The central aim in this research was to assess investors’ accounting information interpretation skills. Therefore, it was assessed how different profit components would affect the future profitability of publicly traded Brazilian companies. Through a sample of all Brazilian public companies listed in Bovespa from 1995 to 2010, the research results demonstrated that current accruals were incapable of explaining future abnormal return behavior in the firms analyzed. In addition, no significant abnormal returns were reached in an accruals-based investment strategy, indicating that investors would be capable of interpreting and pricing accounting data.
Departing from the seminal work by [
Among the data included in the financial statements, there is a trend for investors, analysts and creditors to recommend some information more recurrently like profit-related information for example. That is so because, despite possible social implications, businesses are primarily based on the owners’ conviction that company operations can result in profits. In profit analysis, [
According to [
Investors should consider the reduction in the future profitability level as a result of a shorter persistence of current accruals than cash flow items [8,7], on the other hand, detected negative associations between current accruals and abnormal future returns in the companies. In other words, although empirical evidence indicates that high accrual levels enhance the probability of reductions in future profitability, investors’ expectations—reflected in stock prices—demonstrated that they have not incurporated information about the persistence of accruals in their analyses.
Studies that investigated the effect of accruals on profit persistence and company value in general have focused on developed markets like the United States and the United Kingdom [
A series of distinguished characteristics marks the Brazilian market. Historically, in its institutional environment, fiscal rules have strongly influenced the financial statements. The State issued accounting standards in response to its fiscal and tax needs, what deviates the statements’ focus from the provision of useful information to investors and other users [10,11] defends that, in countries where the impact of accounting information is comparatively weaker, like in Brazil, the functional fixing in the final profit reported is less common or less important in stock pricing, which in turn interferes in accrual anomalies.
The above considerations and a transition period in the financial statements, which used to be based on standards, towards a more principle-based accounting, reveal a unique opportunity to analyze investors’ ability to interpret Accounting data, particularly the disclosure of profit information. Thus, the general aim of this study was to verify whether accounting anomalies in accruals are detected in the Brazilian capital market. To assess this objective, the research was developed to check whether different profit components influence profit persistence and the market’s reactions. The study was conducted based on the method by [
The specific aims are to verify whether the current accrual level is related to the profit behavior and market value in future periods and to detect whether investment strategies based on accruals produce statistically significant abnormal returns.
After presenting the research objectives, it is interesting to express the research hypotheses in analytic terms. The first hypothesis, on the persistence of accruals, serves as a premise to evaluate the other hypothesis. Thus:
H1: The current profits resulting from adjustments to the accrual base of accounting are less persistent than the profits deriving from cash flow components.
The other tests refer to the accrual anomaly itself, that is, that investors would not understand the nature of profits, resulting in asset pricing evaluation errors. Therefore, the second research hypothesis is that:
H2: The expected profits embedded in current stock prices do not correctly reflect available information on accounting profit.
An extension of the second research hypothesis established refers to the formation of investment strategies. It can be formulated as follows:
H2,1: A negotiation strategy based on the profit accrual level produces positive abnormal returns.
Traditional financial literature is based on the semistrong form of the capital market efficiency hypothesis. The premise is that asset prices embed all publicly available information in a logical and instantaneous form, making it impossible for investors to systematically reach extraordinary gains. The empirical results on the anomaly of accruals challenge the market efficiency paradigm in its semi-strong form, suggesting that abnormal returns can be reached through investment strategies based on publicly available information [
In other words, if anomalous accruals are detected, it is demonstrated that investment strategies based on accounting data support decision making, allowing capital market agents to detect investment alternatives with a greater possibility of gains. In that sense, investors and financial market portfolio managers will obtain additional empirical evidence and theoretical arguments for the financial statement information efficiency in companies whose stocks are traded on the primary and seconddary markets of the São Paulo Stock Exchange (BM&F Bovespa).
The theoretical platform can be separated in four subsections. First, a short summary on profit, its elements and importance is presented. Then, financial anomalies are defined, explaining how they challenge the Market Efficiency Hypothesis by [
According to [
Basically, cash flows and accruals constitute profits. The cash flow components comprise revenues and expenses that corresponded to actual inflows or outflows of financial resources during the period. On the other hand, accruals are temporary cash flow adjustments [
Accruals permit compliance with the main foundations of the accrual base of accounting: the revenue realization principle and the matching principles between revenues and expenses in the same period. Cash flows, on the other hand, cannot comply with these principles, as they join revenues and expenses accrued in different periods in the same statement. In that sense, [
[
In capital market studies, in general, the efficient market hypothesis is assumed in its semi-strong form. [
Inquiries about support for the argument in favor of investors’ total rationality encouraged the creation of a group of behavioral fans, in which the principle of limited rationality replaced that of limited rationality [
Financial anomaly can be considered a seemingly unexplained systematic event [
The seminal research developed by [
Based on the estimated regressions and constitution of theoretical portfolios, [
In view of the results those authors found, the conjecture was established that the market would be incapable of correctly pricing accrual information. These findings triggered subsequent studies that investigated the anomaly of accruals, which implied the development of new research methods and a considerable range of foci, samples and empirical evidence available on the theme.
The hypothesis established [
[
Using a method distinct from [12,23] evaluated the functional fixing hypothesis of the final accounting profit. The author demonstrated that the profitability of companies that exhibited extreme accrual levels continued across two consecutive years and that, despite this continuity, companies experienced abnormal returns, suggesting that these returns do not result from the reversal of accruals. Thus, using distinct methods, the findings by [
[
The deep review realized by [
[
The methodological procedures adopted in the study are described in this chapter, in which the sample selection criterion is treated in the first subsection. Then, the procedures to detect outliers are explained, as well as the operational definitions of the study variables. Next, the estimated models to investigate the research hypotheses are presented.
The sample used in this study consisted of Brazilian publicly traded companies with stocks traded on BM&FBovespa between 1995 and 2010, excluding financial companies. As three-monthly observations tend to vary as a result of possible seasonal spreads and accounting principles, we decided to use the companies’ annual reports. The financial information used was taken from the Economática® database and variables were updated to December 31st 2010, using the Brazilian Institute of Geography and Statistics’ (IBGE) Extended Consumer Price Index—IPCA.
To compose the sample, only companies with complete data for all study variables across the study period were considered. In addition, corrective measures were adopted for the outliers. Using box plot graphs, the presence of outliers was evaluated in the dependent variable. Thus, any values beyond the inter-quartile interval were excluded from the sample. On the other hand, to detect outliers in the independent variables, the Mahalanobis distance was used [
Significantly fewer observations were available to assess the second study hypothesis. That was so because the stocks of a range of companies listed on BM&FBovespa are not frequently traded, which makes it impossible to calculate return in the period after the presentation of the financial statements. As a result, less data were available to assess the second hypothesis, resulting in a final sample of 243 companies and 1291 observations.
The measure used to represent company profits was Earnings before Interest and Taxes—EBIT). The used of the EBIT excludes non-recurrent items, such as extraordinary items, discount operations and special items, as well as non-operating profit (present in statements before law 11.638/2007). Non-recurrent items are problematic, as the information needed to decompose them between cash flows and accruals are not always available. The exclusion of these items permits more consistent access to the persistence of cash flow components and accruals. Profits were standardized in relation to mean total assets [
where:
= Current profits before interests and taxes
= Lagged Total Assets
= Current Total Assets Total accruals were calculated through the balance sheet focus ([
where:
= Variation in Current Assets
= Variation in Cash and cash equivalents
= Variation in Current Liabilities
= Variation in Short-term debts
= Variation in taxes payable
= Variation in depreciation and amortization expenses
= Lagged Total Assets
= Current Total Assets The cash flows, on the other hand, were estimated by the difference between profits and accounting accruals, as shown in Equation (4).
Returns were calculated for an annual period. The start of the estimation window was established at the end of the fourth months after closing off the previous year. This procedure assumes a lag between the end of the year and the date on which the financial statements are disseminated [
With and indicating stock prices on t and t–1, respectively.
The abnormal return was given by the actual return obtained by security minus the expected normal return, given, which is the conditioning information for the normal performance model. For a stock i and event date t, the abnormal return was defined as shown in Equation (6):
The expected return was calculated through the financial asset pricing model CAPM. The Beta of each stock was obtained using the Economática database. The riskfree return rate of the economy was represented by the Selic (basic interest rate in Brazilian economy). The proxy of the market return was the Ibovespa index return. The CAPM is expressed in Equation (7):
To assess the risk, the Beta of the sample companies’ stock was estimated. As an alternative, the Ibovespa return was included as a dependent variable. In addition, two additional measures were used: Book-to-Market and return on assets (ROA). Traditional authors like [
To test the influence of the activity sector on the relation between current accruals and future returns, dichotomous dummy variables were inserted to identify the sector each company belongs to (dsector). In this case, 20 sectors were considered, following the Economática classification, as displayed in
In view of the argument that the investment anomaly would cover the effect of the anomaly of accruals [
To evaluate whether accruals and cash flow components affect predicted future profits differently, regressions were estimated, considering the sample companies’ behavior during the years covered in the analysis. Separating the two main profit components, a possible variation in their persistence was investigated, according to Equation (8):
If the regression demonstrates significant differences between and, the hypothesis of different prediction powers between the two profit components will be confirmed. In order to actually test the anomaly of accrual, however, a relation needs to be established between stock returns and past accruals. A negative and statistically significant relation between accruals and stock returns during the immediately subsequent period
demonstrates investors’ inability to understand the reversible property of accruals. To be able to test this relation, control variables were included in the model (Equation (9)), based on [7,23].
In this regression model, measures the ability of accruals to predict future returns. When is not statistically equal to zero, there are signs that an abnormal return can be reached in a strategy that involves the selection of firms based on the exhibit accruals level.
Portfolios were constituted according to the level of accruals the firms exhibited. To evaluate whether companies with a higher level of accruals exhibited significantly lower abnormal returns that companies with low accrual levels, the following was done:
1) Classification of firms according to the level of accruals reported over the years for each year analyzed;
2) Segregation of firms into five portfolios of equal weight, based on the quintiles calculated for each firm;
3) Measurement of abnormal returns for portfolios with low as well as with high levels of accruals;
4) Comparison among returns reached through a test of differences of means.
This chapter presents the analyses and discussion of results. The first hypothesis established, about less persistent profits related to accruals than to cash flow components, represents a premise for the assessment of the other hypotheses. The other tests refer to the anomaly of accruals itself.
Regression analysis with panel data was used to evaluate the first study hypothesis. First, Chow’s test was applied to compare available models for panel data. F-statistics (482, 2, 872) reached 2.26 (Prob > F = 0.000), which permits the rejection of the null hypothesis that the model that would best adapt to the data would be the estimation of the model through the Pooled Ordinary Least Squares —POLS), demonstrating the superiority of the fixed effects model, at a confidence level of 95% (
The Hausman test offered an objective criterion for the decision between the fixed effect model and the random effect model. Hausman test statistics reached 455.44 (Prob > = 0.000), which permitted the rejection of the null hypothesis, again leading to the conclusion that the most adequate model for the sample under analysis is the fixed effect model.
The regression results demonstrated that the accruals are less persistent than the cash flow components. These differences are perceived in a more transparent way based on the standardized coefficients, in which accruals and cash flows presented standardized coefficients of 0.43 and 0.53, respectively. To evaluate whether the coefficients were statistically different, the hypothesis of equality between the coefficients through the F-test is appropriate. The results of that test proved, at a confidence level of 95%, that the coefficient of the accruals related to profit persistence was significantly different from the coefficient related to the cash flow components, with F-statistics (1, 2699) = 35.73 (Prob > F = 0.000).
The obtained results regarding profit persistence differed from Sloan’s findings (1996). In that author’s study, the cash flow components and accruals presented persistence coefficients bordering on 0.8 and 0.5, respectively, when calculating regressions in deciles. The coefficients detected in this study, however, are similar to [
Chow’s test was again applied, which permitted the conclusion that the fixed effect model was superior to the minimum least squares models for the study sample with regard to the second model under evaluation. F-test statistics (482, 2, 872) reached 2.26 (Prob > F = 0.000). At a 95% confidence level, this result induces the rejection of the null hypothesis that the model that would best adapt to the data would be the POLS model.
Like in the first model, to identify which of the fixed and random effects models would be the most adequate for the sample data, Hausman’s test was performed. As a result of a test statistics equal to 16.75, the null hypothesis was rejected. Thus, the result analysis concentrated on the fixed effects models presented in
The first variable to be excluded from the model, the least significant in the regression, was exactly the study variable, i.e. accruals standardized by total assets. This exclusion leads to the rejection of the second study hypothesis. That is, a negative association between the current accruals and future abnormal returns of the companies under analysis cannot be proven. Thus, as opposed to what was detected in international literature, no empirical evidence was found that, after the arrival of new profit information, investors would revise their expectations, raising stock prices with low accrual levels and generating significant abnormal returns.
The growth variable was the second variable excluded from the model, resulting in the final model presented. The Book-to-Market and ROA variables were significant and their signs were in line with international literature. The results signaled that, the greater the company’s valuation in the market (exhibiting a lower Book-toMarket indicator) and the higher its profitability (with a higher ROA rate), the higher the company’s future returns. In addition, the regression results demonstrate that larger companies would present a lower risk of insolvency, which would entail a lower risk and, therefore, a lower risk premium (return).
As an additional way to evaluate the effects of adjustments in the accrual base of accounting in period t on future returns, asset portfolios were constituted based on the accrual levels the companies exhibited [
The returns of the specific companies and the market index (Ibovespa) were calculated on the stock prices obtained from Economática. The portfolios were reached by adding up the returns of their individual stocks. To compare the mean returns, the t-test for two samples was used, always preceded by the variance analysis between the samples, a premise inherent in the application of the t-test.
According to the t-test, it is observed that the mean returns of portfolios with low accrual levels did not significantly differ from companies with high accrual levels. To reach definitive results on the effect of accruals on returns, however, the expected return part should be removed from the assets, permitting the analysis of the unexpected part only. To estimate the expected return part, the asset pricing model CAPM was used, which incorporates the variability of the security in view of market returns in its analyses.
To compare the abnormal return between the two stock portfolios, in which the first included observations with low accrual levels in the profits, and the second behaviors at the other end, the t-test for two samples was again processed, as indicated in
Evidence for the purpose of arbitrage and for the development of investment strategies based on accrual information goes against expectations. In the year after the portfolio was established, no positive abnormal returns were identified for the portfolio with low accrual levels, differently from [
This study investigated the relation between companies’ accrual level and the market’s perception of publicly available profit information. Literature about accounting users’ perception on the different profit components is still incipient, a factor that demands further research to enhance knowledge about the understanding of accounting information in the Brazilian capital market.
The first research hypothesis established was that accruals were less persistent than cash flows. The estimated regression demonstrated that the persistence of the accrual components was significantly lower than the cash flow component, with coefficients of 0.43 and 0.53, respectively. Statistically different coefficients between the cash flow items and the profit components resulting from adjustments in the accrual bases, besides the shorter persistence of the second item, led to the acceptance of the first hypothesis.
The second hypothesis was established to detect whether Brazilian investors understand the relation demonstrated in the first study hypothesis, i.e. that the persistence of accrual-related profits was shorter than that of profits related to cash flow items. Thus, it was evaluated whether they attribute different weights to the accrual components and cash flow components of profits. Among the variables included in the estimated model to explain future stock returns, the accrual variable showed the most limited explanatory capacity, as the statistical tests demonstrated. The lack of statistical significance led to the rejection of the hypothesis, showing that there was no evidence to conclude that the investors did not understand the profit components. In addition, a lack of significant differences among economic sectors was demonstrated, as all coefficients of the dichotomous variables that served to identify the effect of companies’ activity sectors were not statistically significant.
Expanding the second hypothesis, it was verified whether negotiation strategies based on the accrual level would produce positive abnormal returns. Through the establishment of portfolios based on accrual information and inter group comparison, additional evidence was obtained, which demonstrated that investment strategies based on accrual information were unable to produce statistically significant abnormal returns.
In summary, in line with [
In view of evidence on the occurrence of the anomaly of accruals, following [
Nevertheless, gaps remain, including the role of investors’ level of sophistication and financial asset pricing errors. Departing from the premise that investors do not process relevant information on profit components and overestimate the accrual effect on future profits, it seems probable that this behavior is less recurrent in company
stocks traded by less sophisticated investors. That is so because this type of investor tends to understand accounting information in a less detailed way, mainly profit information, thus revealing a lesser ability to adjust their future profit estimates in function of changes between profit proportions [
Another opportunity for research on the effect of accounting information in the Brazilian capital market would be the segregation of accruals into broad categories, according to equity elements, i.e. between elements deriving from assets and from liabilities, leveling accruals according to their reliability level [
Finally, research that provides results on the presence of the anomaly of accruals in emerging markets is highly relevant. In those markets, weak investor protection reduces the efficacy of accounting information for decision-making purposes. Thus, a general model to evaluate the anomaly of accruals in different countries could capture the effect of institutional differences and the legal regimen (code law/common law), indicating how these factors influence investors’ ability to interpret Accounting information.