Download Analysis of Industry Net Operating Assets (NOA) and Its Impact on Future Industry Earnings and more Lecture notes Accounting in PDF only on Docsity!
NET OPERATING ASSETS AS A PREDICTOR
FOR FUTURE STOCK RETURNS
-----AN INDUSTRY ANALYSIS
DISSERTATION
Presented in Partial Fulfillment of the Requirement for The Degree Doctor of Philosophy in the Graduate School of the Ohio State University By Yinglei Zhang, M.A.
The Ohio State University 2005
Dissertation Committee: Professor Siew Hong Teoh, Advisor Approved by Professor Anne Beatty Professor Kewei Hou _______________________________ Professor Douglas Schroeder Advisor Graduate Program in Accounting and MIS
iii
To my parents, Dacheng Zhang and Shuxian Liu and my husband, Guang Yang
iv
ACKNOWLEDGMENT
I am deeply grateful to my advisor, Siew Hong Teoh, for her intellectual support, encouragement ,and on-going enthusiasm in research at every step of my program. I also wish to thank the other members of my committee, Anne Beatty, Kewei Hou and Douglas Schroeder, for their stimulation, helpful comments, and valuable time spent with me. Finally, I thank my parents and my husband for their love, their emotional support and their patience.
vi
TABLE OF CONTENTS
Page Abstract………………………………………………………………………………..ii Dedication………………………………………………………………………….....iii Acknowledgment……………………………………………………………………..iv Vita……………………………………………………………………………...…….v List of Tables…………………………………………………………………………ix List of Figures……………………………………………………………………..…xi Chapters:
- Introduction…………………………………………………………..………..
- Industry NOA as a Misperception Index………………………………….. .. 2.1 Efficient Market Hypothesis and Behavioral Finance………………. 2.2 Limited Attention and the NOA Effect……………………………… 2.3 Industry Variations of NOA and Future Stock Returns………...........182.3.1 What Causes Cross Industry Variations of NOA?…………... 2.3.2 The Industry Common Component of NOA and Future Stock Returns………………………………….… 2.4 The Industry Adjusted Component of NOA and Future Stock Returns……………………………………………….…….….
- Sample Selection and Descriptive Data…………………………..…………. 3.1 Sample ……………………………………………………………..... 3.23.3 Industry Classification…………………………………………….. ..28Financial Variable Measurement…………………………………..... 3.4 Calculation of Abnormal Returns………………………….………..
vii
- 3.5 Descriptive Statistics………………………………………………....
- Empirical Tests……………………………………………….……………...
- for Future Industry Earnings…………………………….…………... 4.1 The Implications of Industry NOA and Industry Accruals
- 4.2 Portfolio Tests…………………………………………..……………
- 4.2.14.2.2 Industry Portfolio Tests………………….............…………...44Random Industry Portfolio Tests………….………...….……
- 4.2.3 Industry-Adjusted Profits…………………….…….…….…..
- 4.3 The Fama-Macbeth Regression Tests...……………….…….….……
- Mispricing or Risk?..........................................................................................
- 5.1 The Persistence of Industry NOA Strategy………………………......
- 5.2 The Sharpe Ratio……………………………………………………..
- 5.3 The Statistical Arbitrage Test………………………………………..
- 5.4 The Industry NOA Strategy and The Business Cycle……………….
- Robustness Analyses and Discussion…………………………………...…...
- 6.1 Sample Selection Bias and the Influence of Extreme Stock Returns..806.1.1 Sample Selection Bias…………………………………….….
- 6.1.2 The Impact of Influential Observations….…………………..
- 6.2 The Influence of Transaction Costs and Small Size Firms……….….
- 6.2.1 The Influence of Transaction Costs………………..…...........
- 6.3 6.2.2Are Industry NOA and Industry Accruals One Effect or Two?..........86 The Influence of Small Size Firms…………………….…….
- Issuance and Merge and Acquisition Activities?................................. 6.4 Is the Industry NOA Effect Driven by the Cluster of Equity
- Conclusion…………………………………………………………………...
- Appendices A Supplementary Tables……………………………………….
- A.1 Definition of Industry Classification………………………………...
- A.2 Alternative Industry Classification—Portfolio Tests……………….
- A.3 Alternative Industry Classification—Regression Tests…………….
ix
LIST OF TABLES
Table Page Table 3.1 Mean Values of Industry Characteristics for Each Industry Portfolio…………………………………….….… Table 3.2 Frequency of an Industry in Extreme Industry NOA and Industry Accruals Portfolio and the Intra-Industry NOA and Intra-Industry Accruals Strategies..………………………..…… Table 3.3 Mean Values of Industry Characteristics for Portfolios Sorted by Industry NOA and Industry Accruals …………….………………… Table 3.4 Pearson (Spearman) Correlation Coefficients above (below) the Diagonal…………………………………………………………. Table 4.1 The Implications of Current Industry NOA and Industry Accruals for Future One to Four Year ahead Industry Earnings ……………... Table 4.2 Average Monthly (Abnormal) Returns for Portfolios Sortedby Industry NOA, Industry Accruals, Random Industry NOA and Random Industry Portfolios One Year after the Portfolios’ Formation……...……………………………...…………. Table 4.3 Average Monthly (Abnormal ) Returns for Portfolios Sortedby Industry-adjusted NOA, Industry-adjusted Accruals, NOA and Accruals One Year after the Portfolios’ Formation…………..... Table 4.4 Fama-Macbeth Monthly Regressions of Stock Returns on Industry NOA, Industry Adjusted NOA, Industry Accruals,Industry-Adjusted Accruals and other Characteristics……………....
Table 4.5 Fama-Macbeth Monthly Regressions of Industry Portfolio Returns on Industry NOA, Industry Accruals and Other Industry Characteristics……………………………………………...
x
Table 5.1 The Constrained Mean Statistical Arbitrage Test……….…..………. Table 5.2 Regression of 12-Month ahead Growth in GDP on 12-month Compounded Hedge Returns of Industry NOA Strategy andFama-French Factors………………………………..……………….
Table 6.1 Average Monthly Raw Returns for Portfolios Sorted First by Industry Size, then by Industry NOA One Year after the Portfolios’ Formation…………………………………………..………………... Table 6.2 Average Abnormal Monthly Returns for Portfolios Sorted by Industry NOA and Industry Momentum Simultaneously One Year after the Portfolios’ Formation …………………………...…… Table 6.3 Average Abnormal Monthly Returns for Portfolios Sorted by Industry NOA for Equity Issuance and/or M&A vs. Non- Equity Issuance and/or M&A Sub-Sample One Year after the Portfolios’ Formation………………………………..………….…… Table A.1 Definition of Industry Classification……………..…………………. Table A.2 Alternative Industry Classification—Portfolio Tests………………. Table A.3 Alternative Industry Classification—Regression Tests…….………
CHAPTER 1
INTRODUCTION
The Statement of Financial Accounting Concept No.1 specifies that one of the objectives of financial reporting is to provide information that is useful to current and potential investors in making rational investment decisions. Implicit in this objective is an overall societal goal of facilitating the efficient functioning of capital markets. How efficiently investors use financial statement information affects how efficiently resources are allocated in the economy, and is therefore an important concern for accounting regulators when setting reporting standards. A large body of empirical research has, by now, documented that accounting variables predict stock returns.^1 If risk effects are adequately controlled for in these studies, these results suggest that investors do not process accounting information efficiently. In this study, I focus on two of these variables, accruals and net operating assets.
(^1) See Ou and Penman (1989a and b), Bernard and Thomas (1989), Abarbanell and Bushee (1991), Lev and Thiagarajan (1993), Sloan (1996), Frankel and Lee (1998), Teoh, Welch, and Wong (1998a and b),Fairfield, Whisenant, and Yohn (2003), Hirshleifer et al. (2004), as well as the comprehensive surveys of Kothari (2000), and Daniel, Hirshleifer, and Teoh (2002).
Sloan (1996) reports annual abnormal profits of about 10%, using a trading strategy based on operating accruals, and Hirshleifer, Hou, Teoh and Zhang (2004) report abnormal profits of 15% (annualized), using a trading strategy based on net operating assets. In this paper, I examine whether these abnormal profits derive from the firm idiosyncratic or the common industry component of accruals and net operating assets. Firms within an industry tend to be highly correlated in several respects: they exhibit similar behaviors in business operations and accounting choices; they operate in the same regulatory environments; they are similarly sensitive to macroeconomic shocks and are exposed to similar supply and demand fluctuations. Therefore, a firm’s balance sheet may be industry dependent for a number of reasons. First, industries are endowed with different production functions and are characterized by various lengths of operating cycles, which I define as the cash-accruals-cash conversion cycle.^2 Second, industries are subject to different guidance of accounting regulations such as the choice of capitalizing vs. expensing. Moreover, firms within the same industry share similar economic shocks and their balance sheets may contain information about the business environment that is common to the industry. Choy (2003) documents that the level of net operating assets (NOA) inversely predicts a firm’s ability to meet analyst’ forecasts (Barton and Simko, 2003), and the predictive ability of NOA largely derives from industry variations in net operating assets.
(^2) In textbook presentations, the days of account receivables turnover plus the days of inventory turnover minus the days of account payables turnover is a proxy for the cash-accruals-cash conversion cycle.
Several accounting anomalies have been explained based upon the premise that investors are not able to fully assimilate the financial information available to them due to their limited cognitive processing abilities (Hirshleifer and Teoh, 2003). A balance sheet contains information beyond that contained in an income statement, which is useful for evaluating the financial prospects of firms. Hirshleifer et al. (2004) argue that if investors with limited attention anchor only on some salient cues (such as income statement earnings), but fail to appreciate the rich information contained in both accounting value added (operating income) and cash value added (free cash flow), then a parsimonious variable derived from a balance sheet, net operating assets—defined as the difference between all operating assets and all operating liabilities — measures the extent to which operating/reporting outcomes provoke excessive investor optimism. They find that the financial position of a firm with high net operating assets indicates a lack of sustainability of recent good earnings performance. Consistent with their argument, they find that net operating assets (scaled by lagged total assets)^3 negatively predict future stock returns at the firm level. I will refer to this finding as the NOA effect. In this paper, I report robust evidence that both the industry common and the firm-specific components^4 of NOA are strong negative predictors of future stock returns in a 1964-2002 sample. A trading strategy based upon buying the lowest industry NOA portfolio and selling short the highest industry NOA portfolio is profitable at 0.73% per (^3) I will refer to lagged total assets scaled net operating assets and operating accruals as NOA and Accruals respectively in the rest of the paper. (^4) I will use “industry-common”, “industry-wide” and ‘inter-industry” interchangeably throughout the paper to refer to the equal average of an attribute within industry. I will also use “industry adjusted”, “firm-specific” and “intra-industry” interchangeably to refer to the difference between a firm’s attribute and its corresponding industry common attribute.
month, highly significant, both economically and statistically. To ensure that the industry effect is not spuriously driven by grouping firms based on the characteristics documented to predict future returns at the firm level, namely NOA here, I adopt the random industry portfolio test design (Moskowitz and Grinblatt, 1999) to isolate the key role of industries in the industry NOA strategy. A trading strategy based upon random industry portfolios, which preserve the same magnitude of NOA as the industry NOA portfolios, but contain stocks from various industries, generate considerably less profits than that based upon industry NOA information. At the same time, the average hedge profits from the industry adjusted NOA strategy significantly decline by about one quarter from those of their unadjusted counterparts-NOA, indicating a loss of valuable information about future stock returns contained in the industry common component of NOA, The above three portfolio tests are complementary with each other and paint the same picture—both the industry common component and the industry adjusted component of NOA contain information that is not fully impounded into current stock prices. In contrast, I find that the Accruals effect (Sloan, 1996) is entirely driven by the firm-specific component of Accruals, and the industry common component of Accruals is not associated with future returns. A trading strategy utilizing industry Accruals information can produce almost nil profit, which is only 0.04% per month and insignificant. Meanwhile, a strategy based upon the industry adjusted (firm-specific) component of Accruals earns exactly the same profits as the one based on Accruals. The empirical results of this study have implications for several strands of research in the accounting and finance literature. First, it identifies a new conditional role
The results of this paper also extend the accounting literature on the usefulness of fundamental analysis in predicting stock returns. While there is abundant empirical evidence demonstrating the success of fundamental analysis in predicting future returns at the firm level, there exists no systematic evidence that fundamental analysis can be effective in identifying mispriced industry groups. This seems surprising given that industry analysis is often the second major step, after a market analysis, for a top-down analysis of a firm. Furthermore, there are more than 600 funds in the US, specializing in industry or economic sector investments (Investment 2002). The results of this study have implications for sector investors who seek mean-variance optimization. Furthermore, the existence of the industry NOA trading profits provides additional support for behavioral models in explaining return anomalies. The industry level anomalous return pattern documented in this study is hard to reconcile with risk- based explanations because there is no wide support for an industry risk factor in the asset pricing literature, also because the latest technology for isolating risk versus mispricing explanations has been employed. The documented industry NOA strategy is profitable in 31 out of 38 years during the sample period. The annualized Sharpe ratio based on the characteristics adjusted returns of the industry NOA strategy is 0.92, 2. times higher than the Sharpe ratio for holding the market for the same sample period, indicating a reward to risk that is very attractive relative to holding the stock market as a whole. Moreover, the industry NOA strategy survives the statistical arbitrage test introduced recently by Hogan, Jarrow, Teo and Warachka (2004), which is designed to encompass the specification of any equilibrium asset pricing model to distinguish
between risk premium and mispricing explanations for abnormal trading profits. I also document a counter-cyclical relationship between the industry NOA payoff and future GDP growth; this raises the doubt that macroeconomics factors are explanations for the documented industry NOA effect. Instead, the results are more consistent with the behavioral prediction based on limited investor attention. The results suggest that investors are unable to adjust adequately for industry common information contained in financial statements. To the extent that industry-common information is more readily accessible to market participants than the firm-specific information, the industry-NOA abnormal trading profits are more troubling for market efficiency proponents. There has been a long time concern about the potential sampling errors and econometric problems associated with anomaly findings (Kothari, 2001). Kraft, Leone and Wasley (2005) document that after eliminating extreme return observations at the 1% level, there is an inverted U-shaped relationship between future one year buy-and-hold size-adjusted abnormal returns (BHAR) and magnitude of Accruals or NOA. Thus, the hedge returns from Accruals and NOA strategies dramatically drop. They argue that this finding challenges the behavioral explanation that anomalies are likely due to investors’ inabilities to process accounting information. After repeating the industry portfolio test by deleting the observations with extreme returns, I don’t observe the inverted U shape relationship between the magnitude of industry NOA and future abnormal returns. This implies the documented industry NOA effect is not driven by observations with extreme returns.
