Resultados contábeis gerenciados e desempenho do modelo Ohlson para avaliação de investimentos em empresas brasileiras

Abstract

This research aims to verify whether investing decisions, based on supposedly managed public accounting information may lead to abnormal returns rates other than those returns that could be obtained from decisions based on supposedly non-managed publicaccounting information. Two hypothesis were tested, the first affirmed the long-term returns provided by companies that manage their earnings positively are lower than other companies returns, and the second affirmed the long-term returns provided by companies that managetheir earnings negatively are higher than other companies returns. Therefore, several investment portfolios were build based on the valuation model developed by Ohlson (1995), hereby OM, considering the 2013 year end, and taking into accounting the different levels which public companies deal with discretionary accruals on your financial statements between the 2009 and 2013 years end. The returns obtained from these portfolios were assessed by the 2014 and 2015 years end. The OM was operationalized following the research of Ota (2002). In order to detect discretionary accruals, this research made use of the model proposed by Ye (2006), since no other national applications of this model were found. The research sample contemplated companies integrating Brazilian Index, IBrX-100, by December 30th, 2013. From all these companies, 72 matched all research methods restrictions and were classifiedaccording with Ye (2006) model into three different categories depending on the level of financial statements discretionary accruals. From these companies, 69 generated the necessary information in order to proceed with the OM valuation process and compose the investmentportfolios. The methodological procedures lead to four different approaches regarding the classifications yield by Ye (2006) model and by the theoretical values estimated by OM. Research results indicated that companies that manage earnings positively and negatively might alter their own value as perceived by Ohlson Model. These results wereverified when Ye (2006) model was estimated using ordinary least squares (OLS) method to panel data, and the CAPM was used in order to estimate the capital cost used as discount rate in Ohlson Model. The same results were not corroborated for other methodologicalapproaches used. Other results also demonstrated that: (i) the Ye (2006) model has shown a better and information criteria (AIC and BIC), when compared with other models for discretionary accruals detection, such as Modified Jones Model and the Performance Matching Model; (ii) The utilization of panel data regression specific models seems to be afeasible alternative in order to estimate the parameters of Ye (2006) model. This research used the fixed effects regression model; and (iii) The results have demonstrated that comprehensive income, replacing the net income, in the total accruals estimation (Ye model) and residualincome (OM) is also a feasible alternative, since comprehensive income have been integrating Brazilian general accepted accounting practices (BR GAAP) since 2008.