof income at year t − 1; DPR, varies across columns; DPR ∗ DΔIt − 1, interaction variable firms with negative changes of income at year t − 1; DPR ∗ ΔIt − 1, inter-action variable firms with changes of net income at year t − 1; DPR ∗ DΔIt − 1 ∗ ΔIt − 1, interaction variable between firms with negative changes of income atyear t-1. We consider net income as defined by Brazilian GAAP. The first column refers to all firms in our sample. The second column considers Brazilian firmswith traded public debt. The third and fourth columns present the results of the regression for the most and the less leveraged (measured by the amount of publicdebt divided by average total debt) firms, respectively. The fifth and sixth columns present the results of the regression for the most and the less leveraged (mea-sured by the amount of public debt divided by average equity) firms, respectively. The seventh column reports results for firms that present covenants on theirdebt. Finally, the eighth and ninth columns present the results considering only the subsample of firms more leveraged than the average firms (measured by publicdebt divided by total debt and public debt by equity, respectively).⁎ represents significance at 10%.⁎⁎ represents significance at 5%.⁎⁎⁎ represents significance at 1%.


(2005) report in the UK. Additionally private firms seem to anticipate economic gains because α2 + α6 is negative, but in-dividually not significant. Table 3, column 2 show the results for public firms considering those without public traded debt(DPR = 1) and those with public traded debt (DPR = 0). The positive coefficient on α7 for public firms without public debtactually indicate that presenting less negative earnings reversion these firms are actually less conservative. However Table 3,column 7 shows that covenants do not cause differences on earnings reversions (α7 is not significant). On other hand thecoefficients α2 and α2 + α6 may indicate earnings reversion for positive earnings changes. The remaining columns ofTable 3 consider the public leverage ratio as a possible aspect to differentiate asymmetric loss recognition by Brazilian publicfirms. Table 3, columns 3 and 5 presents the results for the most public leveraged firms in this segment. In this case, theleverage ratio was calculated as the amount of public debt divided by the total debt in the year. The sample of firms formingthe group of the most leveraged firms comprises firms with leverage ratio above the average for the total sample of publicfirms. Similarly, Table 3, column 4 and 6 present the estimates for firms with leverage ratio below the average. Table 3, col-umns 8 and 9 also consider the leverage ratio as mentioned before, but we consider the sample of the most leverage ratiofirms in our regression. Actually on columns 8 and 9, DPR equals the leverage ratio. The results show that leverage ratio isnot related to transitory gain or loss components on earnings. The coefficient α7 is not statistically significant in any of ourspecifications that consider leverage.

Table 3, columns 4 and 6, which present the regression results for the less leveraged firms show that the interaction variable ofchanges in earnings are significant, what indicate the presence of ALRT (α3 is negative) for these group of firms. This is according



Table 4Pooled OLS regression of accruals on operating cash flows for all firm-years. Dependent variable measured as total accruals. DPR variable varies across columns.

ACt = β0 + β1DOCFt + β2OCFt + β3DOCFt ∗ OCFt + β4DPR + β5 DPR ∗ DOCFt + β6DPR ∗ OCFt + β7 DPR ∗ DOCFt ∗ OCFt + υt

1 2 3 4 5 6 7 8 9All firms(DPR = 1for privatefirms; 0for publicfirms)

Publicfirms(DPR = 1for firmswith nopublicdebt; 0otherwise)

Mostleveragedfirms(publicdebt/totaldebtaverage)

Lessleveragedfirms(publicdebt/totaldebtaverage)

Mostleveragedfirms(publicdebt/equityaverage)

Lessleveragedfirms(publicdebt/equityaverage)

Firms withpublic debt(DPR = 1for firmswithoutcovenants; 0otherwise)

Moreleveraged(publicdebt/totaldebt) than theaverage (DPRis the leverageratio)

Moreleveraged(publicdebt/equity)than theaverage (DPRis theleverageratio)


Intercept (β0) ? 0.046*** 0.058*** 0.053*** 0.041*** 0.029*** 0.046*** 0.008 0.041*** 0.045***DOCFt (β1) ? −0.054*** −0.035* −0.075*** −0.051*** 0.0087496 −0.061*** 0.021 −0.050*** −0.052***OCFt (β2) − −0.769*** −0.806*** −0.900*** −0.698*** −0.593*** −0.784*** −0.459*** −0.688*** −0.769***DOCFt ∗ OCFt (β3) + −0.026 0.133 0.231 −0.223** −0.043 −0.023 0.243 −0.249*** −0.034DPR (β4) ? −0.046*** −0.018 0.054*** 0.019 −0.004DPR ∗ DOCFt (β5) ? 0.088*** −0.018 −0.086** −0.002 0.011DPR ∗ OCFt (β6) ? 0.620*** 0.051 −0.386*** −0.432*** 0.035DPR ∗ DOCFt ∗ OCFt(β7)

− −0.441*** −0.182 −0.510* 0.416** −0.032

β2 + β3 −0.795 −0.673 −0.669 −0.921 −0.636 −0.807 −0.216 −0.936 −0.803β2 + β6 −0.149 −0.756 −0.845 −1.120 −0.734β2 + β3 + β6 + β7 −0.617 −0.805 −1.112 −0.953 −0.800

Adj. R2 0.274 0.582 0.517 0.650 0.453 0.594 0.628 0.725 0.595F-Stat 1323.48*** 84.57*** 36.39*** 253.41*** 77.71*** 149.74*** 47.13*** 239.47*** 85.18***Cross-section obs. 5104 311 167 279 118 304 85 311 311Years 11 11 11 11 11 11 11 11 11No. of obs 55,096 2700 721 1979 474 2226 770 2555 2696

Variables: ACt, accruals at year t scaled by total assets at the end of the year t − 1, estimated as follows: ACt = (ΔCAt − ΔCLt − Δcasht + Δdebtt − Deprt) / TotalAssetst − 1. where ACt represents accruals at year t, ΔCAt measures the change in current assets at year t, ΔCLt represents changes in current liabilities at year t,Δcasht represents changes in cash or cash equivalents at year t, Δdebtt computes the changes in short term debt at year t, Deprt represents the depreciationand amortization at year t; DOCFt equals 1 if operating cash flow at year t is negative or 0 otherwise; OCFt, operating cash flow at year t scaled by total assetsat the end of the year t − 1, estimated as OCFt = (OIt + Deprt − ACt) / Total Assetst − 1, where OIt represents the operating income at year t, and ACt is accrualsat year t; DOCFt ∗ OCFt, interaction variable for negative operating cash flows at year t; DPR, varies across columns; DPR ∗ DOCFt, interaction with firms with neg-ative operating cash flows; DPR ∗ OCFt, interaction variable with operating cash flows; DPR ∗ DOCFt ∗ OCFt, interaction between firms with negative operating cashflows. The first column refers to all firms in our sample. The second column considers Brazilian firms with traded public debt. The third and fourth columns presentthe results of the regression for the most and the less leveraged (measured by the amount of public debt divided by average total debt) firms, respectively. Thefifth and sixth columns present the results of the regression for the most and the less leveraged (measured by the amount of public debt divided by average eq-uity) firms, respectively. The seventh column reports results for firms that present covenants on their debt. Finally, the eighth and ninth columns present the re-sults considering only the subsample of firms more leveraged than the average firms (measured by public debt divided by total debt and public debt by equity,respectively).⁎ represents significance at 10%.⁎⁎ represents significance at 5%.⁎⁎⁎ represents significance at 1%.


the results of Table 3, columns 8 and 9 (α3 is negative and significant). The most leveraged firms (Table 3, columns 3 and 5) donot present significant α3 what may indicate ALRT is not a feature for these firm's earnings.

Summarizing, our results indicate that Brazilian institutional environment does not incentive firms to report quality earnings (mea-sured by ALRT). In this case, debt market's demands do not seem to provide enough incentives to firms report earnings with timelinessloss recognition.

4.3. Public and private loss accruals behavior in Brazilian companies

We apply the accruals-based method (Ball and Shivakumar, 2005) to investigate the relations of operating cash flows and ac-cruals for public and private firms. We estimate the following regression for our sample:

PleasBraz

ACt ¼ β0 þ β1DOCFt þ β2OCFt þ β3DOCFt � OCFt þ β4DPRþ β5DPR � DOCFtþβ6DPR � OCFt þ β7DPR � DOCFt � OCFt þ υt

ð3Þ

where ACt are the accruals at year t scaled by total assets at the end of the year t − 1, DOCFt equals 1 if operating cash flow at year t isnegative or 0 otherwise, OCFt represents the operating cash flow at year t scaled by total assets at the end of the year t − 1, DOCFt ∗ OCFt




is the interaction variable for negative operating cash flows at year t, DPR ∗ DOCFt represents the interaction of private firms and firmswith negative operating cash flows, DPR ∗ OCFt represents the interaction variable for private firms and operating cash flows andDPR ∗ DOCFt ∗ OCFt is the interaction between private firms and firms with negative operating cash flows. Similarly to Eq. (2), DPR as-sume different meanings across our other regression specifications on Table 4. On Table 4, column 2 DPR equals 0 for firms with publicdebt and 1 otherwise. On Table 4, columns 3, 4, 5 and 6 DPR is not defined considering our regressions specifications on subsamples. OnTable 4, column 7 DPR equals 1 for firms without covenants and 0 otherwise. Finally on Table 4, columns 8 and 9 DPR is defined asfirms' leverage ratio.

Consistent with prior research (Healy, 1985; Jones, 1991; Dechow et al., 1995) total accruals are estimated as follows:ACt=(ΔCAt−ΔCLt−Δcasht+Δdebtt−Deprt) where ACt represents accruals at year t, ΔCAt measures the change in current

assets at year t, ΔCLt represents changes in current liabilities at year t, Δcasht represents changes in cash or cash equivalentsat year t, Δdebtt computes the changes in short term debt at year t, Deprt represents the depreciation and amortization atyear t. We estimate operating cash flow as OCFt = OIt + Deprt + DTt − ΔWCt − ÑIt where OIt represents the operating in-come at year t, Deprt is the depreciation and amortization at year t, DTt are the direct taxes expensed at year t, ΔWCt is thechange in operating working capital at year t and ÑIt represents the non-operating income at year t. This procedure is nec-essary because under the Brazilian corporate law the statement of cash flow was not a mandatory financial statement in oursample period.

Eq. (3) allows an asymmetric relation between accruals and cash flows that differs between private and public firms(Ball and Shivakumar, 2005). It relates operating cash flows to accruals considering the idea that negative cash flows sig-nalizes poor performance and managers would recognize, immediately, the present value of all expected future losses re-lated to the current event. According to Ball and Shivakumar (2005, p.94) this model provides for both roles of accruals:mitigation of noise in cash flow and asymmetric recognition of unrealized gains and losses. Asymmetric loss recognitionpredicts that β3 should be positive, while β2 is expected to be negative if accruals ease negative serial correlation incash flows. The other coefficients cannot be predicted in the sense our aim is to test if there is, or not, significant differ-ences between asymmetric loss recognition between private and public firms. In this case, we expect β7 negative in theless ALRT hypothesis for private firms.

Table 4 presents results from Eq. (3), which allows an asymmetric relation between accruals and cash flow levels thatdiffers between private and public firms. In the same sense discussed above for the income, the endogeneity problem maybe present in Eq. (3). Ball and Shivakumar (2005, p.117) recognize that endogeneity is a treat in the accruals model. Theymention that the standard 2SLS simultaneous equations model is inappropriate, because the dummy variable for listingstatus (the dependent variable in the choice regression) is interactive in the regressions. So, they use a Heckit approachto control for any bias resulting from self selection of listing status. After their analysis, the authors comment that(2005, p. 117):

PleasBraz

“(…) controlling for endogeneity has little effect on the estimated coefficients. The coefficient of primary interest (b7) is - 0.42 (t= -6.82)when endogeneity is controlled for, but for the same sample it is -0.45 (t=-7.33) under standard OLS. Further, the adjusted R2 isidentical in the standard OLS and the two stage approaches.We find little benefit from using the two-stage approach, and focus on theresults from standard OLS regressions, which permits a larger sample.”

Following Ball and Shivakumar's (2005) argument and for the sake of comparison of results, we also estimate our modelsconsidering pooled OLS regressions. The coefficient β3 on Table 4, column 1 shows that there is no significant difference inanticipate recognition with positive or negative cash flows for public firms, i.e. there is no asymmetric loss recognition. Pri-vate firms, although present a negative and significant β7, as expected, have accruals' coefficient (β2 + β3 + β6 + β7 =−0.61) which compares to public firms (β2 + β3 = −0.79), both negative; so, the complete set of Brazilian firms do notpresent signs of conditional conservatism. The positive and significant β6 indicate that private firms may use accruals to an-ticipate gains recognition. In this case private and public firms seem to have different incentives for accrual's recognition.However results also show low accounting quality, considering that public firms (benchmark group) and private firms donot present asymmetric loss recognition.

Considering the result that indicate public firms do not use accruals to inform future losses, the effect of public fund raising viadebt, as shown in Table 4, column 2, slightly improves accounting quality for firms with public debt. The coefficient β3 is positive,besides is not significant to differentiate accruals behavior in the presence of negative cash flows. On other hand we cannot con-clude that Brazilian public firms with traded debt and public firms with no traded debt present different accounting quality (β7 isnot significant). Table 4, column 7 shows that firms with covenants and negative cash flows present a smaller positive correlationbetween accruals and cash flows; without differentiate from firms with positive cash flows (β3 not significant). Additionally, thesignificant coefficient β7 shows that firms without covenants on their debt have a higher negative correlation between accrualsand negative cash flows. This indicates lower level of anticipation of asymmetric losses recognition for this group; however,the firms with covenant already present no sign of ALRT. Finally, the negative β2 + β6 show that firms with covenants havelower negative correlation with positive cash flows, what may indicate the anticipated recognition of gains, i.e. higher level ofearnings management.

The leverage ratio does not seem to influence the behavior of public firms in relation to ALRT. Table 4, columns 3, 4, 5, 6, 8 and9 show that the coefficient β3, that differentiate the anticipation of recognition in the presence of negative cash flows, is usuallynegative and when positive is not significant. However there is evidence (Table 4, column 8) that more public leveraged public



Table 5Pooled OLS regression of cash flow from operations on lagged net income and lagged change in net income for all firm-years.

OCFt + 1 = γ0 + γ1It − 1 + γ2IΔI b 0 + γ3IΔI ≥ 0 ∗ ΔIt + γ4IΔI b 0 ∗ ΔIt + γ5DPR + γ6DPR ∗ It − 1 + γ7DPR ∗ IΔI b 0 + γ8DPR ∗ IΔI ≥ 0 ∗ ΔIt + γ9DPR ∗ IΔI b 0 ∗ ΔIt + εt

1 2 3 4 5 6 7 8 9All firms(DPR = 1 forprivate firms; 0for public firms)

Public firms(DPR = 1 for firmswith no publicdebt; 0 otherwise)

Most leveragedfirms (publicdebt/total debtaverage)

Less leveragedfirms (publicdebt/total debtaverage)

Mostleveragedfirms (publicdebt/equityaverage)

Less leveragedfirms (publicdebt/equityaverage)

Firms with public debt(DPR = 1 for firmswithout covenants; 0otherwise)

More leveraged (publicdebt/total debt) than theaverage (DPR is theleverage ratio)

More leveraged (publicdebt/equity) than theaverage (DPR is theleverage ratio)


Intercept (γ0) ? 0.118*** 0.161*** 0.117*** 0.126*** 0.156*** 0.112*** 0.177*** 0.130*** 0.118***It − 1 (γ1) + −0.100** 0.1113049 −0.0967048 −0.124* 0.1862867 −0.106** 0.619*** −0.1037307 −0.0844625IΔI b 0 (γ2) ? −0.044*** −0.045*** −0.043* −0.0476*** −0.053*** −0.044*** −0.0472673 −0.048*** −0.041***IΔI ≥ 0 ∗ ΔIt (γ3) + −0.011* 0.009** −0.010* −0.1172916 −0.1488565 −0.011* −0.1145969 −0.1154565 −0.010*IΔI b 0 ∗ ΔIt (γ4) + −0.093* 0.063 −0.118* −0.0129429 0.1099974 −0.097* 0.0389991 −0.0204379 −0.0430932DPR (γ5) ? 0.0012313 −0.060*** −0.019884 0.073* 0.008*DPR ∗ It − 1 (γ6) − 0.464*** −0.234** −0.595*** −0.09077 0.0448234DPR ∗ IΔI b 0 (γ7) ? −0.035*** 0.006 0.0004111 −0.1052759 0.0008981DPR ∗ IΔI ≥ 0 ∗ ΔIt(γ8)

− 0.013** −0.023** 0.1199705 −0.966*** −0.0038512

DPR ∗ IΔI b 0*ΔIt(γ9)

− 0.170** −0.158 −0.0768384 −0.1037964 0.1554709

Adj. R2 0.1154 0.0599 0.0547 0.0351 0.0613 0.0433 0.0788 0.1239 0.0447F-Stat 87.51*** 9.85*** 5.48*** 6.42*** 5.07*** 8.99*** 10.6*** 8.98*** 7.49***Cross-sectionobs.

5075 283 142 246 104 267 78 283 283

Years 9 9 9 9 9 9 9 9 9No. of obs 44,953 2090 553 1537 377 1713 584 1998 2089

Variables: IΔI b 0; IΔI ≥ 0 are indicator variables of negative (positive) changes on net income at year t. Other variables as defined on Tables 3 and 4.⁎ represents significance at 10%.⁎⁎ represents significance at 5%.⁎⁎⁎ represents significance at 1%.

12A.C.Coelho

etal./Emerging

Markets

Reviewxxx

(2017)xxx–xxx

Pleasecite

thisarticle

as:Coelho,A.C.,etal.,D

eterminants

ofasymmetric

lossrecognition

timeliness

inpublic

andprivate

firmsin

Brazil...,Emerg.M

ark.Rev.(2017),http://dx.doi.org/10.1016/j.emem

ar.2017.02.002



firms anticipate losses when cash flow is negative done by positive and significant coefficient β7 indicating, only in this case, theexpected relation between negative cash flows and accruals.

4.4. Opportunism of managers in private and public firms

We estimate Eq. (3) to assess the relation between past earnings and future cash flows in order to evaluate manager oppor-tunism across public and private companies as follows:

PleasBraz

OCFtþ1 ¼ γ0 þ γ1It−1 þ γ2IΔIb0 þ γ3IΔI≥0 � ΔIt þ γ4IΔIb0 � ΔIt þ γ5DPR

þγ6DPR � It−1 þ γ7DPR � IΔIb0 þ γ8DPR � IΔI≥0 � ΔItþγ9DPR � IΔIb0 � ΔIt þ εt

ð4Þ

where IΔI b 0; IΔI ≥ 0 are indicator variables of negative (positive) changes on net income at year t. Eq. (4) separates the effects ofprivate (DPR = 1) or public firms (DPR = 0) with negative or positive net income on future operating cash flows. Similarly toEqs. (2) and (3) DPR varies across our other regression specifications on Table 5. On Table 5, column 2 DPR equals 1 for firmswithout public debt and 0 otherwise. On Table 5, columns 3, 4, 5 and 6 DPR is not defined considering our regressions specifica-tions on subsamples. On Table 5, column 7 DPR equals 1 for firms without covenants and 0 otherwise. Finally on Table 5, columns8 and 9 DPR is defined as firms' public leverage ratio.

Dechow and Schrand (2004) relate earnings quality to three desirable features of earnings: i) to reflect current performance,ii) to be useful for predicting future performance and iii) to accurately annuitize intrinsic firm value. In this fashion, Eq. (4) mea-sures earnings quality by identifying private or public firms' earnings (positive or negative) relation to future cash flows. Mana-gerial opportunism refers to the current incorporation of extraordinary negative transitory items in earnings foreseeing benefitson future earnings announcements. This process, called “big bath” accounting, allow managers to report higher earnings in futureperiods due to the dual aspect of accounting. Ball and Shivakumar (2005) relate that if negative transitory earnings componentscome up from earnings management they are cosmetic and will not predict future cash flows. Otherwise, positive and negativeearnings changes will be positively correlated with realized future cash flows.

Table 5, column 1 show that current and past earnings relate negatively to future operating cash flows for public firms andcurrent and past earnings relate positively to future operating cash flows for private firms. This result can be explained by thestrong use of accounting earnings for taxes purposes for Brazilian private firms. Tax rules are more cash based than accrualbased, which make this relation positive. This result shows that accounting earnings of Brazilian public firms have small predictivequality to project cash flow, indicating opportunistic behavior to their managers. Moreover, Table 5, column 2 shows that firmswith public debt present current and past earnings with predictive quality to future operating cash flows only for positive changesin earnings (coefficient γ4) The public firms without public debt have also small predictive quality to project cash flow. Table 5,column 7 shows that firms with public debt with covenants present positive coefficients indicating a contemporaneous relation-ship between earnings and cash flows. Finally, the leverage ratio (Table 5, columns 3, 4, 5, 6, 8 and 9) do not contribute to dis-criminate accounting quality of earnings predicting cash flows.

4.5. Robustness test

Ball and Shivakumar (2005) argument that private and public firms may have different investment and financing policiesresulting from their size difference. Usually, public firms are bigger and the effects of size and industry on the estimated coeffi-cients may not be linear. To check the robustness of our results we consider a sub-sample of private and public firms matchedon the basis of size, return on assets and fiscal year-end. In each year and for each public firm, we choose, without replacement,a matched firm from all private firms with the same fiscal year end. This procedure yields 852 matched pairs with the actualnumber for a particular analysis depending on data availability. The mean (median) return on assets for the matched sample is3.2% (3.4%) for public firms and 1.9% (3.0%) for private firms, and the mean (median) total assets is R$ 8.3 billion (R$ 2.5 billion)for public firms and R$ 4.2 billion (R$ 844 million) for private firms.

For this matched sample, as reported on Table 6, panel A, our hypothesis hold. The coefficient α7 in the earnings reversionmodel is not significant, as the results exhibited on Table 3, showing the same level of reversal of income decreases in privatecompanies and in public companies in Brazil. Table 6, panel A also shows that for the matched sample there is no reversion ofnegative changes on profits (α3 = 0.40) and no reversion for profits from private firms (α6 = 0.49).

Table 6, panel B, shows similar results for the full sample and for the matched sample. The accruals model, which allows anasymmetric relation between accruals and cash flow levels to differ between private and public firms, results in an insignificantcoefficient β3 indicating there is no difference in anticipate recognition with positive or negative cash flows for public firms, i.e.there is no asymmetric loss recognition. Private firms, although present a negative and significant β7, as expected, have negativeaccruals' coefficient (β2 + β3 + β6 + β7 = −1.02) which compares to public firms (β2 + β3 = −0.46). Generally the resultspresented in Table 6 panel B conduct to the same conclusion from the results from Table 4, Column 1.

These results are coherent with our hypothesis and show no significant difference between public and private companies inBrazil unlike Ball and Shivakumar (2005) report in the UK.



Table 6Robustness test.

Panel A

Pooled OLS regression of change in earnings on lagged change in earnings for matching firm-years. Dependent variable measured as change in net income.

ΔIt = α0 + α1DΔIt − 1 + α2ΔIt − 1 + α3DΔIt − 1 ∗ ΔIt − 1 + α4DPR + α5DPR ∗ DΔIt − 1 + α6DPR ∗ ΔIt − 1 + α7DPR ∗ DΔIt − 1 ∗ ΔIt − 1 + εt

Predict Coef.

Intercept (α0) ? −0.013DΔIt − 1 (α1) ? 0.000ΔIt − 1 (α2) 0 −0.059DΔIt − 1 ∗ ΔIt − 1 (α3) − 0.399***DPR (α4) ? 0.019DPR ∗ DΔIt − 1 (α5) ? −0.006DPR ∗ ΔIt − 1 (α6) ? 0.494***DPR ∗ DΔIt − 1 ∗ ΔIt − 1 (α7) + 0.031α2 + α3 0.340α2 + α6 0.435α2 + α3 + α6 + α7 0.865Adj. R2 0.732F-Stat. 189.31***No. of obs 536

Panel B

ACt = β0 + β1DOCFt + β2OCFt + β3DOCFt ∗ OCFt + β4DPR + β5 DPR ∗ DOCFt + β6DPR ∗ OCFt + β7 DPR ∗ DOCFt ∗ OCFt + υt

Predict Coef.

Intercept (β0) ? 0.368***DOCFt (β1) ? −0.0455OCFt (β2) − −0.565***DOCFt ∗ OCFt (β3) + 0.101DPR (β4) ? −0.270***DPR ∗ DOCFt (β5) ? 0.0332DPR ∗ OCFt (β6) ? 0.224*DPR ∗ DOCFt ∗ OCFt (β7) − −0.778***β2 + β3 −0.464β2 + β6 0.325β2 + β3 + β6 + β7 −1.018Adj. R2 0.474F-Stat 62.22No. of obs 852

Matching Sample: DPR = 1 for private firms; 0 for public firms.⁎ represents significance at 10%.⁎⁎ represents significance at 5%.⁎⁎⁎ represents significance at 1%.


5. Conclusions

In this paper we investigate the asymmetric timeliness of loss recognition between Brazilian public and private firms and dif-ferent features of debt financing. Ball and Shivakumar (2005) found that British public firms present earnings of superior qualitythan private firms because of differential demand for informative accounting reports. The authors argue that the market for cor-porate financial reporting demands highly informative earnings for public firms. Earnings of public firms are used in contractswith managers, debt agreements, covenants and other governance mechanisms while earnings for private firms are more likelyto be used for tax and other regulatory needs. In this paper we conduct a similar study in a complete different setting fromBall and Shivakumar (2005). We investigate the differences between the quality of earnings between Brazilian public and privatefirms. Unlike the UK, the Brazilian environment does not provide different incentives for the public and private firms. Brazil is acode law developing country where investors are poorly protected and firms base their financing on insider deals. We conjecturethat in this scenario public firms will not face superior incentives than private firms to provide informative reports. Our resultsconfirm our main expectation and show that there is no significant difference in ALRT for Brazilian public and private firms asBall and Shivakumar (2005) reported for the UK. We also show that ALRT is not significantly affected by firm's issuing activityin debt markets, public leverage and in the presence of covenants. We also detected strong signals of opportunistic behaviorfor both samples of firms in Brazil. Our results confirm this notion and show that generally there is no difference in the recogni-tion of gains and losses for private and public firms in Brazil. Few differences arise on ALRT between these two groups due to debtfinancing. Our results contribute to a growing literature on international accounting which tries to understand the determinants ofthe properties of published accounting reports focusing on incentives created by the market demand for corporate financialreporting.




Our study is subject to some limitations. First we consider a private database (SERASA) that is widely used in Brazil by banks,but is not widespread among researchers. Despite many institutions and banks rely on these data to perform credit analysis thedata is likely to suffer measurement errors. Second, we consider a specific legal form of private firms in Brazil (SAcf), but there areother legal forms of private firms that we do not have access to data. Third, because of limited information about private firms,our set of independent variables basically reflects financial characteristics of firms. Fourth, our results may not generalize toother private firms from different emerging markets.

Future research can adopt a cross-country analysis on the determinants of asymmetric loss recognition timeliness in public andprivate firms considering IFRS adopters and different institutional environments in a way that a deeper panel data analysis can beimplemented.

References

Ahmed, A.S., Duellman, S., 2007. Accounting conservatism and board of director characteristics: an empirical analysis. J. Account. Econ. 43, 411–437.Ali, A., Hwang, L., 2000. Country-specific factors related to financial reporting and the value relevance of accounting data. J. Account. Res. vol. 38 (1), 1–21 (Spring).Anderson, C.W., 1999. Financial contracting under extreme uncertainty: an analysis of Brazilian corporate debentures. J. Financ. Econ. 51, 45–84.Ball, R., Shivakumar, L., 2005. Earnings quality in UK private firms. J. Account. Econ. 33, 66–98.Ball, R., Kothari, S.P., Robin, A., 2000a. The effect of institutional factors on the properties of accounting earnings. J. Account. Econ. 29, 1–51.Ball, R., Robin, A., Wu, J., 2000b. Accounting standards, the institutional environment and issuer incentives: effect on timely loss recognition in China. Asia-Pac.

J. Account. Econ. 29, 1–51.Ball, R., Robin, A., Wu, J., 2003. Incentives versus standards: properties of accounting income in four East Asian countries. J. Account. Econ. 36, 235–270.Ball, R., Robin, A., Sadka, G., 2008. Is financial reporting shaped by equity or debt markets? An international study of timeliness and conservatism. Rev. Acc. Stud. 13,

168–205.Basu, S., 1997. The conservatism principle and the asymmetric timeliness of earnings. J. Account. Econ. 78–90.Bebchuk, L.A., Hamdani, A., 2009. The elusive quest for global governance standards. Univ. Pa. Law Rev. 157, 1263–1317.Bebchuk, L.A., Weisbach, M., 2010. The state of corporate governance research. Rev. Financ. Stud. 23, 939–961.Black, B., 2001. Strengthening Brazil's securities markets. J. Comm. Econ. Financ. Law (Available at www.ssrn.com).Brito, G.A.S., Martins, E., 2013. Accounting conservatism and the cost of bank credit in Brazil. Braz. Bus. Rev. vol. 1 (10), 27–48.Burgstahler, D., Hail, L., Leuz, C., 2006. The importance of reporting incentives: earnings management in European private and public firms. Account. Rev. 81 (5),

983–1016.Bushman, Robert M., Piotroski, Joseph D., 2006. Financial reporting incentives for conservative accounting: the influence of legal and political institutions. J. Account.

Econ. 42, 107–148.Chen, F., Hope, O.K., Li, Q., Wang, X., 2011. Financial reporting quality and investment efficiency of private firms in emerging markets. Account. Rev. 86 (4), 1255–1288.Chong, A., Lopez-de-Silanes, F., 2007. Investor Protection and Corporate Governance – Firm Evidence Across Latin America. Stanford University Press and International

Development Bank.Dechow, P.M., Schrand, C.M., 2004. Earnings Quality. The Research Foundation of CFA Institute.Dechow, P.M., Sloan, R.G., Sweeney, Amy P., 1995. Detecting earnings management. Account. Rev. vol. 70 (2), 193–225.Dechow, P., Ge, W., Schrand, C., 2010. Understanding earnings quality: a review of the proxies, their determinants and their consequences. J. Account. Econ. 50,

344–401.Durnev, A., Kim, E.H., 2005. To steal or not to steal: firm attributes, legal environment, and valuation. J. Financ. 60, 1461–1493.Fan, J.P.H., Wong, T.J., 2002. Corporate ownership structure and the informativeness of accounting earnings in East Asia. J. Account. Econ. 33, 401–425.Gopalan, R., Jayaraman, S., 2012. Private control benefits and earnings management: evidence from insider controlled firms. J. Account. Res. 50 (1), 117–157.Healy, P.M., 1985. The effect of bonus schemes of accounting decisions. J. Account. Econ. 7, 85–107 (Rochester).Hung, M., 2000. Accounting standards and value relevance of financial statements: an international analysis. J. Account. Econ. 30, 401–420.Jones, J.J., 1991. Earnings management during import relief investigations. J. Account. Res. vol. 29 (2), 193–228 Chicago.KMPG, 2010. A Governança Corporativa e oMercado de Capitais: Um panorama das corporações brasileiras na BM&FBOVESPA e nas Bolsas norte-americanas. Available

at. http://www.kpmg.com.br.La Porta, R., Lopes-de-Silanez, Shleifer A., 2000. Agency problems around the world. J. Financ. vol. 55, 471–517 (Feb.).La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 1999. Corporate ownership around the world. J. Financ. 54, 471–517.La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 2002. Investor protection and corporate valuation. J. Financ. 57 (3), 1147–1170.LaFond, R., Watts, R., 2008. The information role of conservative financial statements. Account. Rev. 83, 447–479.Nassif, A., 2007. National Innovation System and Macroeconomic Policies: Brazil and India in Comparative Perspective. United Nations Conference on Trade and De-

velopment (Discussion Papers).Nobes, C., 1998. Towards a General Model of the Reasons for International Differences in Financial Reporting. vol. 34. Abacus (September iss. 2).Okimura, R.T., Silveira, A.M., Rocha, K.C., 2007. Estrutura de propriedade e desempenho corporativo no Brasil. Revista de Administração Contemporânea Online 1,

119–135.Pae, J., Thornton, D., Welker, M., 2005. The link between earnings conservatism and the price-to-book ratio. Contemp. Account. Res. 22, 693–717.Pelucio-Grecco, M.C., Geron, C.M.S., Grecco, G.B., Lima, J.P.C., 2014. The effect of IFRS on earnings management in Brazilian non-financial public companies. Emerg.

Mark. Rev. 21, 42–66.Shleifer, A., Vishny, R., 1997. A survey of corporate governance. J. Financ. vol LII (2).Studart, R., 2000. Financial opening and deregulation in Brazil in the 1990's: moving towards a new pattern of development financing. Q. Rev. Econ. Finance 40, 25–44.Sunder, S., 1997. Theory of Accounting and Control. South-Western Publishing.Watts, R.L., 2003a. Conservatism in accounting part I: explanations and implications. Account. Horiz. 17 (3), 207–221.Watts, R.L., 2003b. Conservatism in accounting part II: evidence and research opportunities. Account. Horiz. 17 (4), 287–301.Young, M.N., Peng, M.W., Ahlstrom, D., Bruton, G.D., Jiang, Y., 2008. Corporate governance in emerging economies: a review of the principal–principal perspective.

J. Manag. Stud. vol.1 (45), 196–220.Zhang, Y., Uchida, K., Bu, H., 2013. How do accounting standards and insiders' incentives affect earnings management? Evidence from China. Emerg. Mark. Rev. 16,

78–99.


http://refhub.elsevier.com/S1566-0141(17)30069-9/rf0005
































http://www.kpmg.com.br























emerging markets review - fucape business school · determinants of asymmetric loss recognition...

Documents