اطلاعات در ترازنامه برای بازده آتی سهام: شواهدی از دارایی های خالص عملیاتی

In this paper, we show that the negative relation of net operating assets (NOA) with future stock returns first documented by Hirshleifer et al. (2004) applies to both net working and investing pieces of NOA, while it is mostly driven by asset NOA components. Predictability of returns is significant only for their unexpected parts (unrelated to past sales growth) and not uniform across different industries. We also find that only high (low) NOA firms with asset expansion (contraction) and weak (strong) background of profitable investments exhibit negative (positive) abnormal returns. Our evidence suggests that the NOA anomaly may be present due to a combination of opportunistic earnings management and agency related overinvestment.

In this paper, we investigate the informational content of firms' balance sheets for future stock returns, extending a particular line in the literature1 and offering a number of novel results. Our focus is on the level of net operating assets that represents the cumulation over time of the difference between operating income (accounting profitability) and free cash flow (cash profitability). In other words, the level of net operating assets is a cumulative measure of total accruals — a measure of balance sheet bloat. Hirshleifer, Hou, Teoh, and Zhang (2004) find that the level of net operating assets, scaled by lagged total assets (NOA, hereafter), is a strong negative predictor of future stock returns for at least three years after the balance sheet information is released. They name this finding the “sustainability effect”, by recognizing that an accumulation of accounting income without an accumulation of free cash flows raises doubts about the sustainability of current earnings performance. As a result, the level of net operating assets can be used as a measure of earnings quality and thus, may also have predictive power for future stock returns. In fact, Hirshleifer et al. (2004) argue that high NOA is an indicator of a rising trend in current profitability that is unlikely to be sustained in the future causing investors, with limited attention, who focus in accounting income to make flawed decisions: overvaluing (undervaluing) firms with high (low) NOA. This leads to the, so called, NOA “anomaly” whereby firms with high (low) NOA experience negative (positive) future abnormal stock returns. They also provide evidence that NOA is a more comprehensive measure of investor's over-optimism about the sustainability of current earnings performance that captures information over and above than contained in working capital accruals and total accruals. They claim that NOA is superior to accruals because it captures all cumulative past changes between accounting and cash profitability, rather the most recent change. The economic rationale of the predictive power of NOA for future stock returns is, however, still a controversial issue. Several explanations can be put forward, but the existing literature has not yet distinguished among them. An accounting-based explanation follows the opportunistic earnings management hypothesis2: high NOA could be exploited as managers manipulate earnings upwards (see Barton & Simko, 2002). High NOA may reflect high levels of account receivables as managers book sales prematurely and high levels of inventory as managers allocate more overhead expenses to inventory than to cost of goods sold (or fail to write down obsolete inventory). Similarly, high NOA could capture high levels of fixed assets as firm executives book inappropriate expenses to property, plant and equipment and intangibles or select depreciation/amortization schedules that are not based on the underlying useful life and salvage value of fixed assets. High NOA may also reflect low operating liabilities as firm executives reduce deferred revenues, warranty liabilities, accrued expenses or change actual assumptions and discount rates required for the estimation of pension liabilities. When earnings management reverses, the market is disappointed and downwardly revises its valuation. Xie, 2001 and Chan et al., 2006 employ different models to decompose accounting accruals into their discretionary portion (i.e., unexpected portion driven by managerial discretion) and non-discretionary portion (i.e., expected or normal portion), and find that the discretionary portion predicts returns, but the non-discretionary portion does not. As such, the NOA anomaly could be driven by investors' misunderstanding of opportunistic earnings management. There are four more competing explanations for the NOA anomaly. The first of them is based on the agency-related overinvestment hypothesis: NOA could be derived as managers invest in value-destroying projects to serve their own interests (similar to the arguments in Jensen, 1986). When the market learns that such expenditures dissipate value, stock prices tend to be corrected downwards. There is substantial empirical evidence that the market corrects its initial misunderstanding of information about overinvestment, as this information is recorded in various accounting measures of capital investment, as in the: (a) growth rate in capital expenditures (Titman, Wei, & Xie, 2004); (b) abnormal growth rate in capital expenditures (Wei & Xie, 2008); (c) total accruals (Dechow, Richardson, & Sloan, 2008); (d) asset growth rate (Chan et al., 2008); (e) abnormal asset growth rate (Titman, Wei, & Xie, 2009b). Thus, the NOA anomaly could be driven by investors' misunderstanding of overinvestment. The other two explanations are based on the idea that the NOA effect may stem from the same patterns of investor behavior to other asset pricing regularities. In particular, one can follow the observation that a high NOA may contain adverse information about firm's business conditions. High NOA could reflect high levels of receivables as firms have problems in converting them into actual cash flow or compelled to offer more generous credit terms to support sales. Similarly, high NOA may capture high levels of inventory as a consequence of a relative slowdown in sales growth. High NOA could also reflect high levels of fixed assets as a consequence of replacement of obsolescent fixed assets or investment with transient payoff. According to Chan et al., 1996, Abarbanell and Bushee, 1998, Piotroski, 2000 and Chan et al., 2006, investors often respond slowly or underreact, to information contained in various accounting measures. Thus, the NOA anomaly could arise as the market initially underreacts to adverse information about firm's business conditions and subsequently corrects this underreaction resulting in lower stock returns. Further in this line, one can think that (by definition) NOA reflects all cumulative past changes between accounting and cash profitability, which in turn tend to rise with sales. Firms with high NOA are more likely to have high past growth in sales. Lakonishok, Shleifer, and Vishny (1994), postulate that investors extrapolate the strong past growth rates of firms to form optimistic expectations about their future performance. When growth rates mean-revert in the future, investors are negatively surprised by the performance of growth firms. La Porta, 1996, La Porta et al., 1997, Chan et al., 2003 and Chan and Lakonishok, 2004 document empirical evidence consistent with investors' “errors in expectations” hypothesis. Insofar as high-NOA firms share common attributes with growth firms, they may also be subject to similar valuations errors and experience disappointing returns in the future. The above hypotheses are not mutually exclusive and probably co-exist. Managers of firms that face a slowdown in business conditions may have additional motives to manipulate earnings upwards in order to meet analyst forecasts (see Iatridis, 2011 and Iatridis and Kadorinis, 2009). Similarly, firm executives with “empire building” tendency may have additional motives to inflate earnings to be less likely subjected to market scrutiny (see Polk & Sapienza, 2009). These motives could be stronger as investors and analysts extrapolate past trends in growth rates to form expectations about future growth rates (see Chan et al., 2006). The last possible explanation could be that high NOA firms are less risky than low NOA firms, and thus earn lower risk premia. Hirshleifer et al. (2004) argue that to the extent that NOA proxy for growth, its predictive power for future movements in stock prices could also reflect risk. This argument suggests that as firms increase their investment activity, thereby raising NOA, they face a less risky business environment.3 Recent theoretical papers also suggest that expected returns should systematically decline in response to increasing investment. Berk, Green, and Naik (1999) document that as firms invest, the importance of growth options relative to existing assets declines and consequently reduces equity risk. Further, according to the q-theory of investment ( Cochrane, 1991, Cochrane, 1996, Li et al., 2009 and Liu et al., 2009), firms respond to a reduction in cost of capital by increasing investment. Anderson and Garcia-Feijoo, 2006 and Xing, 2008 provide empirical support for the theoretical relationship. In this line, Wu, Zhang, and Zhang (2010) find after controlling for investment, the magnitude of the NOA effect on stock returns is reduced by more than 60%. In contrary, Li and Zhang (2010) show that limits-to-arbitrage proxies, rather than q-theory with investment frictions, are more appropriate in explaining the NOA effect in stock returns. 4 As such, whether the NOA anomaly represents rational risk premium or market inefficiency is still debatable (see also a related discussion in Hirshleifer et al., forthcoming). All the above combined lead us to focus on the NOA anomaly in order to get a deeper understanding of its underlying causes. In particular, we conduct a series of tests that address the merits of our essential motivation (distinguishing between these possible explanations) and clarify the nature of the NOA effect on stock returns. First, we examine whether different forms of net operating assets are related with future stock returns. For this purpose, we consider firm-level cross-sectional regressions of raw stock returns on NOA and NOA components — after controlling for total accruals (TACC) and investigate abnormal returns (size-adjusted returns and alphas from factor models) on portfolios based on the magnitude of NOA and NOA components. NOA are decomposed into net working capital assets (NWCA) and net non current operating assets (NNCOA), following Hirshleifer et al. (2004) claim that an important distinction could be based on the underlying business activity that NOA capture. In other words, NOA are divided into cumulative operating accruals and cumulative investment. Then, recognizing that an another important distinction could based on the underlying benefits and obligations that NOA represent, we decompose NWCA into working capital assets (WCA) and liabilities (WCL), while NNCOA into non current operating assets (NCOA) and liabilities (NCOL). The importance of each explanation varies across the above selected decompositions. The investment piece of NOA may be a better indicator of agency cost problems or risk in growth options relative to cumulative operating accruals. The asset side of NOA is more subject to managerial discretion (e.g., with respect to the timing and magnitude of revenue and expense recognition) relative to the liability side of NOA.5 At the same time, the effects of adverse changes in business conditions should be relatively uniform across different NOA components. The predictions of some explanations could differ for some NOA components, especially on the liability side. Lower operating liabilities should be associated with poor stock price performance in the future to the extent they reflect opportunistic earnings manipulation or lower operating risk.6 On the other hand, higher operating liabilities should be associated with poor stock price performance to the extent they pick up credit difficulties. Second, we disaggregate NOA and NOA components into their expected and unexpected parts to investigate their portfolio returns. The decomposition will be made using a modified version of the model of Chan et al. (2006) that is based on sales growth. The expected part reflects NOA and NOA components associated with sales growth, while the unexpected part picks up NOA and NOA components associated with opportunistic earnings management and/or a slowdown in business conditions. Thus, if errors-in-expectations about future growth are the driving force of the NOA effect, then only portfolios on the expected part of NOA and NOA components should generate predictable stock returns. On the other hand, if opportunistic earnings management and/or relative slowdown in business conditions are the underlying culprits of the NOA effect, then only portfolios on the unexpected part of NOA and NOA components should experience significant stock returns. Third, we investigate industry portfolio returns on NOA and NOA components using the Fama and French (1997) classification scheme. Industries share differences in production function, guidance of accounting rules and other characteristics which affect scope for both earnings manipulation and wasteful spending. For instance, cash-based industries where earnings roughly match cash flows may be less subject to managerial discretion. Similarly, non-capital intensive industries could be less susceptible to managerial consumption of perquisites. Thus, if opportunistic earnings management and/or agency related overinvestment are the culprits behind the NOA anomaly, portfolio returns on NOA and NOA components should vary across industries. On the other hand, if negative changes in business conditions are the force driving the NOA anomaly, portfolio returns on NOA and NOA components should be relatively uniform across industries.7 Fourth, we examine whether returns on NOA portfolios vary with asset growth and past return on equity. Asset growth (AG) can be used as a measure of managerial choices concerning investment, while past return on equity (ROE) as a measure of managerial ability to generate profits from past capital investments. Chan et al. (2008) point out that NOA has a growth feature due to the deflation by lagged total assets and show that about half of the predictive power of NOA for future stock returns overlaps with that of asset growth. Further, they present evidence that across firms with high asset growth, only those with low past return on equity experience negative abnormal stock returns. As such, a detailed analysis of the predictive power of NOA for future stock returns, conditional on asset growth and past return on equity could highlight the implications of the agency cost problem. Under the overinvestment hypothesis, high (low) NOA firms with asset expansion (contraction) and weak (strong) background of profitable investments should have more poor (strong) future stock returns. Our robustness checks are organized in two directions. First, as in Hirshleifer et al. (2004) and other studies we cannot rule out the possibility that our definition of NOA could omit operating items or include financial items. To address this issue, we investigate the performance of portfolios on NOA and NOA components by considering an alternative NOA definition that is based on the selection of specific operating assets and liabilities. Second, recognizing that abnormal returns from portfolios don't necessarily imply the rejection of market efficiency, since they could be due to mismeasured risk if the model of market returns is invalid, we also apply the statistical arbitrage test of Hogan, Jarrow, Teo, and Warachka (2004) to portfolios based on NOA and NOA components. This test circumvents the joint hypothesis dilemma of market efficiency tests since its definition is not contingent upon a specific model of market returns. Our findings can be summarized as follows: As Hirshleifer et al. (2004) do, we show a negative relation of the level of NOA with future stock returns and that this relation remains strong after additionally controlling for total accruals. In addition, we find a negative association for both cumulative operating accruals and cumulative investment, something that was not explored in Hirshleifer et al. (2004). Our results indicate that this relation is mostly, if not exclusively, driven from asset NOA components. Hedge portfolios (i.e., buying low NOA firms and shorting high NOA firms) based on the magnitude of these NOA components earn positive abnormal returns. On the contrary, liability NOA components lead, almost in all cases, in insignificant returns. This evidence appears inconsistent with a rational pricing interpretation. Another new finding is that only the unexpected part of NOA and NOA components (except liability components) adds on the predictability of stock returns. Further, the stock price performance of NOA and NOA components is not found to be uniform across different industries. As such, it does not seem to be the case, that investors' errors in expectations about future growth or underreaction to adverse business conditions, are the sole underlying drivers of the NOA anomaly. At the same time, our evidence suggests that across high (low) NOA firms only those that have high (low) asset growth and/or low (high) past return on equity experience significant stock returns. Note that our results are robust with respect to the alternative NOA definition. Similarly, hedge portfolios on NOA and those NOA components that generate positive returns are also found to constitute statistical arbitrage opportunities. Our findings could contribute in several ways to the existing literature. In particular, they indicate that investors' misperceptions of firms with bloated balance sheets are associated with the effects of operating assets on the sustainability of current earnings performance. They also suggest that the NOA anomaly may be present due to a combination of opportunistic earnings management and agency related overinvestment. Further, our evidence suggests that possible occurrence and generalizability of the NOA anomaly should be directly linked with cross-country differences in accounting regimes and institutional structures (e.g., accounting standards, auditor influence, cultural acceptability, legal tradition and shareholder protection). Our evidence also recommends that regulators and standard-setters across different countries should focus in providing an information-rich financial reporting and capital market setting. Finally, our results have a practical implication for investment managers, in the current economic and financial environment. Superior returns to NOA investment strategies (i.e., hedge portfolios) should not be limited to good or bad states of the economy, since these strategies are not fundamentally riskier. At the same time, during bad economic periods when earnings management (see Conrad, Cornell, & Landsman, 2002) and overinvestment may be lower (see Thomas, 2006), the NOA effect is expected to be less pronounced. If one applies a strategy based on a symmetric treatment of low/high NOA firms it will result in lower strategy performance, compared to an asymmetric approach were less weight will be placed in shorting high NOA firms. The remainder of the paper is organized as follows: Section 2 presents data, sample formation and methodology. In Section 3 we provide our empirical results. Section 4 summarizes and concludes the paper.