The Effect of Abnormal Accruals on Stock Price Performance: Evidence from the Johannesburg Securities Exchange A Research Report presented to The Graduate School of Business University of Cape Town In partial fulfilment of the requirements for the Masters of Business Administration Degree By Pyi S. Maung December 2013 Supervised by: Chipo Mlambo 1 1
Plagiarism Declaration I know that plagiarism is wrong. Plagiarism is to use another s work and pretend that it is one s own. I have used a recognised convention for citation and referencing. Each significant contribution and quotation from the works of other people has been attributed, cited and referenced. I certify that this submission is all my own work. I have not allowed and will not allow anyone to copy this research project with the intention of passing it off as his or her own work. SIGNED: 11 December 2013 Pyi Maung 2 2
Abstract This study looks at the potential existence of the abnormal accrual anomaly by forming an accrual based portfolio using the Sloan (1996) hedge portfolio and the Hafzalla et al (2010) percent accruals method in scaling accruals. These two methods were used to formulate two different portfolios with differing long and short positions of a set group of 32 stocks that are listed on the JSE and part of the ALSI Top 40. The study found that the Sloan hedge portfolio failed to achieve positive significant returns in four of the five investment periods and the negative returns were significantly different to the benchmark returns. The Hafzalla et al percent accruals portfolio managed to achieve positive returns in all five of the investment periods but only outperformed the benchmark in two investment periods. However, this may be enough to indicate that the accrual anomaly may exist on the JSE and that it can be exploited by using this particular hedge portfolio method going forward. 3 3
Table of Contents ABSTRACT... 3 ACKNOWLEDGEMENT... 5 1. INTRODUCTION... 6 1.1 OBJECTIVES OF THE STUDY... 7 1.2 RESEARCH QUESTIONS AND SCOPE... 8 1.3 RESEARCH ASSUMPTIONS AND ETHICS... 9 2. LITERATURE REVIEW... 10 2.1 THE ABNORMAL ACCRUAL ANOMALY... 10 2.2 OVERVIEW OF THE JOHANNESBURG SECURITIES EXCHANGE... 12 2.3 ABNORMAL RETURNS ON THE JSE... 13 2.4 THE ABNORMAL ACCRUAL ANOMALY ON THE JSE... 15 2.5 ACCRUALS AND STOCK PRICE RETURNS... 17 2.6 CASH FLOWS VERSUS ACCRUALS... 19 2.7 DISCRETIONARY AND PERCENT ACCRUALS... 20 2.8 CONCLUSION... 22 3. RESEARCH METHODOLOGY... 23 3.1 RESEARCH APPROACH AND STRATEGY... 23 3.2 RESEARCH DESIGN, DATA COLLECTION METHODS AND RESEARCH INSTRUMENTS... 24 3.3 SAMPLING... 27 3.4 DATA ANALYSIS METHODS... 29 4. RESEARCH FINDINGS, ANALYSIS AND DISCUSSION... 30 4.1 RESEARCH FINDINGS... 30 4.2 RESEARCH ANALYSIS AND DISCUSSION... 32 4.3 RESEARCH LIMITATIONS... 35 5. RESEARCH CONCLUSIONS... 36 6. FUTURE RESEARCH DIRECTIONS... 37 7. APPENDICES... 38 APPENDIX 1: TRADITIONAL OPERATING ACCRUALS PORTFOLIOS FROM 2008 TO 2013... 38 APPENDIX 2: PERCENT OPERATING ACCRUALS PORTFOLIOS FROM 2008 TO 2013... 43 8. REFERENCES... 48 4 4
Acknowledgement This report is not confidential and may be used freely by the Graduate School of Business. I would particularly like to thank my supervisor Dr Chipo Mlambo for her invaluable help and guidance in the development of this report. Thanks also to Mr Mohau Matlala and Mr Shwetank Rastogi for providing much of the data required for this report and for their invaluable advice. A special word of thanks to my family and my girlfriend Tarryn, for their unwavering support throughout this year and whilst working on this research paper. 5 5
1. Introduction Investors are fixated on current earnings and base their investment decisions on the current performance of a particular stock regardless of the composition of the stocks earnings. There is a wealth of literature on the subject of mispricing and the role accruals and cash flows in current earnings [Sloan (1996), Collins and Hribar (2000), Xie (2001)]. The accrual anomaly was first documented by Sloan (1996) and arises as a result of investors failing to price in the differing persistence of accruals and cash flows. Sloan (1996) states that high accruals in current earnings are negatively related to future stock price returns, and high cash flows in current earnings are positively correlated with future stock price returns. Therefore, a stock with current earnings that have a higher accrual component than a cash flow component will have lower future returns, and vice versa. Sloan (1996) also found that accruals show mean reversion quicker than cash flows do, which meant that earnings comprised of high accruals were essentially unsustainable. In addition, Bradshaw, Richardson and Sloan (2001) and Barth and Hutton (2004), all found that accruals are indeed less persistent than cash flows. Therefore, it follows that a large stock price appreciation in a particular period may not indicate a similar appreciation in the following year. Hence, the appreciation could potentially be a result of high accruals, which would in turn lead to a decrease in earnings in the following period.in addition, Houge and Lougran (2000) find that investors tend to commit a cognitive error because they fail to fully value the information contained within cash flows and accruals, and these investors also tend to overreact to accrual earnings. In essence investors fail to recognize the significance of the different components of earnings (net income) and their specific impact on stock price performance.wilson (1986) finds that the general consensus in the market is that stock value is a function of discounted future cash flows. However, there has been some debate as to the usefulness of the accruals component of earnings in determining stock values. 6 6
Following on from his observations Sloan (1996) found that by using a low-risk hedge strategy that simultaneously purchases stocks with low accruals/ high cash flows, and sells short stocks with high accruals/ low cash flows, a significant positive annual abnormal hedge return can be realised by capturing the mispricing and earning a return of around 10% in the following year. The strategy is based on the assumption that investors react negatively when earnings are lower than expected, and positively when earnings are higher than expected. However, Green, Hand and Soliman (2009) argue that the accrual anomaly has decayed over time on US stock markets and the returns from this strategy are no longer positive. This is an interesting study as it confirms the existence of the anomaly and the fact that positive returns were being generated for some time, this in turn opens the door to the possible existence and success of such a strategy to be employed on the Johannesburg Securities Exchange (JSE). 1.1 Objectives of the Study Copyright The objectives can be stated as follows: UCT To investigate the possible existence of the accrual anomaly on the JSE by looking at measure of accruals and their relation to stock price performance. To determine whether or not an accruals-based (percent accruals and / or traditional accruals) strategy yields significant returns on the JSE. The Hafzalla, Lundholm and Van Winkle (2010) percent accruals method will be compared to the traditional operating accruals measure in order to scale stocks and form portfolios that aim to achieve a potential return on the JSE. 7 7
1.2 Research Questions and Scope Does the JSE exhibit an accrual anomaly as evidence by traditional and percent operating accrual measures? Does the Hafzalla et al (2010) percent accruals strategy outperform the traditional operating accruals based strategy of Sloan (1996), and the JSE benchmark? Could an accruals-based strategy, based on the Hafzalla et al (2010) percent operating accruals measure, yield significant returns on the JSE? This is the first study investigating the existence of the accrual anomaly in South Africa and the subsequent implications of such an anomaly on the JSE may prove to be interesting. The findings of this research may be of importance to investors, decision makers, portfolio managers, analysts and other stock market players as it shows a different perspective to stock price performance evaluation, and may assist in a better understanding of the components of current earnings. This study also aims to contribute to the existing literature and add a different perspective on the existence of the accrual anomaly in an emerging market. 8 8
1.3 Research Assumptions and Ethics The research makes the following assumptions: The Sloan (1996) study is accurate, regardless of the fact that Sloan used the indirect balance sheet method to calculate accruals. o Hribar and Collins (2002) found previous literature calculated accruals by using balance sheet [(i.e. Sloan (1996)], this contaminated the calculation of the accruals. The JSE is weak-form efficient in terms of EMH, the research by Mabhunu (2004) supports this view, and this dictates that investors cannot consistently outperform the market by using historical price data. The findings of Pincus, Rajgopal and Venkatachalam (2007) and Koerniadi and Tourani-Rad (2007) are correct in terms of the claim that the accrual anomaly is more likely to occur in countries that possess certain institutional and accounting structures. o From this perspective South Africa could potentially be one of the countries where the accrual anomaly is more likely to occur. The cash flow method used to identify accruals is the best method that will yield the least amount of selection errors and bias. o There may very well be a better method but for the purposes of this study we assume the statement of cash flows contains the best information of accruals. The abnormal accrual strategy is formed freely, without taking transaction costs into account and it is instantaneous, without a time lag/ delay. o Transactions costs could potentially eliminate any gains from the hedge strategy; the same can apply for timing delays or lags in the market processing buy/sell orders at portfolio rebalancing dates. Due to the nature of the research being predominantly quantitative, the possible ethical implications are limited. 9 9
2. Literature Review 2.1 The Abnormal Accrual Anomaly Market anomalies are defined as; a result that is inconsistent with asset pricing theories, Schwert (2003). Abnormal accruals can be categorised as such an anomaly, and as previously mentioned Sloan (1996) was the first to document it. Sloan (1996) found that (high) accruals are negatively correlated with future stock returns and that the predictability of stock returns is correlated to the differences in the persistence of the cash flow and accrual components of earnings. Collins and Hribar (2000) found evidence that support Sloan s (1996) accrual anomaly, they found that the market tends to overreact on earnings (net income) that has a large accrual component, and this overreaction is subsequently corrected in the following year as the market learns that the earnings are in fact unsustainable. As a result of this finding, there is a possibility to take advantage of this mispricing in the market and to generate an abnormal hedge return. Bradshaw et al (2001) also found evidence that showed how firms that have unusually high working capital accruals were more likely to experience a decline in their future earnings performance.this is also indicative of a market anomaly or an overreaction, and may also point towards an aspect of earnings management. Goodman (2007) supports this view as he found that future cash flows were likely to increase if accruals are below an expected value that is predicted by a model, and vice versa. Richardson, Sloan, Soliman and Tuna (2004) found that accruals are associated with an even greater mispricing than what was stated by Sloan (1996), and they show the magnitude of this security mispricing with regards to accruals. However, this evidence is subject to the specific market, because a particular market could potentially overreact more than another one, it may also depend on market size as this is likely to affect the magnitude of the mispricing and the extent of its impact. 10 10
Interestingly, there is a wealth of research that does not necessarily agree with Sloan (1996). Zach (2004) argues that the return pattern found in the abnormal accrual anomaly is a result of the correlation between firm specific characteristics and accruals, Kraft, Leone and Wasley (2006) claim that the anomaly is due to a look ahead bias and outliers, Khan (2008) claims the accrual anomaly can be explained by a miss-specified risk model and Wu, Zhang and Zhang (2010) believe that it can be explained by the q-theory of time varying discount rates. Kothari, Loutskina and Nikolaev (2006) argue that the agency theory / hypothesis is an explanation for the accrual anomaly, they claim that under this hypothesis the accrual anomaly is in fact a byproduct of overvaluation and accruals are not the cause of mispricing. Desai, Rajgopal, and Venkatachalam (2004) investigated the possibility of the accrual anomaly being nothing more than a value-glamour anomaly based on glamour stocks performing well. However, they only discovered a partial match between the two anomalies and this in turn means that the accrual anomaly does in fact exist from their point of view. Mashruwala, Rajgopal and Shevlin (2006) do not dispute the existence of the accrual anomaly however they claim that investors cannot fully implement the strategy as they are limited by factors such as the presence of arbitrage risk. Lev and Nissim (2006) in their study found that the accrual anomaly still exists, they further claim that its magnitude has not declined over time and that this effect will last for quite some time, but they agree with Mashruwala et al (2006) in stating that investors will be limited by factors such as transaction costs. Additionally, both studies found evidence that the accrual anomaly in the US is concentrated in companies stocks that are likely to have high transactions costs. Green et al (2009) find that the accrual anomaly has diminished over time in the US, because the anomaly has been exploited by prominent hedge funds that have hired researchers away from academia in order to work for the funds. They believe that the hedge funds have the resources and they are the most likely investors to target market inefficiencies such as the accrual anomaly. As a result of this they claim that the hedge returns to Sloan s (1996) anomaly are currently not positive, however Green et al (2009) do not predict whether or not the returns will be insignificant for the foreseeable future. 11 11
2.2 Overview of the Johannesburg Securities Exchange The JSE was established in 1887 as a stock exchange and is licensed under the Securities Services Act of 2004. It is the only stock exchange in South Africa and has been operating as a place for trading of financial products for over 125 years. And it is the oldest exchange in Africa. The JSE was the 19 th largest exchange by market capitalisation as of 31 May 2013, with a total market capitalisation of US$ 895,545 million. As of June 2013 there were 394 companies listed on the JSE, 340 of which were domestic listings and 54 of which were foreign listings, (Johannesburg Securities Exchange, 2013). Jefferis and Smith (2004) claim that the JSE accounts for 79% of total African stock market capitalisation, they also find that the historical liquidity of the JSE is due to the large stock ownership by a small number of large conglomerate companies which are mainly mining houses. Gilbertson and Goldberg (1981) find that the mining and industrial sectors of the JSE make up a significant portion of the market capitalisation, they claim that the factors affecting these sectors can be very different over time and need to be taken into account when analysing stock price movements. In addition to this, they also state that the JSE is unlike other international markets because the mining companies form a large part of the total market capitalisation. The main point of interest for this research is the FTSE/ JSE Top 40 Index because the data used to analyse the existence of the accrual anomaly will be drawn from this index. The Top 40 Index consists of the 40 largest companies listed on the exchange, ranked by full market value. It is a subset of the FTSE/ JSE All Share Index. The Top 40 Index reflects the performance of the South African stock market, and the companies in this index are generally large companies that have widely traded and marketable stocks. The movement of the stocks in this index represent the movement of the market as a whole. A wide range of industries such as, telecommunications, mining and banking are included in this index, (Johannesburg Securities Exchange, 2013). 12 12
2.3 Abnormal Returns on the JSE Fama, Fisher, Jensen and Roll (1969) state that under the Efficient Market Hypothesis (EMH) there are three forms of market efficiency; weak form, semi-strong form and strong form, in addition to this Fama et al (1969) defines market efficiency as the extent to which the available information in a market is reflected into stock prices, at the same time adjusting independently and rapidly to new information. Mabhunu (2004) found the JSE to be weak form in terms of EMH but nevertheless found that the JSE was more efficient than other African markets. Rono (2013) finds that the JSE is efficient in terms of new information releases to the market, and in addition to this the study showed that the market was able to adjust instantly and without bias to earnings announcements. Khaneman and Tversky (1973) found evidence of a behavioural bias where investors would overweight recent data when making forecasts, Page and Way (1992) discovered the existence of investor overreaction on the JSE and there have been many studies since that confirm the existence of such an overreaction. Hsieh and Hodnett (2011) define an overreaction as a situation where stock prices systematically overshoot as a result of the investors in the market, and they claim that the reversals of stock prices should be predictable, therefore opening the door for arbitrage investment strategies. Cubbin, Eidne, Firer and Gilbert (2006) state that market efficiency is essentially challenged by the occurrence of mean reverting returns because this indicates that future returns could potentially be determined by using historical prices. They further state that this phenomenon has been observed in international markets and this is due mainly to the bias involved in the investors decision making process. The Cubbin et al (2006) findings were consistent with Page and Way (1992) as they found the existence of mean reversion of stock returns, essentially indicating that the value approach to investing by taking a contrarian view could yield positive results on the JSE.Hsieh and Hodnett (2011) find that investor overreaction may be attributable to an overweighting on recent or new information and an underweighting of long-term fundamental information by irrational investors, essentially investors ignore the more important information in favour of the more recent. 13 13
Banz (1981) was the first to note the market capitalisation or size effect, he investigated this on the NYSE, essentially this effect states that small market capitalisation stocks outperform large market capitalisation stocks, after adjusting for risk. However, Waelkens and Ward (1997) find that large firms may provide superior performance on the JSE, when compared to small firms. Auret and Cline (2011) also agree with the above as they found no indication of a size effect or a small firm effect on the JSE. In addition to this, Jefferis and Smith (2004) discovered that large capitalisation indices are more liquid than small capitalisation indices on the JSE. De Villiers, Lowing and Pettits (1984) suggest that possible reasons for this phenomenon lie in the fact that the JSE is dominated by institutional investors, the inability of institutions to invest abroad, the extensive cross holdings and the relatively low turnover rate when compared to other exchanges. The inability of institutions to invest abroad is due to exchange controls on capital flows, Jefferis and Smith (2004) find that this restriction on companies and institutional investors has essentially kept capital inside South Africa. However, Bailey and Gilbert (2007) find that liquidity plays a role in the mean reversion of returns on the JSE, they find that only illiquid stocks (low P/E) experience mean reversion and therefore the abnormal returns from mean reversion is effectively eliminated when these stocks are excluded. In addition to this Bailey and Gilbert (2007) find that the easiest way to earn abnormal returns from mean reversion in liquid stocks is to short them, this shows that arbitrage opportunities may exist for high P/E liquid stocks and large investors. 14 14
2.4 The Abnormal Accrual Anomaly on the JSE Koerniadi and Tourani-Rad (2007) found evidence that suggests that the accrual anomaly is a contradiction to the efficient market hypothesis and market efficiency as a whole. Therefore, the existence of the accrual anomaly may dictate a less than efficient market and vice versa. This evidence alludes to the possibility of an accrual anomaly existing on the JSE, due to the nature of the JSE market as mentioned above and evidenced by the findings of Koerniadi and Tourani-Rad (2007). Pasaribu (2009) found that an explanation for the accrual anomaly could be based on the underlying assumption that the market is inefficient because investors fail to classify the differences between the persistence of accruals and cash flows, this is in line with the findings of Sloan (1996) and Hsieh and Hodnett (2011). As a result of the lack of information investors generally do not have the ability to price in the full effects of accruals and cash flows, the two components of earnings, on a particular stock. Also, the excessive optimism of investors in the market leads to the investment in high accrual stocks because they are more attractive due to the fact that they seem to have higher earnings (net income), and by doing so these investors are essentially driving up the price of these high accrual stocks. Hoffman (2012) further substantiates Koerniadi and Tourani-Rad s (2007) view in his study on the Johannesburg Securities Exchange, where he found that stock return anomalies do exist on the JSE and that this provides evidence that the EMH cannot be accepted in strong or semistrong form, when referring to the JSE. Pincus, Rajgopal and Venkatachalam (2007) claim that the accrual anomaly is more likely to occur in countries that possess certain institutional and accounting structures, namely a tradition of common law, low concentration of stock ownership and extensive use of accrual accounting. Koerniadi and Tourani-Rad (2007) agree with the above and in addition found that the occurrence of the accrual anomaly is also correlated with extensive use of accruals accounting and weak shareholder protectionisms. Pincus et al (2007) sampled 20 countries within which they discovered the accrual anomaly worldwide in pooled samples and found concentrated evidence in Australia, Canada, U.K. and U.S. 15 15
Koerniadi and Tourani-Rad (2007) discovered evidence of the accrual anomaly in New Zealand. Clinch, Fuller, Govendir and Wells (2012) performed a study on the existence of the accrual anomaly in Australia, they state that Australia has a developed capital market which is similar to the US, however the Australian market includes a numerous small firms and is more resource based. Clinch et al (2012) found evidence of the accrual anomaly in Australia, however this evidence was weaker than that found in by Sloan (1996). LaFond (2005) found that the accrual anomaly occurs globally as he investigated the existence in 17 different countries in his study, namely Australia, Belgium, Canada, Denmark, France, Germany, Hong Kong, Italy, Japan, the Netherlands, Norway, Singapore, Spain, Sweden, Switzerland, the U.K., and the U.S. However, he discovered that the factor driving the anomaly varies between different international markets, he found that managerial discretion, analyst following and ownership structures were the most influential in the 17 countries. Interestingly, South Africa and the JSE share some of the characteristics that were stated by Clinch et al (2012), in terms of being resource based and having developed capital markets, in addition to this South Africa also has the characteristics described by Pincus et al(2007) and Koerniadi and Tourani-Rad (2007). And South African companies share some of the factors that drive the accrual anomaly as discovered by LaFond (2005). Therefore, this could indicate that there is a likelihood of the existence of the accrual anomaly on the JSE and in South Africa. 16 16
2.5 Accruals and Stock Price Returns Eldomiaty and Abdelazim (2004) found in their research that many investors find that an earnings measure using the accruals basis is a better predictor of a firm s future cash flows, and is therefore a better method for calculating stock price returns. Dechow (1994) agrees with this view and further states that accruals contribute to the value of earnings predictions for investors, he believes accruals (as opposed to cash flows) are a superior measure of firm performance. But, DeFond and Park (2001) argue that stocks with high accruals should be priced differently by the market, this is due mainly to the fact that abnormal working accruals are less persistent and have a reversing nature. Pasaribu (2009) expands on this and found that higher average returns are earned by low accruals firms. DeFond and Park (2001) further reinforce the point by claiming that as a result of the lack of persistence of abnormal accruals they tend to have little or no effect on lifetime earnings and therefore should have little to no impact on stock price performance, especially in an efficient market. They also found evidence that markets do in fact anticipate the reversing implications of abnormal accruals, but there is no evidence to suggest that the earnings surprises, as a result of the lack of persistence of abnormal accruals, are reflected in the current stock prices. In addition to this, DeFond and Park found that the market only anticipates on average between 19 percent to 23 percent of abnormal accruals pricing implications. Bradshaw, Richardson and Sloan (2001)found that investors expectations of stock prices do not adequately reflect the possible earnings issues represented by firms with high accruals, it follows from the literature that it is these firms that are the most likely to experience future earnings issues. DeFond and Park (2001) agree with this view as they found evidence through analysis of stock returns that markets do not fully comprehend the pricing implications that arise from abnormal accruals at earnings announcement dates. On the other hand, Liu and Qi (2006) found that there is an information barrier between informed traders and average investors, they claim that the informed investors make better decisions because they can better understand the value implications of accruals and can make better assessments of earnings quality. They found evidence that supports this view as informed investors earned abnormal returns and Liu and Qi (2006) claim this is a result of these investors having information that is not easily accessible to everyone. 17 17
Houge and Lougran (2000) claim that investors fixate on current earnings, and by doing so commit a cognitive error as they fail to fully value the information contained in cash flows and accruals. Sloan (1999) agrees with this view, as he believes that the marks place a lot of emphasis on earning with accruals intact, evidenced by the markets focus on EPS numbers. Faseruk (2008) found that the valuation of firms and stocks is of great importance to analysts, researchers and investors. Faseruk (2008) also found that the financial press has extensive information and news with the potential to influence a firm s stock directly or indirectly. Therefore, it is up to the investors and/or the market to use and interpret this information and news to form views on a particular stock or the market as a whole. But to do so means that the investors cannot fixate purely on current earnings, Cheng, Liu and Thomas (2012) found that capital misallocation could occur as a result of mispriced accruals, as a result of its potential to cause adverse economic consequences they are seen as a material aspect of stock valuation. Therefore, it follows that if investors were to interpret each component of earnings independently, assuming that they have all the relevant information available to them, then it is clear that the accrual and cash flow anomalies can be identified and be used to capture unique market mispricing. Bradshaw et al (2001) claim that a limitation to Sloan s (1996) study is that the relation between the predictability of earnings declines and stock prices may in fact be attributable to unidentified risk factors or unknown research design flaws. Interestingly, Rayburn (1986) found that accruals in general perform differently as a signal or predictor of future cash flows, because the different components have differing persistence and also provide different information on potential future cash flows. Barth, Cram and Nelson (2001) agree with Rayburn (1986) as they found that the market valuations for inventory were lower than that of cash flows and even other accruals such as receivables, an explanation for this would be the uncertainty with regards to the potential future cash flow that can be generated by current inventory, it is perceived to be a more uncertain outcome. In addition to this, Francis (2008) found that accruals such as receivables were associated with a higher persistence due to their potential to increase earnings, more so than reductions in accounts payable which were essentially a cost reduction and therefore investors perceive changes in receivables to be more valuable. 18 18
Francis (2008) found evidence of market valuations of receivable accruals being higher than that of current accruals. Dreman (2008) asserts that cash is king, and investing is similar to personal finances where cash is favoured. This is a basic notion but nonetheless a key consideration in a market where current earnings may not necessarily be reinforced by cash flows. Dreman also states that if there was a comparison between two stocks that had similar management, outlook and markets then the stock with the higher cash flows is usually the more rewarding. Interestingly, Block (1999) found that financial analysts actually rank current earnings as a more important valuation tool than cash flows and the emphasis on current earnings over cash flows is common throughout the finance community. Bradshaw et al (2001) observed that analysts earnings forecast do not incorporate possible future earnings declines that are related to high accruals. Therefore investors may not have been made aware of future earnings problems caused by high accruals, as analysts do not alert investors to the possible problems and this in turn makes it even more difficult to identify anomalies. 2.6 Cash Flows versus Accruals Houge and Lougran (2000) found that because cash flows are more persistent than accruals they are subject to less manipulation and represent higher quality of earnings. Eldomiaty and Abdelazim (2004) agree with this view as they found that the accrual-based earnings can easily be managed and manipulated, but cash flows cannot. However, given this fact Eldomiaty and Abdelazim believe that accruals may contain important information about future cash flows. Teoh, Welch and Wong (1998) found that positive abnormal accruals are used in earnings management to inflate firm valuations. Collins and Hribar (2000) also agree with this view and state that the current level of accruals is negatively related to the abnormal returns over the following year, and that investors/ the market tend to underestimate the persistence of the cash flow component. 19 19
Sloan (1996) argues that the accrual anomaly coexists with the cash flow anomaly, and Collins and Hribar (2000) report how both the anomalies are more robust when using quarterly data as oppose to annual data used by Sloan (1996). However, Houge and Lougran (2000) found evidence about the characteristics of accrual stocks and how they are actually different to those of cash flow stocks, and the accrual anomaly appears to arise mainly from the poor performance of high accrual stocks. Houge and Lougran (2000) found that the market consistently underestimates the long-term persistence of cash flows and the transitory nature of accruals, and they also claim that the accrual anomaly stems from firms with poor performance who have high accruals because they are more likely to manage their earnings. Sloan (1996) found that the persistence of operating cash flows were greater than that of operating accruals. Pfeiffer and Elgers (1999) found evidence that was consistent with Sloan s (1996) findings, they discovered that operating cash flows resulted in consistently higher valuations than valuations for current and non-current accruals. Cheng, Liu and Schaefer (1996) examined the persistence issue further and found that the relative weight placed on cash flows increases as the persistence of earnings decreases. Barth, Beaver, Hand and Landsman (1999) found that when investors are forecasting or calculating firm value they should incorporate and interpret both the accrual and cash flow components of earnings. Sloan (1999) agrees with this to an extent and states even though the characteristics of accruals and cash flows may be similar it is still important to see if the market anticipates any of the subtle differences. 2.7 Discretionary and Percent Accruals Xie (2001) found that the accruals mispricing reported by Sloan (1996) is a result of discretionary accruals as opposed to non-discretionary accruals. Subramanyam (1996) found evidence that reinforces the view that the markets attach value to discretionary accruals, in addition to this Subramanyam also found evidence to support the belief that discretionary accruals are a predictor of future earnings. This supports the evidence found by Sloan (1996) and takes it a step further to specify the specific component of accruals that is responsible for the anomaly and mispricing. 20 20
Koerniadi and Tourani-Rad (2007) provides additional evidence on Xie s (2001) findings in their study on the accrual and cash flow anomaly, they observed that firms with high discretionary accruals experience significant negative future stock returns.according to Kothari (2001) discretionary accruals are used hand in hand with earnings management, therefore the market possibly overestimates the effect of earnings management through abnormal accruals. This is an interesting point and relates to Sloan (1999) placing more emphasis on the pricing issue due to the fact that managers of firms have some discretion over accruals and do indeed use them in earnings management. Xie (2001) also found that as a result the market overprices these high accrual stocks based on discretionary accruals and when combined with the lower persistence of accruals, this leads to poor/ lower future earnings. Hribar and Collins (2002) made a key discovery, they claim that many of the previous literature employed the indirect method of calculating accruals through the use of the balance sheet [(i.e. Sloan (1996)], and in doing so the calculation of discretionary accruals had become contaminated. Therefore, this resulted in incorrect conclusions of the existence of earnings management. Ideally the cash flow statement should be used when calculating accruals, this finding opens the door to new potential measures for the abnormal accrual anomaly. Hafzalla, Lundholm and Van Winkle (2010) developed a formula for scaling/ ranking accruals, called percent accruals. They showed that firms in the top 10% of negative accruals made annual returns of 5.5% above the market from 1998 to 2008. Hafzalla et al (2010) also showed how firms with high accruals made annual returns that were 6.1% below the market, they found that by using a strategy where they short the high accrual stock and buy the low accrual stock, one could have made an annual return of 11.6% above the market return. This method of percent accruals potentially represents the next step in Sloan s (1996) abnormal hedge strategy and provides an interesting way of scaling abnormal accruals. 21 21
2.8 Conclusion The accrual anomaly and the abnormal hedge strategy of Sloan (1996) is a result that goes against asset pricing theories, but in addition to this it may also go against the EMH of Fama et al (1969). The effect that this anomaly has on stock price returns has been documented around the world by Pincus, Rajgopal and Venkatachalam (2007). In addition to this Koerniadi and Tourani-Rad (2007) also found that the occurrence of the accrual anomaly is correlated with extensive use of accruals accounting, a tradition of common law, weak shareholder protectionisms and low share-ownership concentration. This is an interesting observation, one that may very well apply to the JSE and have significant implications in the investigation of the accrual anomaly on the JSE. Interestingly, Koerniadi and Tourani-Rad (2007) performed a study in New Zealand and generated an annual abnormal hedge return of 16%. However this was achieved through sorting firms by magnitude of cash flows. And it is a fact that cash flows are more persistent than accruals. The use of the percent accruals method discovered by Hafzalla et al (2010) may yield some interesting results. Prior research on this anomaly has been performed on developed markets and it would be interesting to see what the outcome would be on an emerging market stock exchange such as the JSE, especially with the increased volatility due to the size of the market in South Africa and the macroeconomic factors that companies listed on the JSE are exposed to. If the occurrence of the accrual anomaly is in fact correlated to the country-specific factors mentioned above then the JSE should potentially exhibit the accrual anomaly to some extent. 22 22
3. Research Methodology 3.1 Research Approach and Strategy The approach to this study involves deductive reasoning as the literature is used to form a hypothesis that this study aims to test on the JSE. The research method is quantitative as it involves collecting stock price and financial statement data, and analysing it using formula from the literature, in order to determine the potential existence of the accrual anomaly on the JSE. This approach and strategy was chosen because it is relevant to investigating the possible existence of the accrual anomaly in South Africa and on the JSE, it has already been identified in other markets and documented in the literature by Sloan (1996), Pincus et al. (2007) and Koerniadi and Tourani-Rad (2007), to name a few. They have made use of an abnormal hedge return strategy to take advantage of the mispricing, as a result of the accrual anomaly. Interestingly, Green et al (2009) find that the returns from the Sloan (1996) hedge strategy are currently not positive in the US, however there has been no investigation undertaken in the literature on the JSE or South Africa. Therefore, it will be interesting to test for the potential existence of the abnormal accrual anomaly regardless of those we claim it exists and those who claim that it does not because they have all looked at international markets and not South Africa. The assumptions underlying this approach for the purposes of this study are the following: The JSE is weak-form efficient in terms of EMH, the research by Mabhunu (2004) supports this view. The findings of Pincus, Rajgopal and Venkatachalam (2007) and Koerniadi and Tourani-Rad (2007) are correct and relevant to South Africa in terms of the claim that the accrual anomaly is more likely to occur in countries that possess certain institutional and accounting structures / processes. The cash flow method used to identify accruals is the best method that will yield the least amount of selection errors and bias. The abnormal accrual strategy is formed freely, without taking transaction costs into account and it is instantaneous, without a time lag / delay. 23 23
3.2 Research Design, Data Collection Methods and Research Instruments The research design will be in an experimental form, this study involves collecting accrual data from financial statements, ranking companies based on the level of accruals and then forming a hedge portfolio where stocks will be long or short depending on their level of accruals. There are many possible methods that one could use to calculate accruals, however, Hribar and Collins (2002) found that using the indirect method of calculating accruals (i.e. using the statement of financial position) could lead to errors in the conclusions. Therefore, a key issue for the purposes of this study is to make use of the statement of cash flows to identify accruals, as opposed to using the indirect method of calculating accruals through the balance sheet. Interestingly, Hribar and Collins (2002) found that market mispricing accruals tests, when the balance sheet approach was used, could lead to an understatement of the mispricing by approximately 35%. And correctly measuring accruals (i.e. using the statement of cash flows) results in much greater accruals mispricing anomalies than when using the statement of financial position method, this is due to the fact that the indirect method of calculating accruals can misclassify extreme accrual firms. The Sloan (1996) method (equation 1) is the traditional way of measuring accruals, as found in the literature, the formula is given in equation (1): Traditional Operating Accruals = (Net Income Cash from Operations) / Total Assets (1) All the three variables (earnings, cash flows and accruals) are standardized by the fiscal year book value of total assets. Richardson et al (2004) formulated a comprehensive measure of accruals, given in equation (2): Traditional Total Accruals = [Net Income (Net Dividends and Distributions to/from Equity holders + increase in the cash balance)] / Total Assets (2) 24 24
These traditional methods to calculate accruals as used by Sloan (1996) and Richardson (2004) have proven to be useful in prior studies, however they scale accruals by a measure of firm size, i.e. total assets. However, Hafzalla et al (2010) chose to scale accruals by the absolute value of earnings, the percent accruals method is given in equation (3) and (4): Percent Operating Accruals = (Net Income Cash from Operations) / Net Income (3) And, Percent Total Accruals = [Net Income (Net Dividends and Distributions to/from Equity holders + increase in the cash balance)] / Net Income (4) This study will employ the operating accruals methods to scale accruals and the corresponding stocks, both the traditional method, equation (1), and the Hafzalla et al (2010) percent operating accruals formula, equation (3), will be employed. The purpose for the use of both these measurements is to compare the difference in return from the two different ways of scaling abnormal accruals. The Hafzalla et al (2010) method potentially represents the next step in Sloan s (1996) abnormal hedge strategy and provides an interesting way of scaling abnormal accruals. In addition to this the percent accruals formula has been proven to provide significant returns to a hedge portfolio, when compared to the traditional method of scaling accruals. Hafzalla et al (2010) found that a trading strategy using percent accruals, as shown above in equation 3 and 4, yields a larger annual hedge return than traditional accrual measures, this is evidenced in the literature where Sloan (1996) managed to generate a 10.4% annual abnormal hedge return while Hafzalla et al (2010) generated a 16% annual abnormal hedge return. Pasaribu (2009) found that higher average returns are earned by low accruals firms, therefore subsequent to the scaling of the stocks, this study will also look at the accruals and only invest in the low accrual firms within the scale, also in terms of the percent accruals scale we aim to focus on the firms close to zero as this means low accruals. Using the two operating accruals methods mentioned above, this study aims to compute two different portfolios for each period, with any number of stocks depending on the scaling and interpretation of accruals. 25 25
These portfolios will be called; TRAD = scaled using traditional operating accruals with 32 stocks, strategy based on the specific nature of the scaling and accruals (discussed in analysis section) PACC = scaled using percent operating accruals with 32 stocks, strategy based on the specific nature of the scaling and accruals (discussed in analysis section) When composing or rebalancing the portfolios the consideration of which firms to potentially be long in and which to be short in is determined by the scaling measures, as done in Hafzalla et al (2010). However, as a result of the fact that our sample size is only 32 companies it makes sense to also look at the accruals as mentioned above, because the scaling bands may not be very wide as seen in the Hafzalla et al study. We then compare the performance of the traditional operating accruals portfolios to the benchmark, which is the annual return on our 32 stocks, and then we compare it to the percent operating accruals portfolios. We will be long in the stocks that have a negative percent / traditional accruals measure and a low measure of abnormal accruals (the numerator of the Hafzalla et al formula has to be close to zero), and we will be short in the stocks that have a very positive measure, and if the stocks have very large abnormal accruals. The justification for this, as documented by Hafzalla et al (2010), that the firms in the lowest deciles of percent operating accruals have positive cash from operations but their negative accrual pushes their net income back down towards zero, also if cash from operations were negative then a negative accrual would push net income further from zero. Hafzalla et al (2010) claims that the greater return achieved is as a result of the fact that the percent accruals method focuses on the composition of earnings, rather than the size of the firm (i.e. net income vs. total assets). This is an important implication as it is consistent with Sloan s (1996) and Houge and Lougran s (2000) view that investors tend to fixate on current earnings, and by doing so commit a cognitive error as they fail to fully value the information contained in cash flows and accruals. Francis (2008) finds that the inference regarding the market s valuation based on accruals and cash flows needs to simultaneously consider different sources of variation. Also, this study uses data from the JSE Top 40 Index, which includes the 40 largest companies on the JSE by market capitalisation, therefore the method of scaling accruals by net income is a logical formula to use for this study. 26 26
3.3 Sampling The JSE Top 40 Index is a representative measure of the performance of the overall JSE / South African stock market, and the companies in this index are generally large companies that are widely traded (highly liquid) and marketable stocks. Therefore, it makes sense to use this as a sample for the overall stock market population of the JSE, especially when considering that the movement of the stocks in this index represent the movement of the market as a whole, as well as representing a wide range of industries such as, telecommunications, mining and banking, (Johannesburg Securities Exchange, 2013). Rank Security Mkt Cap (ZAR) 1 BRITISH AMERICAN TOB PLC 1 126 678 201 546 2 SABMILLER PLC 851 308 895 029 3 BHP BILLITON PLC 629 747 471 839 4 COMPAGNIE FIN RICHEMONT 470 791 800 000 5 MTN GROUP LTD 344 226 449 460 6 ANGLO AMERICAN PLC 330 520 836 594 7 NASPERS LTD -N- 309 631 513 189 8 SASOL LIMITED 291 590 515 967 9 STANDARD BANK GROUP LTD 179 578 774 725 10 VODACOM GROUP LTD 166 725 245 700 11 KUMBA IRON ORE LTD 165 975 011 958 12 FIRSTRAND LTD 165 600 225 143 13 OLD MUTUAL PLC 151 549 735 007 14 ABSA GROUP LTD 107 322 126 725 15 SHOPRITE HOLDINGS LTD 100 547 512 441 16 SANLAM LIMITED 99 540 000 000 17 REMGRO LTD 95 201 328 496 18 ASPEN PHARMACARE HLDGS LTD 94 366 737 328 19 NEDBANK GROUP LTD 91 756 457 491 20 ANGLO AMERICAN PLAT LTD 88 725 340 494 21 BIDVEST LTD 84 933 860 491 22 ANGLOGOLD ASHANTI LTD 69 392 989 398 23 IMPALA PLATINUM HLGS LTD 66 875 317 744 24 MONDI PLC 49 144 164 525 25 INVESTEC PLC 43 021 415 368 26 WOOLWORTHS HOLDINGS LTD 60 623 405 626 27 MEDICLINIC INTERNAT LTD 59 642 697 153 28 TIGER BRANDS LTD 58 741 855 922 29 EXXARO RESOURCES LTD 56 181 241 585 30 RMB HOLDINGS LTD 56 171 671 044 31 ASSORE LTD 49 768 499 430 32 DISCOVERY LTD 47 602 294 650 33 STEINHOFF INT HLDGS LTD 46 669 197 453 34 GROWTHPOINT PROP LTD 44 879 454 363 35 GOLD FIELDS LTD 44 820 918 190 36 INTU PROPERTIES PLC 44 621 786 083 37 IMPERIAL HOLDINGS LTD 44 287 762 206 38 MASSMART HOLDINGS LTD 42 221 425 911 39 TRUWORTHS INT LTD 39 981 228 711 40 REINET INV SOC ANON 39 384 198 486 Table 1: JSE ALSI Top 40 Index as at 30 June 2013 27 27
This study includes the JSE Top 40 stocks as at 30 June 2013, shown in Table 1 above. These are the largest companies listed on the JSE by market capitalisation. In addition to this, these are the most relevant companies at a particular point in time, i.e. the start of the research which was June 2013. And in terms of the accrual anomaly model these companies are relevant because we are looking to make returns in the future but we first need to find out if the anomaly does in fact exist on the JSE. The sample size was reduced to 32 companies, from the original 40, because we have excluded financial companies in order to be consistent with Hafzalla et al (2010) and other prior studies. The excluded financial firms are shown in Table 1 above. Gilbertson and Goldberg (1981) find that the mining and industrial sectors of the JSE make up a significant portion of the market capitalisation this may be true for the JSE as a whole, however it not necessarily true for the JSE Top 40 Index. Table 1above shows that this study actually includes only 9 mining companies and 4 industrial companies from the total sample of 32 companies. This means that around 40% of our sample data is from the mining and industrial sectors. The data was sourced from the S&P Capital IQ database, all figures are quoted in millions of South African Rands (ZAR), unless otherwise stated, and this data was used to construct our accrual measures. The data was collected for a 5-year period from the end of June 2009 to June 2013 for the 32 companies, because not all the companies have the same year end date we have chosen to rebalance the portfolio once a year at the end of June, this takes into account year ends for all companies and serves as a mid-point in the calendar year. The only key issue was that our dataset could not go back further than 5 years because 6 of our 32 companies were only listed on the JSE between 2006 to 2008. These six companies are quite large and would have made a significant difference to our portfolio, the companies are as follows; Kumba Iron Ore Limited Listed November 2006 Mondi Limited Listed July 2007 Compagnie Fin Richemont Listed October 2008 Reinet Investments S.C.A. Listed October 2008 British American Tobacco Listed October 2008 Vodacom Limited Listed May 2009 28 28