Options trades, short sales and real earnings management

Abstract

We study the link between measures of stock options’ volatility and firms’ real earnings management (RM). We hypothesise that RM causes uncertainty in the value of a firm’s common stock and, as a result, increases the volatility spread and skew of the firm’s options. Spread and skew proxy for investors’ uncertainty in the value of the options underlying a stock. Consistent with our hypothesis, we find an association between a firm’s use of RM, and the volatility spread and skew in the firm’s options, more precisely in its put options. We also study the link between short selling and the extent of RM but do not find a consistent relationship between the two.

Keywords:

• stock options
• short selling
• real earnings management
• earnings quality

JEL Classification:

• G14
• M41
• D82

Acknowledgements

The authors would like to thank Juan Manuel Garcia Lara (the Editor) and two anonymous referees for helpful comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 There are two earnings management categories – i.e. accruals manipulation (AM) and real earnings manipulation (RM). AM does not alter a firm’s operations, and occurs when there is a change in a firm’s financial reporting choices. For instance, a firm may opt to change its depreciation policy or inventory valuation method simply to depict better earnings figures. However, the current paper’s focus does not include AM. Thus, earnings management, real earnings management, real activities manipulations and RM are used interchangeably in the paper unless otherwise specified.

2 Options traders are likely to benefit from the higher volatility in options’ prices induced by real earnings management.

3 Vuolteenaho (2002), and Callen and Segal (2004) find that earnings news drive equity returns.

4 Vega shows the effect of a change in stock price volatility on the option price. The higher the effect of RM on the stock price volatility, the greater will be the change in the option price.

5 Hirshleifer et al. (2011) argue that since it is either harder or more costly to sell a stock short than to go long, arbitrageurs can more easily exploit underpricing than overpricing, which would suggest that short sellers are less likely to trade overpriced stocks due to RM. Richardson (2003) finds that short sellers do not exploit the overvaluation of firms associated with high accruals. Our findings suggest that they do not seem to arbitrage mispricing due to RM either.

6 Options allow the investor to benefit from the same increases in wealth as the shareholders do but at a fraction of the cost. Investors in options pay only the premium, which is only a fraction of the share price.

7 Fang et al. (2016) find that stocks that were exempt from the short-sale price tests between 2005 and 2007 were associated with lower discretionary accruals. The exemption reduced the cost of shorting these stocks and, therefore, it provided short sellers with the necessary incentives to scrutinise the discretionary accruals of the pilot firms. The authors find that upon expiration of the program, discretionary accruals reverted to pre-program levels. All along, non-pilot stocks that were not exempted from the short-sale price tests did not exhibit a decline in discretionary accruals during the pilot period. Thus, absent the incentive (i.e., the reduced cost of shorting due to the exemption), there is no relationship between short selling and discretionary accruals, a finding that is in line with ours.

8 http://scholar.rhsmith.umd.edu/sbrown/pin-data

9 Our findings are robust to regression methods (OLS, GMM and 3SLS) and to various subsamples.

10 Also see Arellano and Bover (1995), Blundell and Bond (2000), and Bond (2002).

11 Delta refers to the change in the options price following a change in the stock price.

12 R&D refers to Research and Development Expenses; and SG&A refers to Selling, General and Administrative Expenses.

13 Results of estimating equations (4)(4) $PRO{D}_{i,t}/A_{i,t-1}={\beta }_0+{\beta }_1\left(\frac{1}{A_{i,t-1}}\right)+{\beta }_2\left(\frac{\mathrm{Sale}{\mathrm{s}}_{i,t}}{A_{i,t-1}}\right)+{\beta }_3\left(\frac{\mathrm{\Delta }\mathrm{Sale}{\mathrm{s}}_{i,t}}{A_{i,t-1}}\right)+{\beta }_4\left(\frac{\mathrm{\Delta }\mathrm{Sale}{\mathrm{s}}_{i,t-1}}{A_{i,t-1}}\right)+{\mu }_{i,t},$(4) and (5)(5) $\frac{DIS{X}_{i,t}}{A_{i,t-1}}={\gamma }_0+{\gamma }_1\left(\frac{1}{A_{i,t-1}}\right)+{\gamma }_2\left(\frac{Sale{s}_{i,t-1}}{A_{i,t-1}}\right)+e_{i,t},$(5) , as well as the estimation of accruals management and unexpected real earnings management, are available from the authors upon request.

14 We multiply the residuals from equation (5)(5) $\frac{DIS{X}_{i,t}}{A_{i,t-1}}={\gamma }_0+{\gamma }_1\left(\frac{1}{A_{i,t-1}}\right)+{\gamma }_2\left(\frac{Sale{s}_{i,t-1}}{A_{i,t-1}}\right)+e_{i,t},$(5) by -1 (i.e., -1 × RM_DISX) such that the higher the residuals, the larger the amount of discretionary expenditures cut by the firm to inflate reported earnings. This transformation aligns the interpretation of the sign of RM_DISX with that of RM_PROD, i.e., positive values of both equate to real activities manipulations. Thus, we are able to add them (i.e., RM_PROD + the transformed value of RM_DISX) to obtain the composite score RM.

15 The estimations of U_RM and AM are available upon request from the authors.

16 Papers that use the database include Feng et al. (2014) and Goyal and Saretto (2009).

17 We also analyse the findings by quarter, i.e. quarter -4 to -1 with quarter 0 representing the one in which we calculate RM, and our findings stay the same.

18 We rerun all regressions using basic OLS and our findings stay the same. We estimate equation (1) by correcting the standard errors for firm clustering effect following Rogers (1994), which allows for intragroup correlation by relaxing the usual requirement that the observations are independent. Thus, while the observations are independent across firms, they are not necessarily independent within a firm. We also control for quarterly and industry fixed effects.

19 The uncertainty in valuation brought about by RM can go both ways. For instance, Gunny (2010) argues that engaging in RM is not opportunistic but enables the firm to signal expected superior performance. Conversely, RM may lead to overvalued securities followed by a correction in prices. We test the hypothesis that short-selling would be related to real earnings management under scenarios where there is less ambiguity on its adverse effects. We create subsamples whereby real earnings management is detrimental to the firm value and would provide short sellers with the incentive to short-sell the shares of the firms. We then run equation (1)(1) $\begin{array}{rl}& Y_{i,t}={\alpha }_0{Y}_{i,t-1}+{\alpha }_{1}R{M}_{i,t}+{\alpha }_{2}A{M}_{i,t}+{\alpha }_{3}PI{N}_{i,t}+{\alpha }_{4}LM{C}_{i,t}+{\alpha }_{5}B{M}_{i,t}+{\alpha }_6{R}_{i,t}+{\alpha }_{7}V{R}_{i,t}\\ & +{\alpha }_{8}Leverag{e}_{i,t}+{\alpha }_{9}Illiquidit{y}_{i,t}+{\alpha }_{10}C{S}_{i,t}+{\alpha }_{11}P{S}_{i,t}+{\alpha }_{12}\mathrm{Institutional}\mathrm{Ownershi}{\mathrm{p}}_{i,t}\\ & +{\alpha }_{13}\mathrm{Number}\mathrm{of}\mathrm{Analyst}{\mathrm{s}}_{i,t}+{\epsilon }_{i,t}\end{array}$(1) using OLS in each subsample to test the effect of U_RM (unexpected real earnings management) on SHINT (the level of short interest). Scenario 1: First, we focus on the ‘abnormally’ high and low levels of real earnings management (U_RM). We surmise that ‘abnormally’ high levels of real earnings management would suggest that the intentions to manipulate earnings is less ambiguous. In the subsample of positive U_RM (i.e., the subsample where real earnings management is more detrimental to firm value), there is no association between SHINT and U_RM (as well as in various other subsamples). Scenario 2: Second, we focus on the ratio of book-to-market, which serves as a useful indicator of future stock return, i.e., high (low) values of the ratio are associated with more (less) favourable future stock returns. In various subsamples comprising of firms with the lowest book-to-market ratios, the relationship between SHINT and the independent variable of interest, i.e., U_RM is not statistically significant. Scenario 3: Third, we consider research and development expenditures. To the extent that R&D is directly related to future stock returns, we hypothesize that minimal R&D expenditure would be damaging. Thus, we break the sample into two, i.e., firms with above median R&D/Sales and those with below median R&D/Sales and hypothesize that the use of RM in the subsample of below-median R&D/Sales would be unambiguously detrimental and short sellers would have clear options to short-sell. However, the coefficient of U_RM in the regression of SHINT for the subsample of firms with below median R&D/Sale is statistically insignificant. Similarly, the coefficient is not significant in the tercile of firms with the lowest R&D/Sales. Conversely, the relationship between each of options spread and skew, and U_RM is positive and highly significant in every subsample mentioned above. As a result, our main findings stay the same, i.e., there exists a statistically significant association between options volatility and real earnings management but not between short-selling and real earnings management.