The Accruals Anomaly: A 50/50 chance game? Evidence from the Iberian Stock Markets

The Accruals Anomaly: A 50/50 chance game? Evidence from the Iberian Stock Markets Catarina Borges Dissertation for Master in Finance Supervisor: Prof. Júlio Fernando Seara Sequeira da Mota Lobão 2015 i Biographical introduction Catarina Borges is an economist, graduated from Aveiro University, currently engaging the Master in Finance at the Economics and Management School of Oporto University. Since early time, she understood the importance of invest not only in her technical skills but also in her soft skills. During her time in Aveiro, beyond the bachelor degree in Economics, she was also a member of the Economic Students’ Club. In her second year of studies, she gave voice to the economics students, being responsible to defend and help to ensure the best needs of her colleagues. She also cooperated with the Aveiro’s Students Association in several projects. In FEP, while taking the master degree, she joint to FEP Finance Club and made some voluntary work to spread financial literacy among layers younger. Currently, she is an intern in an accounting office to stronger and consolidate her knowledge in the area since the accounting background is fundamental to unfold functions in finance. ii Acknowledgements I am extremely thankful to my supervisor for all of the guidance, support and motivation through the entire year. His experience allied to his constructive criticisms and suggestions were essential to conclude my dissertation. I am highly thankful to my family, for all the patience and encouragement that they showed during this intense year. All the support in times of trouble was vital to successfully finish my master dissertation. I am thankful to my employer for all the flexibility given to me, allowing me to align my work schedule with my master’s obligations. I am also thankful to my helpful colleagues that inspire and help me solving data and methodology problems, improving, consequently, my work. Last but not least, I would like to express my gratitude to my friends for the cheer moments in times of stress. iii Abstract Being one of the most widely studied anomalies, the Accruals Anomaly, firstly documented by Sloan (1996), is the negative relation between accounting accruals and future stock returns. Trying to overcome some gaps in the literature, we study the relationship between managers and investor’s behavior and the evolution of stock markets to understand their different contribution to the anomaly in the first place. It is our objective to understand if such close markets as Portugal and Spain behave differently in a way that contributes to the (non)existence of this anomaly and to recognize if both managers and investors play an active role in Accruals Anomaly. The study was conducted by collecting a year-sample of both Portuguese and Spanish public companies, in the two different periods – before and during the financial crisis. We show that 1) there is a negative relation between accruals and future returns, but only in Portugal; 2) it is possible to exploit the anomaly in Portugal and yield abnormal returns (20,4% in first year); 3) the financial crisis impacted negatively the returns achieved by Portuguese investors; 4) the overall results are consistent with earnings manipulation in Portugal and to lower income predictability in Spain due to an inflexible accounting system to record accruals. Key-words: Accruals, Anomaly, Cash flows, Earnings, Market Efficiency JEL-Codes: G14, D53, M12, M41 iv Sumário Executivo Sendo uma das anomalias mais amplamente estudadas, a Anomalia dos Accruals, documentada pela primeira vez por Sloan (1996), estabelece a relação negativa entre accruals e retornos futuros. De forma a superar algumas falhas na literatura, decidímos estudar a relação entre o comportamento de gestores e investidores e a evolução nos mercados financeiros para perceber as suas diferentes contribuições para a anomalia. É nosso objetivo compreender se mercados tão próximos se comportam tão diferentemente ao ponto de contribuir para a (não)existência da anomalia e ao mesmo tempo reconhecer se os gestores e investidores desempenham um papel ativo na Anomalia dos Accruals. O estudo foi realizado através da recolha de uma amostra de dados anuais em empresas públicas portuguesas e espanholas, em dois períodos diferentes - antes e durante a crise financeira. Mostramos que 1) existe uma relação negativa entre accruals e retornos futuros, mas apenas em Portugal; 2) é possível explorar a anomalia em Portugal e produzir retornos anormais (20,4% no primeiro ano); 3) o impacto da crise financeira afetou negativamente os rendimentos obtidos por investidores portugueses; 4) os resultados, na generalidade, são consistentes com manipulação de resultados em Portugal e com baixa previsibilidade dos resultados em Espanha devido a um sistema de contabilidade inflexível para o registo de accruals. Key-words: Acréscimos, Anomalia, Cash flows, Resultados, Eficiência de Mercado JEL-Codes: G14, D53, M12, M41 v Table of Contents 1. Introduction ..................................................................................................................... 1 2. Review of Literature ....................................................................................................... 4 2.1 A Broader Definition of Accruals ................................................................................. 4 2.2. Discretionary and Non-Discretionary Accruals ........................................................... 6 2.3 Causes of the Anomaly .................................................................................................. 7 2.3.1 The Earnings Fixation Hypothesis ........................................................................... 7 2.3.2 The Different Accruals Components ....................................................................... 7 2.3.3 The Earnings Management Hypothesis ................................................................... 8 2.3.4 The Market Participants Misinterpretation ............................................................ 10 2.3.5 Growth Hypothesis ................................................................................................ 11 2.3.6 The Methodology Adopted .................................................................................... 11 2.3.7 The “Irrationality” of Agents ................................................................................. 11 2.3.8 The Risk Factor ...................................................................................................... 12 2.3.9 Anomaly as a Collateral Effect of Other Anomalies ............................................. 12 2.3.10. The Legal System ................................................................................................ 13 3. Development of Hypothesis ......................................................................................... 14 4. Data and Methodology ................................................................................................. 17 5. Empirical Results .......................................................................................................... 24 6. Conclusions................................................................................................................... 38 7. References ..................................................................................................................... 41 8. Annexes ........................................................................................................................ 45 vi Table Index Table 1: Mean (Median) Values of the Ranking Accruals Quartile Portfolios .................. 24 Table 2: Results of estimating an OLS Regression to examine the impact of Current Earnings on Future Earnings performance .......................................................................... 26 Table 3: Results of estimating an OLS Regression to examine the impact of Cashflows and Accruals on Future Earnings performance .................................................................... 27 Table 4: Results derived from a Nonlinear Weighted Least Squares estimation to analyze the information content of Current Earnings in Future Earnings and the respectively impact on Stock Price performance ................................................................................................. 29 Table 5: Results from estimating a Nonlinear Weighted Least Squares Regression to examine the impact of information contained in Cashflows and Accruals on Stock Price performance ......................................................................................................................... 30 Table 6: Abnormal Size-Adjusted Returns of Equal Weighted Accruals Portfolios Time Series .................................................................................................................................... 32 Table 7: Results derived from an Nonlinear Weighted Least Squares Estimation of Cashflows’, Normal Accruals's and Abnormal Accruals's Market Pricing respecting to their one-year-ahead Earnings ............................................................................................. 36 1 1. Introduction The accounting statements disclosed by companies are the major source of information for investors when trying to figure out where to invest. Therefore, understanding financial statements and the story that they told us is a major imperative. By comprehending the relationship between the earnings released and the profitability of stocks, investors can profit from taking the best decisions. However, it is important to notice that, sometimes, decisions based on earnings can be biased either because of manager’s earnings manipulation or because investors fail to identify the different persistence in earnings components. The anomaly plays, therefore, an important role in the literature. The accruals anomaly is the negative relation between accounting accruals – the non-cash component of earnings – and future stock returns. The literature supports that earnings that have a higher (lower) component of accruals when comparing to operating cashflows tend to poorly (wealthy) perform in the future. So, it is really important to understand if company’s earnings are based on real cash inflows or if they are based in revenue appreciation from managers’ questionable accounting practices. Firms that register a high (low) level of accruals have less (more) certain earnings and consequently should earn lower (higher) market returns. Although Vale (2013) found evidence of the anomaly in Portugal and Papanastasopoulos (2013) in European Equity Markets, the questions and conclusions that their work raised are very different from the ones that we formulated. So, this particular study assumes great importance when we try to understand if such close markets behave differently in a way that contribute to the (non) existence of this anomaly. Richard G. Sloan was the first to introduce the anomaly, in 1996. The author, in his paper, refers that the market participants over-weight the accrual component of earnings and consequently, under-weight the cash flow component when setting prices. Thus, the market fails to incorporate the different persistence associated to those two components showing that investors and market participants fixate on earnings when analyzing financial statements and companies’ earnings – the Accruals Anomaly was born. Sloan (1996) goes even further and mentions that this accruals anomaly is the responsible for the market misprices firms allowing the implementation of strategies that get abnormal 2 returns: the market forecast higher (lower) future earnings from continuous operations to firms with higher (lower) levels of accruals. This topic has been very controversial, dividing the scientific community, either because some authors affirms that the anomaly exists and will persist in time and others do not; either because there are no consensus in the literature regarding the causes of the anomaly. As it was mentioned before, Sloan (1996) raised for the first time the Earnings Fixation Hypothesis to explain the anomaly. Others (Hribar, 2000; Xie, 2001; Thomas and Zhang, 2002; Chan et al., 2006) point the Earnings Management Hypothesis as the main cause to the problem, justifying the anomaly with the manipulation of the discretionary part of the accrual’s component by managers. The bad quality of the earnings, the aggregation of the earnings’ components and even the human behavior of analysts appear in the literature as causes of the anomaly. The main aim, with this study, beyond examine if the anomaly is present in Portugal and Spain is to analyze the different agent’s – investors and managers – contribution to the anomaly. By studying investors, we will try to understand if they are rational when evaluating earnings and if they can distinguish earnings components and their different levels of persistence. By studying managers, we will try to assess if earnings management has an important role in the anomaly by trying to perceive if they manipulate a particular component of accruals – the discretionary accruals. The adoption of two different periods – before and during the financial crisis – will allow us to understand if the financial crisis is, somehow, responsible to spread or mitigate the anomaly. It is also an objective, by analyzing such different markets, to understand what may be the reasons that explain the appearance of the anomaly in the first place. The study would not be completed without exploiting the anomaly and trying to understand whether there is any particular strategy that can produce abnormal returns. As it was mentioned before, this research approaches both the investors and managers side. To study investors, we will start by calculating the firms’ accruals since the ones that appear in the balance sheet can contain errors according to Collins and Hribar (2000). With this information, we will be able to aggregate the data by ranking firms by accruals into 9 normal accruals. Both tests indicate the overpricing of abnormal accruals as “more severe”, with this mispricing only occurring because of market’s inability to correctly assess the persistence of such accrual component. Is that lack of settings and tools that opens door to managers manipulate earnings, like in Initial Public Offerings or Seasonal Equity Offerings. Beneish and Vargus (2002) also studied earnings manipulation and the impact of insider trading on earnings quality and in accrual’s valuation. In their work, they evaluated whether a signal based on insider trading can somehow predict one-year-ahead earnings persistence and consequently help identify firms with high and low earnings quality, with earnings quality being the capability of a firm sustain their income in the future. Following Sloan’s techniques, they also tested, by examining the market, if the accruals pricing in the firms with higher and lower earnings quality is different and, if so, if the low earnings quality earnings are associated with earnings management. From their research, they conclude that if managers expect the earnings to persist, they have an incentive to purchase their own firm stock. Prior authors mention that this “buy” happens before the stock prices increase and investors can see that insider buying as a positive signal of earnings quality. However, what may in fact could be happening is that managers keep to themselves information related “to the effect of income-increasing accruals on their firm’s prospectus either to strategically increase their stake in the firms’ equity or to abstain from selling” (Beneish and Vargus, 2002, p.758). The rationale behind it is quite simple: if managers manipulate earnings to cover bad performance by income-increasing accruals, it is logical that as they are aware of the poorly future performance, they sell their stock. What the authors conclude was that “income-increasing accruals are significantly more persistent for firms with abnormal insider buying and significantly less persistent for firms with abnormal insider selling, relative to firms for which there is no abnormal insider trading” (Beneish and Vargus, 2002, p.756).. Moreover, they also stated that insider trading information is particular useful a priori to identify if income-increasing accruals’ quality is high or low. To summarize, the authors gave us important insights in their work. Firstly, they found evidence that the mispricing in the accruals can be mainly attributable to the mispricing of income-increasing accruals. Secondly, they discovered that incomeincreasing accruals tend to be overpriced when managers adopt the strategy of abnormal selling and well-priced when managers take part in an abnormal buying. Thirdly, market 10 participants tend to “overprice income-increasing accruals when firm’s top executives do not trade, but the overpricing is less severe than when managers engage in abnormal selling” (Beneish and Vargus, 2002, pp. 756-757). Fourthly and finally, market participants tend to price and look to all increasing accruals as if they are of good quality. Investors can’t distinguish if an insider selling is encouraged by liquidity or by information. It is also possible that investors “discount insider selling signals that are contrary to their beliefs about a firm’s earnings” (Beneish and Vargus, 2002, p. 757). Leuz et al. (2003) also approach the earnings management but from a different angle and provide evidence that earnings management is less of a concern in common law countries due to high investor protection. Kang et al. (2006) also use discretionary accruals as a topic of investigation and they tried to understand whether aggregate discretionary accruals or aggregate total accruals are the better predictor of stock returns. They conclude that both the aggregate normal accruals and equal-weigh aggregate discretionary accrual do not provide considerable forecasting power. On the other side, value-weighted aggregate discretionary accruals provide strong forecasting power. 2.3.4 The Market Participants Misinterpretation Bradshaw et al. (2001) stated that sell side analysts are not able to distinguish between the different characteristics of the earnings components. In fact, the market participants are not able to predict the earnings reversal on extreme accruals, with firms with high and low accruals reporting the highest forecasting errors on earnings. Barth and Hutton (2004) also investigate the topic and conclude that analysts, overall, can’t interpret in a correct way the information that is related to accruals within the reported earnings. Consequently, an investment strategy that is supported on the information provided by these particular market participants and in accruals’ information is able to yield abnormal returns that proved to be statistically significant. Other studies, like the one from Collins et al. (2003), mention that there’s a particular group of investors (the more sophisticated ones) that are able to interpret the different information contained in earnings. However, this was only verified in a portion of the several procedures that were tested. 11 2.3.5 Growth Hypothesis Zhang (2007), following the footsteps of those authors, tested whether the accrual anomaly is attributable to growth as opposed to earnings persistence – the Growth Hypothesis. Basically, Zhang (2007) examine if the accrual anomaly is mainly motivated by the fundamental growth information contained in accruals contrary to accounting distortion or managerial discretion. In his research, Zhang (2007) discovered that accruals show strong (insignificant) power predicting future stock returns in companies where accruals capture strong (weak) growth information. Moreover, when decomposing accruals into growthrelated and growth-unrelated components, the author found evidence that the first has more power when predicting future stock returns than the latter. So, he concluded that the accrual anomaly is mainly attributable to the growth information that is contained within the accruals. 2.3.6 The Methodology Adopted Kraft et al. (2006) refer that the anomaly could be explained by several boundaries – selection biases, data treatment and data errors – in the methodology. However, this not means that the anomaly disappear when we account for that boundaries. The authors conclude that the anomaly is still present but is driven mainly by the high accruals deciles. Similarly, Desai et al. (2004) found evidence that the anomaly seems to be concentrated in the highest accruals firms. 2.3.7 The “Irrationality” of Agents Several authors argue that investors act in an irrational way and, consequently, cause the anomaly. However, as it is known, this topic – rationality against irrationality of investors – divides the financial community, casting disagreement and opposing opinions. In favor of rationality were Wu et al. (2007). Assuming that we interpret accruals as a working capital investment, the authors conjecture that firms adjust their accruals to discount rate changes in an optimal way based on the q-theory. Wu et al. (2007) conclude that associated to higher discount rates are the less profitable investments and lower accruals (vice-versa) – Optimal Investment Hypothesis. The rationale behind the theory is that when the discount rates increase, there are less viable and profitable projects so we 12 verify a decreasing in accruals. Allied to this, there’s an increase in future returns on average because the higher discount rate means higher expected returns going forward. As a result, accruals negatively predict future returns. The same authors, by studying the Optimal Investment Hypothesis from different perspectives, reach the conclusion that since investment is a value driver of the accrual effect, by controlling such variable we should expect a reduction in the magnitude of the accruals anomaly. They even go further and state that investment is probably more significant when comparing to earnings in driving the anomaly. By studying the optimal investment hypothesis the authors also claim that since we verify a negative relation between investment an discount rates, accruals should covary negatively with the ex ante estimates of the discount rate. Last but not least, they also defend that “the deterioration of the accrual effect in recent years might be temporary and likely to mean-revert in the near future” (Wu et al., 2007, p. 177). 2.3.8 The Risk Factor Khan (2005) approach the topic from a risk perspective. For him, the differences found in portfolios with high and low levels of accruals can be explained by differences in risk associated to each. Khan (2005) found evidence that firms with low accruals and consequently higher returns are the ones in which the bankruptcy risk is higher. He also revealed that accruals are more related with risk proxies – economic and financial distress characteristics – rather than risk itself. 2.3.9 Anomaly as a Collateral Effect of Other Anomalies There are several studies in the literature that look at the accrual anomaly as an effect of broader anomalies. Collins and Hribar (2000) tried to discover if the accruals anomaly can be, somehow, explained by the post-earnings announcement drift documented by Ball and Brown (1968) and Bernard and Thomas (1989, 1990). The results showed that these two anomalies are independent from each other. The research of Desai et al. (2004) and Fairfield et al. (2003a, b) suggested that the accrual anomaly might be related to growth. According to Desai et al. (2004), the anomaly is 13 somehow linked with another one: the Value-Glamour Anomaly, already documented in the literature. Fairfield et al. (2003a) argued that the accruals anomaly is a particular case of a “greater” anomaly based on growth in net operating assets. As so, they decided to explore the relation between the accruals anomaly and the growth anomaly. This anomaly reflects the market’s lack of ability to correctly assess firms’ growth prospects. They conclude that the mispricing of accruals is not the main responsible for the accruals anomaly as they find out that the accruals and cash flow components do not exhibit significant differences in persistence. In matter of fact, they defend that the accruals anomaly is part of a broader anomaly – the Growth Anomaly. They even refer in their article that Sloan’s (1996) findings are a special case of a more general effect – the growth effect – that can be explained by diminishing marginal returns to investment and/or conservative accounting. Disagreeing from Fairfield et al. (2003) were Shon and Zhou (2005). They suggested that both anomalies are actually different from each other. 2.3.10. The Legal System Many authors in the literature defend that the law prevailing in the country can also be an incentive for the appearance of the anomaly. Dechow et al. (2011b) found that the anomaly appears to be stronger in common law countries than in code law countries (Portugal and Spain), with Pincus el al. (2007) being more extreme and saying that the anomaly only appears in common law countries. The latter explain that in code law countries we verify a lower asymmetry of information between stakeholders and firm executives and consequently, in those countries there are a better understanding of earnings and its components. Differently, LaFond (2005) and Leippold and Lohre (2010) found evidence in both civil (code) and common law countries. 14 3. Development of Hypothesis Understanding financial information is key for success when applying a particular strategy instead of another. In fact, it is quite important that any investor, when analyzing earnings, is able to distinguish between cashflows and accruals since they have associated different levels of persistence and have different implications for the assessment of future earnings. Bernstein (1993) advise us to trust more on CFO and less on accruals since the latter is subject to a higher level of distortion: “the accrual system, which produces the income number, relies on accruals, deferrals, allocations and valuations, all of which involve higher degrees of subjectivity than what enters the determination of CFO” (Sloan, 1996, p. 291). As so, we can attest the quality of earnings by analyzing the weight of these two earnings’ components in the overall income that the firms releases, with the earnings supported by casflows being of more quality than the ones relying on accruals. Following this reasoning, it is also expected that the current earnings performance is likely to persist if earnings are based more on cashflows than on accruals. To test whether this is true in the markets under study, the first testable hypothesis will try to infer about the persistence: H1: The current earnings performance is declining as the accrual component of earnings increases and increasing as the cashflows component of earnings increases. By studying this hypothesis, we will be able to understand investors’ behavior towards earnings: can investors distinguish the different components of earnings and its different persistence levels or they act as having some kind of myopia which makes them not to recognize the characteristics associated to each? Similarly to Sloan (1996), the unbiased behavior of stock prices lead us to build another hypothesis to test. By decomposing results into its components, we intend to analyze in which extent stock prices reflect the different properties of accruals’ and cashflows’ components of earnings. Undeniably, if investors can’t distinguish the different components of earnings, they will suffer from forecasting errors and therefore misprice securities, meaning that stocks with high (low) level of accruals will be purchased for the major (small) part of investors, driving the price up (down) and consequently overvaluing (undervaluing) the stock. Indeed, this mispricing will be corrected in the future when the 15 persistence does not follow the investor’s expectations, meaning that investors that do not look at the different components of earnings instead of abnormal returns can suffer from losses. Ball and Brown (1968) found a positive relation between earnings and stock prices that was justified by the ability of earnings to reflect all the relevant information of a company. However, many authors defend that this positive relation between price and earnings is only due to the incapacity of investors distinguish between earnings components, leading us to formulate the second testable hypothesis: H2: Stock prices fail to reflect the higher persistence associated to the cashflow component of earning and the lower persistence associated to the accrual component of earnings. With this hypothesis, we will try to test in which extent prices deviates from the rational expectations model. So, following Sloan (1996), we will test the “naïve expectations model” (investors fixate on earnings) against the null hypothesis of market efficiency. If investors can’t distinguish the different levels of persistence that is associated to each component of earnings, it is expected that they will overprice stocks with a higher accrual component and underprice stocks that have a lower component of accruals, creating, therefore, an arbitrage opportunity. This leads us to the third testable hypothesis: H3: A trading strategy that combines a long and a short position in stock firms with the lowest and highest reported accruals, respectively, can generate positive abnormal returns. The results will evidence if investors can exploit this market flaw, yielding abnormal returns. The last topic that our study will address is related with abnormal accruals and earnings manipulation. Similarly to Xie (2001), we will try to find evidence of earnings management, leading us to formulate the fourth and final hypothesis: H4: The market correctly prices abnormal accruals regarding to their one-yearahead firm earnings. Recall that, according to Xie (2001) earnings manipulation is made through abnormal accruals (also denominated discretionary accruals). 16 Recurring to the Mishkin (1983)’s test, we will be able to compare the theoretical abnormal accruals computed by the Modified Jones Model with the abnormal accruals priced by the market. By using this technique, we can conjecture if the lack of accruals persistence documented by Sloan (1996) is due to the abnormal accruals (market assessment of persistence) or to the normal accruals (historical earnings persistence). If results show that the valuation attributable to abnormal returns by the market is significantly larger than its forecasting coefficient, then we can conclude that the market is overpricing abnormal accruals. On the other hand, if the valuation of the market is smaller than the forecasting coefficient, the market is underpricing abnormal accruals. Based on the outcomes, Xie (2001) attributed any market’s mispricing as a failure to properly distinguish different levels of persistence, as the forecasting coefficient is a measure of abnormal accruals’ persistence according to Freeman et al. (1982) and Sloan (1996). 17 4. Data and Methodology To conduct our empirical analysis, we consider all public firms in Portugal and Spain, quoted in PSI Geral and Bolsa de Madrid with all available information, in the period of 1995 to 2013 5 . Both accounting variables and stock prices were extracted from the Datastream database. From the sample were excluded 1) banks, companies that belonged to the financial sector and life insurance companies since they do not have the financial information needed to calculate accruals; 2) anonymous sports societies due to their very specialized accounting systems; 3) companies that were unclassified; 4) companies quoted in another market different from the origin market and 4) companies with missing information in general or insufficient data to calculate accruals. Moreover, for the sake of consistency, we will only include in our sample firms with the same fiscal year, otherwise accounting information will be available in the market in two different moments in time. This approach is consistent with the one followed by Sloan but different from Xie’ (2001). Instead, the author preferred to adjust the sample and form 12 hedge portfolios, one for each fiscal year-end month and conduct the hedge portfolio test separately for each of the 12 fiscal year-end months. In the end, my sample is composed by 33 Portuguese companies and 93 Spanish companies, in a total of 125 firms and 2,299 firm-year observations. This study implies the formulation of several hypotheses and depends on econometric inference to achieve results. Following Sloan (1996), after defining the testable hypothesis and the sample to which our study applies, we need to compute the variables that assume particular relevance in our study – Earnings, Accruals and Cashflows. Earnings is the operating income after depreciation defined in Datastream as the “earnings of a company before interest expense and income taxes. It is calculated by taking the pretax income and adding back interest expense on debt and subtracting interest capitalized”. As Dechow et al. (1995) and Sloan (1996), we calculate accruals as follow: 5 Initially, the period of the analysis started in 1994 and ended in 2014 but do to the fewer observation in the extreme years, we decide to shorten the sample. 18 ( ) ( ) (1) where, , is the accruals in period t; , is the change in current assets in period t; , is the change in cash/cash equivalents in period t; , is the change in current liabilities in period t; , is the debt included in current liabilities in period t; , is the income taxes payable in period t; , is the depreciation and amortization expenses in period t. In this definition, it is excluded the debt in current liabilities because it is related to financing transactions and only operating transactions should be considered. Moreover, income taxes payable are also not considered to maintain the consistency with the concept of earnings that was used by the authors in the empirical tests. All the variables were extracted directly from the database with the exception of shortterm debt and depreciations since the variables were not available per se. The short-term debt was calculated as the difference between total debt and long term debt and depreciations were calculated as the difference between EBITDA (Earnings Before Interest, Taxes, Depreciations and Amortizations) and EBIT (Earnings Before Interest and Taxes). Following Sloan, cashflows were calculated as the residual part of earnings and accruals. Since the empirical analysis requires both cross-sectional and temporal comparisons (panel data), we scale the following three components by the average total assets (Sloan, 1996): 6 (2) (3) 6 T refers to the variation between the period t and t-1. 25 The Panel II shows the decomposition of accruals into its main components. We can notice that current assets and depreciation expenses increase as accruals increase, corroborating Sloan (1996)’s findings. However, current assets seem to be the ones that impact accruals the most since they assume, generally, the greater percentage on accruals. The current liabilities and current assets according to Sloan (1996) have a negative relation, results that our research also corroborates, being quite intuitive to say that growing firms usually experience high working capital requirements since this is the most inexpensive way for a firm to grow. Our data shows that firms with increasing accruals register both an increase in current assets and decrease in current liabilities, increasing this way working capital requirements. Depreciations increase linearly as we move down quartiles. To be able to identify the outliers, we use the excel function TRIMMEAN. This function calculates the mean, excluding a particular percentage of data, half from the top and half from the bottom. In our particular case, we decide to exclude 20% of the total firm sample. The number of observations to exclude was achieved by simply multiplying the number of observations by the percentage, with the largest and smallest observations being the ones that were removed. If the TRIMMEAN and the mean were equal, we did not exclude any. Testing H1 The first testable hypothesis, as it was stated in previous sections, expects to confirm the different persistence levels of earning’s components. Following Sloan (1996) 11 , we compute the following regressions: (12) (13) By estimating equation (12), we were able to assess in which magnitude the earnings of the next year are a reflection of the present earnings. However, since we also wanted to infer about the different persistence levels of earnings, we computed equation (13), disintegrating earnings in their two components – accruals and cashflows. Table 2 and 3 illustrate the impact of current earnings on future performance and the average level of earnings persistence of each component of accruals, respectively. 11 Sloan (1996) build this hypothesis following the work of Freeman et al. (1982). 26 Table 2: Results of estimating an OLS Regression to examine the impact of Current Earningsa on Future Earnings performance Coefficients St. Error T-statistic Significance Panel I: Regressions using actual values – Portugal α0 0,003 0,014 0,203 0,840 α1 0,706 0,186 3,798 0,000 Model Significance: 0,000 R2 0,093 Test α1=0 Marginal Siginificance 0,000 Panel II: Regressions using actual values – Spain α0 0,009 0,005 1,731 0,084 α1 0,683 0,040 1,697 0,000 Model Significance: 0,000 R2 0,446 Test α1=0 Marginal Siginificance 0,084 a Earnings: Income of a company before interest expense and income taxes, It is calculated by taking the pre-tax income and adding back interest expense on debt and subtracting interest capitalized. This variable was deflated by the average total assets. Overall results show that earnings are mean-reverting, with the average persistence parameter, α1, being 0,706 and 0,683 for Portugal and Spain, respectively. By testing if α1=0, the results allow us to conclude that, in both countries, the earnings performance is transitory, for a significance level of 10%, meaning that current earnings explain future earnings performance. Results presented in Table 3 show us that in both in Portugal and in Spain, < , reflecting the lower persistence of accruals regarding cashflows. In Portugal, the accruals persistence coefficient, γ1, is -0,127, against the cashflows persistence coefficient, γ2, of 27 0,009. In Spain, γ1 is 0,547 against γ2 of 0,659, respectively. Moreover, when testing if the coefficients are equal (γ1 = γ2), we reject the null hypothesis for both markets, when a significance level of 10% is considered, confirming the different persistence levels associated to the different earnings components. These findings are consistent with the ones already documented in the literature. Table 3: Results of estimating an OLS Regression to examine the impact of Cashflowsa and Accrualsb on Future Earningsc performance Coefficients St. Error T-statistic Significance Panel I: Regressions using actual values - Portugal γ0 0,040 0,010 4,052 0,000 γ1 -0,127 0,104 -1,217 0,226 γ2 0,009 0,022 0,407 0,684 Test if γ1 = γ2 Marginal Significance 0,0983 Panel II: Regressions using actual values - Spain γ0 0,007 0,005 1,403 0,162 γ1 0,547 0,046 1,202 0,000 γ2 0,659 0,042 1,572 0,000 Test if γ1 = γ2 Marginal Significance 0,001 a Cashflows: difference between Earnings (defined as earnings of a company before interest expense and income taxes. It is calculated by taking the pre-tax income and adding back interest expense on debt and subtracting interest capitalized) and Accruals. Both variables were deflated by the average total assets. c Earnings: income of a company before interest expense and income taxes. It is calculated by taking the pre-tax income and adding back interest expense on debt and subtracting interest capitalized. This variable was deflated by the average total assets. b Accruals: difference between the change in non-cash assets, the current liabilities and the depreciation expenses. This variable was deflated by the average total assets. 28 Testing H2 The second testable hypothesis tries to detect if prices reflect the different features of earnings components. Both Sloan (1996) and Xie (2001) recur to the Mishkin (1983)’s test to test the rational expectations theory, so, following their guidelines we will test together the following regressions: (14) ( ) (15) And (16) ( ) (17) By building this equation system, we can compare the results of the forecasting equations (14) and (16) with the ones obtained from the valuation equations (15) and (17), allowing us to trace the extent in which the market follows its fundamental principles. The results from estimating equations (14) and (15) are illustrated in Table 4. By testing , we seek to discover if stock prices anticipate this difference in earnings persistence performance. In other words, we are testing for efficiency of markets. In Portugal, the earning’s coefficient in the forecasting equation, α1, is 0,727 and the earning’s coefficient in the valuation equation, α1*, is 1,614. In Spain, the earning’s coefficient in the forecasting equation, α1, is 0,681 and the earning’s coefficient in the valuation equation, α1*, is -1,806. However it seemed that the market was overpricing (underpricing) earnings in Portugal (Spain), there isn’t statistical significance inference to reject H0, meaning that our results don´t classify the markets as being inefficient. This hypothesis was not also rejected by Sloan (1996) in his study. The results from estimating equations (16) and (17) are illustrated in Table 5. 29 Table 4: Results derived from a Nonlinear Weighted Least Squares estimation to analyze the information content of Current Earningsa in Future Earnings and the respectively impact on Stock Priceb performance Estimate St. Error T-statistic Significance Panel I: Regressions using actual values of financial statement variables – Portugal α0 0,001 0,014 0,061 0,952 α1 0,727 0,186 3,918 0,000 α1* 1,614 0,708 2,280 0,023 βc 1,168 0,514 2,273 0,024 Testing Market Efficiency α1 = α1* Marginal Significance 0,206 Panel II: Regressions using actual values of financial statement variables – Spain α0 0,009 0,005 1,665 0,096 α1 0,681 0,044 1,543 0,000 α1* -1,806 2,514 -0,718 0,473 β 0,664 0,650 1,021 0,308 Testing Market Efficiency α1 = α1* Marginal Significance 0,323 a Earnings: income of a company before interest expense and income taxes. It is calculated by taking the pre-tax income and adding back interest expense on debt and subtracting interest capitalized. This variable was deflated by the average total assets. b Abnormal Returns: computed as the difference between the raw buy-hold return and the return of a similar market portfolio. To calculate the returns we, per Sloan, use the adjusted price to include dividends and other liquidation distributions. The return cumulation period begins after 4 months in which the financial variables were calculated. c β: a valuation multiplier. Since we do not reject the null hypothesis of efficiency, the results shown in the table below that confirm should allow us to conclude that , meaning that the efficiency of markets should guarantee that stock prices reflect the different impact that 30 accruals and cashflows have on earnings. However, although we verify this premise to Spain, the same does not holds in Portugal 12 . Table 5: Results from estimating a Nonlinear Weighted Least Squares Regression to examine the impact of information contained in Cashflows and Accruals on Stock Price performance Estimate St. Error T-statistic Significance Panel I: Regressions using actual values of financial statement variables – Portugal γ0 0,044 0,012 3,677 0,000 γ1 -0,110 0,126 -0,874 0,383 γ2 0,008 0,027 0,297 0,767 γ1* 2,384 2,556 0,932 0,352 γ2* -0,067 0,256 -0,261 0,794 β 0,519 0,484 1,072 0,284 Testing Market Efficiency γ1 = γ1* and γ2 = γ2* Marginal Significance 0,616 Panel II: Regressions using actual values of financial statement variables – Spain γ0 0,007 0,005 1,423 0,155 γ1 0,547 0,045 1,207 0,000 γ2 0,658 0,042 1,581 0,000 γ1* -1,282 0,946 -1,356 0,176 γ2* -0,482 0,620 -0,778 0,437 β 1,302 0,613 2,124 0,034 Testing Market Efficiency γ1 = γ1* and γ2 = γ2* Likelihood ratio statistic 0,154 The variables used to compute the model were defined/ calculated as follow: - Earnings: income of a company before interest expense and income taxes. It is calculated by taking the pre-tax income and adding back interest expense on debt and subtracting interest capitalized. This variable was deflated by the average total assets. - Accruals: difference between the change in non-cash assets, the current liabilities and the depreciation expenses. This variable was deflated by the average total assets. 12 The results, once again, do not reject the null hypothesis of market efficiency. We believe that even when markets are efficient, there are moments in time that create arbitrage opportunities for the existence of the anomaly. However, is that efficiency that will correct those momentaneous opportunities and erode the abnormal returns. 31 - Cashflows: difference between Earnings and Accruals (both deflated by the average total assets). - Abnormal Returns: computed as the difference between the raw buy-hold return and the return of a similar market portfolio. To calculate the returns we, per Sloan, use the adjusted price to include dividends and other liquidation distributions. The return cumulation period begins after 4 months in which the financial variables were calculated. Similarly to Sloan’s findings in US, Portuguese and Spanish investors do not seem to fixate on earnings since are different from α1. In fact, the Portuguese market attributes a higher persistence to accruals than cashflows, contrary to Spain, with investors treating accruals as a more persistent component of earnings that in fact it is. This means that, in Portugal, the market do not appear to anticipate the lower (higher) persistence in earnings performance that is associated to accruals (cashflows). Consequently, these findings allow us to speculate about the existence of the anomaly in Portugal, prediction that we will be able to confirm by testing H3. Testing H3 To test our third hypothesis, we rank firms according to their accrual magnitude to be able to build equally weighted portfolios (quartiles). By computing the annual buy-hold return for each of the three subsequent years we were able to assess if a strategy that combines a long position in the lowest accruals portfolio with a short position in the highest accruals portfolio yields abnormal returns. The results are presented in Table 6. According to Schwert (2003), “anomalies are empirical results that seem to be inconsistent with maintained theories of asset pricing behavior”, (Schwert, 2003, p. 939). On one hand, researchers look at this phenomenon as a way to challenge literature and principles already defended. On the other, investors see anomalies as an opportunity to profit from a market flaw. So, analyzing the investors side and following Sloan, we tested if the strategy described before yields, indeed, abnormal returns. Panel I evidence decreasing marginal returns in Portugal, with this strategy yielding abnormal returns only in the first year of the strategy (20,4%, -1,1%% and -10,2%). Contrary, in Spain, we observe negative returns in the three periods in study, with these returns being less negative as the years pass by (-26,1%, -12,7% and -1,9%). Therefore, the 32 negative relation between accruals and stock returns – the accruals anomaly – is only present in Portugal (and not in Spain) 13 , as results in H2 suggested. Table 6: Abnormal Size-Adjusted Returnsa of Equal Weighted Accrualsb Portfolios Time Series Portugal Spain t+1 t+2 t+3 t+1 t+2 t+3 Panel I: Abnormal returns covering all period from 1996 to 2013 Q1 0,040 0,012 -0,024 -0,032 -0,139 -0,031 Q4 -0,163 0,023 0,078 0,230 -0,012 -0,013 Hedgec 0,204 -0,011 -0,102 -0,261 -0,127 -0,019 Panel II : Abnormal returns from 2000 (inclusive) to 2006 Q1 0,028 -0,038 -0,162 -0,092 -0,103 -0,196 Q4 -0,090 -0,168 0,150 0,268 -0,046 -0,027 Hedge 0,118 0,130 -0,313 -0,360 -0,057 -0,169 Panel III: Abnormal returns from 2007 (inclusive) to 2013 Q1 0,092 -0,177 -0,107 -0,003 -0,162 0,113 Q4 -0,005 0,059 0,080 -0,033 -0,032 0,151 Hedge 0,096 -0,236 -0,187 0,030 -0,131 -0,038 a Size-Adjusted Returns: the returns were computed as the difference between the raw buy-hold return of the portfolio and the raw buy-hold return of the correspondent size-matched portfolio. The size matched portfolio was created by ranking firms by its market value at the beginning of the year and choosing the observation immediately before as a proxy for the security held in accruals portfolio. b Accruals: difference between the change in non-cash assets, the current liabilities and the depreciation expenses. Accruals were deflated by the Average Total Assets. c Hedge: the hedge strategy consists in adopt a long position in the lowest accruals portfolio with a short position in the highest accruals portfolio. 13 This conclusion takes in consideration all sample. 33 To investigate if the final crisis of 2007 was somehow responsible to spread or mitigate the anomaly, we divide de sample in two periods – before and after/during the financial crisis, considering the returns of the past 12 years. Panel II and Panel III report those results. Prior to the 2007 financial crisis, this strategy, in Portugal, yielded positive returns in the t+1 and t+2 (11,8%, 13,0% and -31,3%), with the return of first year being relative smaller and the return of second year being considerably higher than the ones we verified before (when we analyze all sample). In Spain, we observe that the returns for this period are, similarly to before, negative for all the years in consideration (-36,0%, -5,7% and -16,9%), with this particular time period explaining most of the results obtained in the Panel I. Post 2007, the returns achieved in Portugal decrease considerably, being positive only in the first year (9,6%, -23,6%, -18,7%). In Spain, surprisingly, we notice that this strategy yields positive returns in the first year of the strategy (3,01%) raising us some questions. To be able to understand if the positive results mean that the anomaly is appearing in recent years or not we divide this sample period in two: from 2007 to 2009 and from 2010 to 2013. Non documented results show us that the strategy yielded positive returns from 2007 to 2009, in t+1 and t+2, registering only losses in the t+3 (8,3%, 1,6% and -2,8%). From 2010 to 2013, the returns are negative in the three years (-0,9%, -27,7% and -5,1%), meaning that there’s no evidence of the accrual anomaly in Spain in recent years. The results allow us to conclude that in countries, like Portugal, where the anomaly is present, the 2007 financial crisis had a great impact in the magnitude of the returns earned by investors. In countries like Spain, where we did not find evidence of the anomaly 14 , the strategy cannot be applied and yield abnormal returns. We believe that the existence of the accruals anomaly is connected to the market attributing a higher persistence to accruals than cashflows. Recall that H2 confirmed these results for Portugal contrary to Spain. However, it is important to be aware that the (non) existence of the anomaly can also be a reflection of all of the problems that calculating the returns of this particular strategy bear. Kothari (2001) highlights the extreme skewness of long-term abnormal returns, the survival biases and inference problems arising from cross14 Since we only find positive abnormal returns in the period from 2007 to 2009, we believe that the anomaly is not present in this market. In fact, these positive returns can be connected to other factors extra-anomaly. 34 sectional dependence as the main difficulties faced by anyone trying to compute accruals’ anomaly abnormal returns. Regarding the anomaly as a consequence of the legal system of the country, our results point that the law of the country has weaker explication power in the appearance of the anomaly since both countries are code law countries and have both French origin. Agreeing is LaFond (2005), stating that the anomaly in neither connected to the legal framework the country is inserted to, neither with investor protection or the dependence on accruals accounting. A recent study of Goncharov et al. (2013) concluded that Spanish accruals are not associated with greater insider returns or income predictability, since the country hasn’t a flexible accounting system to record accruals. Moreover, Ball et al. (2000, 2003) argued that, in Spain, information flows easily between small stakeholders. Therefore, it not surprising that we did not found the anomaly in this market. Testing H4 To test if the market prices cashflows and accruals components correctly regarding their one-year-ahead firm earnings, we follow Xie (2001)’s guidelines and we estimate the following regressions using an iterative nonlinear weighted least squares estimation procedure: (18) ( ) (19) where, , is the earnings in the period t+1; , is the cashflows from operations in period t; , is the normal accrual component in period t; , is the abnormal accrual component in period t, , is the sized-adjusted return in period t+1. 41 7. References Alford, A. W. et al. (1994), “Extensions and Violations of the Statutory SEC Form 10-K Filing Requirements”, Journal of Accounting & Economics, Vol. 17(1-2), pp. 229254. Allen, E. J. et al. (2013), “Accrual reversals, earnings and stock returns”, Journal of Accounting & Economics, Vol. 56(1), pp. 113-129. Baber, W. R. et al. (2006), "Stock price reaction to evidence of earnings management: Implications for supplementary financial disclosure", Review of Accounting Studies, Vol. 11(1), pp. 5-19. Ball, R. and P. Brown (1968), “An Empirical Evaluation of Accounting Numbers”, Journal of Accounting Research, pp. 159-178. Ball, R. et al. (2000), “The effect of international institutional factors on properties of accounting earnings”, Journal of Accounting and Economics, Vol. 29, pp. 1–51. Ball, R., A. Robin and J. S. Wu (2003), “Incentives versus standards: properties of accounting income in four East Asian countries”, Journal of Accounting and Economics, Vol. 36, pp. 235–270. Balsam, S. et al. (2002), "Accruals management, investor sophistication, and equity valuation: Evidence from 10-Q filings", Journal of Accounting Research, Vol. 40(4), pp. 987-1012. Barth, M. E. and A. P. Hutton (2004), "Analyst earnings forecast revisions and the pricing of accruals", Review of Accounting Studies, Vol. 9(1), pp. 59-96. Beneish, M. and M. Vargus (2002), "Insider trading, earnings quality, and accrual mispricing", Accounting Review, Vol. 77(4), pp. 755-791. Bernard, V. and J. Thomas (1989, 1990), “Post-Earnings-Announcement Drift: Delayed Price Response or Risk Premium?, Journal of Accounting Research, Vol. 27, Current Studies on the Information Content of Accounting Earnings, pp. 1-36. Bernstein, L. (1993), Financial Statement Analysis, 5th Edition, Homewood, IL:Irwin. Bradshaw, M. T. et al. (2001), "Do analysts and auditors use information in accruals?" Journal of Accounting Research, Vol. 39(1), pp. 45-74. Chan, K. et al. (2006), “Earnings Quality and Stock Returns”, The Journal of Business, Vol. 79, pp. 1041-1082. 42 Collins, D. and P. Hribar (2000), "Earnings-based and accrual-based market anomalies: one effect or two?", Journal of Accounting & Economics, Vol. 29(1), pp. 101-123. Collins, D. W., et al. (2003), "Investor sophistication and the mispricing of accruals", Review of Accounting Studies, Vol. 8(2-3), pp. 251-276. Dechow, P. M et al. (2011a), “Predicting Material Accounting Misstatements”, Contemporary Accounting Research, Vol. 28(1), pp. 17-82. Dechow, P. M. and W. Ge (2006), "The persistence of earnings and cash flows and the role of special items: Implications for the accrual anomaly.", Review of Accounting Studies, Vol. 11(2-3), pp. 253-296. Dechow, P. M. (1994), "Accounting earnings and cash flows as measures of firm performance: The role of accounting accruals", Journal of Accounting and Economics, Vol. 18(1), pp. 3-42. Dechow, P. M. and I. D. Dichev (2002), "The quality of accruals and earnings: The role of accrual estimation errors.", Accounting Review, Vol. 77, pp 35-59. Dechow, P. M., et al. (1995), "Detecting Earnings Management", Accounting Review, Vol. 70(2), pp. 193-225. Dechow, P. M. et al. (2011b), “The Accrual Anomaly”, Working Paper, University of California, Berkeley. Desai, H. M., et al. (2004), "Value-glamour and accruals mispricing: One anomaly or two?", Accounting Review, Vol. 79(2), pp. 355-385. Fairfield, P. M. et al. (2003), "Accrued earnings and growth: Implications for future profitability and market mispricing.", Accounting Review, Vol. 78(1), pp. 353-371. Francis, J. et al. (2005), "The market pricing of accruals quality", Journal of Accounting & Economics, Vol. 39(2), pp. 295-327. Freeman, R., et al. (1982), “Book rate-of-return and predictions of earnings changes: an empirical investigation”, Journal of Accounting Research, Vol. 20, No 2, pp. 639 – 653. Goncharov, I. et al. (2013), “Asymmetric trading by insiders comparing abnormal returns and earnings prediction in Spain and Australia”, Accounting and Finance, Vol. 53(1), pp. 163-184. Healy, P. (1985), “The effect of bonus schemes on accounting decisions”, Journal of Accounting and Economics, Vol. 7, pp. 85-107. 43 Hirshleifer, D. et al. (2004), “Do investors overvalue firms with bloated balance sheets?”, Journal of Accounting and Economics, Vol. 38, pp. 297–331. Hribar, P. and D. W. Collins (2002), "Errors in estimating accruals: Implications for empirical research", Journal of Accounting Research, Vol. 40(1), pp. 105-134. Jones, J. (1991), “Earnings Management During Import Relief Investigations”, Journal of Accounting Research, Vol. 29, pp. 193-228. Kang, Q. et al. (2006), “Predicting stock market returns with aggregate (discretionary) accruals”, EFA 2006 Zurich Meetings Paper. Khan, M. (2005), "Are Accruals Really Mispriced? Evidence from Tests of an Intertemporal Capital Asset Pricing Model", Toronto. Knez, P. J. and M. J. Ready (1997), “On The Robustness of Size and Book-to-Market in Cross-Sectional Regressions”, The Journal of Finance, Vol. 52, pp. 1355–1382. Kothari, S. (2001), “Capital markets research in accounting”, Journal of Accounting and Economics, Vol. 31, pp. 105–231. Kraft, A. et al. (2006), "An analysis of the theories and explanations offered for the mispricing of accruals and accrual components", Journal of Accounting Research, Vol. 44(2), pp. 297-339. Kraft, A. et al. (2007), “Regression-Based Tests of the Market Pricing of Accounting Numbers: The Mishkin Test and Ordinary Least Squares”, Journal of Accounting Research, Vol. 45 (5), pp. 1081-1114. LaFond, R. (2005), “Is the Accrual Anomaly a Global Anomaly?”, MIT Sloan Research Paper No. 4555-05. Leippold, M. and H. Lohre (2010), “Data Snooping and the Global Accrual Anomaly”, EFA 2007 Ljubljana Meetings Paper. Leuz, C. et al. (2002), “Earnings Management and Investor Protection: An International Comparison”, Journal of Financial Economics Vol. 69, pp. 505-527. Mishkin, F. (1983), “A Rational Expectations Approach to Macroeconometrics: Testing Policy Effectiveness and Efficient-markets Models”, Chicago, IL: University of Chicago Press. Papanastasopoulos, G. (2013), “Accounting Accruals and Stock Returns: Evidence from European Equity Markets”, European Accounting Review, Vol. 23, pp. 729-768. Pincus, M. et al. (2007), "The accrual anomaly: International evidence.", Accounting 44 Review, Vol. 82(1), pp. 169-203. Pope, P. (2001), “Discussion of The Relation Between Incremental Subsidiary Earnings and Future Stock Returns in Japan”, Journal of Business Finance & Accounting, Vol. 28(9-10), pp. 1141-1148. Richardson, S. A. et al. (2005), "Accrual reliability, earnings persistence and stock prices", Journal of Accounting & Economics, Vol. 39(3), pp. 437-485. Richardson, S. et al. (2005), 'Accrual reliability, earnings persistence and stock prices', Journal of Accounting & Economics, Vol. 39, No. 3, pp. 437-485. Richardson, S. et al. (2006), “The Implications of Accounting Distortions and Growth for Accruals and Profitability”, The Accounting Review, Vol. 81, No. 3, pp. 713-743. Soares, N. and A. Stark (2011), “Is there an Accruals or Cashflow Anomaly in UK Stock Returns?”, CEFUP, pp. 1-37. Schwert, G. (2003), Anomalies and Market Efficiency, Handbook of the Economics of Finance, Amsterdam: Elsevier Science B.V. Shon, J. J. and P. Zhou (2005), "Does the accruals anomaly result from the market's mispricing of growth in net operating assets?", American Accounting Association Conference. Sloan, R. G. (1996), "Do Stock Prices Fully Reflect Information in Accruals and Cash Flows about Future Earnings?", The Accounting Review, Vol. 71(3), pp. 289-315. Thomas, J. and H. Zhang (2002), “Inventory Changes and Future Returns”, Review of Accounting Studies, Vol. 7, No. 2, pp. 163-187. Vale, Ivo (2013), “The Accruals Anomaly: Evidência do Caso Português”, Master Dissertation, FEP, University of Porto. Wu, J. et al. (2007), “Understanding the Accrual Anomaly”, Working Paper No. 13525, Cambridge University. Xie, H. (2001), "The mispricing of abnormal accruals", Accounting Review, Vol. 76(3), pp. 357-373. Zach, T. (2003), “Inside the Accrual Anomaly”, Olin School of Business, Washington University in St. Louis. Zhang, X. F. (2007), “Accruals, investment, and the accrual anomaly”, Accounting Review, Vol. 82(5), pp. 1333-1363. 45 8. Annexes A. Variables The variables used in our study are described in the table below. Variable Code Datastream Description Cash (WC02003) Represents money available for use in the normal operations of the company. It is the most liquid of all of the company's assets. Current Assets (Total) (WC02201) Represents cash and other assets that are reasonably expected to be realized in cash, sold or consumed within one year or one operating cycle. Generally, it is the sum of cash and equivalents, receivables, inventories, prepaid expenses and other current assets. For non-U.S. corporations, long term receivables are excluded from current assets even though included in net receivables. Current Liabilities (Total) (WC03101) Represent debt or other obligations that the company expects to satisfy within one year. Earnings Before Interest and Taxes (WC18191) Represent the earnings of a company before interest expense and income taxes. It is calculated by taking the pre-tax income and adding back interest expense on debt and subtracting interest capitalized. Long Term Debt (WC03251) Represents all interest bearing financial obligations, excluding amounts due within one year. It is shown net of premium or discount. Income Taxes Payable (WC03063) Represents an accrued tax liability which is due within the normal operating cycle of the company. Market Value For Company (MVC) MVC is the consolidated market value of a company displayed in millions of units of local currency. MVC is the same value as MV for companies with a single listed equity security. However, for companies with more than one listed or unlisted equity security MVC represents: EQUITY A(MV) + EQUITY B(MV) + EQUITY C(MV) etc . Net Receivables (WC02051) Represent the amounts due to the company resulting from the sale of goods and services on credit to customers (after applicable reserves). These assets should reasonably be expected to be collected within a year or within the normal operating cycle of a business. Price (P) The ‘current’ price on Datastream’s equity programs is the latest price available to us from the appropriate market in primary units of currency (except in the case of the UK where price is given in pence). It is the previous day’s closing price from the default exchange except 46 where more recent or real-time prices are available. The ‘current’ prices taken at the close of market are stored each day. These stored prices are adjusted for subsequent capital actions, and this adjusted figure then becomes the default price offered on all Research programs. The actual historical prices can be accessed using the unadjusted price datatype (UP). Prices are generally based on ‘last trade’ or an official price fixing. For stocks which are listed on more than one exchange within a country, default prices are taken from the primary exchange of that country (note that this is not necessarily the ‘home’ exchange of the stock). For Japan and Germany , prices from the secondary markets can be obtained by qualifying the price datatype with an exchange code (see below for details). Property, Plant and Equipment (WC02301) represents tangible assets with an expected useful life of over one year which are expected to be used to produce goods for sale or for distribution of services. Revenues (WC01001) Represent gross sales and other operating revenue less discounts, returns and allowances. Total Assets (WC02999) Represent the sum of total current assets, long term receivables, investment in unconsolidated subsidiaries, other investments, net property plant and equipment and other assets. Total Debt (WC03255) Represents all interest bearing and capitalized lease obligations. It is the sum of long and short term debt. Note: WC stands for Worldscope.