2013年11月12日学术报告通知.pdf

SEC Comment Letters and Insider Sales Patricia Dechow patricia_dechow@haas.berkeley.edu Alastair Lawrence lawrence@haas.berkeley.edu James Ryans james_ryans@haas.berkeley.edu Haas School of Business University of California at Berkeley 2220 Piedmont Avenue Berkeley, CA 94720-1900 November 2013 ABSTRACT: The Securities and Exchange Commission’s (SEC) Division of Corporation Finance reviews company filings and where appropriate, issues comment letters to elicit better compliance with applicable disclosure requirements and Generally Accepted Accounting Principles. Since 2005, the Division’s staff publicly releases comment letters and companies’ responses to those letters no earlier than 45 days (20 business days post 2011) following the review completion. We provide evidence of increases in insider sales following the public disclosure of comment letters pertaining to more material financial disclosure and reporting issues. We also provide evidence of increases in insider sales in the week prior to such disclosures. Moreover, increases in pre-disclosure insider sales are most pronounced for firms with high short interest. Our findings suggest that SEC’s practice of delaying the release of comment letter correspondence does not best serve the interests of outside investors. KEYWORDS: SEC, Division of Corporation Finance, comment letters, insider trading, revenue recognition, financial disclosure, GAAP, short interest. JEL CLASSIFICATION: M41, M48, and K42. We have received valuable comments from Dan Amiram, Philip Berger, Daniel Collins, Sunil Dutta, Ellen Engel, Fabrizio Ferri, Yaniv Konchitchki, Craig Lewis, Michael Minnis, Alexander Nezlobin, Panos Patatoukas, Stephen Penman, Richard Sloan, Jacob Thomas, Wan Wongsunwai, Sarah Zechman, Xiao-Jun Zhang, and workshop participants at Columbia University, University of California at Berkeley, and the University of Illinois at Chicago Accounting Research Conference. We thank Korcan Ak and Henry Laurion for their helpful research assistance. 1. Introduction and background The U.S. Securities and Exchange Commission (SEC) is comprised of the following five major divisions: Corporation Finance, Enforcement, Economic and Risk Analysis, Investment Management, and Trading and Markets. 1 Of these five divisions, the Division of Corporation Finance (hereafter, the Division) is the one that seeks to ensure that issuers’ filings are in compliance with the applicable financial reporting requirements, and historically, it has selectively reviewed issuer filings made under the Securities Act of 1933 and the Securities Exchange Act of 1934. However, as stipulated by the Sarbanes-Oxley Act of 2002, the Division now reviews each issuer’s filings at least once every three years. As a result, the Division has significantly increased its resources and headcount over the past ten years to keep pace with its review requirements. While the Division does not publicly disclose its selection criteria or the specific scope of the reviews, it states that much of its review “involves evaluating the disclosure from a potential investor’s perspective and asking questions that an investor might ask when reading the document” (SEC 2013a). The Division completes many reviews without issuing comments but when it identifies areas for an issuer to better comply with applicable disclosure requirements and Generally Accepted Accounting Principles (GAAP), it provides a comment letter outlining its questions and concerns. The company will then have 10 days to respond to each concern raised in the letter, and the back and forth between the company and the Division will continue until the company has resolved all of the Division’s comments. To increase the transparency of the review process, since May 12, 2005 the SEC has publicly released comment letters and the companies’ response letters after the filing review is complete. Prior to January 1, 2012, the 1 See Appendix A for the organizational structure of the SEC’s five main divisions and over 20 offices. 1 comment letter correspondence was disclosed “no earlier than 45 days” following the review completion; however, subsequent to January 1, 2012 to further increase the transparency of the review process, the Division now releases comment letter correspondence “no earlier than 20 business days” following the review completion (SEC 2013b). When appropriate, the Division of Corporation Finance refers serious violations, misconduct, or fraud noted in the filing review process to the Division of Enforcement. While the transparency of the comment letter review process has significantly increased over the past decade, it is somewhat surprising that the review process still affords companies who were deficient in their regulatory filing requirements, or who may disclose important information in the correspondence process, an extended period of time to retain and potentially trade on this private information. This study examines whether insiders take advantage of the current practice of delaying the release of the Division’s comment letters once the review is finished. Given that many comment letters only relate to minor or limited concerns, we do not anticipate heightened selling activity prior to minor or more trivial comment letters. We investigate insiders’ trades surrounding the public disclosure of comment letters, and specifically, examine whether insiders opportunistically sell their shares prior to the SEC’s public release of comments letters pertaining to important disclosure and reporting issues. Moreover, we predict that the sale of shares prior to the public release of important comment letters will be more pronounced for firms with high short interest. A key component of our research design involves separating important comment letters from trivial ones. Although comment letters can essentially pertain to any type of SEC filing (e.g., 10-Ks, 8-Ks, DEF 14As, S-1s, etc.), we focus on comment letters relating to Form 10-K filings. The 10-Ks provide the most extensive annual disclosures and describe how the company 2 applied GAAP during the year. Hence, reviews pertaining to 10-K filings can include the Division’s requests to fix material disclosure deficiencies and to further explain or adjust how the company applies critical accounting policies. Not all 10-K comment letters pertain to material issues. For example, many 10-K comment letters ask simple clarifying questions or for more information on relatively minor issues. To effectively separate important and trivial comment letters, we seek to identify comment letters that question critical accounting policies and disclosures yet they still reflect a significant portion of all comment letters. Revenue is arguably the most important line item in the financial statements. The Analysts’ Accounting Observer states: “revenue is the largest, most important number on the income statement—and the most amorphous, with less specific disclosures about is constitution than any other line on the income statement” (Ciesielski 2013). Moreover, the Division has continually mentioned revenue recognition as one of the top critical accounting policies addressed in the comment letters. The SEC staff usually requests that the company justify the timing of when revenue is recognized or improve the transparency of the revenue recognition policy through enhanced disclosures and discussions of the main revenue recognition assumptions and estimates (e.g., Ciesielski 2012; SEC 2013c). Furthermore, overstating revenues using inappropriate revenue recognition policies is the most common form of accounting fraud, representing approximately 54 percent of all accounting frauds (e.g., Dechow, Ge, Larson, and Sloan 2011). Hence, as Revenue Recognition Comment Letters represent approximately 20 percent of all comment letters and given that, as a group, they likely reflect the largest number of the most material comment letters, we flag them as likely being more important relative to All Other 10-K Comment Letters. Another method to identify more important comment letters is to identify those comment letters that result in restatements of the financial statements. While 3 resulting restatements are an effective method to identify material comment letters, the restatements will be made prior to the public disclosure of the comment letter correspondence and thus, the incentive for insiders to sell shares prior to the public disclosure will be limited in such circumstances. Hence, it appears that the revenue recognition partition is a more obvious choice for our setting. The empirical results are consistent with increases in insider sales around the public disclosure of the comment letter correspondence for important comment letters and not for more trivial comment letters. The results not only provide evidence of significant increases in insider sales following the public disclosure of more important comment letters, but they also provide evidence of significant increases in insider sales in the five trading days prior to these disclosures. Specifically, we find abnormal increases in insider sales between 37 to 100 percent in the ten days following the public disclosure of Revenue Recognition Comment Letters and abnormal increases between 46 to 65 percent in the five days prior to the public disclosure. Moreover, for firms in the top quartile of short interest, we find abnormal increases in insider sales between 109 to 126 percent in the five days prior to the public disclosure of Revenue Recognition Comment Letters. Furthermore, we find negative cumulative abnormal returns (CARs) of -59 basis points in the 15 days following the public disclosure for Revenue Recognition Comment Letters and positive CARs of 41 basis points in the 15 days following the public disclosure of All Other 10K Comment Letters. It is not overly surprising that there are positive returns following the disclosure of comment letters relating to more trivial concerns as such letters may be viewed as positive news. In particular, given that the Division has to review each issuer’s filings at least once every three years, the disclosure of comment letters identifying trivial concerns suggests a 4 passing grade by the SEC on the company’s financial reporting practices and that another review is not likely for another few years. The mean negative returns for the Revenue Recognition Comment Letter firms, where approximately 54.1 percent of firms have negative returns, highlight that, on average, insiders appear to anticipate the market response to such comment letter disclosures. Our main inferences are robust to alternative specifications and additional analyses as highlighted in Section 4.3. Together, these findings indicate significant increases in insider sales prior to the SEC’s public release of material comments letters and hence, suggest that some insiders take advantage of the SEC’s practice of delaying the release of comment letter correspondence. Our findings extend prior research suggesting that insiders effectively time the issuance or sale of their equity compensation (e.g., Keown and Pinkerton 1981; Penman 1982; Lee, Mikkelson, and Partch 1992; Harlow and Howe 1998; Lie 2005; Chen and Lo 2006; Li and Zhang 2006; Jagolinzer 2009; Cohen, Malloy, and Pomorski 2012) and highlight a new setting where it appears that insiders take advantage of their material private information. Given that we use a fairly crude partition to separate trivial comment letters from more important ones and that we only focus on comment letters pertaining to 10-Ks, our findings likely provide a lower bound estimate of the extent of insider trading prior to the disclosure of material comment letters. Moreover, the study’s findings support calls from the investment industry to eliminate the disclosure delay (e.g., SEC Insight 2004; Bloomberg 2013). The remainder of this paper is organized as follows. Section 2 describes the Division’s filing review process and our main predictions, while Section 3 describes the sample. Section 4 reports the results of our main analyses and Section 5 concludes. 5 2. Division of Corporation Finance reviews and predictions In the financial reporting world, the term “comment letter” actually has two distinct meanings. The more well-known use of “comment letter” pertains to letters written by the public to the Financial Accounting Standards Board (FASB) or the SEC highlighting feedback on proposed accounting and reporting standards. The lesser known use of “comment letter” relates to the SEC’s comment letters where the Division identifies and outlines areas for an issuer to better comply with GAAP and the applicable disclosure requirements. In this study, we will refer to the latter meaning. While there is little disclosure surrounding the Division’s review process, including the filings its staff focus on, the majority of SEC comment letters for U.S. issuers relate to annual and quarterly financial reports (Form 10-Ks, Form 10-Qs—45 percent). The other main type of comment letters pertain to material news releases (Form 8-Ks—11 percent), proxy statements (e.g., DEF 14A—5 percent), and registration and prospectus filings (Form S-1s, S-2s, and S-3s— 21 percent).2 The remaining comment letters pertain to over 50 of the more miscellaneous yet required EDGAR filings. We exclusively focus on comment letters relating to Form 10-K filings given that such letters are likely to reflect the more material reviews where the SEC staff questions and critiques firms’ critical financial reporting disclosure and reporting practices. While comment letters related to registration filings could also include similar discussions as 10K related comment letters given that full annual financial statements are presented in prospectuses, we do not investigate such filings as many of the lock-up periods will be in effect during the Division’s review. Furthermore, given that the critical accounting policies are only 2 SEC Comment Letter form type percentages were calculated using Audit Analytics SEC Comment Letter Database as of September 2013. 6 discussed in detail in the 10-Ks (annual reports) and not the 10-Qs (quarterly reports), we choose to focus on the 10-K related comment letters. As previously mentioned, we identify more important 10-K comment letters as those that question and relate to the firm’s revenue recognition policy. See Appendix B for randomly selected examples of both Revenue Recognition Comment Letter and All Other 10-K Comment Letter correspondence between the respective firms and the Division. We limit the randomization to sample from shorter letters for the sake of presentational brevity. The revenue recognition partition, classifying approximately 20 percent (80 percent) of comment letters as more important (more trivial), will in no doubt misclassify important comment letters as trivial comment letters and vice-versa. However, as the revenue recognition policy across all firms has the most pervasive effects on the financial statements, the Staff’s investigation into these policies should be effective in providing significant cross-sectional variation in terms of the importance of the letters between the two groups. Moreover, improperly classifying more and less important comment letters will only add noise and reduce the magnitude of the directional effects, and hence, highlight that the results serve as a lower bound in terms of the economic magnitude of insider sales prior to the public disclosure of the comment letter correspondence. Despite the fact that the Division’s filing review process plays a critical role in supporting the Commission’s main “mission to protect investors, maintain fair, orderly, and efficient markets, and facilitate capital formation” (SEC 2013d), there has been little research investigating the Division’s review process. Early research examines the determinants of receiving comment letters (e.g., Ertimur and Nondorf 2006; Ettredge, Johnstone, Stone, and Wang 2011; Robinson, Xue, and Yu 2011; Cassell, Dreher, and Myers 2013) or the benefits of the comment letter review process (Johnston and Petacchi 2012). However, prior research does 7 not focus on the appropriateness of the Division’s practice of delaying the public disclosure of the comment letter correspondence following the review completion nor whether insiders take advantage of this opportunity. In this study, we investigate whether insiders take advantage of the opportunity to sell shares prior to and at the time of the public release of more material comments letters. We contend that insiders will face a risk-return trade-off. They will trade off the expected benefit from trading prior to the review disclosure relative to the potential cost of being caught and successfully prosecuted. Below we discuss three attributes of the comment letter review process that are likely to affect this trade-off. First, comment letter correspondence disclosures are discrete events that are likely to go under the radar of investors and the media. Thus insiders are likely to view the probability that their trades will be detected as low. Specifically, it is difficult to identify in a timely manner new comment letter correspondences for a firm of interest. The only way to find comment letter correspondence, without subscribing to a private data provider, is to go to the SEC’s Electronic Data Gathering Analysis and Retrieval (EDGAR) system and search for “UPLOAD” files to find SEC issued comment letters and “CORRESP” files to find company responses to the letters. Investors can also check EDGAR daily releases for “UPLOAD” and “CORRESP” files. However, once at each company’s page, EDGAR lists “UPLOAD” and “CORRESP” files according to their filing date, and an investor needs to examine the header of the raw text file to identify the disclosure date. Hence, without an appreciation of the idiosyncrasies inherent in the comment letter disclosure process, it is difficult to assess the timeliness of such disclosures. Moreover, it appears that the Division’s findings, even when serious, do not make the news 8 headlines. These circumstances suggest that insider trading activities are unlikely to gain attention at the time of comment letter releases. Second, many comment letters address issues that are of little value-relevance. While a significant portion of the letters relate to financial reporting or disclosure practices that could have a material impact on the financial statements, the majority of comment letters tend to reflect minor issues. Therefore, unlike the average earnings or merger and acquisition announcements that are likely to impact stock prices, the average news in comment letters is small. Moreover, even if there is a negative market reaction to important comment letters an insider could view the risk of trading as worth taking for two reasons. First, authorities must detect and then decide to pursue the case against the insider (to our knowledge there have been no such cases). Second, the authorities must prove that the insider traded on non-public information. Comment letters generally relate to already available public disclosures. Therefore, it is quite possible that defendant lawyers could successful argue that the insider did not believe he or she was trading on private information. Thus, insiders could view the expected cost, even if detected, to be low. Third, firms are typically reviewed every three years and only receive comment letters when there is an issue of concern, so receiving a comment letter is a relatively rare and unpredictable event. Therefore, it is unlikely that companies have strict policies preventing employee or director trading during the time between the completion of the comment review process and the public release of the comment letters. Thus insiders could view the risk of being reprimanded or fired for trading on comment letter information as being low. In addition, while insiders can fairly accurately anticipate the SEC’s public disclosure date given its comment letter disclosure rule, insiders could argue that they were unaware of when the disclosure would occur and thus, the timing of their sale had nothing to do with the subsequent disclosure. 9 Therefore, insiders with knowledge of bad news are likely to be able to sell their shares with confidence that the probability of facing significant penalties is low. Thus, consistent with Kyle (1985) we expect that some insiders will exploit their advantage and earn “positive profits” by trading on their private information of the Division’s review. P1: Insider sales increase prior to the disclosure of important SEC comment letters. Short sellers are sophisticated investors who use financial reporting information to evaluate earnings quality and valuation based strategies in order target overpriced firms (e.g., Dechow, Hutton, Meulbroek, and Sloan 2001; Curtis and Fargher 2008; Karpoff and Lou 2010; Hirshleifer, Teoh, and Yu 2011). The press has recently highlighted that short-oriented hedge funds are active consumers of the SEC comment letter correspondence. For example, a recent Bloomberg article highlights that much of Muddy Waters LLC research, which has triggered billions of dollars of losses for Chinese stocks listed on U.S. markets, is initially informed and motivated by SEC comment letters (Bloomberg 2013). 3 Thus, short sellers appear to be an investor group that is effectively utilizing comment letters. Corporate executives and insiders are typically concerned with the effects of short sellers and frequently take legal action against them.4 Lamont (2012, p. 1) explains that “firms use a variety of methods to impede short selling, including legal threats, investigations, lawsuits, and various technical actions intended to create a short squeeze.” This behavior suggests that executives are aware and concerned with the amount of short positions in their stock. Moreover, as short sellers appear to track comment letters, insiders will anticipate that the negative 3 http://www.bloomberg.com/news/2013-02-19/muddy-waters-secret-china-weapon-is-on-sec-website.html 4 A very public example is the case of Overstock.com CEO Patrick Bryne who campaigned against naked short selling and filed lawsuits against the largest U.S. securities firms (see Forbes 2007). 10 information in comments letters will more likely be noticed and publicized by short funds upon the public release on EDGAR. Thus the presence of high short interest could increase the risks of selling prior to the public disclosure due to the increasing probability that the sale will be noticed. However, the presence of high short interest is likely to also indicate more material accounting issues that are not fully impounded in the current stock price. Therefore, the presence of short-sellers could be indicative of greater expected benefits to insiders from trading before the comment letter process is disclosed. Thus, whether insiders in stocks with high short interest choose to trade before comment letter releases depends on the risk return trade-off. We make the following prediction: P2: 3. Increases in insider sales prior to the disclosure of important SEC comment letters is more pronounced for firms with high-short interest. Sample Since 2005, the Division has publicly released comment letters and companies’ responses to the comment letters. The comment letters (form type “UPLOAD”) and company responses (form type “CORRESP”), which can be found on the SEC’s EDGAR filing system, are posted approximately 45 calendar days after the completion of the Division’s review (reduced to 20 business days post 2011). The EDGAR system provides the date each letter was sent, however, only in the raw text file header does it provide the date the comment letter or correspondence was disclosed to the public in the EDGAR system (Audit Analytics field “FILE_DIS_DATE”). Our sample is taken from Audit Analytics, which records the disclosure date, and transcribes and codes the issues identified in the comment letters. Exhibit 1 depicts the 10-K comment letter timeline and highlights that the median review lasts 131 calendar days and is disclosed 46 11 calendar days pre-2012 (20 business days during 2012) after the review completion. Moreover, it highlights that during the comment letter review period, less than one percent of firms are required to restate their financial statements. Apart from these restatements, the outcomes of the comment letter review process are either that (1) the SEC determines that the companies’ responses are sufficient or (2) that the SEC requests changes to companies’ future financial statements. Table 1, Panel A describes the sample of comment letters used in this analysis. We first limit our analysis to comment letters related to a 10-K. Since there are often multiple correspondences in the same series, we identify a “conversation” as the series of comments and responses relating to an initial comment letter (Audit Analytics field “CL_CON_ID”). Column (1) identifies the number of unique conversations in the period from 2006 to 2012, which encompasses 14,770 10-K related comment letters. We further search the issues raised by the SEC to identify revenue recognition related comment letters (Audit Analytics field “LIST_CL_ISSUE_PHRASE”). There are 4,768 such letters in the 2006-2012 time period. Our analysis requires that we match these comment letters to both the Thomson Reuters Insider transaction database and to Compustat for fundamental variables, using the Central Index Key (CIK) of the comment letter firm. Insider sales are from the Thomson Reuters Insider database, and we record daily sales for officers and directors with firms matched by CUSIP.5 Insider sales are measured as the daily percent of shares outstanding sold by insiders, with non-zero values winsorized at the 1% level (SHARES from Thomson Reuters insider database divided by 5 Insiders are defined as sales by officers and directors from the Thomson Reuters insider database (ROLECODE of: “CEO”, “D”, “O”, “H”, “DO”, “OD”, “VC”, “OB”, “OP”, “OT”, “CB”, “AV”, “CFO”, “CI”, “CO”, “CT”, “EVP”, “OX”, “P”, “S”, “SVP”, or “VP”). 12 SHROUT from CRSP scaled as a percent of shares outstanding).6 The requirement for merging both the comment letter and insider sale data to Compustat provides a sample described in Column (2) of 8,996 total 10-K comment letters and 1,771 Revenue Recognition Comment Letters. Finally we require that the comment letter firms have stock return data available in CRSP for at least 30 days before and after the comment letter disclosure date. We use two approaches to control for the expected level of insider sales. First, we use a methodology similar to Lie (2005), and take five observations of the same firm with a disclosure date randomly chosen from between six and three months prior to the comment letter disclosure date and between three and six months after the comment letter disclosure date. Second, in robustness analyses, we use the comment letter firm’s five closest peers matched by FamaFrench industry groups, fiscal year-end date, and market capitalization. For a control observation to be selected it must also have CRSP stock return data for the 30 days before and after the randomly selected disclosure date. The combination of a valid comment letter observation and at least one control period observation results in a final sample in Column (3) of 7,964 total 10-K comment letters and 1,560 Revenue Recognition Comment Letters. Table 1, Panel B highlights the distribution of comment letters across the 49 FamaFrench industry groupings. The first column, % Compustat, presents the industry composition of all Compustat firm-years from 2005 to 2012 and % Comment Letters, presents the industry composition of our full comment letter sample. Comparing these two columns highlights that industry composition of Compustat and comment letter firms appears fairly similar. It does appear that financial, retail, and utility firms receive a relatively greater percentage of comment 6 We also investigate insiders’ buying activities, which are very limited, and repeat all of our main analyses using net sales. All inferences are robust to using insider sales net of buys instead of total insider sales. 13 letters, and software and telecommunications receive a relatively lower percentage of comment letters then suggested by the overall Compustat industry compositions. The third column, % Revenue Recognition, presents the percent of comment letters in each industry group that are revenue recognition related. Software firms have the highest proportion of Revenue Recognition Comment Letters (42.1 percent), followed by hardware firms (39.0 percent), agriculture firms (36.8 percent), pharmaceutical firms (34.0 percent), and electrical equipment firms (33.9 percent). However, given that the majority of industries received a significant number of Revenue Recognition Comment Letters, no one particular industry accounts for more than 14 percent of all Revenue Recognition Comment Letters. 4. Results 4.1. Descriptive Statistics Table 2 provides descriptive statistics for all comment letter firms and separately presents the descriptive statistics for Revenue Recognition Comment Letter firms and All Other 10-K Comment Letter firms. We measure insider sales as the percent of shares outstanding sold by insiders. The mean daily insider sales for the full 10-K Comment Letter sample 30 days before and after the comment disclosure date, as a percentage of shares outstanding, is 0.0016, whereas the mean daily insider sales for the Revenue Recognition Comment Letter firms and All Other 10-K Comment Letter are 0.0025 and 0.0014, respectively. For descriptive purposes we also report the mean (median) dollar amount of daily insider sales which is $60,111 ($0) and $68,374 ($0) for Revenue Recognition Comment Letter and All Other 10-K Comment Letter firms, respectively. Over the 61 day window this corresponds to approximately $3.7 and $4.2 million of total insider sales for Revenue Recognition Comment Letter and All Other 10-K Comment Letter 14 firms, respectively. The mean (median) assets of firms receiving 10-K comment letters is $11.9 billion ($1.1 billion) and the mean (median) revenues are $4.6 billion (0.6 billion). Moreover, the mean (median) market capitalization is $5.8 billion ($0.9 billion), the mean (median) book-tomarket ratio is 0.71 (0.50), the mean (median) earnings-to-price ratio is -0.06 (0.04), the mean (median) 30 day CAR leading up to the comment letter disclosure is -0.001 (-0.006), and the mean (median) short interest is 4.94 percent (3.33 percent). Relative to the full Compustat population during our sample period, comment letter firms have a slightly smaller mean asset value than the Compustat mean of $13.5 billion but have a larger median asset value than the Compustat median of $0.6 billion. Moreover, comment letter firms have slightly higher revenues and market capitalizations than the Compustat means (medians) of $3.9 billion ($0.3 billion) and $4.4 billion ($0.4 billion), respectively. Furthermore, comment letter firms have lower book-to-market ratios than the Compustat mean (median) of 0.72 (0.56) and higher short interest than the Compustat mean (median) of 4.08 percent (2.40 percent). Hence, the descriptives highlight that the attributes of comment letter firms closely align with those of the whole Compustat population, which is consistent with the mandate to review all companies’ filings every three years and the notion that the comment letters issued by the Division are not overly biased towards specific types of firms. While the mean total assets are smaller for Revenue Recognition Comment Letter firms ($8.4 billion) than for All Other 10K Comment Letter Firms ($12.7 billion), the two groupings appear fairly balanced across most other characteristics. Of the total 7,964 10-K related comment letters in our sample, 1,560 (19.6 percent) represent Revenue Recognition Comment Letters and 6,404 (80.4 percent) represent All Other 10-K Comment Letters. 15 4.2. Results Table 3 presents the insider sales for the 30 days before and after the comment letter disclosure date for the Revenue Recognition Comment Letter firms. Following Lie (2005) we use the firm as its own control, using five randomly selected event dates, if available, between 3 to 6 calendar months before and after the event day, which in our case is the comment letter disclosure date. Panel A presents the results for full sample and Panel B presents the results for the high short interest sample (top quartile of short interest). We report one-sided t-tests as we hypothesize that comment letter firm insiders will sell more stock around the comment letter disclosure than at other times during the year. Panel A highlights insignificant differences at the five-percent level between comment letter and control period insider sales for the Revenue Recognition Comment Letter firms during days -30 to -6 and days +11 to +30; however, there are significant differences between comment letter and control period insider sales during days -5 to +10. Specifically, the mean insider sales, as a percentage of shares outstanding, during days -30 to -21, -20 to -11, and -10 to -6 are 0.0017, 0.0016, and 0.0021, respectively, and are insignificantly different from the control period sales during this period of 0.0018. However, between days -5 to +10 the comment letter insider sales, as a percentage of shares outstanding, increase to 0.0027, 0.0024, and 0.0032 for days -5 to -1, 0 to +5, and +6 to +10, respectively—reflecting increases over normal insider sales of 46.1 percent, 42.4 percent, and 107.3 percent, respectively. Hence, while the majority of abnormal insider sales occur after the public disclosure, consistent with our first prediction, a significant portion of insiders unload shares in the week prior to the public disclosure. 16 To provide an indicator of the dollar magnitudes of insider sales, first note from Table 2, that the mean market value of these firms is $4.9 billion. Therefore during the ten days prior to disclosure, cumulative insider sales are approximately $1.8 billion (((0.000021 x $4.9 billion x 5 days) + (0.000027 x $4.9 billion x 5 days)) x 1,560 firms), with the mean firm selling approximately $1.2 million ($1.8 billion / 1,560 firms). Moreover, to provide an indicator of the dollar magnitude of abnormal insider sales, for the mean firm, insiders are selling approximately $294,900 more shares ((0.000003 x $4.9 billion x 5 days) + (0.000009 x $4.9 billion x 5 days)) than we would expect in the ten days before the disclosure, reflecting total abnormal insider sales of approximately $460 million ($294,900 x 1,560 firms). During days +11 to +20 and +21 to +30 insider sales decrease to 0.0020 and 0.0021, respectively, as they revert to normal levels. We expect that insider selling prior to material comment letter disclosures will be more pronounced in the presence of short-oriented hedge funds. Panel B focuses on the Revenue Recognition Comment Letter firms in the highest quartile of short interest. The results indicate that there are insignificant differences at the five-percent level between comment letter and control period insider sales during days -30 to -11. Specifically, the mean insider sales, as a percentage of shares outstanding, of firms with high short interest during days -30 to -21 and -20 to -11 are 0.0026 and 0.0027, respectively, which are insignificantly different from the respective control period sales of 0.0026. However, insider sales for high short interest firms substantially increase in the ten trading days prior to the comment letter disclosure to 0.0033, and 0.0053 for days -10 to -6, and -5 to -1, respectively—corresponding to increases over normal levels by 44.9 percent (p < 0.05) and 108.6 percent (p < 0.01), respectively. Therefore during the ten days prior to disclosure, cumulative insider sales are approximately $322 million (((0.000033 x $1.7 billion x 5 days) + (0.000053 x $1.7 billion x 5 days)) x 441 firms), with the mean firm selling 17 approximately $731,000 ($322 million / 441 firms). Moreover, for the mean firm, excess insiders are approximately $323,000 ((0.000010 x $1.7 billion x 5 days) + (0.000028 x $1.7 billion x 5 days)) more per firm than we would expect in the ten days before the disclosure, reflecting total abnormal insider sales of approximately $142 million ($323,000 x 441 firms). In contrast to the full sample, we do not find any evidence of heightened insider sales in the five days (0 to +5) following the comment letter disclosure, as it appears that insiders in high short firms unload the majority of their shares prior to the comment letter disclosure. In days +6 to +10 and +21 to +30, we again find heighten insider sales of 61.8 percent (p < 0.05) and 32.5 percent (p < 0.05), respectively. Table 4 repeats the analysis in Table 3 while focusing on All Other 10-K Comment Letter firms. In the full sample in Panel A, we do not find evidence of significant increases in insider sales for All Other 10-K Comment Letter firms between days -30 to +30. All insider sales during the comment letter periods are insignificantly different than those during the control periods. Panel B of Table 4 presents the insider sales of high short firms for All Other 10-K Comment Letters. While we find some evidence of heighten insider sales in days -20 to -11 and days +21 to +30 of 19.5 and 21.1 percent, respectively, we again find insignificant increases in insider sales for all Other 10-K Comment Letters between days -10 to +20. This evidence is consistent with our expectation that insiders will only unload shares prior to the public disclosure of more important comment letters. Moreover, Tables 3 and 4 provide evidence suggesting that our revenue recognition partition is successful in identifying more important comment letters. Panels A and B of Figure 1 plot the differences in insider sales presented in Table 3 for all Revenue Recognition Comment Letter firms and for Revenue Recognition Comment Letter firms in the top quartile of short interest. Panels C and D of Figure 1 plot the differences in 18 insider sales for All Other 10-K Comment Letter firms and for All Other 10-K Comment Letter firms in the top quartile of short interest presented in Table 4. Panels A and B of Figure 2 plot the cumulative daily insider trading of high short interest firms during the 30 days prior to 30 days following the comment letter disclosure date relative to control observations for Revenue Recognition Comment Letters and All Other 10-K Comment Letters, respectively. Panel A highlights insider sales for firms receiving Revenue Recognition Comment Letters are fairly consistent with normal levels until approximately ten days prior to the comment letter disclosure. At this point, comment letter insider sales begin to significantly diverge from control period observations with approximately 23 percent greater cumulative insider sales in the 30-day window up to the disclosure date. The spread between comment letter insider trades and control observations continues to slowly increase in the 30 days following the comment letter disclosure but the majority of the spread forms in the 10 days prior to the comment letter disclosure. Panel B illustrates that for the All Other 10-K Comment Letters, insiders do not sell appreciably more stock than control period observations, with approximately 2 percent greater sales by the disclosure date. Overall, the findings in Tables 3 and 4 and Figures 1 and 2 provide evidence consistent with our predictions that insiders opportunistically sell their shares prior to the SEC’s public release of comments letters pertaining to significant disclosure and reporting issues, and that this practice is more pronounced for firms with high short interest. In Table 5, we present regressions examining the determinants of pre-disclosure insider sales using both comment letter and firm characteristics. In Columns (1) to (3), the dependent variable is the mean daily insider sales as a percentage of shares outstanding for days -5 to -1 and in Column (4) the dependent variable is insider sales as a percentage of shares outstanding for days -10 to -1. The coefficient on the Revenue Recognition indicator variable of 0.0014 (p < 19 0.01) in Column (1) suggests that insider sales, as a percentage of shares outstanding, is 0.0014 higher in the five trading days prior to the comment letter disclosure date for Revenue Recognition Comment Letters than for All Other 10-K Comment Letters. This finding is consistent with the inference from Tables 3 and 4 that our revenue recognition partition is successful in identifying more important comment letters as suggested by insider sales. While the Book to Market ratio is insignificant, the coefficients on Earnings to Price and Log Total Assets are 0.0003 and -0.0001, respectively, and significant (p < 0.10 and p < 0.01 respectively), suggesting that pre-disclosure insider sales are more pronounced for higher earnings to price and smaller firms. Moreover, CAR -30 to -1 is positive (0.0061) and significant (p < 0.01). A standard deviation increase in Earnings to Price or CAR -30 to -1 corresponds to increases in insider sales, as a percentage of shares outstanding, of 0.0003 or 0.0009, respectively, while a standard deviation increase in Total Assets, corresponds to a decrease of -0.0002. Column (2) includes the High Short Interest indicator variable and yields similar inferences for all other variables as in Column (1). The coefficient of 0.0016 (p < 0.01) on High Short Interest is positive and significant, suggesting that firms in the top quartile of short interest have higher insider sales, as a percentage of shares outstanding, of 0.0016 for all types of 10-K comment letters. In Column (3) we add an interaction term between the Revenue Recognition indicator variable and the High Short Interest indicator variable. While the coefficients on the Revenue Recognition and High Short Interest variables decline to 0.0006 (p < 0.10) and 0.0010 (p < 0.01), respectively, the interaction term between the two variables is 0.0027 (p < 0.05), confirming Table 3’s result that pre-disclosure insider sales are most pronounced for firms receiving Revenue Recognition Comment Letters that also have high short interest. Inferences in Column (4) are similar to those in Column (3) when the dependent variable is insider sales as a 20 percentage of shares outstanding for days -10 to -1. Together, these findings suggest that while valuation multiples, firm size, and past returns are determinants of pre-disclosure insider sales, high short interest appears to be a very important firm attribute in explaining pre-disclosure insider sales. In Table 6, we examine the cumulative abnormal returns (CARs) following the comment letter disclosure date. Panel A presents the CARs for the Revenue Recognition Comment Letters and Panel B presents the CARs for All Other 10-K Comment Letters. We calculate CAR as the cumulative daily stock return less the appropriate CRSP capitalization portfolio return using the prior fiscal year-end portfolio assignment. Panel A highlights that CARs for Revenue Recognition Comment Letters as of day +1 are -23.2 basis points (p < 0.05), and continue to decrease to -43.7 basis points (p < 0.05) and -59.0 basis points (p < 0.05) as of day +5 and +15, respectively. By day +15 approximately 54.1 percent of firms have negative CARs. Figure 3 Panel A provides a plot of these results. This analysis indicates that there is a negative and significant market response to the public disclosure of Revenue Recognition Comment Letter correspondence. Panel B of Table 6 provides results for the All Other 10-K Comment Letters sample. We find positive CARs in the weeks following the comment letter disclosure date. The CAR as of day +1 is 1.3 basis points (p > 0.50) but increases to 18.2 basis points (p < 0.10), 39.2 basis points (p < 0.01), and 40.5 basis points (p < 0.01) by days +5, +10, and +15, respectively. By day +15 approximately 49.4 percent of firms have positive CARs. These findings, (plotted in Figure 3, Panel B) confirm that our comment letter partition is effective in separating important comment letters from those that are more trivial. Furthermore, the positive returns following the 21 disclosure of comment letters identifying trivial concerns suggests that the market views such findings as a passing grade on the company’s financial reporting practices. 4.3. Robustness Tests We examine the robustness of our main results by performing additional analyses using alternative specifications and methods. First, in our main analyses, we determine abnormal insider sales using the firm as its own control by randomly selecting five control observations (when available) between 3 and 6 months before and after the comment letter disclosure date. However, using the firm as its own control matches different periods and doesn’t effectively control for macroeconomic and financial reporting events that could influence insider sales. Hence, we also calculate abnormal insider sales using the insider sales of the comment letter firm’s five closest peers matched by Fama-French industry groups, fiscal year-end date, and closest market capitalization within a -50 to +100 percent range.7 Table 7 presents the insider sales for the 30 days before and after the comment letter disclosure date using the peer matching for our full sample. Table 7, Panel A highlights that insider sales for the Revenue Recognition Comment Letter firms are insignificantly different from their peers from days -30 to -11 but are significantly higher than their peers from days -10 to +30. Specifically, insider sales are 30.9 percent (p < 0.05), 65.0 percent (p < 0.01), 76.4 percent (p < 0.01), and 123.6 percent (p < 0.01) higher than their peer sales during days -10 to -6, -5 to -1, 0 to +5, and +6 to +10, respectively. In terms of dollar amounts, this suggests that for excess insider sales for the mean firm is $367,500 more shares ((0.000005 x $4.9 billion x 5 days) + (0.000010 x $4.9 billion x 5 days)) than we would expect in the ten days before the 7 The extent of pre-disclosure sales are more pronounced when using only one control firm, as using five control firms provides a lower estimate of abnormal trading than does one control firm. 22 disclosure, reflecting total abnormal insider sales of approximately $514 million ($367,500 x 1,398 firms). Table 7, Panel B presents insider sales for the All Other 10-K Comment Letters. Except for days -30 to -21, 0 to +5, +21 to +30 where comment letter insider sales are 12.4 percent (p < 0.05), 19.3 percent (p < 0.05) and 11.7 percent (p < 0.05) higher than peer sales during these windows, comment letter insider sales during all other windows surrounding the comment letter disclosures are insignificantly different than peer insider sales. Moreover, consistent with Table 4, Panel A there is no evidence of heighten pre-disclosure sales by insiders in the 20 days prior to the disclosure date for All Other 10-K Comment Letters. Overall, the evidence of the pre-disclosure insider sales for Revenue Recognition Comment Letters is more pronounced using the firm’s peers to infer abnormal insider sales. Second, we limit our sample to those comment letter disclosures where the actual disclosure is released within five days of the anticipated SEC disclosure date. Specifically, we examine insider sales for comment letters released only during the window from five days prior to five days after the expected release date, which is 45 calendar days prior to 2012, and 20 business days thereafter. Table 8, Panels A and B present the results for the full Revenue Recognition Comment Letter sample and the high short interest Revenue Recognition Comment Letter sample, respectively. Although the full sample is reduced by approximately 34 percent, we still find increases in insider sales in the -5 to -1 days prior to disclosure of 42.8 percent. Moreover, in Panel B, the increase in insider sales for the high short interest Revenue Recognition Comment Letters in the -5 to -1 days prior to disclosure is 125.5 percent (up from 108.6 percent in Table 3). These results suggest that in some settings we underestimate the amount of insider selling in our main analyses, since a portion of the comment letters are delayed 23 or accelerated from the generally expected disclosure date, eliminating the ability for insiders to sell or for us to identify selling in the period immediately prior to disclosure. Third, we examine Form 8-K disclosures for the firms with insider sales in the predisclosure window. If firms issue press releases (Form 8-Ks) that effectively summarize the key issues in the comment letters prior to the public disclosure of the comment letter correspondence, then going forward, insiders will not be trading on their private information of the SEC’s filing review process. We reviewed Revenue Recognition Comment Letter firms accounting for more than 90 percent of the total abnormal insider sales in the five days prior to the public disclosure of the comment letter correspondence and checked to see whether any of these firms issued an 8K relating to the comment letter correspondence between the dates of the filing review completion and the public disclosure of the correspondence. We found that none of the firms that we selected disclosed an 8-K previewing the comment letter correspondence, suggesting that details of the comment letter correspondence are not disclosed by companies prior to the SEC’s public disclosure. Moreover, if accelerated filers or large accelerated filers have unresolved SEC comments outstanding for at least 180 days as of the fiscal year-end date, then in Section 1B of the 10-K these firms must disclose that there are unresolved staff comments. There were only two Revenue Recognition Comment Letter firms with insiders selling shares that were required to make such disclosures; however, upon reviewing their 10-Ks we found that both of these firms failed to make the appropriate Section 1B disclosures. Hence, these firms not only violated SEC’s disclosure requirements, but their insiders appear to trade on non-public information concerning the comment letter review findings. Fourth, we also limit our sample to those comment letter disclosures without restatements between the first comment letter date and the disclosure date. We find that our main inferences 24 are unchanged and that the pre-disclosure trading documented in our main analyses is not driven by firms restating their financial statements during the review and then trading on this public information prior to the comment letter disclosure. Fifth, we investigate the types of insiders engaging in sales prior to the comment letter disclosure date. We find that other officers (all officers excluding the CEO and CFO) account for 44.2 percent of all insider sales (as a percentage of the number of sales) during the 10 trading days prior to the comment letter disclosure, followed by CEOs accounting for 27.1 percent, other directors (all directors excluding officers) accounting for 22.8 percent, and CFOs accounting for 5.9 percent. As a result, it appears that both executives, including the CEO, and directors are responsible for the observed insider sales patterns. Lastly, as insiders have regular black-out periods before earnings announcements and in turn, have heightened sales following earnings announcements, we examine whether comment letter disclosures abnormally overlap with earnings announcements. We find evidence that the timing of comment letter disclosures is random relative to earnings announcements as there are no abnormal levels of earnings announcements during days -30 to +30 period and as the median earnings announcement is -45 days prior to the comment letter disclosure. Hence, this analysis confirms that our findings are not driven by post earnings announcement insider sales. 5. Conclusion The SEC’s Division of Corporation Finance reviews company filings and where appropriate, issues comment letters to elicit better compliance with applicable financial reporting standards. The SEC however, affords companies who were deficient in their regulatory filing 25 requirements a significant amount of time after the completion of the Division’s review to sit on private information concerning the review findings before this information is publicly disclosed. This study provides empirical evidence suggesting that corporate insiders take advantage of this opportunity and unload shares prior to the public release of SEC comment letter correspondence pertaining to material financial disclosure and reporting issues. We highlight that this practice appears most pronounced for firms with significant short-oriented hedge fund attention, which we proxy for using the top quartile of short interest. While the results suggest that insiders significantly increase equity sales in the ten days prior to the comment letter disclosure date, the findings cannot speak to whether such trading is inappropriate or illegal. Our results provide strong evidence suggesting that the current practice of delaying the real-time disclosure of SEC comment letter correspondence should be revisited and support calls from the investment industry that the disclosure delay appears to best serve the interests of corporate insiders. Moreover, the results have implications for corporate boards in regards to whether insider sale black-out periods should be implemented prior to the comment letter disclosure date. 26 REFERENCES Bloomberg. 2013. Muddy Waters Secret China Weapon Is On SEC Website. February 19, 2013. Cassell, C., L. Dreher, and L. Myers. 2013. Reviewing the SEC’s review process: 10-K comment letters and the cost of remediation. Working Paper, University of Arkansas. Cheng, Q., and K. Lo. 2006. Insider trading and voluntary disclosures. Journal of Accounting Research 44 (5): 815-848. Ciesielski, J. 2012. The Analyst’s Accounting Observer—What Keeps The SEC Busy – 2013. December 18, 2012. ——–. 2013. The Analyst’s Accounting Observer—2012 Annual Reports: A Peek Inside. March 8, 2013. Cohen, L., C. Malloy, and L. Pomosrki. 2012. Decoding inside information. Journal of Finance 67 (3): 1009-1043. Curtis, A., and N. Fargher. 2008. Does short-selling amplify price declines or align stocks with their fundamental values? Working Paper, University of Utah. Dechow, P., W. Ge, C. Larson, and R. Sloan. 2011. Predicting material accounting misstatements. Contemporary Accounting Research 28 (1): 17-82. ——–, A. Hutton, L. Meulbroek, and R. Sloan. 2001. Short-sellers, fundamental analysis, and firm returns. Journal of Financial Economics 61 (1): 71-90. Ertimur, Y., and M. Nondorf. 2006. IPO firms and the SEC comment letter process. Working Paper, Duke University. Ettredge, M., K. Johnstone, M. Stone, and Q. Wang. 2011. The effects of firms size, corporate governance quality, and bad news on disclosure compliance. Review of Accounting Studies 16 (4): 866-889. Forbes. 2007. Naked Short Victim Strikes Back. February 2, 2007. Harlow, W.V., and J. Howe. 1998. Leveraged buyouts and insider nontrading. Financial Management 22 (1): 109-118. Hirshleifer, D., S.H. Teoh, and J. Yu. 2011. Short arbitrage, return asymmetry, and the accrual anomaly. Review of Financial Studies 24 (7): 2429-2461. Jagolinzer, A. 2009. SEC Rule 10b5-1 and insiders’ strategic trade. Management Science 55 (2): 224-239. Johnston, R., and R. Petacchi. 2012. Regulatory oversight of financial reporting: Securities and Exchange Commission comment letters. Working Paper, Purdue University. Karpoff, J., and X. Lou. 2010. Short sellers and financial misconduct. Journal of Finance 65 (5): 1879-1913. 27 Keown, A., and J. Pinkerton. 1981. Merger announcements and insider trading activity: An empirical investigation. Journal of Finance 36 (4): 855-869. Kyle, A. 1985. Continuous auctions and insider trading. Econometrica 53 (6): 1315-1335. Lamont, O. 2012. Go down fighting: Short sellers vs. firms. Review of Asset Pricing Studies 2 (1): 1-30. Lee, S., W. Mikkelson, and M. Partch. 1992. Managers’ trading around stock repurchases. Journal of Finance 47 (5): 1947-1961. Li, O.Z., and Y. Zhang. 2006. Financial restatement announcements and insider trading. Working Paper, University of Notre Dame. Lie, E. 2005. On the timing of CEO stock option awards. Management Science 51 (5): 802-812. Penman, S. 1982. Insider trading and the dissemination of firms’ forecast information. Journal of Business 55 (4): 479-503. Robinson, J., Y. Xue, and Y. Yu. 2011. Determinants of disclosure noncompliance and the effect of the SEC review: Evidence from the 2006 mandated compensation disclosure regulations. The Accounting Review 86 (4): 1415–1444. Securities and Exchange Commission (SEC). 2013a. Division of Corporation Finance: Filing Review Process. Available at: http://www.sec.gov/divisions/corpfin/cffilingreview.htm ——–. 2013b. SEC Staff to Release Filing Review Correspondence Earlier. Available at: http://www.sec.gov/divisions/corpfin/cfannouncements/edgarcorrespondence.htm ——–. 2013c. Summary by the Division of Corporation Finance of Significant Issues Addressed in the Review of the Periodic Reports of the Fortune 500 Companies. Available at: http://www.sec.gov/divisions/corpfin/fortune500rep.htm ——–. 2013d. The Investor’s Advocate: How the SEC Protects Investors, Maintains Market Integrity, and Facilitates Capital Formation. Available at: http://www.sec.gov/about/whatwedo.shtml SEC Insight Inc. 2004. Response to “Proposal to Publicly Release Staff Comment Letters and Non-Confidential Response Letters. September 30, 2004. 28 EXHIBIT 1 SEC Comment Letter Timeline Company responds and receives follow up comment letters Median = 131 calendar days Company receives initial comment letter Company required to restate financial statements (less than 1 percent of sample) Potential Insider Trading Window 2012 Median = 20 business days Pre-2012 Median = 46 calendar days Company receives letter indicating review is complete Company letter correspondence disclosure date Outcomes: (1) Company’s responses are sufficient (2) SEC requests changes to company’s future financial statements 29 APPENDIX A SEC Organizational Chart (Source: http://www.sec.gov/images/secorg.pdf) 30 APPENDIX B Panel 1: Sample Revenue Recognition Comment Letter Mr. C. Bradford Richmond Darden Restaurants, Inc. August 29, 2011 Page 2 methodology for being able to reasonably and objectively determine the amount of gift card breakage in addition to the estimated time period of actual gift card redemption. We may have further comment upon receipt of your response. August 29, 2011 Via E-Mail Mr. C. Bradford Richmond Chief Financial Officer Darden Restaurants, Inc. 1000 Darden Center Drive Orlando, Florida 32837 Re: Darden Restaurants, Inc. Form 10-K for the year ended May 29, 2011 Filed July 22, 2011 File No. 001-13666 We urge all persons who are responsible for the accuracy and adequacy of the disclosure in the filing to be certain that the filing includes the information the Securities Exchange Act of 1934 and all applicable Exchange Act rules require. Since the company and its management are in possession of all facts relating to a company’s disclosure, they are responsible for the accuracy and adequacy of the disclosures they have made. In connection with responding to our comment s, please provide, in writing, a statement from the company acknowledging that:  the company is responsible for the adequacy and accuracy of the disclosure in the filing;  staff comments or changes to disclosure in response to staff comments do not foreclose the Commission from taking any action with respect to the filing; and  the company may not assert staff comments as a defense in any proceeding initiated by the Commission or any person under the federal securities laws of the United States. Dear Mr. Richmond: We have reviewed your filing and have the following comments. In some of our comments, we may ask you to provide us with information so we may better understand your disclosure. Please respond to this letter within ten business days by amending your filing, by providing the requested information, or by advising us when you will provide the requested response. If you do not believe our comments apply to your fact s and circumstances or do not believe an amendment is appropriate, please tell us why in your response. You may contact Effie Simpson at (202) 551-3346, or in her absence, Jean Yu at (202) 551-3305 if you have questions regarding comments on the financial statements and related matters. Please contact the undersigned with any other questions at (202) 551-3750. Sincerely, After reviewing any amendment to your filing and the information you provide in response to these comments, we may have additional comments. Form 10-K for the fiscal year ended May 29, 2011 Exhibit 13 Unearned Revenues, page 30 /s/ Linda Cvrkel Linda Cvrkel Branch Chief 1. We note that you recognize breakage for unused gift card amounts in proportion to actual gift card redemptions. We also note that during fiscal 2010 you changed your estimate of gift card breakage and adjusted unearned revenue with a corresponding reduction in gift card breakage of $20.4 million as a result of a significantly higher trend in gift card redemption. In light of such current consumer redemption behavior, the increase in gift card sales and that unearned revenue represented $200 and $172.7 million of your balance sheet at May 31, 2011 and 2010, respectively, please tell us and expand your unearned revenue policy footnote to disclose the estimate value or percentage of gift card sales that you recognize as breakage for each period presented and the period over which breakage is recognized. Furthermore, as part of your response please describe for us your 31 APPENDIX B Panel 2: Sample All Other 10-K Comment Letter Martin M. Ellen Dr Pepper Snapple Group, Inc. April 6, 2011 Page 2 April 6, 2011 Via E-mail Martin M. Ellen Executive Vice President and Chief Financial Officer Dr Pepper Snapple Group, Inc. 5301 Legacy Drive, Plano, TX 75024 Re: Dr Pepper Snapple Group, Inc. Form 10-K Filed February 22, 2011 File No. 001-33829 Dear Mr. Ellen: We have reviewed your filing and have the following comment. In some of our comments, we may ask you to provide us with information so we may better understand your disclosure. In responding to our comments, please provide a written statement from the company acknowledging that: • the company is responsible for the adequacy and accuracy of the disclosure in the filing; • staff comments or changes to disclosure in response to staff comments do not foreclose the Commission from taking any action with respect to the filing; and • the company may not assert staff comments as a defense in any proceeding initiated by the Commission or any person under the federal securities laws of the United States. Please contact Damon Colbert at (202) 551-3581 or Pamela Howell at (202) 551-3357 with any questions. Sincerely, /s/ Pamela Howell for John Reynolds Assistant Director Please respond to this letter within ten business days by amending your filing, by providing the requested information, or by advising us when you will provide the requested response. If you do not believe our comment applies to your facts and circumstances or do not believe an amendment is appropriate, please tell us why in your response. After reviewing any amendment to your filing and the information you provide in response to this comment, we may have additional comments. General We note the disclosure on page six that on February 26, 2010 you completed the licensing of certain brands to PepsiCo and the disclosure on page seven that on October 4, 2010 you completed the licensing of certain brands to Coca-Cola. We note that these licensing agreements have not been filed as exhibits. Please confirm that you will file these exhibits with your next periodic report or explain why you believe these exhibits are not material. We urge all persons who are responsible for the accuracy and adequacy of the disclosure in the filing to be certain that the filing includes the information the Securities Exchange Act of 1934 and all applicable Exchange Act rules require. Since the company and its management are in possession of all facts relating to a company’s disclosure, they are responsible for the accuracy and adequacy of the disclosures they have made. 32 FIGURE 1 Mean Daily Insider Sales Relative to Comment Letter Disclosure Date This figure presents the mean daily insider sales relative to the comment disclosure date for Revenue Recognition Comment Letters and All Other 10-K Comment Letters. Panel A illustrates that for firms with Revenue Recognition Comment Letters, insider sales are more pronounced around the disclosure of the comment letters, including increased insider sales in the five days before the comment letter disclosure. Panel B illustrates that for firms with Revenue Recognition Comment Letters combined with high short-interest (top 25%), insider sales are even more pronounced around the disclosure of the comment letters, including increased insider sales in the five days before the comment letter disclosure. These same patterns are not evident in Panel C and Panel D for All Other 10-K Comment Letter firms in either the full or high short interest samples. Insider sales are measured as the percent of shares outstanding sold by insiders each day, with non-zero values winsorized at the 1% level (SHARES from Thomson Reuters insider database divided by SHROUT from CRSP scaled as a percent of shares outstanding). The Revenue Recognition Comment Letter full sample comprises 1,560 observations and the high short interest sample comprises 441 Revenue Recognition Comment Letter observations. The All Other 10-K Comment Letter full sample comprises 6,404 observations and the high short interest sample comprises 1,557 All Other 10-K Comment Letter observations. Control period observations are comprised of up to 5 observations from the same firm with event dates selected randomly from between 3 to 6 months before and after the comment letter disclosure date. 33 FIGURE 2 Cumulative Daily Insider Sales for High Short Interest Firms Relative to Comment Letter Disclosure Date This panel presents the cumulative difference in daily insider sales from 30 days prior to 30 days following the comment disclosure date for Revenue Recognition Comment Letters and All Other 10-K Comment Letters, in the presence of high short interest. Panel A illustrates that for firms with Revenue Recognition Comment Letters, insider sales begin to diverge noticeably from control period insider sales beginning 10 days prior to disclosure, with approximately 23% greater cumulative insider sales by the disclosure date. Panel B illustrates that for the All Other 10K Comment Letter sample, comment letter firms do not sell appreciably more stock than the control period observations, with approximately 2% greater sales by the disclosure date. The high short interest sample comprises 441 Revenue Recognition observations and 1,557 All Other 10-K Comment Letter observations. Control period observations are comprised of up to 5 observations from the same firm with event dates selected randomly from between 3 to 6 months before and after the comment letter disclosure date. 34 FIGURE 3 Cumulative Abnormal Returns Relative to Comment Letter Disclosure Date This figure shows the cumulative abnormal returns for the 15 days following the 10-K related comment letter disclosure. Panel A shows that cumulative abnormal returns for Revenue Recognition Comment Letter firms are negative for the days subsequent to comment letter disclosure. Panel B shows that All Other 10-K Comment Letters have positive cumulative abnormal returns subsequent to comment letter disclosure. Cumulative abnormal return is measured as the daily total return less the appropriate CRSP capitalization portfolio return using the prior fiscal yearend portfolio assignment. 35 TABLE 1 Panel A: Comment Letter Sample Distribution 1) Unique 10-K comment letter conversations 2006 2007 2008 2009 2010 2011 2012 Total All 10-K Related Revenue Recognition 2,045 1,347 1,671 2,299 2,666 2,388 2,354 14,770 1,168 769 633 560 524 521 593 4,768 2) Subset of 1 with valid matches to Thomson Reuters Insiders and Compustat All 10-K Revenue Related Recognition 1,187 864 1,112 1,451 1,587 1,344 1,451 8,996 342 242 230 235 229 230 263 1,771 3) Subset of 2 with CRSP data during event period and at least one valid control firm. All 10-K Revenue Related Recognition 1,016 774 1,000 1,301 1,443 1,235 1,195 7,964 289 219 205 211 204 215 217 1,560 This table shows the sample formation and the disclosure-year distribution of comment letter conversations pertaining to Form 10-K filings. Comment letter conversations are a related series of SEC Division of Corporation Finance comment letters (Form “UPLOAD”s) and company response letters (Form “CORRESP”s). Subset 1 includes all unique conversations in Audit Analytics Comment Letter database for the years selected, with any duplicate conversations eliminated. Subset 2 includes all those conversations from Subset 1 that can be matched to both the Thomson Reuters Insiders database and the Compustat Fundamentals Annual database. Subset 3 is the final comment letter sample used in this analysis, where comment letters in Subset 2 can be matched to at least one and five control observations of the same firm, with a random disclosure date generated in the window of 3 to 6 months before and after the comment letter disclosure date. 36 TABLE 1 Panel B: Comment Letter Industry Distribution % Compustat % Comment Letters Aero 0.5 % Revenue Recognition 0.5 27.9 % Compustat % Comment Letters Insur 3.2 % Revenue Recognition 4.0 6.3 Agric 0.3 0.2 36.8 LabEq 2.0 2.0 26.8 Autos 1.4 1.4 15.8 Mach 2.6 2.8 21.7 Banks 11.8 8.3 5.0 Meals 1.4 1.2 8.2 Beer 0.3 0.2 - MedEq 3.0 2.9 31.7 BldMt 1.3 1.7 15.2 Mines 0.7 0.5 7.3 Books 0.5 0.4 23.5 Oil 4.4 4.7 7.0 Boxes 0.2 0.3 - Other 0.6 0.6 21.3 BusSv 4.5 4.6 19.2 Paper 1.0 1.1 9.5 Chems 2.0 2.2 14.6 PerSv 1.1 1.2 24.7 Chips 6.2 5.4 29.8 RlEst 0.9 0.7 16.7 Clths 1.1 1.2 16.7 Rtail 3.8 4.8 24.2 Cnstr 0.9 1.3 23.5 Rubbr 0.5 0.7 9.6 Coal 0.3 0.4 5.7 Ships 0.2 0.3 26.1 Drugs 7.0 5.9 34.0 Smoke 0.1 0.1 - ElcEq 1.6 1.5 33.9 Soda 0.3 0.3 20.0 FabPr 0.2 0.2 29.4 Softw 7.4 6.4 42.1 Fin 5.9 8.4 10.0 Steel 1.2 1.4 16.5 Food 1.4 1.4 16.7 Telcm 3.3 2.3 11.1 Fun 1.1 1.1 12.8 Toys 0.5 0.5 22.5 Gold 0.8 0.2 16.7 Trans 3.0 2.5 22.7 Guns 0.2 0.3 28.6 Txtls 0.2 0.3 - Hardw 1.8 2.1 39.0 Util 2.5 3.3 7.6 Hlth 1.4 1.5 27.6 Whlsl 2.7 3.0 23.1 Hshld 1.1 1.3 19.2 This table illustrates the distribution of firms, comment letters, and revenue recognition comment letters by FamaFrench 49-industry group. % Compustat is the percentage of firm-years in each group in Compustat for the years 20062012. % Comment Letters is the percentage of comment letters in each group in our sample (n = 7,964). % Revenue Recognition indicates the percentage of comment letters in the industry group that are coded as revenue recognitionrelated (the overall sample distribution of revenue recognition letters is 19.6%). 37 TABLE 2 Descriptive Statistics All 10-K Related Comment Letter Firms (n = 7,964 Unique Firms = 3,572) Mean Median Insider sales, daily (% shares outstanding) Insider sales, daily ($) Total assets ($MM) Revenues ($MM) Market capitalization ($MM) Book-to-market ratio Earnings-toprice ratio CAR -30 to -1 Short interest (% shares outstanding) All Revenue Recognition Comment Letter Firms (n = 1,560 Unique Firms = 1,263) Mean Median All Other 10-K Comment Letter Firms (n = 6,404 Unique Firms = 3,244) Mean Median 0.0016 0 0.0025 0 0.0014 0 66,756 0 60,111 0 68,374 0 11,881 1,115 8,367 497 12,736 1,326 4,643 624 3,783 421 4,852 686 5,769 852 4,915 626 5,978 918 0.709 0.504 0.566 0.412 0.743 0.532 -0.062 0.042 -0.052 0.032 -0.064 0.045 -0.001 -0.006 0.003 -0.005 -0.001 -0.007 4.94 3.33 5.43 3.64 4.83 3.28 This table presents the descriptive statistics of the all 10-K comment letter firms in our sample, as well as the subsets of Revenue Recognition Comment Letter firms and All Other 10-K Comment Letter Firms, for the fiscal year prior to the comment letter disclosure date. Insider sales (% shares outstanding) is calculated as the daily shares sold, from Thomson Reuters’ Insider database (SHARES where TRANCODE is “S”), by insiders, divided by the number of shares outstanding in CRSP (SHROUT) scaled as a percent of shares outstanding. Insiders are defined as sales by officers and directors from the Thomson Reuters insider database (ROLECODE of: “CEO”, “D”, “O”, “H”, “DO”, “OD”, “VC”, “OB”, “OP”, “OT”, “CB”, “AV”, “CFO”, “CI”, “CO”, “CT”, “EVP”, “OX”, “P”, “S”, “SVP”, or “VP”). Insider sales ($) is calculated as the daily shares sold in dollars by officers and directors from the Thomson Reuters insider database (ROLECODE of: “CEO”, “D”, “O”, “H”, “DO”, “OD”, “VC”, “OB”, “OP”, “OT”, “CB”, “AV”, “CFO”, “CI”, “CO”, “CT”, “EVP”, “OX”, “P”, “S”, “SVP”, or “VP”). Total assets is the prior fiscal year-end total asset value from Compustat (AT). Total revenues is the prior fiscal year revenues from Compustat (SALE). Market capitalization is the prior fiscal year-end market capitalization from Compustat (CSHO * PRCC_F). Book-to-market is the ratio at the comment letter disclosure date (Compustat SEQ / (CRSP PRC * CRSP SHROUT / 1000)). Earnings to price ratio is prior fiscal year end operating earnings divided by the market capitalization at the comment letter disclosure date ((Compustat IB) / (CRSP PRC * CRSP SHROUT / 1000)). CAR -30 to -1 is calculated as the daily total return less the appropriate CRSP capitalization portfolio return using the prior fiscal year-end portfolio assignment, for the 30 days prior to the comment letter disclosure date. Short interest, as a percentage of shares outstanding, is calculated as the number of shares sold short prior to the disclosure date from the Compustat short interest database, divided by shares outstanding from CRSP (SHORT/(SHROUT*1000)*100%). 38 TABLE 3 Differences in Insider Sales between Comment Letter Firms and Control Period Observations Revenue Recognition Comment Letter Firms Comment letter insider sales Control period insider sales Difference % Diff. t-stat p-value 0.0017 0.0016 0.0021 0.0027 0.0018 0.0018 0.0018 0.0019 -0.0001 -0.0001 0.0003 0.0009 -7.4 -7.5 17.0 46.1 -0.844 -0.858 1.349 2.735 0.801 0.805 0.089 0.003 0.0024 0.0032 0.0020 0.0021 1,560 0.0017 0.0015 0.0019 0.0018 0.0007 0.0016 0.0001 0.0002 42.4 107.3 6.5 12.7 2.586 4.788 0.691 1.307 0.005 0.000 0.245 0.096 0.0026 0.0027 0.0033 0.0053 0.0026 0.0026 0.0023 0.0025 -0.0001 0.0000 0.0010 0.0028 -2.7 1.8 44.9 108.6 -0.189 0.118 1.899 3.255 0.575 0.453 0.029 0.001 0.0025 0.0034 0.0031 0.0035 441 0.0027 0.0021 0.0025 0.0026 -0.0002 0.0013 0.0006 0.0008 -5.7 61.8 23.8 32.5 -0.295 2.134 1.427 1.883 0.616 0.016 0.077 0.030 Panel A: Full Sample Pre-Disclosure Insider Sales Insider sales during days -30 to -21 Insider sales during days -20 to -11 Insider sales during days -10 to -6 Insider sales during days -5 to -1 Post-Disclosure Insider Sales Insider sales during days 0 to +5 Insider sales during days +6 to +10 Insider sales during days +11 to +20 Insider sales during days +21 to +30 Number of observations Panel B: High Short Interest Sample Pre-Disclosure Insider Sales Insider sales during days -30 to -21 Insider sales during days -20 to -11 Insider sales during days -10 to -6 Insider sales during days -5 to -1 Post-Disclosure Insider Sales Insider sales during days 0 to +5 Insider sales during days +6 to +10 Insider sales during days +11 to +20 Insider sales during days +21 to +30 Number of observations This table presents differences in insider sales between comment letter firms and control period observations for Revenue Recognition Comment Letter firms. Insider sales are measured as the daily percent of shares outstanding sold by insiders, with non-zero values winsorized at the 1% level (SHARES from Thomson Reuters insider database divided by SHROUT from CRSP scaled as a percent of shares outstanding). T-tests reflect differences in mean daily insider sales, as a percentage of shares outstanding, between comment letter firms and control period observations. Revenue Recognition Comment Letter firms have significantly greater sales in the -5 to -1 day period prior to the comment letter disclosure date than in control periods. Specifically, in Panel A, Revenue Recognition Comment Letter firms’ insiders in the full sample sell 46.1 percent more stock in the five days prior to disclosure than the control period insider sales. In Panel B, Revenue Recognition Comment Letter firms’ insiders in the high short interest sample (top 25% of short interest during the disclosure month) sell 108.6 percent more stock in the five days prior to disclosure than the control period observations. Ttests are one-sided, as we hypothesize that comment letter firm insiders will sell more stock than control period observations around the comment letter disclosure. Control period observations are comprised of up to 5 observations from the same firm with event dates selected randomly from 3 to 6 months before and after the comment letter disclosure. 39 TABLE 4 Differences in Insider Sales between Comment Letter Firms and Control Period Observations All Other 10-K Comment Letter Firms Comment letter insider sales Control period insider sales Difference % Diff. t-stat p-value 0.0014 0.0013 0.0013 0.0012 0.0013 0.0013 0.0013 0.0013 0.0001 0.0000 0.0000 -0.0001 4.6 -2.0 3.6 -8.6 0.821 -0.381 0.442 -1.296 0.206 0.648 0.329 0.902 0.0014 0.0013 0.0013 0.0013 6,404 0.0013 0.0014 0.0013 0.0013 0.0001 -0.0001 0.0000 0.0000 4.3 -4.1 -0.2 3.4 0.611 -0.568 -0.040 0.635 0.270 0.715 0.516 0.263 0.0018 0.0021 0.0018 0.0019 0.0018 0.0017 0.0018 0.0020 -0.0001 0.0003 0.0001 -0.0001 -4.2 19.5 3.5 -2.9 -0.449 1.735 0.260 -0.244 0.673 0.041 0.398 0.596 0.0017 0.0020 0.0018 0.0022 1,557 0.0018 0.0021 0.0018 0.0018 -0.0001 -0.0001 0.0000 0.0004 -4.3 -4.5 -0.5 21.1 -0.364 -0.348 -0.061 1.954 0.642 0.636 0.524 0.025 Panel A: Full Sample Pre-Disclosure Insider Sales Insider sales during days -30 to -21 Insider sales during days -20 to -11 Insider sales during days -10 to -6 Insider sales during days -5 to -1 Post-Disclosure Insider Sales Insider sales during days 0 to +5 Insider sales during days +6 to +10 Insider sales during days +11 to +20 Insider sales during days +21 to +30 Number of observations Panel B: High Short Interest Sample Pre-Disclosure Insider Sales Insider sales during days -30 to -21 Insider sales during days -20 to -11 Insider sales during days -10 to -6 Insider sales during days -5 to -1 Post-Disclosure Insider Sales Insider sales during days 0 to +5 Insider sales during days +6 to +10 Insider sales during days +11 to +20 Insider sales during days +21 to +30 Number of observations This table presents differences in insider sales between comment letter firms and control period observations for All Other 10-K Comment Letter firms. Insider sales are measured as the daily percent of shares outstanding sold by insiders, with non-zero values winsorized at the 1% level (SHARES from Thomson Reuters insider database divided by SHROUT from CRSP scaled as percent of shares outstanding). T-tests reflect differences in mean daily insider sales, as a percentage of shares outstanding, between comment letter firms and control period observations. For All Other 10-K Comment Letter firms, there is no significant increase in insider sales in the -5 to -1 day period in either the full sample in Panel A or the high short interest sample (top 25% of short interest during the disclosure month) in Panel B. T-tests are one-sided, as we hypothesize that comment letter firm insiders will sell more stock than control firms around the comment letter disclosure. Control observations are comprised of up to 5 observations from the same firm with event dates selected randomly from between 3 to 6 months before and after the comment letter disclosure date. 40 TABLE 5 Regression of Pre-Disclosure Insider Sales on Comment Letter Type and Short Interest Dependent variable: Insider Sales % of Shares Outstanding Revenue Recognition -5 to -1 Days (1) -5 to -1 Days (2) -5 to -1 Days (3) -10 to -1 Days (4) 0.0014*** (0.0004) 0.0014*** (0.0004) 0.0006* (0.0003) 0.0005** (0.0002) 0.0016*** (0.0004) 0.0010*** (0.0003) 0.0009*** (0.0002) 0.0027** (0.0014) 0.0018** (0.0009) High Short Interest Revenue Recognition * High Short Interest Earnings to Price 0.0003* (0.0002) 0.0004** (0.0002) 0.0004** (0.0002) 0.0004** (0.0002) Book to Market 0.0003 (0.0004) 0.0003 (0.0004) 0.0003 (0.0004) 0.0002 (0.0004) Log Total Assets −0.0001*** (0.0001) −0.0001*** (0.00005) −0.0001*** (0.00005) −0.0001*** (0.00003) CAR -30 to -1 0.0061*** (0.0016) 0.0061*** (0.0016) 0.0062*** (0.0016) 0.0036*** (0.0008) Constant 0.0020*** (0.0005) 0.0016*** (0.0004) 0.0018*** (0.0004) 0.0018*** (0.0004) Observations Adjusted R2 F Statistic 7,964 0.012 20.94*** 7,964 0.017 23.42*** 7,964 0.019 22.51*** 7,964 0.017 20.02*** This table presents OLS regression results for insider sales in the periods from -5 days to -1 day before the comment letter disclosure date, and from -10 days to -1 day before the disclosure date. The dependent variable is Insider Sales % of Shares Outstanding. Insider sales are measured as the daily percent of shares outstanding sold by insiders, with nonzero values winsorized at the 1% level (SHARES from Thomson Reuters insider database divided by SHROUT from CRSP scaled as a percent of shares outstanding). The independent variables of interest are High Short Interest (indicator equaling “1” if in top 25% of short interest during the disclosure month), Revenue Recognition (indicator equaling “1” if a Revenue Recognition Comment Letter), and the interaction between these two variables. Control variables are the Earnings to Price ratio, Book to Market ratio, Log Total Assets, and Cumulative Abnormal Returns in the 30 days prior to comment letter disclosure (CAR -30 to -1). Model (1) omits the High Short Interest variable, and shows that having a Revenue Recognition Comment Letter is significantly positively associated with increased insider sales in the -5 to -1 day period before the disclosure date. Model (2) includes High Short Interest, and indicates that it is a determinant of increased insider sales. Model (3) adds the interaction of Revenue Recognition and High Short Interest, and finds that the combination of High Short Interest and a Revenue Recognition Comment Letter is a strong predictor of insider sales in the -5 to -1 day period before the disclosure date. Model (4) provides similar inferences as Model (3), but for the -10 to -1 day period before the disclosure date. Robust standard errors. *, **, *** indicate significance at the 0.10, 0.05, and 0.01 levels, respectively, using two-tailed tests. 41 TABLE 6 Significance of Cumulative Abnormal Returns for Comment Letter Firms Comment Letter Firms CAR (%) t-stat p-value -0.232 -0.437 -0.473 -0.590 1,560 -2.154 -2.361 -2.036 -2.112 0.031 0.018 0.042 0.034 0.013 0.182 0.392 0.405 6,404 0.230 1.904 2.928 2.795 0.818 0.057 0.003 0.005 Panel A: Revenue Recognition Comment Letters CAR at day +1 CAR at day +5 CAR at day +10 CAR at day +15 Number of observations Panel B: All Other 10-K Comment Letters CAR at day +1 CAR at day +5 CAR at day +10 CAR at day +15 Number of observations This table presents cumulative abnormal returns for comment letter firms following the comment letter disclosure date. T-tests reflect differences in cumulative abnormal returns from zero. Revenue Recognition Comment Letter firms have a negative CAR in the 0-15 days following comment letter disclosure, whereas All Other 10-K Comment Letter firms have a positive CAR in the 0-15 days following comment letter disclosure. CAR is calculated as the daily total return less the appropriate CRSP capitalization portfolio return using the prior fiscal year-end portfolio assignment. 42 TABLE 7 Differences in Insider Sales Using Matched Firms Sample Comment letter insider sales Panel A: Revenue Recognition Comment Letters Pre-Disclosure Insider Sales Insider sales during days -30 to -21 Insider sales during days -20 to -11 Insider sales during days -10 to -6 Insider sales during days -5 to -1 Post-Disclosure Insider Sales Insider sales during days 0 to +5 Insider sales during days +6 to +10 Insider sales during days +11 to +20 Insider sales during days +21 to +30 Number of observations Control firm insider sales Difference % Diff. t-stat p-value 0.0017 0.0017 0.0021 0.0024 0.0015 0.0016 0.0016 0.0015 0.0002 0.0001 0.0005 0.0010 15.8 7.5 30.9 65.0 1.334 0.667 1.930 2.986 0.091 0.252 0.027 0.001 0.0027 0.0030 0.0021 0.0022 1,398 0.0015 0.0014 0.0014 0.0016 6,237 0.0012 0.0017 0.0007 0.0006 76.4 123.6 45.0 39.1 3.522 4.557 3.252 3.030 0.000 0.000 0.001 0.001 0.0013 0.0012 0.0013 0.0012 0.0012 0.0012 0.0012 0.0013 0.0001 0.0001 0.0001 -0.0001 12.4 8.1 10.3 -8.6 1.841 1.232 1.096 -1.185 0.033 0.109 0.137 0.882 0.0014 0.0013 0.0013 0.0013 5,761 0.0012 0.0011 0.0012 0.0012 26,062 0.0002 0.0001 0.0001 0.0001 19.3 11.5 6.6 11.7 2.203 1.277 1.091 1.825 0.014 0.101 0.138 0.034 Panel B: All Other 10-K Comment Letters Pre-Disclosure Insider Sales Insider sales during days -30 to -21 Insider sales during days -20 to -11 Insider sales during days -10 to -6 Insider sales during days -5 to -1 Post-Disclosure Insider Sales Insider sales during days 0 to +5 Insider sales during days +6 to +10 Insider sales during days +11 to +20 Insider sales during days +21 to +30 Number of observations This table presents differences in insider sales between comment letter firms and matched control firms. Insider sales are measured as the daily percent of shares outstanding sold by insiders, with non-zero values winsorized at the 1% level (SHARES from Thomson Reuters insider database divided by SHROUT from CRSP scaled as a percent of shares outstanding). T-tests reflect differences in mean daily insider sales, as a percentage of shares outstanding, between comment letter firms and control firms. Revenue Recognition Comment Letter firms have significantly greater sales in the -5 to -1 day period relative to the comment letter disclosure date. Specifically, Revenue Recognition Comment Letter firms’ insiders sell 65.0 percent more stock in the five days prior to disclosure than control firm insiders. For All Other 10-K Comment Letter firms, there is no significant increase in insider sales in the -5 to -1 day period. T-tests are one-sided, as we hypothesize that comment letter firm insiders will sell more stock than control firms around the comment letter disclosure. Control firms are matched on Fama-French industry groupings, fiscal year-end date, and closest market capitalization within a -50 to 100 percent range. Five control firms are selected for each comment letter firm. 43 TABLE 8 Differences in Insider Sales between Comment Letter Firms and Control Period Observations Comment Letters when Disclosed as Expected Comment letter insider sales Control period insider sales Difference % Diff. t-stat p-value 0.0016 0.0015 0.0019 0.0027 0.0018 0.0021 0.0021 0.0019 -0.0002 -0.0006 -0.0001 0.0008 -13.0 -29.1 -6.7 42.8 -1.297 -3.339 -0.510 2.161 0.903 1.000 0.695 0.015 0.0021 0.0027 0.0019 0.0020 1,034 0.0018 0.0017 0.0015 0.0016 0.0002 0.0010 0.0004 0.0004 12.9 60.7 29.0 24.4 0.842 2.756 2.083 1.770 0.200 0.003 0.019 0.038 Panel A: Revenue Recognition Comment Letters Pre-Disclosure Insider Sales Insider sales during days -30 to -21 Insider sales during days -20 to -11 Insider sales during days -10 to -6 Insider sales during days -5 to -1 Post-Disclosure Insider Sales Insider sales during days 0 to +5 Insider sales during days +6 to +10 Insider sales during days +11 to +20 Insider sales during days +21 to +30 Number of observations Panel B: High Short Interest Revenue Recognition Comment Letters Pre-Disclosure Insider Sales Insider sales during days -30 to -21 Insider sales during days -20 to -11 Insider sales during days -10 to -6 Insider sales during days -5 to -1 Post-Disclosure Insider Sales Insider sales during days 0 to +5 Insider sales during days +6 to +10 Insider sales during days +11 to +20 Insider sales during days +21 to +30 Number of observations 0.0018 0.0023 0.0035 0.0057 0.0021 0.0030 0.0027 0.0025 -0.0004 -0.0006 0.0009 0.0032 -17.5 -21.3 33.4 125.5 -1.132 -1.336 1.217 2.827 0.871 0.909 0.112 0.002 0.0024 0.0038 0.0029 0.0036 278 0.0024 0.0022 0.0019 0.0020 0.0000 0.0016 0.0010 0.0016 -0.6 71.1 51.5 78.7 -0.023 1.906 2.054 2.771 0.509 0.028 0.020 0.003 This table presents differences in insider sales between comment letter firms and control period observations when comment letters are disclosed close to the expected timeframe (between 40 and 50 calendar days after the closing comment letter prior to 2012, and between 15 and 25 business days thereafter). Insider sales are measured as the daily percent of shares outstanding sold by insiders, with non-zero values winsorized at the 1% level (SHARES from Thomson Reuters insider database divided by SHROUT from CRSP scaled as a percent of shares outstanding). T-tests reflect differences in mean daily insider sales, as a percentage of shares outstanding, between comment letter firms and control period observations. Panel A shows that all Revenue Recognition Comment Letter firms have significantly greater sales in the -5 to -1 day period relative to the comment letter disclosure date. Specifically, Revenue Recognition Comment Letter firms’ insiders sell 42.8 percent more stock in the five days prior to disclosure than the control period observations. Panel B shows that in the presence of high short interest (top 25% of short interest during the disclosure month), Revenue Recognition Comment Letter firms have significantly greater sales in the -5 to -1 day period relative to the comment letter disclosure date. Specifically, Revenue Recognition Comment Letter firms’ insiders sell 125.5 percent more stock in the five days prior to disclosure than the control period observations. T-tests are one-sided, as we hypothesize that comment letter firm insiders will sell more stock than control firms around the comment letter disclosure. Control observations are comprised of up to 5 observations from the same firm with event dates selected randomly from between 3 and 6 months before and after the comment letter disclosure date. 44