Options Trading And Earnings Management - ACFR

Second, while both short selling and options trading can promote price efficiency, very few studies examine the relative information benefits of these two types of trading.5 We show that options trading volume has a significantly negative impact on earnings management after controlling for short sale activities, suggesting that active trading in options markets contains additional information. This result sheds light on the relative information gathering and processing abilities of investors in these two markets. Third, we contribute to the literature on the determinants of earnings management. Prior research has documented various factors that affect managers’ earnings management behavior including financial performance, leverage, board structure, compensation structure, audit quality, investor protection, and regulations.6 We find that active derivative market trading is another important determinant of earnings management. The rest of the paper is organized as follows. Section 2 reviews the related literature and develops our hypotheses. Samples and the main measures of options trading and earnings management are described in section 3. Section 4 provides our research design and main empirical results. Section 5 addresses the endogeneity concern. Sections 6 and 7 investigate underlying mechanisms and rule out alternative explanations. Section 8 concludes. 2. LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT 2.1 Informational Role of Options Markets This paper fits within the literature that studies the informational role of the options market. A traditional view treats options as redundant derivative securities that can be replicated by stocks and bonds when the markets are perfect in information and frictions, and therefore, should not convey additional information beyond the underlying security markets (Black and Scholes 1973). In imperfect markets, a growing body of literature suggests the contrary. Exchange-traded options may provide investors with 5 One exception is DeLisle, Lee, and Mauck (2015). The authors find that options traders are relatively more informed than both traditional equity and short traders. 6 See Dechow, Ge, and Schrand (2010) for a review of the determinants of earnings management. 5

various benefits such as lower transaction costs (e.g., Black 1975; Back 1993; Biais and Hillion 1994), increased opportunity for leverage (e.g., Back 1993; Biais and Hillion 1994), more investment strategies, and opportunities to circumvent short-sale constraints. Because of these benefits, informed investors may prefer to trade in the options markets rather than the stock market (e.g., Chowdhry and Nanda 1991; Easley, O’Hara, and Srinivas 1998). Consistent with this notion, a series of studies focus on the relationship between options markets and the underlying stock markets, and find that parameters in the options markets have predictive power for stock market performances (e.g., Pan and Poteshman 2006; An, Ang, Bali, and Cakici 2014; Hu 2014). Besides the general lead-lag relationship with the stock markets, prior studies document the informational advantage of options markets using various settings such as earnings announcements (e.g., Jennings and Starks 1986; Amin and Lee 1997), analyst revisions (e.g., Hayunga and Lung 2014), mergers and acquisitions (e.g., Cao, Chen, and Griffin 2005; Chan, Ge, and Lin 2015; Hao 2016), and so on. Recently, several studies explore how options trading promotes shareholder interest. For example, Roll et al. (2009) show that higher options trading volumes increase firm valuation. Naiker et al. (2013) find that firms with listed options and firms with higher options trading volume are associated with lower implied costs of equity capital. 2.2 Earnings Management This paper is also closely related to the literature on how external factors influence corporate earnings management. The majority of the earnings management literature examines how managers’ incentives such as increased compensation and gains in stock sales (e.g., Bergstresser and Philippon 2006; Beneish and Vargus 2002), and mechanisms that rises internally such as dispersed ownership, independent boards, audit committee expertise (Leuz, Nanda, and Wysocki 2003; Dechow and Dichev 2002; Carcello, Hollingsworth, Klein, and Neal 2006) affect earnings management behaviors. 6

Relatively fewer papers studies the impact of external factors on earnings management. For example, prior literature documents that stronger legal systems and higher investor protection are associated with a lower degree of earnings management (e.g., Shleifer and Wolfenzon 2002; Nenova 2003; Leuz, Nanda, and Wysocki 2003; Cohen, Dey and Lys 2008; Burgstahler, Hail, and Leuz 2006; DeFond, Hung, and Trezevant 2007). Other than legal mechanisms, a few studies investigate market-based factors. One example is Yu (2008), which documents that analysts play a monitoring role and have a disciplining effect on earnings management. More recently, Massa et al. (2015) and Fang et al. (2015) find that the threat of short selling has a mitigating effect on earnings management. 2.3 Hypothesis Development In this paper, we argue that a larger options trading volume decreases managers’ incentives for earnings management. One key feature of options markets is that they attract informed traders. As a result, private information would flow to the options markets and this would result in a facilitated price discovery process of the underlying securities. With more enhanced information efficiency in the stock market, managers’ any manipulation behavior will be less effective (Ali, Li, and Zhang 2015), and the associated benefits will be reduced. Therefore, when facing more active options markets, managers would have less incentives to engage in earnings manipulations. Corporate managers manipulate earnings for various reasons and we focus on two incentives. The first one is job security concern. Managers are concerned with dismissals and, as a result, they would have incentives to manipulate the accounting information to pretend that they have achieved better performance in order to reduce the probability of turnovers. Hermalin and Weisbach (2012) argue theoretically that increased monitoring could encourage information manipulations by managers due to the job security concern. More recently, Ali and Zhang (2015) documents empirical facts that CEOs overstate earnings more significantly in the early stage of their careers and such evidence is consistent with the arguments that 7

CEOs’ earnings manipulation behavior resulting from the concerns with job security. We label this channel as job security concern channel. The second one is gaining private benefits. Managers could also have incentives to inflate earnings in order to reap private benefits. Prior research has documented empirical evidence that is consistent with this channel. For example, Beneish and Vargus (2002) and Bergstresser and Philippon (2006) show that managers have incentives to manipulate earnings to increase the gains in stock sales and their compensation. When options market is more active, both incentives for earnings manipulation are mitigated because of the increased information efficiency in the stock market. Therefore, more active stock options trading is associated with lower earnings manipulations. We label this channel as disciplining channel. Even though these two mechanisms lead to the same prediction regarding how options trading affects earnings manipulations, they have different cross-sectional implications with regard to product market competition and takeover threat. We utilize cross-section tests to examine which mechanism drives the results. If the job security concern channel dominates, then the effect of options trading on earnings management is expected to be more pronounced when the product market competition or the takeover threat is higher. When the product market is more competitive, both the likelihood of forced CEO turnover and its sensitivity to performance increases (Dasgupta, Li, and Wang, 2017). Therefore, managers would have more incentives to manipulate earnings to secure jobs when such external pressure is larger. If options trading makes such behavior less effective, then the effects would be more significant when firms are in an industry with a higher product market competition. A similar argument could be applied to the takeover pressure that a manager faces. Takeover threats would also affect the job security and managers would have more incentives to manipulate earnings as a response to a larger threat (Sul, 2017). As a result, we also expect a larger effect of options trading on earnings management when a firm is more exposed to the corporate control market. 8

However, if the private benefits channel dominates, then the cross-section predictions would be the opposite. Both the product market competition and takeover pressure can be interpreted as external corporate governance mechanisms. Hart (1983) argues that a higher product market competition would induce a manager to work harder because of the bankruptcy threat. Therefore, product market competition implicitly serves as an external governance mechanism and Giroud and Mueller (2010) finds that product market competition substitutes for shareholder governance. When takeover pressure is higher, it implies that shareholders have more rights and have a higher bargaining power relative to the management. The firm is better governed as a consequence. As a result, when either product market competition or takeover threat is higher, managers are more likely to behave and less likely to keep private benefits through manipulating earnings. If options trading serves as an alternative external governance mechanism, then its effect on earnings management would be diminished when other governance mechanisms are in place. 3. DATA AND EMPIRICAL MEASURES We obtain option data from OptionMetrics. Stock return data are from the Center for Research in Security Prices (CRSP), and corporate financial variables are from Compustat. The institutional ownership measure is calculated using the Thomson Reuters database. The Institutional Brokers’ Estimate System (I/B/E/S) is used to construct analyst coverage variables. Since the option data in OptionMetrics have been available only since 1996, our sample period spans from 1996 to 2014. 3.1 Measurement of Options Trading We construct two alternative measures for options trading volume. Following Roll et al. (2009), the first measure of options trading (OPTVOL) is the aggregate annual dollar options trading volume. For each stock trading day, we sum the dollar trading volume of all corresponding option contracts by 9

multiplying the quoted midpoint price and the number of traded contracts.7 Annual dollar options trading volume OPTVOLi,j is the aggregated amount across all trading days for stock i in year j. We follow Roll et al. (2010) and Johnson and So (2012) to construct our second measure of options trading (O/S). For each stock in each trading day, we aggregate the trading volume across all of its corresponding option contracts (1 contract corresponds to 100 shares of underlying stocks). Meanwhile, we obtain the number of traded shares in underlying stocks from CRSP for each stock in each trading day. O/Si,j is defined as the annual daily average of the ratio of the options trading volume over the stock trading volume for each stock i in year j. Different from the first measure, the relative trading volume between these two markets proxies for the intensity of informed trading of options markets over the underlying equity markets (Roll et al. 2010; Johnson and So 2012). 3.2 Measurement of Discretionary Accrual Following the earnings management literature, we use discretionary accruals as our main proxy for earnings management (e.g., Dechow, Sloan, and Sweeney, 1995; Massa et al. 2015). To construct this measure, we first calculate total accruals (TA) for each firm-year. TA is defined as earnings before extraordinary items and discontinued operations less operating cash flows. Conceptually, total accruals contain a non-discretionary component, which is determined by the economic performance of the firm, and a discretionary component, which captures managerial discretions in reported earnings and is the focus of our study. To measure discretionary accruals, we follow Dechow, Sloan, and Sweeney (1995) and employ the modified Jones (1991) model and estimate the following cross-sectional model within each fiscal year and two-digit SIC code: 𝑇𝐴𝑖,𝑡 1 𝑆𝐴𝐿𝐸𝑆𝑖,𝑡 𝐴𝐶𝐶 𝑅𝐸𝐶𝑖,𝑡 𝑃𝑃𝐸𝑖,𝑡 𝛼0 𝛼1 𝛼2 𝛼3 𝜀𝑖,𝑡 𝐴𝑆𝑆𝐸𝑇𝑖,𝑡 1 𝐴𝑆𝑆𝐸𝑇𝑖,𝑡 1 𝐴𝑆𝑆𝐸𝑇𝑖,𝑡 1 𝐴𝑆𝑆𝐸𝑇𝑖,𝑡 1 7 (1) We retain option contracts with maturities longer than five days and with a Saturday expiration (following the third Friday of the expiration month). We also make sure that the open interest and the midpoint price of each contract are larger than 0. 10

ASSETt-1 stands for the total assets at the end of fiscal year t-1. SALES represents the change in sales revenue from the prior year. ΔACC RECi,t is the change in accounts receivable from the prior year. PPEt is the gross value of property, plant, and equipment at the end of fiscal year t. We require at least 10 observations to run each estimation. Our main variable of interest, discretional accruals (DA MJ), is captured by the residual of this regression model.8 3.3 Control Variables Following previous literature, we control for factors that may have an impact on managerial earnings management behaviors (e.g., Yu 2008; Massa et al. 2015). SIZE is the natural log value of market capitalization. GROWTH is the change of assets scaled by lagged assets. TOBINQ is the market value of assets scaled by total assets. LEV is the ratio of total debt to total assets. RET is the natural logarithm of firm-specific buy-and-hold returns over the fiscal-year period, and RETSTD is the annualized standard deviation of monthly stock returns. EXFIN is the sum of net proceeds from equity financing and debt financing scaled by total assets. The proceeds from equity financing are measured by net cash received from the sale of common and preferred stock less cash dividends paid. The proceeds from debt financing are measured by net cash received from the issuance (or reduction) of short- and long-term debt. NANALYST is the natural logarithm of the number of financial analysts plus one. IO is the total equity holdings by institutional investors as a percentage of the total number of outstanding shares. ILLIQ is natural logarithm of the daily average of Amihud's (2002) measure, which is calculated as the absolute value of the stock return divided by the dollar trading volume on a given day. BIGN is a dummy variable that equals one if the firm is audited by any of the Big 4 or 5 auditors. S&P500 is a dummy variable that equals one if firm i is included in the S&P 500 index for the entire year. Both year and firm fixed effects are included in all regressions. 8 To ensure the robustness of our tests, we also use alternative measures for earnings management. See section V for details. 11

3.4 Descriptive Statistics Table 1 presents summary statistics for the variable of interests and control variables. Panel A outlines the descriptive statistics for the whole sample, while Panel B covers firms with positive options trading volume. The whole sample in Panel A contains 37,417 firm-year observations from 1996 to 2014, covering 5,751 unique firms. As a subset of the whole sample, our positive options volume sample contains 3,274 firms with 19,840 firm-year observations. [Insert Table 1] All the dependent and independent variables show reasonable variation. Consistent with prior studies (e.g., Roll et al. 2009), we find that compared with the whole sample in Panel A, firms with positive options volume are larger firms. The mean (median) market capitalization of firms with positive options volume is 1070.335m ( 984.795m). Our options measures are comparable to those documented in prior studies (e.g., Roll et al. 2009; Johnson and So 2012). The mean dollar options volume is 34.96 million dollars (65.63 million dollars) for the full sample (firms with positive options volume). The sample mean of O/S in Panel B is 5.3%, which suggests that about 18.9 times more equity round lots than option contracts are traded. For the whole sample (positive options volume sample), the average DA MJ is 0.7% (0.8%) of the beginning balance of total assets. Potentially because of the difference in firm size, Panel B shows that firms with positive options volume are more likely to be covered in the S&P 500 index, have a higher growth rate and better stock performance, have more analysts following and higher institutional ownership, and are more liquid and more likely to be audited by a Big 4 audit firm. [Insert Table 2] Table 2 presents pairwise correlations among earnings management measures, options trading measures, and control variables. Similar to Table 1, Panel A covers the full sample while Panel B covers firms with positive options volume. We find that discretionary accruals are negatively correlated with dollar options volume and relative options volume in both samples, consistent with the notion that active options trading has a reducing effect on upward earnings management. Since this table only presents simple 12

pairwise correlations, the effect of options trading activities on earnings management needs to be further investigated in the multivariate regressions. 4. EMPIRICAL RESULTS 4.1 Options Trading Volume and Earnings Management To empirically test how active options trading affects managerial earnings management behavior, we follow Roll et al. (2009) and regress DA MJ on two alternative options trading volume measures (OPTVOL and O/S) and control variables. For this test, we utilize both the full sample and the sample of firms with listed options: DA MJi,t β0 β1Options Trading Volumei,t β2SIZEi,t-1 β3GROWTHi,t β4TOBINQi,t β5LEVi,t β6RETi,t β7RETSTDi,t β8CFVOLi,t β9EXFINi,t 1 β10NANALYSTSi,t β11IILLIQi,t β12BIGNi,t Year FE Firm FE υi,t (2) In this regression, the variable of interest is Options Trading Volume, which can represent either OPTVOL or O/S. We include the same control variables as those in the previous section. Both firm and year fixed effects are controlled in all regression specifications. [Insert Table 3] The results for the full sample and the positive options volume sample are reported in panels A and B of Table 3, respectively. For both samples, we find both OPTVOL and O/S are significantly and negatively correlated with DA MJ, indicating that active options trading has a downward impact on earnings management. To put the coefficient estimates in economic terms, the whole sample results suggest that a one standard deviation increase in OPTVOL (O/S) is associated with 4.4% (2.9%) standard deviation less earnings management.9 The positive options volume sample shows similar economic significance: a one standard deviation increase in OPTVOL (O/S) results in a 6.6% (3.6%) standard deviation reduction in discretionary accruals. 9 The economic significance is calculated as the regression coefficient multiplied by the one standard deviation change in OPTVOL (O/S), scaled by the standard deviation

In this paper, we explore potential real impacts of options trading and specifically investigate how active options trading affects managers' earnings management behavior. We focus on earnings management because earnings contain crucial financial information, and how they are reported affects resource allocation in the capital markets.