Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Shareholder litigation as a disciplining device: evidence from firms' financial policies
(USC Thesis Other)
Shareholder litigation as a disciplining device: evidence from firms' financial policies
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
SHAREHOLDER LITIGATION AS A DISCIPLINING DEVICE: EVIDENCE FROM FIRMS' FINANCIAL POLICIES by Vuk Talijan A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulllment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (BUSINESS ADMINISTRATION) May 2018 Copyright 2018 Vuk Talijan Dedication To Ashley and to my parents. ii Acknowledgements I am especially grateful to my adviser, Kevin J. Murphy, as well as my committee members Gerard Hoberg, Oguzhan Ozbas, and Yoon-Ho Alex Lee for providing guidance and advice throughout the writing of my dissertation. I would also like to thank Joshua Ackerman, Kenneth Ahern, Tom Chang, Wayne Ferson, Ayse Imrohoroglu, Michelle Lowry, Julia Schwartz and participants at the USC Marshall School of Business Finance Brownbag, the Edinburgh Conference on Legal Institutions and Finance, the FMA Doctoral Student Consortium, and the FMA Annual Meeting for helpful comments and discussion. Cornerstone Research and Stanford Law School are the source of the class action data. The views expressed in this dissertation are my own and do not represent in any way the views of Cornerstone Research or Stanford Law School. iii Table of Contents Dedication ii Acknowledgements iii List Of Tables vi List Of Figures viii Abstract ix Chapter 1: Empirical Findings 1 1.1 Institutional Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.2 Data and Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 1.3 Experimental Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.3.1 In re IPO Securities Litigation . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.3.2 Class Action Fairness Act of 2005 . . . . . . . . . . . . . . . . . . . . . . . 13 1.3.3 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.3.4 Econometric Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 1.4.1 Dierence Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 1.4.2 Dierence in Dierences Model . . . . . . . . . . . . . . . . . . . . . . . . . 26 1.4.3 Triple Dierences Model: IPO Firms . . . . . . . . . . . . . . . . . . . . . . 29 1.4.4 Dierence in Dierences Models Using the IPO Sample . . . . . . . . . . . 33 1.4.5 IPO Proceeds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 1.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 1.7 Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 1.8 Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Chapter 2: Theoretical Framework 67 2.1 The Option-like Nature of Shareholder Litigation . . . . . . . . . . . . . . . . . . . 72 2.2 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 2.2.1 The Arrival of News . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 2.2.2 Shareholder Litigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 2.3 Opportunism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 2.4 Cash and Investment in Relation to Negative News . . . . . . . . . . . . . . . . . . 80 2.5 Debt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 2.7 Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 2.8 Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 iv Bibliography 98 Appendix 101 Proof of Lemma 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 Proof of Proposition 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Proof of Lemma 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 Proof of Proposition 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Proof of Proposition 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Proof of Lemma 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 v List Of Tables 1.1 Summary statistics for sued rms by industry and by irregularity . . . . 52 1.2 Summary statistics for sued rms by ruling and by exchange . . . . . . . 53 1.3 The combined eect of In re IPO and CAFA on lawsuit lings and lawsuit dismissals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 1.4 Incidence of lawsuit lings and dismissals among IPO rms with high stock volatility and high stock turnover . . . . . . . . . . . . . . . . . . . . . 55 1.5 Financial policies and shareholder litigation { dierence coecients . . . 56 1.6 Financial policies and shareholder litigation { DD coecients . . . . . . . 57 1.7 Cash and shareholder litigation { DDD coecients using IPO rms . . . 58 1.8 Investment and shareholder litigation { DDD coecients using IPO rms 59 1.9 Debt and shareholder litigation { DDD coecients using IPO rms . . . 60 1.10 Shareholder litigation and nancial policies of IPO rms { DD coe- cients using rms alleged to be involved in the IPO allocation scheme . 61 1.11 Shareholder litigation and nancial policies of IPO rms { DD coe- cients using litigious industries . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 1.12 Shareholder litigation and nancial policies of IPO rms { DD coe- cients using stock volatility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 1.13 Shareholder litigation and nancial policies of IPO rms { DD coe- cients using stock turnover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 1.14 IPO proceeds of rms involved in the IPO allocation scheme . . . . . . . 65 1.15 Financial policies and shareholder litigation in the period after 2004 . . 66 2.1 Summary of probability related notation . . . . . . . . . . . . . . . . . . . . 95 vi 2.2 Summary of cash related notation . . . . . . . . . . . . . . . . . . . . . . . . . 96 2.3 Summary of remaining notation . . . . . . . . . . . . . . . . . . . . . . . . . . 97 vii List Of Figures 1.1 Average cash, investment, and debt trends for sued and nonsued rms surrounding the start of shareholder litigation risk . . . . . . . . . . . . . . 47 1.2 Timeline of a typical shareholder class action . . . . . . . . . . . . . . . . . 48 1.3 Epanechnikov kernel densities of cash for sued rms . . . . . . . . . . . . . 49 1.4 Epanechnikov kernel densities of investment for sued rms . . . . . . . . . 50 1.5 Epanechnikov kernel densities of debt for sued rms . . . . . . . . . . . . . 51 2.1 Payo from a hypothetical put writing strategy. . . . . . . . . . . . . . . . . 91 2.2 Payo from buying a security which has a hypothetical put baked into its price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 2.3 The setup of the model with the arrival of news at t = 1 and with no shareholder litigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 2.4 The setup of the model with the arrival of news at t = 1 and with shareholder litigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 viii Abstract Shareholder litigation risk varies across time and across rms. When shareholder litigation risk is high, it can increase (decrease) a rm's cash and investment before (after) a lawsuit ling. When shareholder litigation risk is low, little to no impact occurs. A quasi-natural experiment using two legal shocks, In re IPO and CAFA, supports a causal interpretation. Shareholder litigation risk can also impact a rm's debt around the time of a lawsuit ling, but the empirical results are less clear-cut. In addition to the empirical results, I oer a theoretical framework to help explain the changes in cash, investment, and debt around the time of a lawsuit ling. The framework posits that an entrepreneur, at risk of a lawsuit ling, may save cash as a precautionary measure. When the cash accumulates and a lawsuit ling does not occur, an entrepreneur increases invest- ment in hopes of superior future performance. But if future performance wanes notwithstanding, shareholders then le a lawsuit against their entrepreneur. When cash is limited, debt may act as an alternative precautionary measure against a lawsuit ling. Finally, the theoretical framework emphasizes a benet of shareholder litigation { the option to le a lawsuit against an entrepreneur encourages shareholders to fund projects and to retain entrepreneurs. ix Chapter 1 Empirical Findings Little is known about the eect of shareholder litigation on a rm's nancial policies. While prior literature has documented an increase in cash and investment prior to shareholder litigation (Arena and Julio, 2015; McTier and Wald, 2011), I show the results to be sensitive to the time period examined and the identication strategy used. After all, the eect of shareholder litigation on a rm's nancial policies is an equilibrium outcome. The outcome depends on a rm's shareholder litigation risk, which varies across time and across rms. A central contribution of this dissertation is to show that the eect of shareholder litigation on a rm's nancial policies re ects variation in shareholder litigation risk. Shareholder litigation can have a rst-order impact on a rm's nancial policies. But the rst-order impact arises if and only if the risk of shareholder litigation is high: for example, when shareholder litigation insurance is limited or expensive. The risk of shareholder litigation consists of two components: the likelihood of litigation and the expected costs of litigation. A high likelihood coupled with high expected costs of shareholder litigation suggests a need for precautionary measures. So before a lawsuit ling, the ocers and directors of a rm may adjust nancial policies in precaution of shareholder litigation. After a lawsuit ling, the nancial policies of a rm re ect the expected costs of shareholder litigation. The expected costs are direct as well as indirect. Direct costs include payments of a deductible to shareholder litigation insurers as well 1 as settlement payments (in excess of the insurance limit) and super uous legal expenses; indirect costs include materially adverse changes in contracts with suppliers or customers. I examine three nancial policies: cash, investment, and debt. Cash is dened as cash and short-term investments for rmi at scal yeart divided by total assets for rmi at scal yeart1. Investment is dened as capital expenditures for rm i at scal year t divided by net property, plant, and equipment for rmi at scal yeart1. Debt is dened as the sum of book-value short- and long-term debt for rm i at scal year t divided by total assets for rm i at scal year t 1. Figure 1.1 depicts average cash, investment, and debt in the years surrounding a shareholder class action ling. I focus on shareholder class actions due to their scope. Any publicly traded rm can succumb to a shareholder class action. Shareholder class actions are costly for rms (Gande and Lewis, 2009). Finally, the data on shareholder class actions are publicly available. Figure 1.1 uses shareholder class actions from 1996 to 2014, and shows trends for sued as well as matched nonsued rms. Sued rms appear to increase the three nancial policies before and decrease the three nancial policies after a lawsuit ling; matched nonsued rms do not exhibit commensurate changes in the three nancial policies. I show, however, that the increases before a lawsuit ling and the decreases after a lawsuit ling depend on the time period and the identication strategy. Specically, the increases before a lawsuit ling are attributable to years preceding 2001; the decreases after a lawsuit ling are attributable to years preceding 2005. Furthermore, all of the increases before a lawsuit ling and most of the decreases after a lawsuit ling are attributable to IPO rms. Of course, any empirical examination linking shareholder litigation to rms' nancial policies faces a critical hurdle: shareholder litigation as well as nancial policies are endogenous. To ad- dress the endogeneity, I provide a novel design which identies the eect of shareholder litigation on cash and investment. I use two exogenous shocks that create variation in the risk of share- holder litigation. The rst shock is the December 2000 revelation of an IPO allocation scheme in which issuers and their underwriters allocated IPO shares conditional on allocants' promises to 2 purchase additional shares in the aftermarket. The promises to purchase additional shares were not disclosed in issuers' prospectuses. Under Section 11 of the Securities Act of 1933, investors can sue issuers and underwriters for such material omissions in a prospectus. The revelation of the scheme led to an important set of legal cases dubbed In re Initial Public Oering Securities Litigation. Henceforth, I refer to the revelation as In re IPO. The second shock is the enactment of the Class Action Fairness Act of 2005. CAFA aimed to prevent class action abuses. It expanded the criteria for federal jurisdiction of class actions: historically, when federal courts preside over a class action, the class action is less prone to abuse. The Act also directed courts to scrutinize corporate class action settlements, which were deemed excessive. 1 While In re IPO and CAFA both reduce a rm's risk of shareholder litigation, the eect of each diers. In re IPO targets rms that recently underwent an IPO. Before In re IPO, an issuer's shareholder litigation risk depended, among other factors, on its exposure to the allocation scheme. The greater the exposure, the greater the incentive of an issuer's ocers and directors to take precautionary measures against a lawsuit ling: for example, to retain cash. Furthermore, IPO rms generally have underdeveloped legal divisions and lower insurance coverage against shareholder litigation, both exacerbating their need for precautionary measures. Because In re IPO reveals the allocation scheme to exist on a massive scale, it stops its use altogether. So after In re IPO, a new issuer's exposure to the allocation scheme eectively goes to zero. CAFA, on the other hand, aects all rms by reducing their expected costs of shareholder litigation. CAFA curtails excessive settlements and limits plaintis' attorneys fees; in turn, po- tential plaintis and their attorneys have less incentive to sue. When a rm's expected costs of shareholder litigation fall, so too does the rm's demand for insurance against such litigation. A 1 CAFA's guidelines for federal jurisdiction do not apply to shareholder class actions involving \covered secu- rities" (i.e. securities sold on a national exchange). Reason being that the guidelines for removal, from state to federal jurisdiction, of class actions involving covered securities had already been specied in the Securities Litigation Uniform Standards Act of 1998. CAFA's requirements for greater scrutiny of corporate class action settlements, however, does apply to covered securities. 3 rm incurs two main expenses in obtaining shareholder litigation insurance: an annual premium and a per ling deductible (or retainer). Because the per ling deductible can range in millions of dollars, the eect of CAFA is most likely to appear after a lawsuit ling. In re IPO and CAFA are powerful instruments. Together, they increase the likelihood of a lawsuit dismissal and decrease the likelihood of a lawsuit ling. After In re IPO and CAFA, the fraction of dismissals rises by 92.0%; the number of annual lings falls by 26.2%. To identify the eect of shareholder litigation on cash, investment, or debt, I employ a triple dierences regression using an indicator variable equal to 1 for the two years of data that imme- diately follow a rm's IPO. In re IPO and CAFA serve as \treatments" in the triple dierences regression. The results show IPO rms to be responsible for all of the cash increase in the two years before a lawsuit ling and for practically all of the cash decrease in the two years after a lawsuit ling. Furthermore, most of the cash changes surrounding a lawsuit ling occur before In re IPO. In re IPO dampens the cash increase of IPO rms in the two years before a lawsuit ling and has no eect on the cash of more established rms. The results show CAFA to aect all rms and to dampen the cash decrease in the two years after a lawsuit ling. The investment results parallel the cash results. To further isolate shareholder litigation risk as the source of variation in an IPO rm's cash and investment, I examine cross-sectional variation by shareholder litigation risk in a sample of IPO rms alone. In the sample of IPO rms alone, I use four separate sources of cross-sectional variation. The rst source is an indicator variable \IPO Allocation" equal to 1 for rms alleged to be involved in the scheme that led to In re IPO. Firms accused of being involved in the scheme have high ex ante exposure to the scheme and high risk of shareholder litigation; rms not accused of being involved in the scheme are far enough removed from the scheme to avoid implication, are lucky enough to avoid implication, or are not involved in the scheme. So on average, the group of rms with the IPO Allocation variable set to 1 should have a higher risk of shareholder litigation than the group of rms with the IPO Allocation variable set to 0. The second source 4 is industry litigiousness. By construction litigious industries have a higher risk of shareholder litigation. The third source is stock volatility. A rm with high stock volatility has a higher likelihood of sustaining a drop in its stock price. Because lawsuit lings often coincide with falling stock prices, rms with volatile stocks have an increased risk of shareholder litigation (Lowry and Shu, 2002; Gande and Lewis, 2009). The fourth source is stock turnover. The greater the stock turnover, the greater the number of potential plaintis. Because shareholder damages are increasing in the number of plaintis, plaintis' attorneys have incentive to target rms with high stock turnover (Lowry and Shu, 2002; Arena and Julio, 2015). The regressions, using the sample of IPO rms alone and any one of the four cross-sectional variables, help conrm that before In re IPO an IPO rm's cash and investment change as a result of shareholder litigation risk. The increases in cash and investment before a lawsuit ling and the decreases in cash and investment after a lawsuit ling are concentrated among IPO rms alleged to be involved in the IPO allocation scheme, among IPO rms in litigious industries, among IPO rms with high stock volatility, and among IPO rms with high stock turnover. Despite the results, an alternative explanation looms: the scheme, itself, could have produced high cash for rms that engaged in the scheme. For example, IPO rms that engaged in the scheme could have obtained greater IPO proceeds, directly increasing cash. In turn, investment could mechanically rise as a consequence of the greater IPO proceeds. To test the alternative explanation, I compare the IPO proceeds of IPO rms alleged to be involved in the scheme against those of IPO rms not alleged to be involved in the scheme. IPO rms alleged to be involved in the scheme have the same proceeds as IPO rms not alleged to be involved in the scheme. Yet IPO rms alleged to be involved in the scheme hoard cash and in ate investment, hinting at a causal link between shareholder litigation risk and an IPO rm's cash and investment in the period before In re IPO. The results on debt are less clear-cut. Debt increases in the two years before a lawsuit ling, and the increase occurs among more established rms. Limiting the sample to IPO rms alone, 5 however, reveals IPO rms at higher risk of shareholder litigation to decrease debt in the two years before a lawsuit ling. Together the results suggest that debt may be an alternative to cash as a precautionary measure against the risk of shareholder litigation. When cash is used to build a precautionary buer, less debt is required; when cash is sparse, debt is used. IPO rms are replete with cash from IPO proceeds, so they may avoid any additional burdens of debt. While shareholder litigation has been a largely unexplored topic in both theoretical and em- pirical nance, several papers have examined the eects of shareholder litigation on rm out- comes. Drake and Vetsuypens (1993) discuss dierences in IPO returns of sued and nonsued rms. The authors nd no evidence that underpricing reduces litigation risk. Lowry and Shu (2002), however, account for the endogeneity bias in the previous study. The authors explore two eects, dubbed as \insurance" and \deterrence" eects. The insurance eect captures the use of IPO underpricing as a way to insure against litigation risk. The deterrence eect captures the notion that higher underpricing also lowers expected litigation costs. The authors present evi- dence supporting both eects. Hanley and Hoberg (2012) extend the study of IPO underpricing by examining the tradeo between underpricing and strategic disclosure as potential hedges against litigation risk. The authors nd that strategic disclosure serves as an eective hedge against all types of lawsuits. Moving away from the relation between IPO strategy and litigation risk, DuCharme et al. (2004) discuss increases in abnormal accruals around stock oers. Sued rms have higher abnormal accruals. Gande and Lewis (2009) show that lawsuit lings closely follow trigger events. Examples of trigger events include operating or accounting irregularities. Together, trigger events and lawsuit lings entail a signicant fall in market value. Other papers have examined the eects of litigation on corporate governance. Both Ferris et al. (2007) and Appel (2016) claim that derivative suits mitigate agency con icts. Ferris et al. (2007) use a matching criteria to show that derivative suits are associated with improvements in 6 a rm's board of directors. Appel (2016) uses the passage of universal demand laws, which signif- icantly obstruct lawsuits against directors, to show that such laws are associated with decreases in corporate governance. Finally, a few papers have examined the use of directors' and ocers' (D&O) liability insur- ance and its impact on a rm's actions. Chalmers et al. (2002) nd a negative relation between D&O insurance coverage and post-IPO stock price performance for a sample of IPO rms during 1992-1996. Lin et al. (2011) provide evidence that managers of rms with higher D&O insurance coverage make inferior acquisition decisions { they pay higher premiums and obtain lower syner- gies. Lin et al. (2013) show that rms with higher D&O insurance coverage engage in greater risk taking and have a higher likelihood of nancial restatements. These actions translate into higher loan spreads for rms with higher D&O insurance coverage. My dissertation emphasizes that the eects of litigation on rms depend on the legal environ- ment, which changes over time. Often the legal environment in the US depends on the whims of those in political power. I show that these whims can have far-reaching consequences. While In re IPO and CAFA are both foregone events, their lessons remain meaningful. Recently, the House of Representatives passed the Fairness in Class Action Litigation Act (FICALA). The Act sets stricter standards for class certication and further limits plaintis' attorneys fees. In that regard, the Act favors corporate defendants: it further reduces a rm's need for precautionary measures before a lawsuit ling and its expenses after a lawsuit ling. A number of consumer pro- tection, labor, and civil rights organizations oppose the Act. Should the political climate change in favor of these organizations, there is reason to believe that the measures enacted by CAFA and FICALA could be reversed. If a reversal occurs, then once again we should expect litigation to have a signicant impact on a rm's cash, investment, and debt. In general, the impact of litigation on rms would be easier to observe and evaluate with litigation insurance data. Unfortunately, US rms are not required to publicly disclose their 7 litigation insurance premiums, deductibles, and coverage limits. The availability of such data is key to future research exploring the eects of litigation on rm behavior. 1.1 Institutional Background The Securities Act of 1933 and the Securities Exchange Act of 1934 establish rules allowing shareholders to le a class action against a publicly traded rm. Shareholder class actions are complex events. From a legal perspective, statutes are often vague, while the case law is large and esoteric. From a nancial perspective, accounting and nance experts are required to evaluate the scope of the purported violation. From a clerical perspective, a large number of documents need to be obtained and numerous witnesses questioned. Consequently, a majority of class actions settle out of court. A variety of trigger events lead to a shareholder class action: misrepresentation of accounting, operating, or nancing activities; securities violations, such as insider trading or inappropriate initial public oering allocations; merger and acquisition related fraud; or frivolous reasons, such as shareholders' gripe over the eort, skill, or performance of an ocer or director. A trigger event closely precedes a lawsuit ling. After a lawsuit ling, defendants generally have 21 days to respond. Thereafter, the judge determines the lead plainti, who subsequently chooses the lead counsel and helps consolidate all complaints into a single suit. Discovery begins when a case survives a defendant's motion to dismiss. Discovery is the process by which opposing parties gather documents, question witnesses, and request admissions. The process often lasts more than a year. Generally during or after discovery, the lead plainti requests to continue the proceedings as a class. If the request is granted, the class is certied. Once a class is certied, all class members are subject to the same outcome. Henceforth, if particular class members want to bring individual lawsuits, they rst need to opt out of the certied class. 8 When discovery ends, motions are led for summary judgment. If judgment is entered in favor of a party on all motions, then the losing party can immediately appeal. If judgment is not granted on a particular motion, preparation for trial begins, and the remainder of the case concludes in trial court; the losing party can appeal only after nal judgment is entered. Figure 1.2 provides a summary and depicts a typical timeline of a class action. Dierent types of securities lawsuits exist. In addition to shareholder class actions, there are Securities and Exchange Commission (SEC) enforcement actions and shareholder derivative suits. SEC enforcement actions are, as the name states, initiated by the SEC. During the investigation period, the SEC gathers information to gauge whether or not regulatory proceedings are necessary. If they are not, the case is dismissed. If they are, a civil litigation begins and the case enters a regulatory period. Karpo, Lee, and Martin (2008a, 2008b) provide a timeline and an explanation of SEC enforcement actions. While shareholder derivative suits appear similar to shareholder class actions, the two dier in procedure and purpose. The class action involves the consolidation of multiple complaints into a single lawsuit, whereas the derivative suit involves no consolidation. The class action seeks redress for shareholders harmed by a specic defendant, whereas the derivative suit seeks redress for injuries done to the corporation. So a derivative suit requires a valid injury for which the corporation itself could have sued, but failed to do so. 1.2 Data and Summary Statistics Data on shareholder class actions come from Cornerstone Research and Stanford Law School. The data track all shareholder class actions led in Federal Court following the enactment of the Private Securities Litigation Reform Act of 1995. 2 The following are not included in the sample: 2 The Private Securities Litigation Reform Act of 1995 aims to decrease the number of frivolous lawsuits by imposing a higher threshold for ling. The Act requires plaintis to provide evidence of fraud before any pretrial discovery takes place. While limiting frivolous lawsuits, the Act also makes it more dicult to le legitimate lawsuits. 9 lawsuits led in state courts without a parallel federal class action; SEC enforcement actions without a parallel federal class action; and derivative suits without a parallel federal class action. I also exclude Mado-related and mutual-fund market-timing lawsuits. Data on company nancials come from Compustat. I obtain IPO dates using the linking table between Compustat and CRSP. Data on IPO proceeds come from the Securities Data Company (SDC). The population of rms consists of \domestic" rms reporting in US dollars. The sample includes only rms that trade on the NYSE, NASDAQ, and AMEX exchanges at time of lawsuit ling. The sample contains no exchange-traded funds or open-end funds. No rms in the real estate, nancials, utilities, or conglomerates industries are included; as is the case with many studies in nance, these rms are excluded due to the complexity of their accounting structures. Lastly, rms with instances of negative balance-sheet cash or negative revenue are excluded, as are rms with under $1 million in market capitalization. I am then left with a sample of 66,381 rm-year observations involving 7,467 rms, which I call the full sample. From the full sample, I develop a matched sample. The matched sample matches all sued rms to nonsued rms in the same 4-digit SIC industry based on market capitalization at t 1, where t denotes a sued rm's year of lawsuit ling. After matching, I am left with 24,410 rm-year observations involving 1,794 rms. From the matched sample, I obtain a sued sample and an IPO sample. The sued sample contains only sued rms. The sued sample has 11,321 rm-year observations involving 897 rms. The IPO sample consists of the rst two observations that follow each rm's date of going public in the matched sample. The IPO sample contains 2,160 rm-year observations involving 1,080 rms. I reference the dierent samples throughout this chapter. All dependent variables are Winsorised in the full sample at the bottom 0.5% and the top 99.5%. Both the matched and the sued sample contain 1,011 lawsuit lings. The rst lawsuit is led on January 29, 1996. The last lawsuit is led on September 9, 2014. Detailed information for each lawsuit can be found on the Securities Class Action Clearinghouse website. 10 Tables 1.1 and 1.2 provide summary statistics. By ruling, 53.7% of lings are settled, 35.5% are dismissed, and 10.8% are still ongoing. By exchange, 66.1% of sued rms are listed on NASDAQ, 31.9% on NYSE, and 2.0% on AMEX. The high percentage of NASDAQ rms is unsurprising, given that the largest portion (44.3%) of sued rms are in the technology industry. Services (20.2%) and healthcare (17.6%) are the next most representative industries among sued rms. Of reported irregularities, failures of disclosure (accounting and non-accounting) represent the highest percentage (41.7%). Misrepresentations (non-accounting) are the next highest irregularity (31.2%), followed by IPO allocation (13.3%). As discussed throughout the chapter, IPO allocation cases are only reported during the period from 2001 through 2004. Industry and irregularity designations come from Cornerstone Research and Stanford Law School. 1.3 Experimental Design 1.3.1 In re IPO Securities Litigation On December 6, 2000, the Wall Street Journal released a story documenting an IPO allocation scheme in which issuers and their underwriters allocated IPO shares conditional on allocants' promises to purchase additional shares in the aftermarket. 3 The promises to purchase additional shares were not publicly disclosed. In the year following the story, thousands of investors led class action lawsuits against the issuers and their underwriters. In her 2004 ruling, Judge Scheindlin of the US District Court in the Southern District of New York described the events that led to In re IPO: Between January 11 and December 6, 2001, thousands of investors led class action lawsuits, alleging that 55 underwriters, 310 issuers and hundreds of individuals associ- ated with those issuers had engaged in a sophisticated scheme to defraud the investing 3 Pulliam S., Smith R., December 6, 2000. Trade-os: Seeking IPO shares, investors oer to buy more in after-market. The Wall Street Journal, A1. 11 public. In brief, the scheme consisted of a requirement, imposed by the underwriters, that IPO allocants purchase shares in the aftermarket, often at escalating prices, and pay undisclosed compensation. In addition, the underwriters prepared analyst reports that contained inaccurate information and recommendations because the analysts op- erated under a con ict of interest. As a result of the scheme, plaintis allege that they collectively lost billions of dollars. These actions were consolidated before this Court for pre-trial supervision. At the suggestion of the Court, the parties selected six cases to be used as test cases for determining whether these suits can proceed as class actions. In re Initial Public Oering Sec. Litig., 227 F.R.D. 65 (S.D.N.Y. 2004) The 2004 ruling concerned the class certication of six test cases out of the 310 cases connected to IPOs. Rule 23 of the Federal Rules of Civil Procedure outlines the requirements for class certication. In 2004, Judge Scheindlin ruled that only \some showing" of evidence is needed to satisfy Rule 23. As a result, the six test cases could proceed as class actions. Soon after the 2004 ruling, a tentative one billion dollar settlement was proposed between defendants and plaintis. In 2005, however, the 2004 ruling was appealed to the Second Circuit. In 2006, the Second Circuit found Judge Scheindlin's use of the \some showing" standard to be awed and decided that all of the Rule 23 requirements have to be established by facts in order for a class to be certied. 4 The Second Circuit's decision set a more rigorous standard for class certication. Consequently, courts under the Second Circuit's jurisdiction have had to assess evidence and nd every requirement of Rule 23 to be met before certifying a class. The Second Circuit reached an additional holding specic to IPO securities: that the market for IPO securities is not ecient, thus precluding the use of the \fraud-on-the-market" theory as an argument for class certication. The \fraud-on-the-market" theory states that the market price of securities incorporates all publicly available information, including material misrepresentations. So it may be presumed that an investor who buys securities at the market price relies on the misrepresentations. By severing 4 In re Initial Public Oering Sec. Litig., 471 F.R.D. 24 (2d Cir. 2006) 12 the link between market price and misrepresentations, the Second Circuit's holding discards an important argument for certifying a class in lawsuits involving IPO securities. The Second Circuit's ruling triggered a provision in the proposed settlement agreement voiding the one billion dollar settlement. Subsequently, the parties involved in In re IPO agreed to a $586 million settlement in 2009. 1.3.2 Class Action Fairness Act of 2005 CAFA has three main goals. First, it increases the amount in controversy requirement for federal jurisdiction of a class action from $75 thousand to $5 million. While the $75 thousand pertains to at least one individual, the $5 million represents the sum of all plaintis' claims. On average, the change does not necessarily lead to a more rigorous standard: most federal class actions involve a lead plainti whose damages exceed $75 thousand. The change only aects a subset of outlier cases that previously could not have been brought to federal court, because the $75 thousand threshold was not met: for example, Blockbuster, Inc. v. Galeno, 472 F.3d 53 (2d Cir. 2006); Cappuccitti v. DirecTV, Inc., 623 F.3d 1118 (11th Cir. 2010). Second, CAFA changes the requirement that all plaintis be diverse from all defendants, known as \complete diversity," to allow federal jurisdiction where at least one plainti is diverse from at least one defendant. The word \diverse" indicates that the parties involved in a lawsuit are citizens of dierent states. The change aims to limit \forum-shopping" { plaintis exploiting the class action process by targeting friendly state courts, sometimes referred to as \magnet jurisdictions." Rather than sweep all class actions into federal court, CAFA aims to preclude the magnet cases from being led in the rst place. This second goal of CAFA excludes class actions against NYSE, NASDAQ, and AMEX rms; the reason being that the Securities Litigation Uniform Standards Act of 1998 had already set guidelines for removal, from state to federal court, of class actions against NYSE, NASDAQ, and AMEX rms. 13 Third, CAFA enacts new rules to help ensure reasonable settlements and limit excessive plain- tis' attorneys fees. As a result, CAFA aects NYSE, NASDAQ, and AMEX rms by decreasing the incentives of both shareholders and attorneys to le and prolong class actions. This third goal of CAFA is pertinent to my experimental design. 1.3.3 Setup Two conditions must hold for shareholder litigation risk to in uence cash, investment, or debt. First, shareholder litigation risk must have a rst-order impact on cash, investment, or debt at some time. Second, rms must not be able to perfectly hedge against shareholder litigation risk at all times. Alternatively, even if rms could perfectly hedge against shareholder litigation risk, the expense to perfectly hedge should be high. Given the costs of shareholder litigation, it should not be a surprise that the risk of shareholder litigation can have a rst-order impact on a rm's nancial policies. The costs of shareholder litigation, both direct and indirect, are substantial. Engelmann and Cornell (1988) discuss a variety of indirect costs aecting a rm that becomes a defendant in a lawsuit. The indirect costs range from higher transaction costs to reputational penalties. Karpo et al. (2008b) show that reputational penalties of litigation can be over 7.5 times the sum of all penalties imposed by legal institutions. Brochet and Srinivasan (2014) show that directors named in a class action subsequently receive negative recommendations from shareholders and proxy advising rms. Fich and Shivdasani (2007) present evidence that directors, who do not depart from the board of a sued rm, nevertheless experience signicant declines in other board seats held. The second condition asserts that rms must not be able to perfectly hedge against shareholder litigation risk at all times. Arena and Julio (2015) state that in a sample of sued public rms from 1996 to 2006, the average settlement amount is more than three times larger than the average litigation insurance limit. Furthermore, litigation insurance oers protection only for the direct costs of litigation. It does not account for the indirect costs. Karpo et al. (2008a) nd that 14 ocers and directors responsible for nancial misrepresentation suer signicant nancial losses. So ocers and directors certainly have good reason to beware the threat of shareholder litigation. The ocers and directors of a rm are cognizant of the actions that may lead to a lawsuit ling. If they cannot hedge away the risk of shareholder litigation, they may take precautionary measures in anticipation of a lawsuit ling. For example, the ocers and directors of a rm may hoard cash as a buer against shareholder litigation. They may increase investment in an attempt to report exceptional performance and obviate the lawsuit ling. They may raise debt as a deterrent to costly shareholder litigation. If the lawsuit ling nevertheless occurs, cash falls as legal costs are incurred. Investment is curtailed as the rm's resources, including time and cash, run thin. Debt is paid down in order to avoid nancial distress. 1.3.4 Econometric Method To identify the eect of shareholder litigation on a rm's nancial policies, I begin with a dierence in dierences (DD) specication: Cash it = Pre Pre it + PreIn re IPO (Pre it In re IPO t ) + PreCAFA (Pre it CAFA t ) + Post Post it + PostIn re IPO (Post it In re IPO t ) + PostCAFA (Post it CAFA t ) + i + t + it : (1.1) Cash is the dependent variable for rm i during scal year t. Cash is dened as cash and short- term investments for rm i at scal year t divided by total assets for rm i at scal year t 1. In addition to cash, I use investment and debt as dependent variables. Investment is dened as capital expenditures for rm i at scal year t divided by net property, plant, and equipment for rmi at scal yeart 1. Debt is dened as the sum of book-value short- and long-term debt for rm i at scal year t divided by total assets for rm i at scal year t 1. I employ the above specication in the sued sample. I supplement the above specication with a second specication 15 replacing year xed eects t with industry-year xed eects jt , where j denotes a 4-digit SIC industry code. Industry-year xed eects help account for industry-level technology shocks that otherwise could be responsible for the results. I employ the second specication in the matched sample. Finally, I use a third specication which keeps industry-year xed eects jt but removes rm xed eects i . I also employ the third specication in the matched sample. A comparison of second and third specications suggests that the results are not attributable to industry-level technology shocks and are most pronounced when accounting for within-rm variation. Pre and Post are the primary independent variables. Pre equals 1 for the two scal years preceding the date of a lawsuit ling against rmi and 0 otherwise. So two scal years before the date of a lawsuit ling represents the assumed start of shareholder litigation risk. The assumed start of shareholder litigation risk coincides with the average class period start for sued rms: 2.1 years before the date of a lawsuit ling. Post equals 1 for the two scal years following the date of a lawsuit ling and 0 otherwise. A majority of shareholder class actions reach a resolution within two years of a lawsuit ling (Klausner and Hegland, 2010, 2013). The variables Pre and Post gauge whether or not nancial policies change as a result of shareholder litigation risk. In re IPO is an indicator variable equal to 1 for the year 2001 and all years that follow. CAFA is an indicator variable equal to 1 for the year 2005 and all years that follow. In Specication 1.1, the coecient Pre ( Post ) captures the level of cash relative to rm average during the two years preceding (following) a lawsuit ling in the period before In re IPO. The coecient PreIn re IPO ( PostIn re IPO ) captures the eect of In re IPO on Pre ( Post ). In re IPO aects new issuers by revealing an IPO allocation scheme in which issuers and their underwriters allocated IPO shares conditional on allocants promising to purchase additional shares in the aftermarket. Before In re IPO new issuers used the scheme often; after In re IPO new issuers stop using the scheme altogether. As a result, the threat of shareholder litigation to new issuers falls: new issuers have one fewer reason to take precaution against a potential lawsuit 16 ling. If the threat of shareholder litigation impacts a new issuer's nancial policies before a lawsuit ling, then a decrease in the threat of shareholder litigation should dampen the impact. The coecient PreCAFA ( PostCAFA ) captures the eect of CAFA on the combined coef- cients Pre plus PreIn re IPO ( Post plus PostIn re IPO ). CAFA aects all rms by curtailing excessive settlements and limiting plaintis' attorneys fees. As a result, potential plaintis and their attorneys have less incentive to sue. When a rm's expected costs of shareholder litigation fall, so too does the demand for insurance against such litigation. In obtaining shareholder litiga- tion insurance, a rm incurs two main expenses: an annual premium and a per ling deductible (or retainer). Because the per ling deductible can range in millions of dollars, the eect of CAFA is likely to be pronounced after a lawsuit ling. To be more certain that a rm's nancial policy changes surrounding a lawsuit ling arise from shareholder litigation risk and not an alternative reason, I explore cross-sectional variation. I use a triple dierences (DDD) specication: Cash it = Pre Pre it + PreIPO (Pre it IPO it ) + PreIn re IPO (Pre it In re IPO t ) + PreIPOIn re IPO (Pre it IPO it In re IPO t ) + PreCAFA (Pre it CAFA t ) + PreIPOCAFA (Pre it IPO it CAFA t ) + Post Post it + PostIPO (Post it IPO it ) + PostIn re IPO (Post it In re IPO t ) + PostIPOIn re IPO (Post it IPO it In re IPO t ) + PostCAFA (Post it CAFA t ) + PostIPOCAFA (Post it IPO it CAFA t ) + IPO IPO it + IPOIn re IPO (IPO it In re IPO t ) + IPOCAFA (IPO it CAFA t ) + i + t + it : (1.2) IPO equals 1 for the rst two observations that follow rmi's date of going public and 0 otherwise. IPO represents cross-sectional variation by age. As discussed, before In re IPO many IPO rms operated under a greater risk of shareholder litigation. In addition, IPO rms generally have 17 underdeveloped legal divisions and lower insurance coverage against shareholder litigation. So before In re IPO, IPO rms should be most susceptible to shareholder litigation risk and should take the most pronounced precautionary measures in anticipation of a lawsuit ling. In addition to cash, I use investment and debt as dependent variables. I employ Specica- tion 1.2 in the sued sample. I supplement Specication 1.2 with a second DDD specication replacing year xed eects t with industry-year xed eects jt , where j denotes a 4-digit SIC industry code. I employ the second DDD specication in the matched sample. Finally, I use a third DDD specication which keeps industry-year xed eects jt but removes rm xed eects i . I also employ the third specication in the matched sample. The DDD results conrm that most of the nancial policy changes surrounding a lawsuit ling occur among IPO rms and in the period before In re IPO. To further isolate the risk of shareholder litigation as the source of an IPO rm's nancial policy changes, I examine the cross- sectional implications of shareholder litigation risk in a sample of IPO rms alone. In the IPO sample, I examine four sources of cross-sectional variation in the risk of shareholder litigation. The rst source of cross-sectional variation in the risk of shareholder litigation is an indicator variable equal to 1 for rms accused of being involved in the IPO allocation scheme. Cornerstone Research and Stanford Law School designate all cases alleged to be involved in the scheme that led to In re IPO as IPO allocation cases. Firms accused of being involved in the scheme have high ex ante exposure to the scheme and high risk of shareholder litigation. Firms not accused of being involved in the scheme consist of two sets: (1) those actually not involved in the scheme; and (2) those far enough removed from the scheme or lucky enough to avoid implication. So on average, the group of rms with the IPO Allocation variable set to 1 should have greater risk of shareholder litigation than the group of rms with the IPO Allocation variable set to 0. The second source of cross-sectional variation in the risk of shareholder litigation is industry litigiousness. To obtain industry litigiousness, I arrange all 4-digit SIC industries in the full sample from highest to lowest based on one, the total number of lings within each industry and two, the 18 fraction of rms that are sued within each industry. Given the arrangement, litigious industries comprise the uppermost set of 4-digit SIC industries which contains half of all lawsuit lings in the matched sample. By construction litigious industries are at higher risk of shareholder litigation. The variable Lit Ind assigns a value of 1 to rms in litigious industries. The third source of cross-sectional variation in the risk of shareholder litigation is stock volatil- ity. I measure stock volatility for a rm's initial 252 days of trading. Regressions with stock volatility as the cross-sectional variable exclude rms that undergo a lawsuit ling within the rst 252 days of trading or within one month after the 252nd day of trading. A rm with high stock volatility has a higher likelihood of sustaining a drop in its stock price. Because lawsuit lings often coincide with falling stock prices, rms with volatile stocks have an increased risk of shareholder litigation (Lowry and Shu, 2002; Gande and Lewis, 2009). The fourth source of cross-sectional variation in the risk of shareholder litigation is stock turnover. Stock turnover is dened as [1 252 t=1 (1 volume traded t =total shares t )]. Like stock volatility, I measure stock turnover for a rm's initial 252 days of trading. Regressions with stock turnover as the cross-sectional variable exclude rms that undergo a lawsuit ling within the rst 252 days of trading or within one month after the 252nd day of trading. The greater the stock turnover, the greater the number of potential plaintis. Because shareholder damages are increasing in the number of plaintis, plaintis' attorneys have incentive to target rms with high stock turnover (Lowry and Shu, 2002; Arena and Julio, 2015). The rst DD specication using the IPO sample with IPO Allocation as the cross-sectional variable is Cash it = IPO Allocation IPO Allocation i + IPO AllocationIn re IPO (IPO Allocation i In re IPO t ) + t + it : (1.3) A signicant IPO Allocation coecient suggests that an IPO rm alleged to be involved in the IPO allocation scheme retains a dierent level of cash compared to the annual average of all IPO 19 rms. A signicant IPO AllocationIn re IPO coecient suggests that an IPO rm alleged to be involved in the IPO allocation scheme responds to In re IPO more than an average IPO rm. Time xed eects t account for annual changes in cash. Firm xed eects i are not necessary, because the IPO sample consists only of data on the rst two scal years following each rm's date of going public. In addition to cash, I use investment and debt as dependent variables. To use industry litigiousness as the cross-sectional variable, I replace IPO Allocation in Specication 1.3 with Lit Ind. To use stock volatility as the cross-sectional variable, I replace IPO Allocation in Specication 1.3 with Volatility. Volatility is an indicator variable equal to 1 for IPO rms with stock volatility exceeding the median stock volatility of all IPO rms, where stock volatility is measured over the initial 252 days of trading. To use stock turnover as the cross-sectional variable, I replace IPO Allocation in Specication 1.3 with Turnover. Turnover is an indicator variable equal to 1 for IPO rms with stock turnover exceeding the median stock turnover of all IPO rms, where stock turnover is measured over the initial 252 days of trading. 5 The second DD specication using the IPO sample with IPO Allocation as the cross-sectional variable is Cash it = Pre Pre it + PreIPO Allocation (Pre it IPO Allocation i ) + Post Post it + PostIPO Allocation (Post it IPO Allocation i ) + IPO Allocation IPO Allocation i + t + it : (1.4) Again, a signicant IPO Allocation coecient suggests that an IPO rm alleged to be involved in the IPO allocation scheme retains a dierent level of cash compared to the annual average of all IPO rms. A signicant Pre coecient suggests that an IPO rm in the two years before a lawsuit ling retains a dierent level of cash compared to the annual average of all IPO rms. The 5 Specically, the DD specication using the IPO sample with industry litigiousness as the cross-sectional vari- able is Cash it = Lit Ind Lit Ind i + Lit IndIn re IPO (Lit Ind i In re IPOt) + t + it . The DD specication using the IPO sample with stock volatility as the cross-sectional variable is Cash it = Volatility Volatility i + VolatilityIn re IPO (Volatility i In re IPOt)+ t + it . The DD specication using the IPO sample with stock turnover as the cross-sectional variable is Cash it = Turnover Turnover i + TurnoverIn re IPO (Turnover i In re IPOt)+ t + it . In addition to cash, I use investment and debt as dependent variables. 20 coecient on Pre IPO Allocation reports additional cash dierences, aside from those captured by IPO Allocation and Pre , in the two years before a lawsuit ling for an IPO rm alleged to be involved in the IPO allocation scheme. A signicant Post coecient suggests that an IPO rm in the two years after a lawsuit ling retains a dierent level of cash compared to the annual average of all IPO rms. The coecient on Post IPO Allocation reports additional cash dierences, aside from those captured by IPO Allocation and Post , in the two years after a lawsuit ling for an IPO rm alleged to be involved in the IPO allocation scheme. In addition to cash, I use investment and debt as dependent variables. To use industry litigiousness as the cross-sectional variable, I replace IPO Allocation in Specication 1.4 with Lit Ind. To use stock volatility as the cross-sectional variable, I replace IPO Allocation in Specication 1.4 with Volatility. To use stock turnover as the cross-sectional variable, I replace IPO Allocation in Specication 1.4 with Turnover. 6 For the experimental design to identify the eect of shareholder litigation on a rm's nancial policies, three conditions should hold. First, In re IPO and CAFA should represent changes in shareholder litigation risk. Second, the eect of shareholder litigation on a rm's nancial policies should be more pronounced at times of high shareholder litigation risk and for rms with high shareholder litigation risk. Third, factors other than In re IPO and CAFA leading to nancial policy changes from years before to years after In re IPO or CAFA should be controlled. In summary, all coecients in Specications 1.1 and 1.2 should be zero in the absence of shareholder litigation risk. As discussed, In re IPO reduces the risk of shareholder litigation by eliminating the use of the IPO allocation scheme. After In re IPO new issuers have one fewer reason to worry about the threat of shareholder litigation. CAFA reduces the risk of shareholder litigation by curtailing 6 Specically, the DD specication using the IPO sample with industry litigiousness as the cross-sectional vari- able is Cash it = Pre Pre it + PreLit Ind (Pre it Lit Ind i ) + Post Post it + PostLit Ind (Post it Lit Ind i ) + Lit Ind Lit Ind i + t + it : The DD specication using the IPO sample with stock volatility as the cross-sectional variable is Cash it = Pre Pre it + PreVolatility (Pre it Volatility i )+ Post Post it + PostVolatility (Post it Volatility i )+ Volatility Volatility i + t + it . The DD specication using the IPO sample with stock turnover as the cross-sectional variable is Cash it = Pre Pre it + PreTurnover (Pre it Turnover i )+ Post Post it + PostTurnover (Post it Turnover i )+ Turnover Turnover i + t + it . In addition to cash, I use investment and debt as dependent variables. 21 excessive settlements and limiting plaintis' attorneys fees; in turn, potential plaintis and their attorneys have less incentive to sue. So CAFA reduces a rm's expected costs of shareholder litigation. Table 1.3 provides evidence of reductions in the risk of shareholder litigation resulting from In re IPO and CAFA. The table examines two periods: through 2004 and after 2004. In 2005, CAFA was enacted and the initial settlement of In re IPO was proposed. The proposal of the initial settlement allows me to safely assume that by 2005 the dust had settled surrounding the events of In re IPO: for example, no new lings alleging the scheme are led in 2005 and after. Thus, Table 1.3 examines the combined eect of In re IPO and CAFA on lawsuit lings and dismissals. Table 1.3 shows the number of lawsuit lings per year to fall by 26.2% after 2004. The table shows the incidence of lawsuit lings to fall by 26.1% after 2004. In sample, among lawsuits that have been resolved, the fraction of dismissed lawsuits rises from 32.8% through 2004 to 54.1% after 2004. Even so, the rise from 32.8% to 54.1% is an underestimate, because the number of dismissed lawsuits manifests truncation bias toward the end of the sample: it often takes a couple years after the date of a lawsuit ling for a lawsuit to be resolved. To adjust for the truncation bias, I randomly allot dismissals to ongoing lawsuits so that the fraction of dismissals among ongoing lawsuits equals the lesser of the two fractions of dismissals among lawsuits that have been resolved. Because I make the conservative choice and equate to the lesser fraction (i.e. 32.8%), the rise in dismissals after 2004 continues to be an underestimate. To preclude selection bias, I simulate the random allotment 1,000 times. The results for dismissals in Table 1.3 are averages of the 1,000 simulations. The table shows that the ratio of dismissed to total lawsuits rises by 91.9% after 2004. The table also shows the incidence of dismissed lawsuits to increase by 46.7% after 2004. In re IPO alters the risk of shareholder litigation specic to IPO rms { those that recently underwent an IPO. Before In re IPO, an IPO rm could be accused of engaging in the IPO allocation scheme. By eliminating the use of the IPO allocation scheme, In re IPO reduces the risk 22 of shareholder litigation for all IPO rms. In addition, IPO rms generally have underdeveloped legal divisions and lower insurance coverage against shareholder litigation, exacerbating their risk of shareholder litigation. Indeed, the results show IPO rms before In re IPO to be responsible for most of the cash and investment increases before a lawsuit ling; before In re IPO, IPO rms also exhibit the most pronounced cash and investment decreases after a lawsuit ling. Furthermore, the changes in cash and investment, surrounding a lawsuit ling, are more pronounced for IPO rms eventually accused of engaging in the IPO allocation scheme. As mentioned, I examine three additional measures of shareholder litigation risk within a sample of IPO rms alone. One of the additional measures is industry litigiousness. By construc- tion, litigious industries have a higher risk of shareholder litigation. Another is stock volatility. Table 1.4 shows the incidence of lawsuit lings among IPO rms with high stock volatility to be 0.042 greater than the incidence of lawsuit lings among IPO rms with low stock volatility, suggesting that IPO rms with high stock volatility have a higher risk of shareholder litigation. The nal measure is stock turnover. Table 1.4 shows the incidence of lawsuit lings among IPO rms with high stock turnover to be 0.030 greater than the incidence of lawsuit lings among IPO rms with low stock turnover. While IPO rms with high stock turnover do experience a higher incidence of lawsuit dismissals compared to IPO rms with low stock turnover, the dierence is barely signicant. So IPO rms with high stock turnover also appear to have higher shareholder litigation risk. The results show changes in cash and investment, surrounding a lawsuit ling, to be concentrated among IPO rms in litigious industries, IPO rms with high stock volatility, and IPO rms with high stock turnover. Because the experimental design relies on a quasi-natural experiment, the possibility remains that a time-varying omitted variable can explain the results. To mitigate the concern, the regres- sions include rm as well as year or industry-year xed eects. Industry-year xed eects account for time-varying unobserved heterogeneity among industries. They help account for industry-level 23 technology shocks that otherwise could be responsible for the results. None of the regressions in- clude rm-level controls. Shareholder litigation, along with In re IPO and CAFA, may aect such controls and in turn lead to imprecise estimates of the eect of shareholder litigation on a rm's nancial policies. Figure 1.1 oers some evidence debunking a time-varying omitted variable. The gure shows average cash, investment, and debt adjusted by rm and time eects for sued and matched nonsued rms over an 8-year window centered about the assumed start of shareholder litigation risk. With the exception of scal years that entail shareholder litigation risk { the two scal years surrounding a lawsuit ling { the three dependent variables trend closely in parallel between sued and matched nonsued rms. Furthermore, the regression results show that the three dependent variables { cash, investment, debt { respond to In re IPO and CAFA as hypothesized: In re IPO and CAFA reduce the risk of shareholder litigation and move the nancial policies of sued rms closer to those of matched nonsued rms during the scal years conjectured to entail an increased risk of shareholder litigation. Finally, sued rms should not change among the three periods: before In re IPO, after In re IPO, and after CAFA. That is, the type of rm that is sued before In re IPO should be similar to the type of rm that is sued after In re IPO and after CAFA. To test for equality of sued rms among the three periods, I compare the kernel densities of cash across the three periods. I perform the same comparison for investment and debt. The procedure of comparing kernel densities within a sample is similar to the procedure in Seru (2014). Figure 1.3 shows the kernel densities of cash across the three periods. A Kruskal-Wallis test fails to reject the null hypothesis for equality of kernel densities across the three periods at the 5% level. Figure 1.4 shows the kernel densities of investment across the three periods. A Kruskal-Wallis test does reject the null hypothesis for equality of kernel densities across the three periods. Nevertheless, the kernel densities certainly look similar across the three periods, and any dierences among the three kernel densities can be attributed to a time eect that occurs for all rms. Figure 1.5 shows the kernel densities of debt 24 across the three periods. A Kruskal-Wallis test fails to reject the null hypothesis for equality of kernel densities across the three periods at the 5% level. 1.4 Results 1.4.1 Dierence Model Table 1.5 reports the results of the dierence model. The estimation spans the years from 1993 through 2014. Columns 1, 2, and 3 show regressions with cash as the dependent variable. Column 1 uses the sued sample and includes rm and year xed eects. The coecient on Pre suggests that in the two years before a lawsuit ling a rm increases cash away from rm average by 0.197; the coecient on Post suggests that in the two years after a lawsuit ling a rm drops cash to 0.153 below rm average. Column 2 uses the matched sample and includes rm and industry-year xed eects. Now the coecient on Pre suggests that in the two years before a lawsuit ling a rm increases cash away from rm average by 0.186; the coecient on Post suggests that in the two years after a lawsuit ling a rm drops cash to 0.165 below rm average. Column 3 also uses the matched sample but only includes industry-year xed eects. The coecient on Pre suggests that in the two years before a lawsuit ling a rm's cash is 0.297 above the industry-year average; the coecient on Post suggests that in the two years after a lawsuit ling a rm's cash drops to 0.061 below the industry-year average. Columns 4, 5, and 6 of Table 1.5 show regressions with investment as the dependent variable. Column 4 uses the sued sample and includes rm and year xed eects. The coecient on Pre suggests that in the two years before a lawsuit ling a rm increases investment away from rm average by 0.246; the coecient on Post suggests that in the two years after a lawsuit ling a rm drops investment to 0.110 below rm average. Column 5 uses the matched sample and includes rm and industry-year xed eects. The coecient on Pre suggests that in the two years before a lawsuit ling a rm increases investment away from rm average by 0.219; the coecient on Post 25 suggests that in the two years after a lawsuit ling a rm drops investment to 0.125 below rm average. Column 6 also uses the matched sample but only includes industry-year xed eects. The coecient on Pre suggests that in the two years before a lawsuit ling a rm's investment is 0.325 above the industry-year average; the coecient on Post suggests that in the two years after a lawsuit ling a rm's investment drops to the industry-year average. Columns 7, 8, and 9 of Table 1.5 show regressions with debt as the dependent variable. Column 7 uses the sued sample and includes rm and year xed eects. The coecient on Pre suggests that in the two years before a lawsuit ling a rm raises debt away from rm average by 0.022; the coecient on Post suggests that in the two years after a lawsuit ling a rm lowers debt to 0.024 below rm average. Column 8 uses the matched sample and includes rm and industry-year xed eects. The coecient on Pre suggests that in the two years before a lawsuit ling a rm raises debt away from rm average by 0.021; the coecient on Post suggests that in the two years after a lawsuit ling a rm lowers debt to 0.020 below rm average. Column 9 also uses the matched sample but only includes industry-year xed eects. The coecient on Pre suggests that in the two years before a lawsuit ling a rm's debt is 0.024 above the industry-year average; the coecient on Post suggests that in the two years after a lawsuit ling a rm lowers debt to 0.015 below the industry-year average. 1.4.2 Dierence in Dierences Model Table 1.6 reports the results of the DD model. The estimation spans the years from 1993 through 2014. I separate the years into three periods: (1) before In re IPO and before CAFA, (2) after In re IPO and before CAFA, and (3) after In re IPO and after CAFA. Column 1 uses the sued sample and includes rm and year xed eects. Cash is the dependent variable. The coecient on Pre suggests that before In re IPO a rm increases cash, in the two years before a lawsuit ling, away from rm average by 0.413. In re IPO dampens the increase by 85.5%. CAFA has an insignicant impact on cash in the two years before a lawsuit ling. The coecient on Post 26 suggests that before In re IPO a rm drops cash, in the two years after a lawsuit ling, to 0.218 below rm average. In re IPO has an insignicant impact on cash in the two years after a lawsuit ling. CAFA, however, retracts cash in the two years after a lawsuit ling toward rm average by 72.2%. Column 2 of Table 1.6 uses the matched sample and includes rm and industry-year xed eects. Again cash is the dependent variable. The coecient on Pre suggests that before In re IPO a rm increases cash, in the two years before a lawsuit ling, away from rm average by 0.403. In re IPO dampens the increase by 86.6%. CAFA has an insignicant impact on cash in the two years before a lawsuit ling. The coecient on Post suggests that before In re IPO a rm drops cash, in the two years after a lawsuit ling, to 0.190 below rm average. In re IPO has an insignicant impact on the decrease. CAFA, however, retracts cash in the two years after a lawsuit ling toward the rm average by 65.9%. Column 3 of Table 1.6 also uses the matched sample but only includes industry-year xed eects. Again cash is the dependent variable. Column 3 demonstrates the importance of account- ing for within-rm variation. Without accounting for within-rm variation, In re IPO absorbs the eect of CAFA in the two years after a lawsuit ling. So the regression cannot disentangle the eect of In re IPO from that of CAFA. Nonetheless, the regression in column 3 shows that before In re IPO a rm's cash, in the two years before a lawsuit ling, is 0.560 above the industry-year average. Before In re IPO a rm's cash, in the two years after a lawsuit ling, drops to 0.206 below the industry-year average. After In re IPO a rm's cash retracts, in the two years surrounding a lawsuit ling, toward the industry-year average. Column 4 of Table 1.6 uses the sued sample and includes rm and year xed eects. Now investment is the dependent variable. The coecient on Pre suggests that before In re IPO a rm increases investment, in the two years before a lawsuit ling, away from rm average by 0.481. In re IPO dampens the increase by 79.2%. CAFA has an insignicant impact on a rm's investment in the two years before a lawsuit ling. The coecient on Post suggests that before In re IPO a 27 rm drops investment, in the two years after a lawsuit ling, to 0.167 below rm average. In re IPO has an insignicant impact on the decrease. CAFA, however, retracts investment in the two years after a lawsuit ling toward the rm average by 82.9%. Column 5 of Table 1.6 uses the matched sample and includes rm and industry-year xed eects. Again investment is the dependent variable. The coecient on Pre suggests that before In re IPO a rm increases investment, in the two years before a lawsuit ling, away from rm average by 0.445. In re IPO dampens the increase by 84.3%. CAFA has an insignicant impact on a rm's investment in the two years before a lawsuit ling. The coecient on Post suggests that before In re IPO a rm drops investment, in the two years after a lawsuit ling, to 0.172 below rm average. In re IPO has an insignicant impact on the decrease. CAFA, however, retracts investment in the two years after a lawsuit ling toward the rm average by 62.6%. Column 6 of Table 1.6 also uses the matched sample but only includes industry-year xed eects. Again investment is the dependent variable. As with column 3, column 6 demonstrates the importance of accounting for within-rm variation. Without accounting for within-rm variation, In re IPO absorbs the eect of CAFA. So the regression cannot disentangle the eect of In re IPO from that of CAFA. Nonetheless, the regression in column 6 shows that before In re IPO a rm's investment, in the two years before a lawsuit ling, is 0.569 above the industry-year average. Before In re IPO a rm's investment, in the two years after a lawsuit ling, drops to 0.123 below the industry-year average. After In re IPO a rm's investment retracts, in the two years surrounding a lawsuit ling, toward the industry-year average. Column 7 of Table 1.6 uses the sued sample and includes rm and year xed eects. Now debt is the dependent variable. The coecient on Pre suggests that before In re IPO a rm raises debt, in the two years before a lawsuit ling, away from rm average by 0.035. In re IPO and CAFA have no impact on debt in the two years before a lawsuit ling. The coecient on Post suggests that before In re IPO a rm lowers debt, in the two years after a lawsuit ling, to 0.048 28 below rm average. In re IPO and CAFA have no impact on debt in the two years after a lawsuit ling. Column 8 of Table 1.6 uses the matched sample and includes rm and industry-year xed eects. Again debt is the dependent variable. The coecient on Pre suggests that before In re IPO a rm raises debt, in the two years before a lawsuit ling, away from rm average by 0.037. None of the other coecients are signicant. Column 9 also uses the matched sample but only includes industry-year xed eects. None of the coecients in column 9 are signicant. 1.4.3 Triple Dierences Model: IPO Firms To be more certain that a rm's nancial policy changes arise from the risk of shareholder litigation and not an alternative reason, I explore the cross-sectional implications of shareholder litigation risk. I use age as the primary dimension of cross-sectional variation. As mentioned, In re IPO targets rms that recently went public. Before In re IPO many IPO rms, along with their underwriters, allocated IPO shares conditional on allocants' promises to purchase additional shares in the aftermarket. The promises to purchase additional shares were not publicly disclosed. Under Section 11 of the Securities Act of 1933, investors can sue issuers and underwriters for such material omissions. So before In re IPO many IPO rms operated at an increased risk of shareholder litigation. Moreover, IPO rms generally have underdeveloped legal divisions and lower insurance coverage against shareholder litigation. So IPO rms are most likely to use nancial policies as a precautionary tool against the risk of shareholder litigation. Table 1.7 reports the results of the DDD model using an indicator variable equal to 1 for the two years of data that immediately follow a rm's IPO. Cash is the dependent variable. The estimation spans the years from 1993 through 2014. I separate the years into three periods: (1) before In re IPO and before CAFA, (2) after In re IPO and before CAFA, and (3) after In re IPO and after CAFA. It should be unsurprising that an IPO rm's cash is above its overall rm 29 average. IPO rms raise money during an IPO. As a result, their cash levels rise. In re IPO, which coincides with the dot-com crash, reduces the average cash of all IPO rms. Column 1 of Table 1.7 uses the sued sample and includes rm and year xed eects. As mentioned cash is the dependent variable. Column 1 shows that in the two years before a lawsuit ling only IPO rms increase cash: the coecient on Pre IPO suggests that before In re IPO an IPO rm increases cash, in the two years before a lawsuit ling, away from the IPO rm average by 0.802 (i.e. 0.004 + 0.798). Thereafter In re IPO only aects the cash of IPO rms: In re IPO dampens an IPO rm's cash increase in the two years before a lawsuit ling by 108.5%. CAFA provides no additional eect on cash in the two-year period preceding a lawsuit ling. Column 1 of Table 1.7 shows the coecient on Post to be negative and slightly signicant: before In re IPO a rm drops cash, in the two years after a lawsuit ling, to 0.080 below rm average. At the same time, an IPO rm drops cash, in the two years after a lawsuit ling, to 0.811 (i.e. -0.080 - 0.731) below the IPO rm average. In re IPO has no eect on cash in the two years after a lawsuit ling. CAFA, however, has a strong eect on the cash of all rms in the two years after a lawsuit ling: CAFA retracts cash in the two years after a lawsuit ling toward the rm average by 95.0%. CAFA has no additional eect on IPO rms in the two years after a lawsuit ling. Recall, CAFA curtails excessive settlements and limits plaintis' attorneys fees. As a result, CAFA decreases the expected costs of shareholder litigation for all rms. When a rm's expected costs of shareholder litigation decrease, so too does its demand for insurance against shareholder litigation. Because a major expense of shareholder litigation insurance is a per ling deductible, which can range in millions of dollars, it should not be surprising that CAFA's eect on cash arises in the two-year period following a lawsuit ling. Because CAFA targets all rms, it need not have a dierential impact on IPO rms. Column 2 of Table 1.7 uses the matched sample and includes rm and industry-year xed eects. Again cash is the dependent variable. Column 2 shows practically the same results as column 1. Column 3 also uses the matched sample but only includes industry-year xed 30 eects. Column 3 demonstrates the importance of accounting for within-rm variation. Without accounting for within-rm variation, In re IPO absorbs the eect of CAFA. So the regression cannot disentangle the eect of In re IPO from that of CAFA. Nonetheless, the regression in column 3 shows that before In re IPO only IPO rms change cash away from the industry-year average in the two years before a lawsuit ling: an IPO rm increases cash away from the industry- year average by 0.863 (i.e. 0.027 + 0.836) in the two years before a lawsuit ling. Turning to the two years after a lawsuit ling, the coecient on Post shows that before In re IPO a rm drops cash, in the two years after a lawsuit ling, to 0.078 below the industry-year average. But the result is much more pronounced among IPO rms: before In re IPO, an IPO rm drops cash in the two years after a lawsuit ling to 0.693 (i.e. -0.078 - 0.615) below the IPO industry-year average. Table 1.8 reports the results of the DDD model using investment as the dependent variable. The estimation spans the years from 1993 through 2014. Again I separate the years into three periods: (1) before In re IPO and before CAFA, (2) after In re IPO and before CAFA, and (3) after In re IPO and after CAFA. It should be unsurprising that an IPO rm invests more than its overall rm average. Presumably, a rm commences an IPO in order to obtain funds to satisfy its high investment needs. Neither In re IPO nor CAFA signicantly impacts the investment of an average IPO rm. Column 1 of Table 1.8 uses the sued sample and includes rm and year xed eects. Again investment is the dependent variable. Column 1 shows the coecient on Pre to be positive and signicant: before In re IPO a rm increases investment, in the two years before a lawsuit ling, away from rm average by 0.179. More importantly, before In re IPO an IPO rm increases investment, in the two years before a lawsuit ling, away from the IPO rm average by 0.708 (i.e. 0.179 + 0.529). Thereafter In re IPO dampens an IPO rm's investment increase in the two years before a lawsuit ling by 102.0%. Turning to the two years after a lawsuit ling, the coecient 31 on Post IPO reveals that before In re IPO an IPO rm reduces investment to 0.381 (i.e. -0.064 - 0.317) below the IPO rm average. Column 2 of Table 1.8 uses the matched sample and includes rm and industry-year xed eects. Again investment is the dependent variable. Column 2 shows practically the same results as column 1 with two notable exceptions. First, the coecient on Post is signicant indicating that all rms reduce investment in the two years after a ling to 0.093 below their rm average. Second, CAFA does not appear to retract the investment reductions that occur in the two years after a lawsuit ling. Column 3 also uses the matched sample but only includes industry-year xed eects. The regression in column 3 shows that before In re IPO a rm in the two years before a lawsuit ling invests 0.207 more than the industry-year average; meanwhile, before In re IPO an IPO rm shifts investment, in the two years before a lawsuit ling, by an additional 0.512 away from the IPO industry-year average. Thereafter In re IPO moves the investment of all rms in the two years before a lawsuit ling closer to their industry-year average by 0.106; In re IPO moves the investment of IPO rms in the two years before a lawsuit ling closer to their IPO industry-year average by an additional 0.558. Practically none of the coecients for the two-year period following a lawsuit ling are signicant in column 3. Table 1.9 reports the results of the DDD model using debt as the dependent variable. The estimation spans the years from 1993 through 2014. Again I separate the years into three periods: (1) before In re IPO and before CAFA, (2) after In re IPO and before CAFA, and (3) after In re IPO and after CAFA. Column 1 uses the sued sample and includes rm and year xed eects. Column 1 shows the coecient on Pre to be positive and signicant: before In re IPO a rm raises debt, in the two years before a lawsuit ling, away from rm average by 0.037. None of the coecients on the Pre interaction terms are signicant in column 1. The coecient on Post shows that before In re IPO a rm lowers debt, in the two years after a lawsuit ling, to 0.039 below rm average. None of the coecients on the Post interaction terms are signicant in column 1. Column 2 uses the matched sample and includes rm and industry-year xed eects. The results 32 of column 2 are not all that dierent from those of column 1, except that the coecient on Post in column 2 is insignicant. Column 3 of Table 1.9 uses the matched sample and only includes industry-year xed eects. The coecient on Pre indicates that before In re IPO a rm raises debt, in the two years before a lawsuit ling, away from the industry-year average by 0.053. The coecient on Pre IPO, however, suggests that an IPO rm lowers debt, in the two years before a lawsuit ling, to 0.015 (i.e. 0.053 - 0.068) below the IPO industry-year average. This is the rst hint suggesting the relation between debt and shareholder litigation risk to dier among IPO rms versus more established rms. Later I conrm that IPO rms with higher shareholder litigation risk reduce debt. Together the results suggest that debt may be an alternative to cash as a precautionary measure against the threat of shareholder litigation. When cash is used to build a precautionary buer, less debt is required; when cash is sparse, debt is used. Most IPO rms are replete with cash from IPO proceeds, so they may avoid any additional burdens of debt. While my results do not unequivocally show debt to be an alternative to cash as a precautionary measure against the threat of shareholder litigation, the conjecture is an interesting point of inquiry for future empirical work. 1.4.4 Dierence in Dierences Models Using the IPO Sample The prior subsection demonstrates that before In re IPO practically all of the cash increases in the two-year period preceding and decreases in the two-year period following a lawsuit ling occur among IPO rms. Similarly, before In re IPO practically all of the investment increases in the two-year period preceding a lawsuit ling occur among IPO rms. Before In re IPO many IPO rms, along with their underwriters, allocated IPO shares conditional on allocants promising to purchase additional shares in the aftermarket. The promises to purchase additional shares were not publicly disclosed. Under Section 11 of the Securities Act of 1933, investors can sue issuers and underwriters for such material omissions. So before In re IPO many IPO rms operated at 33 an increased risk of shareholder litigation. After In re IPO reveals the existence of the scheme in which IPO allocants promised to purchase additional shares in the aftermarket, the use of the scheme stops. In turn, an IPO rm's actions change: the increases in cash and investment, in the two years before a lawsuit ling, disappear. To further isolate the risk of shareholder litigation as the source of an IPO rm's nancial policy changes, I examine cross-sectional variation in the risk of shareholder litigation in a sample of IPO rms alone. I focus on a sample of IPO rms alone because as the prior subsection reveals, they are responsible for most of the cash and investment uctuations surrounding a lawsuit ling. In the sample of IPO rms, I examine four sources of cross-sectional variation in the risk of shareholder litigation. The rst source of cross-sectional variation in the risk of shareholder litigation is an indicator variable \IPO Allocation" equal to 1 for rms alleged to be involved in the scheme that led to In re IPO. Firms accused of being involved in the scheme have high ex ante exposure to the scheme and high risk of shareholder litigation. Firms not accused of being involved in the scheme consist of two sets: 1) those actually not involved in the scheme; and 2) those far enough removed from the scheme or lucky enough to avoid implication. So on average, the set of cases with the IPO Allocation variable set to 1 should have a higher risk of shareholder litigation than the set of cases with the IPO Allocation variable set to 0. The second source of cross-sectional variation in the risk of shareholder litigation is industry litigiousness. To obtain industry litigiousness, I sort all 4-digit SIC industries in the full sample from highest to lowest based on one, the total number of lawsuit lings within each industry and two, the fraction of rms that are sued within each industry. I then classify the rst set of industries, which contains half of all lawsuit lings in the matched sample, as litigious. By construction litigious industries have a higher risk of shareholder litigation. The third source of cross-sectional variation in the risk of shareholder litigation is stock volatil- ity. I measure stock volatility for a rm's initial 252 days of trading. Regressions with stock volatility as the cross-sectional variable exclude rms that undergo a lawsuit ling within the 34 rst 252 days of trading or within one month after the 252nd day of trading. A rm with high stock volatility has a higher likelihood of sustaining a drop in its stock price. Because lawsuit lings often coincide with falling stock prices, rms with volatile stocks have an increased risk of shareholder litigation (Lowry and Shu, 2002; Gande and Lewis, 2009). The fourth source of cross-sectional variation in the risk of shareholder litigation is stock turnover. Like stock volatility, I measure stock turnover for a rm's initial 252 days of trading. Regressions with stock turnover as the cross-sectional variable exclude rms that undergo a lawsuit ling within the rst 252 days of trading or within one month after the 252nd day of trading. The greater the stock turnover, the greater the number of potential plaintis. Because shareholder damages are increasing in the number of plaintis, plaintis' attorneys have incentive to target rms with high stock turnover (Lowry and Shu, 2002; Arena and Julio, 2015). Table 1.10 reports the DD results using the IPO sample and IPO Allocation as the cross- sectional variable. The estimation spans the years from 1993 through 2014. Columns 1 and 2 use cash as the dependent variable. In column 1, the coecient on IPO Allocation indicates that before In re IPO, an IPO rm later accused of being involved in the IPO allocation scheme retains 1.269 more cash than the annual average. In re IPO reduces the cash holdings of an IPO rm accused of being involved in the IPO allocation scheme by 107.5%. In column 2, the coecient on IPO Allocation indicates that for all years in the sample an IPO rm accused of being involved in the IPO allocation scheme retains 1.564 more cash than the annual average. In the two-year period preceding a lawsuit ling, an IPO rm retains 0.213 more cash than the annual average, while an IPO rm accused of being involved in the IPO allocation scheme retains 1.225 (i.e. 1.564 + 0.213 - 0.552) more cash than the annual average. The coecient on Post indicates that an IPO rm in the two-year period following a lawsuit ling decreases cash to 0.414 below the annual average. The coecient on Post IPO Allocation indicates that an IPO rm accused of being involved in the IPO allocation scheme has a considerably more pronounced decrease in cash. 35 Columns 3 and 4 of Table 1.10 use investment as the dependent variable. In column 3, the coecient on IPO Allocation indicates that before In re IPO, an IPO rm later accused of being involved in the IPO allocation scheme invests 0.944 more than the annual average. In re IPO reduces the investment level of an IPO rm accused of being involved in the IPO allocation scheme by 132.7%. In column 4, the coecient on IPO Allocation indicates that for all years in the sample an IPO rm accused of being involved in the IPO allocation scheme invests 0.970 more than the annual average. In the two-year period preceding a lawsuit ling, an IPO rm invests 0.383 more than the annual average. Turning to the two-year period following a lawsuit ling, the coecient on Post IPO Allocation indicates that an IPO rm accused of being involved in the IPO allocation scheme reduces investment to 0.305 (i.e. 0.970 - 0.116 - 1.159) below the annual average. Columns 5 and 6 of Table 1.10 use debt as the dependent variable. In column 5, the coecient on IPO Allocation indicates that before In re IPO, an IPO rm later accused of being involved in the IPO allocation scheme holds 0.074 less debt than the annual average. In column 6, the coecient on IPO Allocation indicates that for all years in the sample an IPO rm accused of being involved in the IPO allocation scheme holds 0.380 less debt than the annual average. The results contrast the debt results of more established rms { those that did not recently undergo an IPO. More established rms increase debt in the two-year period preceding a lawsuit ling, possibly as a deterrent against shareholder litigation. Together the results of IPO and more established rms suggest that debt may be an alternative to cash as a precautionary measure against the threat of shareholder litigation. When cash is used to build a precautionary buer, less debt is required; when cash is sparse, debt is used. IPO rms are replete with cash, so they avoid any additional burdens of debt. More established rms are likely to have more persistent cash holdings, so they turn to debt for the precautionary buer. The coecient on Post indicates that an IPO rm in the two-year period following a lawsuit ling lowers debt to 0.093 below the annual average. 36 Table 1.11 reports the DD results using the IPO sample and industry litigiousness as the source of cross-sectional variation. The estimation spans the years from 1993 through 2014. Columns 1 and 2 use cash as the dependent variable. In column 1, the coecient on Lit Ind indicates that before In re IPO, an IPO rm in a litigious industry retains 0.876 more cash than the annual average. In re IPO decreases the cash holdings of an IPO rm in a litigious industry by 45.1%. In column 2, the coecient on Lit Ind indicates that for all years in the sample an IPO rm in a litigious industry retains 0.748 more cash than the annual average. In the two-year period preceding a lawsuit ling, an IPO rm holds 0.451 more cash than the annual average. While the coecient on Post is insignicant, the coecient on Post Lit Ind is signicant: an IPO rm in a litigious industry reduces cash in the two years after a lawsuit ling to 0.230 (i.e. 0.748 + 0.002 - 0.520) above the annual average. Columns 3 and 4 of Table 1.11 use investment as the dependent variable. In column 3, the coecient on Lit Ind indicates that before In re IPO, an IPO rm in a litigious industry invests 0.377 more than the annual average. In re IPO does not have a signicant impact on the investment level of an IPO rm in a litigious industry. In column 4, the coecient on Lit Ind indicates that for all years in the sample an IPO rm in a litigious industry invests 0.279 more than the annual average. In the two-year period preceding a lawsuit ling, an IPO rm invests 0.415 more than the annual average. Furthermore, an IPO rm in a litigious industry in the two-year period preceding a lawsuit ling invests 0.932 (i.e. 0.279 + 0.415 + 0.238) more than the annual average. Neither the coecient on Post nor the coecient on Post Lit Ind is signicant. So in the two-year period following a lawsuit ling, investment stands at the annual average irrespective of an IPO rm's industry. Columns 5 and 6 of Table 1.11 use debt as the dependent variable. In column 5, the coecient on Lit Ind indicates that before In re IPO, an IPO rm in a litigious industry holds 0.189 less debt than the annual average. In re IPO dampens the debt of an IPO rm in a litigious industry by 43.4%. In column 6, the coecient on Lit Ind indicates that for all years in the sample an 37 IPO rm in a litigious industry holds 0.170 less debt than the annual average. The coecient on Post indicates that an IPO rm in the two-year period following a lawsuit ling reduces debt to 0.121 below the annual average. As with Table 1.10, the debt results of Table 1.11 contrast the debt results of more established rms. Again debt appears to be an alternative to cash as a precautionary measure against the threat of shareholder litigation. IPO rms use cash as the precautionary buer against the threat of shareholder litigation, while more established rms use debt as the precautionary buer against the threat of shareholder litigation. Table 1.12 reports the DD results using the IPO sample and stock volatility as the source of cross-sectional variation. I consider an IPO rm to have high stock volatility if its stock volatility exceeds the median stock volatility of all IPO rms during the initial 252 days of trading. The estimation spans the years from 1993 through 2014. Columns 1 and 2 use cash as the dependent variable. In column 1, the coecient on Volatility indicates that before In re IPO, an IPO rm with high stock volatility retains 0.905 more cash than the annual average. In re IPO decreases the cash holdings of an IPO rm with high stock volatility by 75.5%. In column 2, the coecient on Volatility indicates that for all years in the sample an IPO rm with high stock volatility retains 0.735 more cash than the annual average. In the two-year period preceding a lawsuit ling, an IPO rm holds 0.329 more cash than the annual average. While the coecient on Post is insignicant, the coecient on Post Volatility is signicant: an IPO rm with high stock volatility reduces cash in the two years after a lawsuit ling to 0.344 (i.e. 0.735 - 0.095 - 0.296) above the annual average. Columns 3 and 4 of Table 1.12 use investment as the dependent variable. In column 3, the coecient on Volatility indicates that before In re IPO, an IPO rm with high stock volatility invests 0.529 more than the annual average. In re IPO does not have a signicant impact on the investment level of an IPO rm with high stock volatility. In column 4, the coecient on Volatility indicates that for all years in the sample an IPO rm with high stock volatility invests 0.441 more than the annual average. In the two-year period preceding a lawsuit ling, an IPO 38 rm invests 0.514 more than the annual average. While the coecient on Post is insignicant, the coecient on Post Volatility is signicant: an IPO rm with high stock volatility reduces investment in the two years after a lawsuit ling to 0.112 (i.e. 0.441 + 0.025 - 0.354) above the annual average. Columns 5 and 6 of Table 1.12 use debt as the dependent variable. In column 5, the coecient on Volatility indicates that before In re IPO, an IPO rm with high stock volatility holds 0.124 less debt than the annual average. In re IPO has no impact on the debt level of an IPO rm with high stock volatility. In column 6, the coecient on Volatility indicates that for all years in the sample an IPO rm with high stock volatility holds 0.117 less debt than the annual average. The coecient on Post indicates that an IPO rm in the two-year period following a lawsuit ling reduces debt to 0.157 below the annual average. Again, it appears as if debt may be an alternative to cash as a precautionary measure against the threat of shareholder litigation. IPO rms use cash as the precautionary buer against the threat of shareholder litigation, while more established rms use debt as the precautionary buer against the threat of shareholder litigation. Table 1.13 reports the DD results using the IPO sample and stock turnover as the source of cross-sectional variation. I consider an IPO rm to have high stock turnover if its stock turnover exceeds the median stock turnover of all IPO rms during the initial 252 days of trading. The estimation spans the years from 1993 through 2014. Columns 1 and 2 use cash as the dependent variable. In column 1, the coecient on Turnover indicates that before In re IPO, an IPO rm with high stock turnover retains 0.679 more cash than the annual average. In re IPO decreases the cash holdings of an IPO rm with high stock turnover by 69.8%. In column 2, the coecient on Turnover indicates that for all years in the sample an IPO rm with high stock turnover retains 0.480 more cash than the annual average. In the two-year period preceding a lawsuit ling, an IPO rm holds 0.329 more cash than the annual average. While the coecient on Post is insignicant, the coecient on Post Turnover is signicant: an IPO rm with high stock turnover reduces 39 cash in the two years after a lawsuit ling to 0.042 (i.e. 0.480 - 0.014 - 0.424) above the annual average. Columns 3 and 4 of Table 1.13 use investment as the dependent variable. In column 3, the coecient on Turnover indicates that before In re IPO, an IPO rm with high stock turnover invests 0.565 more than the annual average. In re IPO reduces the investment level of an IPO rm with high stock turnover by 70.1%. In column 4, the coecient on Turnover indicates that for all years in the sample an IPO rm with high stock turnover invests 0.350 more than the annual average. In the two-year period preceding a lawsuit ling, an IPO rm invests 0.380 more than the annual average. Neither the coecient on Post nor the coecient on Post Turnover is signicant. So in the two-year period following a lawsuit ling, investment stands at the annual average irrespective of an IPO rm's stock turnover. Columns 5 and 6 of Table 1.13 use debt as the dependent variable. In column 5, neither the coecient on Turnover nor the coecient on Turnover In re IPO is signicant. In column 6, however, the coecient on Turnover is slightly signicant. It suggests that an IPO rm with high stock turnover holds 0.048 less debt than the annual average. None of the other coecients in column 6 are signicant. The results on debt in Table 1.13 provide little additional insight. 1.4.5 IPO Proceeds Thus far, the results show the risk of shareholder litigation to vary across time and across rms. When the risk of shareholder litigation is high and shareholder litigation insurance does not ac- count for all of the risk, a rm may supplement insurance with nancial policy changes. As discussed, In re IPO revealed an IPO allocation scheme in which many IPO rms and their un- derwriters allocated IPO shares conditional on allocants' promises to purchase additional shares in the aftermarket. The promises to purchase additional shares were not publicly disclosed. Under Section 11 of the Securities Act of 1933, investors can sue issuers and underwriters for such mate- rial omissions. So before In re IPO many IPO rms operated at an increased risk of shareholder 40 litigation. Since IPO rms generally have underdeveloped legal divisions and lower insurance cov- erage against shareholder litigation, they are likely to take supplemental precautionary measures in anticipation of a lawsuit ling. The results before In re IPO show IPO rms to increase cash and investment before a lawsuit ling. I claim these cash and investment changes to be driven by precautionary motives resulting from the risk of shareholder litigation. For example, an IPO rm retains cash to build a buer; it increases investment to improve performance and obviate a lawsuit ling. But an alternative explanation looms. The IPO allocation scheme, itself, may generate a large amount of cash for IPO rms that engage in the scheme. For example, IPO rms that engage in the scheme may obtain greater IPO proceeds, which directly increase the cash holdings. In turn, investment mechanically rises as a consequence of the greater IPO proceeds. An examination of IPO proceeds, however, reveals the alternative explanation to be unlikely. Table 1.14 compares IPO proceeds of rms alleged to be involved in the IPO allocation scheme with IPO proceeds of all other rms. IPO proceeds include the overallotment option. Data on IPO proceeds come from the Securities Data Company (SDC). Of the 1,080 IPOs in the sample, I obtain 715 matches in SDC. Of the 715 matches, 114 are for rms alleged to be involved in the IPO allocation scheme. Columns 1 and 2 of Table 1.14 show IPO proceeds de ated by total assets. Both columns show no signicant dierence between the IPO proceeds of a rm alleged to be involved in the scheme and the IPO proceeds of an average rm. Columns 3 and 4 show IPO proceeds de ated by market capitalization. Now the IPO proceeds of a rm alleged to be involved in the scheme are signicantly lower than the IPO proceeds of an average rm. This is not a surprise { the scheme aimed to drive up a new issuer's market capitalization in the aftermarket. So the post IPO market capitalization of a rm alleged to be involved in the scheme was articially in ated without aecting its proceeds. Hence the negative coecients in columns 3 and 4. Table 1.14 lends support to the claim that an IPO rm's nancial policy changes before a lawsuit ling are driven by precautionary motives. The table shows the IPO proceeds of an IPO 41 rm, at risk of shareholder litigation, to be no dierent than the IPO proceeds of an average IPO rm. But an IPO rm at risk of shareholder litigation retains a much greater quantity of cash compared to an average IPO rm; also, an IPO rm at risk of shareholder litigation invests far more compared to an average IPO rm. 1.5 Discussion Two papers have considered a question related to mine: Arena and Julio (2015) and McTier and Wald (2011). In this section, I draw a contrast between my results and those of Arena and Julio (2015) and McTier and Wald (2011). Arena and Julio (2015) use the sample period from 1996 to 2006 to show that rms adjust cash holdings and investment policy in order to manage shareholder litigation risk. I show the results to be sensitive to the time period examined and the identication strategy used. The increase in cash and investment prior to a lawsuit ling occurs before In re IPO and among IPO rms. Arena and Julio (2015) also show that a rm's cash increases after a lawsuit ling. My results show the opposite: a rm decreases cash after a lawsuit ling. If a rm uses cash as a buer against the risk of shareholder litigation, then cash should accumulate before a lawsuit ling. After a lawsuit ling, the buer should be used to combat the consequences of shareholder litigation: for example, to pay the deductible for shareholder litigation insurance or to mitigate the ill eects of shareholder litigation. Consider Table 1.15, which examines the eect of a lawsuit ling on cash and investment in the period after CAFA: when shareholder litigation risk is low. Columns 1 and 2 show no increase in cash in the two-year period preceding a lawsuit ling. Contrary to Arena and Julio (2015), columns 1 and 2 do show cash to decrease in the two-year period following a lawsuit ling. So after a lawsuit ling, cash decreases even in the period when the risk of shareholder litigation is low. Column 2 accounts for IPO status { being within two years of going public { and shows the 42 brunt of the decrease in the two-year period following a lawsuit ling to occur among IPO rms. Column 3 shows a rm to increase investment, in the two years before a lawsuit ling, away from rm average by 0.081. But after controlling for IPO status in column 4, the investment increase in the two years before a lawsuit ling becomes practically insignicant. To address endogeneity my identication strategy uses a quasi-natural experiment, whereas Arena and Julio (2015) use a simultaneous equation model. In the simultaneous equation model, the authors proxy for shareholder litigation risk by assigning a dummy variable equal to 1 for rms involved in a shareholder litigation in the following year and 0 otherwise. The authors cite Lowry and Shu (2002) as support for the choice of dummy variable. But the shareholder litigation risk dummy correlates with an important omitted factor { IPO status. Lowry and Shu do not have the issue, because they only examine IPO rms and thus need not control for IPO status. Finally, Arena and Julio (2015) use rm leverage and stock turnover as identifying variables in the simultaneous equation model. Firm leverage is assumed to uniquely relate to cash but not to litigation risk, and stock turnover is assumed to uniquely relate to litigation risk but not to cash. My results show, however, that rm leverage may relate to shareholder litigation risk. In addition, Lin et al. (2013) provide evidence of a relation between shareholder litigation risk and loan spreads. So a relation between rm leverage and litigation risk is not unthinkable. With respect to stock turnover, Arena and Julio (2015) again cite Lowry and Shu (2002) as support for the choice of identifying variable. But Lowry and Shu do not use a rm's stock turnover as an identifying variable. Instead, they use a matched rm's stock turnover as an identifying variable. A rm's stock turnover may relate to its cash, particularly around the time of an IPO. For instance, a rm that reveals new investment opportunities may exhibit high stock turnover and high cash. In my examination, I only use stock turnover as a robustness check. Moreover, I have the added benet of using stock turnover in conjuction with an exogenous legal shock { In re IPO. In re IPO helps gauge the validity of stock turnover as a proxy for shareholder litigation risk. If stock turnover is endogenous and does not proxy for shareholder litigation risk, 43 then its eect on an IPO rm's cash would change little or not at all as a result of In re IPO. Column 1 of Table 1.13, however, shows In re IPO to decrease the eect of stock turnover on an IPO rm's cash by a signicant 69.8%. McTier and Wald (2011) use the sample period from 1996 to 2005 to argue that a rm which overinvests is more likely to succumb to a shareholder class action and that the shareholder class action corrects the overinvestment. As mentioned, the result is sensitive to the time period examined and the identication strategy used. Consider Table 1.15 again, which shows the eect of a lawsuit ling on investment in the period after In re IPO and CAFA: when shareholder litigation risk is low. After controlling for IPO status in column 4, the investment increase in the two years before a lawsuit ling becomes practically insignicant. Finally, I argue that neither high cash nor high investment cause a shareholder class action. Instead, high cash and high investment are a consequence of an anticipated shareholder class action. Suppose high cash caused a shareholder class action, because plaintis' attorneys target rms that have more resources to pay large settlements. Then after In re IPO and CAFA, when it becomes more dicult to obtain a large settlement, plaintis' attorneys would target rms with even greater cash holdings to help ensure an adequate settlement. Suppose high investment caused a shareholder class action, because a shareholder class action corrects overinvestment problems. Then In re IPO and CAFA would have little eect on the correlation between investment and shareholder class actions. That is, shareholder class actions would always coincide with high overinvestment problems. My results run counter to these suppositions. After In re IPO and CAFA, when the risk of shareholder litigation decreases, rms reduce cash and investment before a lawsuit ling. The results suggest that neither cash nor investment cause a lawsuit ling; rather cash and investment change relative to the risk of shareholder litigation. 44 1.6 Conclusion In this chapter, I show shareholder litigation risk to vary across time and across rms. Shareholder litigation can impact a rm's cash and investment if and only if the risk of shareholder litigation is high: for example, when shareholder litigation insurance is limited or costly. I show IPO rms { those within two years of going public { to be responsible for all cash and investment increases before a lawsuit ling. Moreover, the increases are concentrated in the period before In re IPO. Before In re IPO shareholder litigation risk is higher for IPO rms. After a lawsuit ling occurs, IPO rms reduce cash and investment. To further isolate the risk of shareholder litigation as the source of an IPO rm's cash and investment changes surrounding a lawsuit ling, I examine cross-sectional variation by risk of shareholder litigation in a sample of IPO rms alone. I use four sources of cross-sectional variation: (1) an indicator variable for rms alleged to be involved in the IPO allocation scheme that led to In re IPO; (2) an indicator variable for rms in litigious industries; (3) an indicator variable for rms with high stock volatility; and (4) an indicator variable for rms with high stock turnover. Compared to an average IPO rm, an IPO rm alleged to be involved in the IPO allocation scheme has higher cash and higher investment before a lawsuit ling. Also, compared to an average IPO rm, a IPO rm alleged to be involved in the IPO allocation scheme has a more signicant decrease in cash and investment after a lawsuit ling. The same results hold for IPO rms in litigious industries, for IPO rms with high stock volatility, and for IPO rms with high stock turnover. An examination of IPO proceeds further suggests the above linkage, between the risk of share- holder litigation and an IPO rm's nancial policy changes, to be causal. Specically, the IPO proceeds of an IPO rm alleged to be involved in the scheme are not higher than the IPO pro- ceeds of an average IPO rm. So it is not the case that the scheme, itself, generates high cash and high investment for an IPO rm alleged to be involved in the scheme. Rather, an IPO rm 45 alleged to be involved in the scheme chooses to retain cash and to increase investment, suggesting a precautionary motivation related to the risk of shareholder litigation. Debt increases in the two years before a lawsuit ling, and the increase occurs among more established rms. Limiting the sample to IPO rms alone, however, reveals IPO rms at risk of shareholder litigation to decrease debt in the two years before a lawsuit ling. Together the results suggest debt to be an alternative to cash as a precautionary measure against the threat of shareholder litigation. When cash is used to build a precautionary buer, less debt is required; when cash is sparse, debt is used. Many IPO rms are replete with cash from IPO proceeds, so they may avoid any burdens of debt. While the results do not unequivocally show debt to be an alternative to cash as a precautionary measure against the threat of shareholder litigation, the conjecture is an interesting point of inquiry for future empirical work. 46 1.7 Figures Figure 1.1: Average cash, investment, and debt trends for sued and nonsued rms surrounding the start of shareholder litigation risk. Data for all years, involving at least one lawsuit ling, are placed at year 0 on the x-axis. The gure shows average cash, investment, and debt adjusted by rm and time eects for sued and nonsued rms, starting 6 years before and ending 2 years after year 0. Cash is dened as cash and short-term investments for rm i at scal year t divided by total assets for rm i at scal year t 1. Investment is dened as capital expenditures for rmi at scal yeart divided by net property, plant, and equipment for rm i at scal yeart 1. Debt is dened as the sum of book-value short- and long-term debt for rm i at scal yeart divided by total assets for rmi at scal yeart 1. Vertical blue dashed lines denote the assumed start of shareholder litigation risk. Solid black (dashed red) lines represent sued (nonsued) rms. Dashed black lines denote the 95% condence interval for the dierence between sued and nonsued rms. Data are for the years 1993-2014 and are described in Section 1.2. ● ● ● ● ● ● ● ● ● −6 −4 −2 0 2 −0.8 −0.4 0.0 ● ● ● ● ● ● ● ● ● ● Sued Nonsued Cash ● ● ● ● ● ● ● ● ● −6 −4 −2 0 2 −0.8 −0.4 0.0 ● ● ● ● ● ● ● ● ● Investment ● ● ● ● ● ● ● ● ● −6 −4 −2 0 2 −0.28 −0.22 −0.16 ● ● ● ● ● ● ● ● ● Time (annual): 0 denotes year of filing Debt 47 Figure 1.2: Timeline of a typical shareholder class action. The class period, during which the purported violation takes place, precedes the trigger date. The revelation of the purported violation leads to the trigger date. The lawsuit ling closely follows the trigger date. After preliminary motions are led, and the judge chooses a lead plainti, discovery begins. Generally during or after discovery, class certication occurs. Following discovery, summary judgment is made. Should any of the case survive by this point, preparation for trial begins. The remainder of the case concludes in trial court. Class Period Class Period Start Class Period End Trigger Date Filing Date Discovery Period Class Certied Summary Judgment Trial 48 Figure 1.3: Epanechnikov kernel densities of cash for sued rms. Cash is adjusted by industry-time eects. The densities are formed by dividing the full sample according to two legal shocks: (1) the December 2000 revelation of an IPO allocation scheme that led to an important set of legal cases dubbed In re IPO; (2) the 2005 enactment of CAFA. The densities resemble one another. Furthermore, a Kruskal-Wallis test for equality of densities cannot be rejected at the 5% level. Cash is dened as in Figure 1.1. Data are described in Section 1.2. −2 0 2 4 6 8 10 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Kernel densities of cash Before 2001 2001 − 2004 After 2004 49 Figure 1.4: Epanechnikov kernel densities of investment for sued rms. Investment is adjusted by industry-time eects. The densities are formed by dividing the full sample according to two legal shocks: (1) the December 2000 revelation of an IPO allocation scheme that led to an important set of legal cases dubbed In re IPO; (2) the 2005 enactment of CAFA. The densities resemble one another. Investment is dened as in Figure 1.1. Data are described in Section 1.2. 0 5 10 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Kernel densities of investment Before 2001 2001 − 2004 After 2004 50 Figure 1.5: Epanechnikov kernel densities of debt for sued rms. Debt is adjusted by industry-time eects. The densities are formed by dividing the full sample according to two legal shocks: (1) the December 2000 revelation of an IPO allocation scheme that led to an important set of legal cases dubbed In re IPO; (2) the 2005 enactment of CAFA. The densities resemble one another. Furthermore, a Kruskal-Wallis test for equality of densities cannot be rejected at the 5% level. Debt is dened as in Figure 1.1. Data are described in Section 1.2. 0 5 10 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Kernel densities of debt Before 2001 2001 − 2004 After 2004 51 1.8 Tables Table 1.1: Summary statistics for sued rms by industry and by irregularity. Panel A shows summary statistics by industry. Panel B shows summary statistics by irregularity. The industry and irregularity classications are obtained from Cornerstone Research and Stanford Law School. The sample of lawsuit lings spans years 1996-2014. Before 2001 refers to the period before In re IPO; 2001-2004 refers to the period after In re IPO and before CAFA; after 2004 refers to the period after CAFA. In panel A, \other" includes consumer cyclical, energy, and transportation. In panel B, \other" includes broker-market manipulation and patent infringement fraud, while \not specied" includes lawsuit lings for which Cornerstone Research and Stanford Law School did not specify an irregularity. Panel A: Summary statistics by industry Percent Industry Before 2001 2001 - 2004 After 2004 Full Sample Technology 44.4 59.3 30.8 44.3 Services 23.4 15.2 22.6 20.2 Healthcare 11.7 13.6 24.9 17.6 Consumer Goods 11.3 4.4 11.2 8.8 Industrial Goods 5.2 4.4 4.5 4.6 Basic Materials 2.8 3.0 6.0 4.2 Other 1.2 0.0 0.0 0.3 Panel B: Summary statistics by irregularity Percent Irregularity Before 2001 2001 - 2004 After 2004 Full Sample Failure of disclosure (accounting) 41.1 40.4 24.1 34.1 Misrepresentation (non accounting) 55.6 20.5 25.6 31.2 IPO allocation 0.0 37.1 0.0 13.3 Failure of disclosure (non accounting) 0.4 0.3 18.7 7.6 Not specied 0.0 0.0 16.9 6.7 Option dating 0.0 0.0 5.2 2.1 Product liability related fraud 0.0 0.6 2.7 1.3 Mergers & acquisitions 0.8 0.0 2.5 1.2 IPO non allocation 0.8 0.3 1.2 0.8 Insider trading 0.4 0.6 1.2 0.8 Credit crisis 0.0 0.0 1.7 0.7 Other 0.8 0.3 0.0 0.3 52 Table 1.2: Summary statistics for sued rms by ruling and by exchange. Panel A shows summary statistics by ruling. Panel B shows summary statistics by exchange. The ruling and exchange data are obtained from Cornerstone Research and Stanford Law School. The sample of lawsuit lings spans the years 1996-2014. Before 2001 refers to the period before In re IPO; 2001-2004 refers to the period after In re IPO and before CAFA; after 2004 refers to the period after CAFA. Panel A: Summary statistics by ruling Percent Ruling Before 2001 2001 - 2004 After 2004 Full Sample Settled 62.1 70.1 33.8 53.7 Dismissed 37.9 29.1 39.8 35.5 Ongoing 0.0 0.8 26.4 10.8 Panel B: Summary statistics by exchange Percent Exchange Before 2001 2001 - 2004 After 2004 Full Sample NASDAQ 64.5 73.7 60.2 66.1 NYSE 32.7 24.9 37.8 31.9 AMEX 2.8 1.4 2.0 2.0 53 Table 1.3: The combined eect of In re IPO and CAFA on lawsuit lings and lawsuit dismissals. The table examines two periods: through 2004 and after 2004. In 2005, CAFA was enacted and the initial settlement of In re IPO was proposed. The proposal of the initial settlement allows me to safely assume that by 2005 the dust had settled surrounding the events of In re IPO: for example, no new lings alleging the scheme are led in 2005 and after. By splitting the sample at the year 2004, I examine the combined eect of In re IPO and CAFA on lawsuit lings and dismissals. The estimation period is 1996-2013, which denotes years with lawsuit ling data available for the entirety of each year. Data are described in Section 1.2. The number of dismissed lawsuits manifests truncation bias toward the end of the sample, because it often takes a couple years after the date of a lawsuit ling for a lawsuit to be resolved. To adjust for the truncation bias, dismissals are randomly alloted to ongoing lawsuits, so that the fraction of dismissals among ongoing lawsuits equals the fraction of dismissals among lawsuits that have been resolved through 2004. The adjustment is conservative, because the fraction of dismissals among lawsuits that have been resolved through 2004 is smaller than the fraction after 2004: 32.8% through 2004 versus 54.1% after 2004. So the rise in dismissals after 2004 is an underestimate. To preclude selection bias, the random allotment is simulated 1,000 times. The results for dismissals are averages of the 1,000 simulations. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels. Through 2004 After 2004 Dierence Observations Lawsuit lings per year 50:8 37:5 13:3 22 Incidence of lawsuit lings 0:046 0:034 0:012 24,383 (0:004) Ratio of dismissed to total lawsuits 0:332 0:637 0:305 984 (0:037) Incidence of dismissed lawsuits 0:015 0:022 0:006 24,383 (0:002) 54 Table 1.4: The incidence of lawsuit lings and dismissals among IPO rms with high stock volatility and high stock turnover, relative to IPO rms with low stock volatility and low stock turnover. The two dependent variables are: (1) File, an indicator variable equal to 1 if a class action has been led in a given year; and (2) Dismiss, an indicator variable equal to 1 if a class action has been dismissed in a given year. Volatility is an indicator variable equal to 1 for IPO rms with stock volatility exceeding the median stock volatility of all IPO rms, where stock volatility is measured over the initial 252 days of trading. Turnover is an indicator variable equal to 1 for IPO rms with stock turnover exceeding the median stock turnover of all IPO rms, where stock turnover is measured over the initial 252 days of trading. The estimation period is 1996-2013, which denotes years with lawsuit ling data available for the entirety of each year. The adjustment for the truncation bias in the number of dismissals is described in Table 1.3. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels. Dependent variable 1(File) 1(Dismiss) 1(File) 1(Dismiss) (1) (2) (3) (4) Intercept 0:018 0:012 0:024 0:005 (0:004) (0:003) (0:004) (0:002) Volatility 0:042 0:006 (0:008) (0:004) Turnover 0:030 0:008 (0:008) (0:005) Observations 2,062 2,062 2,062 2,062 55 Table 1.5: Financial policies and shareholder litigation { dierence coecients. The nancial policies of rms are measured by three dependent variables: (1) Cash, dened as cash and short-term investments for rm i at scal yeart divided by total assets for rm i at scal year t 1; (2) Invest, dened as capital expenditures for rm i at scal year t divided by net property, plant, and equipment for rm i at scal yeart 1; and (3) Debt, dened as the sum of book-value short- and long-term debt for rm i at scal yeart divided by total assets for rm i at scal year t 1. The independent variable Pre equals 1 for the two scal years preceding a lawsuit ling and 0 otherwise. The independent variable Post equals 1 for the two scal years following a lawsuit ling and 0 otherwise. The xed eects for each regression are noted in the table. Industry-year xed eects use 4-digit SIC codes. The estimation period is 1993-2014. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Cash Cash Cash Invest Invest Invest Debt Debt Debt (1) (2) (3) (4) (5) (6) (7) (8) (9) Pre 0:197 0:186 0:297 0:246 0:219 0:325 0:022 0:021 0:024 (0:045) (0:045) (0:065) (0:033) (0:036) (0:047) (0:009) (0:009) (0:010) Post 0:153 0:165 0:061 0:110 0:125 0:029 0:024 0:020 0:015 (0:024) (0:035) (0:023) (0:017) (0:021) (0:019) (0:009) (0:009) (0:009) Firm FE Yes Yes No Yes Yes No Yes Yes No Year FE Yes No No Yes No No Yes No No Industry-Year FE No Yes Yes No Yes Yes No Yes Yes Observations 11,321 24,410 24,410 11,321 24,410 24,410 11,321 24,410 24,410 R 2 0.400 0.478 0.285 0.341 0.410 0.261 0.541 0.628 0.320 56 Table 1.6: Financial policies and shareholder litigation { DD coecients. The nancial policies of rms are measured by three dependent variables: Cash, Invest, and Debt. The three dependent variables, along with the independent variables Pre and Post, are dened in Table 1.5. In re IPO is an indicator variable equal to 1 for the year 2001 and all years that follow. CAFA is an indicator variable equal to 1 for the year 2005 and all years that follow. The xed eects for each regression are noted in the table. Industry-year xed eects use 4-digit SIC codes. The estimation period is 1993-2014. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Cash Cash Cash Invest Invest Invest Debt Debt Debt (1) (2) (3) (4) (5) (6) (7) (8) (9) Pre 0:413 0:403 0:560 0:481 0:445 0:569 0:035 0:037 0:027 (0:082) (0:078) (0:099) (0:060) (0:067) (0:074) (0:017) (0:016) (0:018) Pre In re IPO 0:353 0:349 0:474 0:381 0:375 0:445 0:021 0:018 0:008 (0:082) (0:086) (0:093) (0:075) (0:084) (0:081) (0:019) (0:019) (0:027) Pre CAFA 0:010 0:018 0:035 0:012 0:002 0:042 0:004 0:015 0:022 (0:038) (0:040) (0:042) (0:047) (0:048) (0:041) (0:014) (0:016) (0:025) Post 0:218 0:190 0:206 0:167 0:172 0:123 0:048 0:031 0:022 (0:037) (0:052) (0:056) (0:037) (0:040) (0:042) (0:018) (0:019) (0:016) Post In re IPO 0:009 0:042 0:177 0:003 0:017 0:101 0:025 0:015 0:002 (0:045) (0:048) (0:050) (0:049) (0:049) (0:046) (0:018) (0:019) (0:021) Post CAFA 0:164 0:153 0:024 0:141 0:097 0:043 0:009 0:0001 0:019 (0:041) (0:050) (0:019) (0:032) (0:037) (0:030) (0:011) (0:012) (0:017) Firm FE Yes Yes No Yes Yes No Yes Yes No Year FE Yes No No Yes No No Yes No No Industry-Year FE No Yes Yes No Yes Yes No Yes Yes Observations 11,321 24,410 24,410 11,321 24,410 24,410 11,321 24,410 24,410 R 2 0.411 0.484 0.295 0.353 0.415 0.269 0.542 0.628 0.320 57 Table 1.7: Cash and shareholder litigation { DDD coecients using IPO rms. The dependent variable Cash, along with the independent variables Pre and Post, is dened in Ta- ble 1.5. In re IPO is an indicator variable equal to 1 for the year 2001 and all years that follow. CAFA is an indicator variable equal to 1 for the year 2005 and all years that follow. IPO equals 1 for the rst two observations following each rm's date of going public and 0 otherwise. The xed eects for each regression are noted in the table. Industry-year xed eects use 4-digit SIC codes. The estimation period is 1993-2014. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Cash Cash Cash (1) (2) (3) IPO 0:646 0:684 0:773 (0:117) (0:110) (0:114) IPO In re IPO 0:240 0:329 0:266 (0:163) (0:126) (0:101) IPO CAFA 0:040 0:033 0:087 (0:144) (0:073) (0:094) Pre 0:004 0:005 0:027 (0:026) (0:042) (0:035) Pre IPO 0:798 0:765 0:836 (0:097) (0:105) (0:099) Pre In re IPO 0:053 0:071 0:036 (0:036) (0:046) (0:043) Pre IPO In re IPO 0:923 0:830 0:911 (0:162) (0:162) (0:202) Pre CAFA 0:035 0:059 0:033 (0:031) (0:032) (0:030) Pre IPO CAFA 0:078 0:148 0:282 (0:191) (0:212) (0:242) Post 0:080 0:070 0:078 (0:042) (0:036) (0:032) Post IPO 0:731 0:760 0:615 (0:152) (0:182) (0:143) Post In re IPO 0:024 0:033 0:057 (0:044) (0:044) (0:027) Post IPO In re IPO 0:078 0:150 0:278 (0:162) (0:128) (0:119) Post CAFA 0:076 0:059 0:018 (0:025) (0:026) (0:019) Post IPO CAFA 0:208 0:274 0:090 (0:189) (0:140) (0:107) Firm FE Yes Yes No Year FE Yes No No Industry-Year FE No Yes Yes Observations 11,321 24,410 24,410 R 2 0.503 0.550 0.403 58 Table 1.8: Investment and shareholder litigation { DDD coecients using IPO rms. The dependent variable Invest, along with the independent variables Pre and Post, is dened in Table 1.5. In re IPO is an indicator variable equal to 1 for the year 2001 and all years that follow. CAFA is an indicator variable equal to 1 for the year 2005 and all years that follow. IPO equals 1 for the rst two observations following each rm's date of going public and 0 otherwise. The xed eects for each regression are noted in the table. Industry-year xed eects use 4-digit SIC codes. The estimation period is 1993-2014. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Invest Invest Invest (1) (2) (3) IPO 0:609 0:543 0:627 (0:065) (0:049) (0:052) IPO In re IPO 0:127 0:192 0:162 (0:166) (0:119) (0:100) IPO CAFA 0:261 0:127 0:131 (0:176) (0:113) (0:098) Pre 0:179 0:159 0:207 (0:032) (0:040) (0:045) Pre IPO 0:529 0:516 0:512 (0:099) (0:108) (0:123) Pre In re IPO 0:084 0:091 0:106 (0:050) (0:056) (0:052) Pre IPO In re IPO 0:638 0:463 0:558 (0:201) (0:225) (0:228) Pre CAFA 0:040 0:024 0:018 (0:041) (0:041) (0:029) Pre IPO CAFA 0:232 0:061 0:268 (0:203) (0:199) (0:203) Post 0:064 0:093 0:035 (0:041) (0:032) (0:027) Post IPO 0:317 0:319 0:287 (0:149) (0:229) (0:225) Post In re IPO 0:016 0:019 0:021 (0:049) (0:041) (0:035) Post IPO In re IPO 0:116 0:058 0:029 (0:193) (0:191) (0:200) Post CAFA 0:066 0:037 0:024 (0:030) (0:036) (0:032) Post IPO CAFA 0:419 0:265 0:355 (0:206) (0:170) (0:163) Firm FE Yes Yes No Year FE Yes No No Industry-Year FE No Yes Yes Observations 11,321 24,410 24,410 R 2 0.419 0.456 0.337 59 Table 1.9: Debt and shareholder litigation { DDD coecients using IPO rms. The dependent variable Debt, along with the independent variables Pre and Post, is dened in Ta- ble 1.5. In re IPO is an indicator variable equal to 1 for the year 2001 and all years that follow. CAFA is an indicator variable equal to 1 for the year 2005 and all years that follow. IPO equals 1 for the rst two observations following each rm's date of going public and 0 otherwise. The xed eects for each regression are noted in the table. Industry-year xed eects use 4-digit SIC codes. The estimation period is 1993-2014. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Debt Debt Debt (1) (2) (3) IPO 0:021 0:012 0:012 (0:023) (0:013) (0:014) IPO In re IPO 0:053 0:035 0:031 (0:057) (0:028) (0:034) IPO CAFA 0:023 0:016 0:034 (0:059) (0:033) (0:039) Pre 0:037 0:037 0:053 (0:015) (0:017) (0:018) Pre IPO 0:015 0:007 0:068 (0:035) (0:030) (0:034) Pre In re IPO 0:017 0:014 0:015 (0:018) (0:019) (0:028) Pre IPO In re IPO 0:019 0:024 0:057 (0:074) (0:055) (0:071) Pre CAFA 0:017 0:025 0:024 (0:015) (0:016) (0:026) Pre IPO CAFA 0:112 0:067 0:005 (0:073) (0:058) (0:075) Post 0:039 0:024 0:020 (0:017) (0:019) (0:017) Post IPO 0:104 0:121 0:026 (0:088) (0:100) (0:056) Post In re IPO 0:016 0:005 0:002 (0:018) (0:019) (0:021) Post IPO In re IPO 0:128 0:156 0:018 (0:106) (0:107) (0:069) Post CAFA 0:008 0:004 0:022 (0:012) (0:012) (0:016) Post IPO CAFA 0:019 0:045 0:079 (0:067) (0:049) (0:061) Firm FE Yes Yes No Year FE Yes No No Industry-Year FE No Yes Yes Observations 11,321 24,410 24,410 R 2 0.543 0.628 0.321 60 Table 1.10: Shareholder litigation and nancial policies of IPO rms { DD coecients using rms alleged to be involved in the IPO allocation scheme. The nancial policies of rms are measured by three dependent variables: Cash, Invest, and Debt. The three dependent variables, along with the independent variables Pre and Post, are dened in Table 1.5. In re IPO is an indicator variable equal to 1 for the year 2001 and all years that follow. IPO Allocation equals 1 for rms alleged to be involved in the IPO allocation scheme that led to In re IPO and 0 otherwise. The xed eects for each regression are noted in the table. Industry-year xed eects use 4-digit SIC codes. The estimation period is 1993-2014. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Cash Cash Invest Invest Debt Debt (1) (2) (3) (4) (5) (6) IPO Allocation 1:269 1:564 0:944 0:970 0:074 0:380 (0:162) (0:119) (0:170) (0:318) (0:038) (0:195) IPO Allocation In re IPO 1:364 1:253 0:021 (0:154) (0:217) (0:038) Pre 0:213 0:383 0:004 (0:076) (0:076) (0:032) Pre IPO Allocation 0:552 0:394 0:324 (0:166) (0:264) (0:209) Post 0:414 0:116 0:093 (0:092) (0:083) (0:036) Post IPO Allocation 1:207 1:159 0:397 (0:171) (0:352) (0:214) Year FE Yes Yes Yes Yes Yes Yes Observations 2,160 2,160 2,160 2,160 2,160 2,160 R 2 0.148 0.153 0.124 0.137 0.021 0.026 61 Table 1.11: Shareholder litigation and nancial policies of IPO rms { DD coecients using litigious industries. The nancial policies of rms are measured by three dependent variables: Cash, Invest, and Debt. The three dependent variables, along with the independent variables Pre and Post, are dened in Table 1.5. In re IPO is an indicator variable equal to 1 for the year 2001 and all years that follow. Lit Ind equals 1 for rms in litigious industries and 0 otherwise. Litigious industries comprise the uppermost set of 4-digit SIC industries, which contains half of all lings in the matched sample, after arranging all 4-digit SIC industries in the full sample from highest to lowest based on one, the total number of lings within each industry and two, the fraction of rms that are sued within each industry. The xed eects for each regression are noted in the table. Industry-year xed eects use 4-digit SIC codes. The estimation period is 1993-2014. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Cash Cash Invest Invest Debt Debt (1) (2) (3) (4) (5) (6) Lit Ind 0:876 0:748 0:377 0:279 0:189 0:170 (0:141) (0:116) (0:097) (0:073) (0:037) (0:030) Lit Ind In re IPO 0:395 0:147 0:082 (0:138) (0:115) (0:038) Pre 0:451 0:415 0:025 (0:129) (0:104) (0:038) Pre Lit Ind 0:112 0:238 0:018 (0:139) (0:142) (0:047) Post 0:002 0:098 0:121 (0:106) (0:100) (0:048) Post Lit Ind 0:520 0:173 0:074 (0:132) (0:119) (0:052) Year FE Yes Yes Yes Yes Yes Yes Observations 2,160 2,160 2,160 2,160 2,160 2,160 R 2 0.175 0.198 0.106 0.146 0.072 0.072 62 Table 1.12: Shareholder litigation and nancial policies of IPO rms { DD coecients using stock volatility. The nancial policies of rms are measured by three dependent variables: Cash, Invest, and Debt. The three dependent variables, along with the independent variables Pre and Post, are dened in Table 1.5. In re IPO is an indicator variable equal to 1 for the year 2001 and all years that follow. Volatility is an indicator variable equal to 1 for IPO rms with stock volatility exceeding the median stock volatility of all IPO rms, where stock volatility is measured over the initial 252 days of trading. I exclude IPO rms that undergo a lawsuit ling within the rst 252 days of trading or within one month after the 252nd day of trading. The xed eects for each regression are noted in the table. Industry-year xed eects use 4-digit SIC codes. The estimation period is 1993-2014. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Cash Cash Invest Invest Debt Debt (1) (2) (3) (4) (5) (6) Volatility 0:973 0:735 0:529 0:441 0:124 0:117 (0:085) (0:063) (0:075) (0:067) (0:029) (0:027) Volatility In re IPO 0:735 0:213 0:048 (0:134) (0:132) (0:042) Pre 0:329 0:514 0:028 (0:089) (0:103) (0:043) Pre Volatility 0:194 0:047 0:027 (0:169) (0:146) (0:046) Post 0:095 0:025 0:157 (0:103) (0:079) (0:043) Post Volatility 0:296 0:354 0:129 (0:125) (0:154) (0:062) Year FE Yes Yes Yes Yes Yes Yes Observations 2,064 2,064 2,064 2,064 2,064 2,064 R 2 0.169 0.177 0.116 0.149 0.036 0.038 63 Table 1.13: Shareholder litigation and nancial policies of IPO rms { DD coecients using stock turnover. The nancial policies of rms are measured by three dependent variables: Cash, Invest, and Debt. The three dependent variables, along with the independent variables Pre and Post, are dened in Table 1.5. In re IPO is an indicator variable equal to 1 for the year 2001 and all years that follow. Turnover is an indicator variable equal to 1 for IPO rms with stock turnover exceeding the median stock turnover of all IPO rms, where stock turnover is measured over the initial 252 days of trading. I exclude IPO rms that undergo a lawsuit ling within the rst 252 days of trading or within one month after the 252nd day of trading. The xed eects for each regression are noted in the table. Industry-year xed eects use 4-digit SIC codes. The estimation period is 1993-2014. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Cash Cash Invest Invest Debt Debt (1) (2) (3) (4) (5) (6) Turnover 0:679 0:480 0:565 0:350 0:048 0:048 (0:134) (0:104) (0:070) (0:051) (0:030) (0:027) Turnover In re IPO 0:474 0:396 0:030 (0:180) (0:091) (0:044) Pre 0:329 0:380 0:013 (0:131) (0:107) (0:054) Pre Turnover 0:062 0:138 0:029 (0:188) (0:172) (0:064) Post 0:014 0:112 0:056 (0:088) (0:182) (0:081) Post Turnover 0:424 0:191 0:022 (0:129) (0:139) (0:106) Year FE Yes Yes Yes Yes Yes Yes Observations 2,064 2,064 2,064 2,064 2,064 2,064 R 2 0.140 0.148 0.124 0.145 0.023 0.024 64 Table 1.14: IPO proceeds of rms involved in the IPO allocation scheme. The de- pendent variables are (1) IPO proceeds including the overallotment option for rm i divided by total assets for rm i in its rst scal year in the sample, and (2) IPO proceeds including the overallotment option for rmi divided by market capitalization for rm i in its rst scal year in the sample. The independent variable IPO Allocation equals 1 for IPOs alleged to be involved in the IPO allocation scheme that led to In re IPO. The xed eects for each regression are noted in the table. The estimation period is 1993-2014. Data on IPO proceeds are obtained from the Securities Data Company (SDC). , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Proceeds Assets Proceeds Assets Proceeds Mrkt. Cap. Proceeds Mrkt. Cap. (1) (2) (3) (4) IPO Allocation 0:017 0:032 0:284 0:225 (0:039) (0:056) (0:061) (0:085) Year FE No Yes No Yes Observations 715 715 715 715 65 Table 1.15: Financial policies and shareholder litigation in the period after 2004. The dependent variables Cash and Invest, along with the independent variables Pre and Post, are dened in Table 1.5. IPO equals 1 for the rst two observations following each rm's date of going public and 0 otherwise. The xed eects for each regression are noted in the table. Industry-year xed eects use 4-digit SIC codes. The estimation period is 2005-2014. Data are described in Section 1.2. Robust (clustered by industry) standard errors are in parentheses. , , indicate statistical signicance at the 1%, 5%, and 10% test levels, respectively. Dependent variable Cash Cash Invest Invest (1) (2) (3) (4) IPO 0:328 0:222 (0:066) (0:063) Pre 0:037 0:005 0:081 0:055 (0:028) (0:017) (0:028) (0:028) Pre IPO 0:107 0:110 (0:137) (0:129) Post 0:070 0:040 0:039 0:027 (0:014) (0:012) (0:028) (0:027) Post IPO 0:344 0:003 (0:095) (0:115) Firm FE Yes Yes Yes Yes Industry-Year FE Yes Yes Yes Yes Observations 11,068 11,068 11,068 11,068 R 2 0.617 0.634 0.439 0.446 66 Chapter 2 Theoretical Framework This chapter develops a model of shareholder litigation as a mechanism for capital market dis- cipline. Litigation disciplines both entrepreneurs and shareholders: entrepreneurs obtain a safe- guard against discontent shareholders; shareholders obtain a safeguard against unscrupulous en- trepreneurs. A rm's nancial policies can bear witness to the eects of litigation. On particular occasions, litigation determines a rm's nancial policies in light of alternative mechanisms by which a rm sets such policies. The alternative mechanisms include bankruptcy costs, agency con icts, or product market interactions, among others. Shareholder litigation impacts a rms nancial policies if and only if the expected costs of litigation are high. The expected costs of litigation are high when litigation insurance is limited or expensive. Furthermore, given a level of litigation insurance, the expected costs of litigation are increasing in the level of asymmetric information between an entrepreneur and shareholders. Con- sequently, the model is particularly useful at describing the actions of a rm undergoing an early stage nancing or an initial public oering. These young rms often have underdeveloped legal divisions and low insurance coverage. Also, they possess high levels of asymmetric information. I demonstrate the eect of shareholder litigation on three nancial policies: cash, investment, and debt. Litigation risk can increase the three nancial policies prior to a lawsuit ling. The nuances of the increase can be understood in the context of the following parable. Suppose Jen 67 just completed culinary school and has aspirations to start a company called Bread Co. Jen obtains nancing for Bread Co from a group of investors. In her prospectus to investors, Jen promises an update of Bread Co's progress in the rst annual report and revenues in the second annual report. In the rst annual report, progress will be either good or bad; in the second annual report, revenues will be either high or low. Assume that the nancial benet or the control rent of managing Bread Co accrues to Jen only after publishing the second annual report: that is, after the resolution of uncertainty. If the rst annual report shows Jen's progress to be bad, her investors may lose condence and sue to mitigate losses. For instance, the investors may claim that the prospectus misrepresented Jen's abilities. If Jen cannot insure against the potential lawsuit ling, then she saves in precaution of the lawsuit ling. By saving she hopes to reach a quick settlement with her shareholders. If a quick settlement occurs, Jen continues managing Bread Co and does not lose her job. At the same time, Jen's investors may give her a second chance and not sue despite the bad progress in the rst annual report. By not suing, Jen's investors hope she rebounds and announces high revenues in the second annual report. But if revenues turn out low in the second annual report, then Jen's investors certainly sue. In order to achieve high revenues, Jen invests all of her remaining cash into Bread Co. So on average, Bread Co exhibits an increase in cash and investment prior to a lawsuit ling. Now suppose Jen has a competitor named Jim who starts the company Pastry Co. Jim also completed culinary school, but his true aspiration lies in owning ne art. Jim uses the nancing for Pastry Co to purchase ne art, which is displayed in Pastry Co's storefront. For instance, Jim buys lower quality our, eggs, or ovens in order to fund his penchant for ne art. Jim's prospectus to investors closely resembles Jen's prospectus, and it contains no mention of ne art. Jen and Jim have the same outcomes. In the rst annual report, each announces bad progress, and each saves in precaution of a lawsuit ling. By saving each hopes to reach a quick settlement with her or his shareholders and to keep her or his job. Suppose neither is sued after the rst 68 annual report: investors hope for a rebound and high revenues in the second annual report. But alas, each announces low revenues in the second annual report and is sued. The court exculpates Jen from blame, citing her low revenue to be a consequence of poor luck: Jen does not pay damages, because she did nothing wrong. The court nds Jim guilty of wrongdoing, citing his low revenue to be a consequence of misconduct: Jim pays damages, because he jeopardized Pastry Co for the personal benet of consuming ne art. The more Jim spends on ne art, the greater the likelihood of being found guilty. Before proceeding, consider the scenario where both announce high revenues in the second annual report. Investors are content and do not sue. Jen and Jim both obtain the satisfaction of successfully managing a business. Consider the satisfaction to be a control rent: a nonpledgeable benet that accrues to an entrepreneur. Jim, however, obtains an additional private benet: the added satisfaction of consuming ne art. The above parable encapsulates my model of shareholder litigation. The model explains both the relevance of litigation to the nancing process and its in uence on a rm's nancial policies. An entrepreneur chooses nancial policies in accordance with his or her expected costs of litigation (e.g. potential loss of employment or wealth as in Karpo et al. (2008a)). When an entrepreneur's expected costs of litigation are high, she or he saves cash in precaution of a premature lawsuit ling. By saving, an entrepreneur hopes to reach a quick settlement with shareholders and to keep her or his job. Sometimes an entrepreneur saves cash in precaution of a premature lawsuit ling, but shareholders do not sue: they hope that future performance exceeds the benets of a premature lawsuit ling. In turn, an entrepreneur increases investment to achieve high future performance. But if future performance wanes notwithstanding, then shareholders certainly sue. They have more to gain from ling a lawsuit than from doing nothing. For simplicity, the above parable contains no mention of debt. When sucient cash is avail- able, it alone acts as a precautionary measure against a premature lawsuit ling. But when cash is 69 limited, debt acts as an alternative precautionary measure: debt immediately reduces sharehold- ers' expected value should a premature lawsuit ling occur. So with debt, shareholders never le a lawsuit prematurely. Instead, they wait for the uncertainty in a rm's outcomes to be resolved. While debt does halt the ling of a premature lawsuit, it also exerts an additional cost on an entrepreneur. The additional cost arises during nancial distress. When debt matures and an entrepreneur cannot pay face value, a lender seizes assets. When a lender seizes assets, an en- trepreneur's control rent evaporates. So debt expands the set of states in which an entrepreneur obtains no control rent. As a result, debt distorts the actions of an entrepreneur who misappro- priates funds { debt pushes a Jim to consume an even greater quantity of ne art. In turn, courts are more likely to identify a Jim. When Jen's cash is limited, she too resorts to debt for more nancing. But because her outcomes are purely exogenous, she simply bears the costs of debt without any change to her actions. To summarize, I emphasize four takeaways, which draw connections between shareholder liti- gation and corporate nance. First, shareholder litigation brings stability to the nancing process. Without litigation, an entrepreneur is reluctant to begin a project, because investors may have an incentive to replace her or him prematurely. So without litigation, innovation is sti ed. Second, litigation improves funding availability by providing a mechanism for capital market discipline. Third, litigation can impact an entrepreneur's choice of cash and investment. On average, cash and investment increase before a lawsuit ling. Fourth, when cash is limited, debt acts as an alternative precautionary measure against a premature lawsuit ling. But in doing so, debt dis- torts the actions of an entrepreneur who commits wrongdoing, thus enabling courts to more easily discern the wrongdoing. More generally, this chapter relates to the literature on the implications of corporate cash holdings. Keynes (1936) suggests that a rm has two motives for holding cash: a transactions and a precautionary motive. The transactions motive suggests that holding cash allows a rm to avoid the costs of raising funds or liquidating assets in order to make payments. The precautionary 70 motive suggests that holding cash allows a rm to seize investment opportunities even when other sources of funding are unavailable or costly. Opler et al. (1999) examine the transactions and precautionary motives to nd evidence supporting a tradeo between the benets and costs of holding cash. The authors nd that rms with lower access to capital markets, better investment opportunities, and riskier cash ows have higher ratios of cash to non-cash assets. Bates et al. (2009) extend the sample period of Opler et al. (1999) to nd added support for the precautionary motive. The authors show that rms increase cash as their cash ows become more volatile. Furthermore, rms that increase cash have lower non-cash components, such as inventories and receivables, that can be converted to cash quickly. These rms also have increased R&D expenditures. Finally, the authors document that IPO rms have greater average cash ratios compared to non-IPO rms. In part, the greater average cash ratios of IPO rms can be attributed to capital raising activities. But as discussed in Chapter 1, IPO rms with high litigation risk have greater average cash ratios compared to IPO rms with low litigation risk. Furthermore, Chapter 1 shows that the relation between cash holdings and litigation risk occurs only among IPO rms. This result extends the ndings of Arena and Julio (2015), who suggest a relation between cash holdings and litigation risk to exist among all rms. All together, the results indicate that litigation risk can play a role in the precautionary motives of IPO rms and can act as a determinant of their cash holdings. Other papers provide theoretical models to explain the transactions and precautionary motives of holding cash. Kim et al. (1998) propose that the existence of capital market imperfections establishes a rationale for holding excess cash. When external nancing is costly, investment in liquid assets is an optimal response for funding future investment opportunities. The authors nd the optimal level of cash holdings to directly relate to the costs of external nancing and the protability of expected investment opportunities. Froot and Stein (1998) provide a model in which nancial institutions face risks that cannot be easily hedged. To help account for these risks, nancial institutions hold more capital and alter investment policies. While their model is 71 specic to nancial institutions, one can see the model apply generally to rms as well. In the context of this chapter, if shareholder litigation cannot be easily hedged, then rms may hold more capital and alter investment policies. As far as I know, this is the rst example of a theoretical model specic to shareholder litigation within a corporate nance setting. 2.1 The Option-like Nature of Shareholder Litigation Before proceeding to the model, I emphasize an added aspect of shareholder litigation { its option-like nature. Recognizing the option-like nature of shareholder litigation may help clarify concepts. Shareholder litigation comprises three types of lawsuits: class actions, derivative suits, and individual shareholder suits. The three types resemble nancial options. Attorneys initiate the option-like contracts at lawsuit ling. Defense attorneys enable the writing of puts on behalf of defendants: the ocers or directors of a rm. Plaintis' attorneys enable the buying of puts on behalf of plaintis: the shareholders of a rm. Defense attorneys receive an immediate payment. Plaintis and their attorneys receive pay- ment only if the lawsuit is won { if the option is in the money. When courts render judgment in a lawsuit, they implicitly set a strike price for the option. If defendants win, the strike price is low: the option is out of the money and plaintis obtain no damages. If plaintis win, the strike price is high: the option is in the money and plaintis obtain damages. Plaintis' attorneys collect a fee from the damages. A security has the option baked into its price. So when plaintis (i.e. shareholders) purchase a security, their overall prot resembles that of a protective put (a.k.a. synthetic long call). As a result, shareholder litigation impacts the equity distribution of returns. For example, the observed volatility of a stock is more pronounced with the option to sue, because the ling of a lawsuit creates direct and immediate costs for the rm. Figures 2.1 and 2.2 demonstrate the payos to defendants and plaintis. 72 2.2 The Model Tables 2.1 through 2.3 summarize all notation in the chapter. The model contains four periods. In the rst period, an entrepreneur seeks equity nancing for her project. In the second period, shareholders obtain news about their entrepreneur. In the third period, cash ows are reported. In the fourth period, all uncertainty becomes resolved and consumption takes place. Shareholders sue in either the second or the third period. Everyone is risk neutral. The riskless interest rate equals zero and no unexpected changes occur in the rate. Att = 0, an entrepreneur has a project but no capital. To initiate the project, an entrepreneur needs K of funding. She reaches out to N investors. The N investors conduct due diligence and obtain information about their entrepreneur: with probabilityf she is good, and with probability 1f she is bad. Att = 0 she has the same expectations about her type; between t = 0 andt = 1 she learns her type with certainty. Bad entrepreneurs are the reason for shareholder litigation. They take actions to capture private benets (or perquisites) at the expense of shareholders. By taking actions to capture private benets, they reduce shareholders' expected return. I call such actions opportunism. In the introduction, Jim represents a bad entrepreneur; his consumption of ne art exemplies opportunism. Throughout the chapter, the words \wrongdoing" or \misbehavior" are synonymous with opportunism. To correct opportunism, shareholder litigation exists. Shareholder litigation aims to discipline entrepreneurs who engage in opportunism. Att = 1 shareholders obtain news about their entrepreneur. The arrival of news resembles the arrival of additional information in the Diamond (1991) model, in which lenders privately obtain additional information about each borrower after making the initial loan. Section 2.2.1 provides details concerning the arrival of news. At t = 2, a good entrepreneur reports cash ows of X H , while a bad entrepreneur reports cash ows of X H with probability and cash ows of X L with probability 1 . Assume 73 X H to be the exogenous cash ows that shareholders attain after suing their entrepreneur and receiving damages. So if an entrepreneur reports X H , shareholders have no need to le a lawsuit. Assume X L to be the cash ows that trigger a lawsuit ling. Of course X H >X L . A number of studies demonstrate that a lawsuit ling often follows a loss in value (Karpo et al., 2008b; Gande and Lewis, 2009). The loss in value results from some negative news: market downturns, poor performance, operational mishaps, or accounting irregularities. Later, I show how X L depends on a bad entrepreneur's opportunistic choices. For the moment, suppose that shareholder litigation does not exist. The N investors fund an entrepreneur's project when K X L + (X H X L )[ +f(1)]; rearranging implies (K X L )=(X H X L ) [ +f(1)]. So the fraction of the potential loss to the total gain of going from the low to the high state needs to be below the probability of gain. An entrepreneur, who develops and manages a project, requires compensation. An entrepreneur agrees to manage a project in return for a control rent C to be obtained at t = 3. A control rent is a nonpledgeable benet that accrues to an entrepreneur: it may be a part of rm value that cannot be unlocked without an entrepreneur's abilities. If shareholders replace their entrepreneur before t = 3, her control rent evaporates. For simplicity, I assume the control rent to have a xed value C, but the value can depend on an entrepreneur's type or reported cash ows. The simplication does not change the results. After funding a project, the N investors become shareholders. 2.2.1 The Arrival of News Shareholders obtain news about each entrepreneur at t = 1. The news arrives privately to share- holders. For example, shareholders learn about their entrepreneur from personal interactions, 74 press releases, or earnings announcements. The news arrives in one of two forms: positive or neg- ative news. Denote the conditional probability of a good entrepreneur given positive or negative news as f p =Pr(good entrepreneurj positive news) f n =Pr(good entrepreneurj negative news): Assume all bad entrepreneurs to provide negative news. Hence every time shareholders obtain positive news, they know their entrepreneur to be good. That is, f p =Pr(good entrepreneurj positive news) = 1: The model's tension arises because a good entrepreneur can have poor luck. Due to poor luck, a good entrepreneur provides negative news just like a bad entrepreneur. Denote e to be the probability that a good entrepreneur has poor luck and provides negative news. Bayes' Law implies that e =Pr(negative newsj good entrepreneur) = f n (1f) f(1f n ) ; (2.1) where f is an entrepreneur's initial probability of being good. I dub a good entrepreneur who has poor luck and provides negative news as \unlucky." In legal jargon, a lawsuit against an unlucky entrepreneur is frivolous and lacks merit. In the introduction, Jen represents an unlucky entrepreneur when she announces poor progress in her rst annual report. In summary, three types of entrepreneurs exist at t = 1: good, unlucky, and bad. The probabilities of each type are exogenously provided. Without assistance, shareholders cannot distinguish between an unlucky and a bad entrepreneur. Figure 2.3 depicts the model's setup, which includes the arrival of news at t = 1 and excludes the possibility of shareholder litigation. 75 2.2.2 Shareholder Litigation For the moment, continue to assume that shareholder litigation does not exist. At t = 1 share- holders have a choice to make: retain or replace their entrepreneur. Replacing their entrepreneur at t = 1 yields shareholders a net present value of L. Of course L<X H . Given positive news, shareholders' expected value at t = 1 is E p 1 = X H . Positive news guarantees an entrepreneur to be good, and a good entrepreneur always reports cash ows X H at t = 2. So shareholders never replace a good entrepreneur. Given negative news, shareholders' expected value att = 1 isE n 1 = maxfL;X H +(1)X L g. Negative news implies an entrepreneur to be either bad or unlucky. At t = 1 the project of a bad or an unlucky entrepreneur provides shareholders with an expected value of X H + (1)X L . If L>X H + (1)X L , shareholders replace their entrepreneur and their value becomes E n 1 =L. If L < X H + (1)X L , shareholders retain their entrepreneur and their value becomes E n 1 = X H + (1)X L . SupposeL>X H +(1)X L , so shareholders choose to replace their entrepreneur. The choice may be inecient, because shareholders do not heed the control rent lost by their entrepreneur. Total value includes an entrepreneur's control rent. So at t = 1, expected total value is X H + (1)X L +C. IfC >LX H (1)X L , then the choice to replace an entrepreneur becomes inecient. Now I introduce shareholder litigation. The opportunity to sue may correct the ineciency. Suppose courts help identify a bad entrepreneur and transfer wealth back to shareholders. When shareholders cannot distinguish between a bad and an unlucky entrepreneur, courts nd a bad entrepreneur guilty with probability s and not guilty (i.e. dismiss a lawsuit) with probability 1s. If shareholders could distinguish between a bad and an unlucky entrepreneur, courts would nd a bad entrepreneur guilty with probability one. 76 Assume courts never nd an unlucky entrepreneur guilty. In reality, however, courts may nd some unlucky entrepreneurs guilty. So the chance of nding an unlucky entrepreneur guilty would not equal zero, but would tend to zero. The chance would tend to zero, because someone who happens to be unlucky and does nothing wrong should be found guilty with low probability. When courts nd an entrepreneur guilty, shareholders obtain damages in the amount of X H X L . Recall,X H are cash ows that shareholders attain after suing their entrepreneur and receiving damages. To le a lawsuit, shareholders must payB, which represents the cost of a lawsuit ling. So given negative news and the opportunity to sue, shareholders' expected value is X H + (1 )X L +(1)[hs(X H X L )B]. h denotes the portion of entrepreneurs who are perceived to be bad after negative news arrives: h = (1f)=[(1f) +fe]. If (1)[hs (X H X L )B] LX H (1)X L , then shareholders retain their entrepreneur. Denition 1 Shareholder litigation implies a cost (e.g. B) of receiving damages (e.g. X H X L ) with some probability (e.g. hs), such that shareholders' expected value of retaining their en- trepreneur is greater than or equal to shareholders' expected value of replacing their entrepreneur. The denition of shareholder litigation implies a viable opportunity to sue. The denition relates to the discussion of legal enforcement in Diamond (2004). Shareholder litigation is a means of providing legal enforcement: entrepreneurs obtain a safeguard against discontent shareholders, and shareholders obtain a safeguard against unscrupulous entrepreneurs. Figure 2.4 depicts the model's setup with the introduction of shareholder litigation. According to Denition 1, shareholder litigation gives shareholders a real option with a non- negative expected value after accounting for opportunity costs. Given the existence of shareholder litigation, Lemma 1 characterizes the condition for ecient investment. Lemma 1 WhenC (1)(B+d), shareholder litigation eliminates the ineciency of replacing an entrepreneur at t = 1. 77 d represents an entrepreneur's unavoidable cost of undergoing a lawsuit ling: for example, the value of time spent dealing with attorneys. For simplicityB andd are xed, exogenous parameters. Details behind Lemma 1 can be found in the Appendix. 2.3 Opportunism Opportunism allows a bad entrepreneur to capture private benets or perquisites in the high state, but reduces shareholders' cash ows in the low state. Shareholders observe neither the opportunism nor the private benets of a bad entrepreneur. Courts discern opportunism with probability s. K is the total funding at t = 0. Suppose that a bad entrepreneur chooses to spend w oppor- tunistically, leaving Kw to be spent on acceptable operating activity I. Opportunism yields X(w) in the high state. Assume the function X(w) to be dierentiable, concave, and increasing in w; also X(w) > X H . A bad entrepreneur reports the exogenous X H with probability and consumes X(w)X H as private benets. Note, opportunism diers from \free cash ow" as dened in Jensen (1986). Free cash ow entails investment that generates a negative net present value. Opportunism, as dened here, entails investment that generates a positive net present value in excess of reported, but does not because a bad entrepreneur hides its potential. If opportunism pays, then why don't all entrepreneurs engage in it? First, to engage in opportunism requires a certain type of personality. Second, it requires a special set of skills. By analogy, consider star athletes who use performance enhancing drugs. In expectation the use of drugs should be a positive net present value investment. However, not every athlete uses drugs. Athletes who use drugs have unique personalities and skills. For instance, they may have innate characteristics that help mask the drug use; they may form networks to acquire the least detectable drugs; or they may know methods to avoid detection. 78 Suppose a bad entrepreneur reports the low state cash ows X L (w) with probability 1. Assume the function X L (w) to be decreasing in w. So shareholders' expected value falls as a result of opportunism. Assume an unlucky entrepreneur to also report X L (w) with probability 1. Of course, an unlucky entrepreneur does not engage in opportunism, she merely has poor luck. But as a result of poor luck, an unlucky entrepreneur reports cash ows in the low state that appear as if she engaged in opportunism. Lastly, suppose s, the probability that a court nds a bad entrepreneur guilty of engaging in opportunsim, to be a function of w. Assume s(w) to be dierentiable, convex, and increasing on the interval (0;w max ], where s(w) 0 and s(w max ) = 1. In summary, a bad entrepreneur faces the following problem: Problem 1 Find a level of opportunism w (or alternatively a level of investment I) that maxi- mizes [C + (X(w)X H )] + (1) [Cds(w) (X H X L (w))] (2.2) subject to KwI (2.3) Recall, every entrepreneur who undergoes a lawsuit ling bears the unavoidable cost d. The solution w implies an optimal level of opportunism that satises X 0 (w ) = (1) [s 0 (w ) (X H X L (w ))s(w )X 0 L (w )]: (2.4) While not shown, the following condition holds: the greater a bad entrepreneur's expected costs of a lawsuit ling, the lesser the opportunism. A bad entrepreneur's expected costs of a lawsuit ling are a function of the unavoidable costd, the probability of being found guiltys(w), and the damages after being found guilty [X H X L (w)]. 79 As discussed in Diamond (1984), contracts may involve pecuniary as well as nonpecuniary penalties. Nonpecuniary penalties entail an entrepreneur's loss that provides no benet to share- holders. The inclusion of nonpecuniary penalties implies that the above problem can include negative values. For example, [Cds(w)(X H X L (w))] may be negative. A more elaborate framework would separately dene pecuniary and nonpecuniary penalties. 2.4 Cash and Investment in Relation to Negative News As discussed, shareholders obtain news about each entrepreneur at t = 1. A portion (1e)f of entrepreneurs provide positive news; the remaining portion, (1f) +fe, provide negative news. An entrepreneur who provides positive news always reports cash ows of X H , so shareholders neither replace her nor le a lawsuit against her. Shareholders may replace or le a lawsuit against an entrepreneur who provides negative news. So an entrepreneur who provides negative news saves cash as a precautionary measure. Recall thatL>X H +(1)X L according to Denition 1. An unlucky entrepreneur provides shareholders with an expected value of X H + (1)X L at t = 1. Shareholders, however, have trouble identifying an unlucky entrepreneur. But suppose an unlucky entrepreneur, at risk of being replaced at t = 1 and of losing her entire control rent, can benet from saving cash at t = 1. Specically, an unlucky entrepreneur saves cash at t = 1 to negotiate a quick settlement with her shareholders. If an unlucky entrepreneur and her shareholders reach a quick settlement att = 1, then she retains her job and continues to manage the project. Thereafter, no cash ows are reported at t = 2. So shareholders obtain the quick settlement at t = 1 and nothing more at t = 2. Suppose an unlucky entrepreneur to face the following problem at t = 1: Problem 2 Find a quick settlement Q that maximizes Q (2.5) 80 subject to db(Q)<d + [X(w max )X H ] (1) [X H X L (w max )] (2.6) 0<QX H + (1)X L (w ) + (1) [hs (X H X L (w ))B] (2.7) Q +B <K: (2.8) b(Q) is an entrepreneur's benet from a quick settlement at t = 1. Constraint 2.6 asserts that an unlucky entrepreneur, who negotiates a quick settlement at t = 1, must obtain at least her t = 1 expected value of the control rent. Constraint 2.6 also guarantees that the threshold, for a bad entrepreneur to mimic an unlucky entrepreneur in her choice of cash at t = 1, lies in the interval [0; 1]. While the threshold need not lie in the interval [0; 1], it is more intuitive for it to do so. The threshold is discussed in the proposition below. Constraint 2.7 asserts that the quick settlement cannot provide shareholders with less than zero or more than their expected value at t = 1. Constraint 2.8 asserts that an entrepreneur who provides negative news must be able to save enough cash to pay shareholders' cost of a lawsuit ling at t = 1. Assume that a bad entrepreneur, who undergoes a lawsuit ling at t = 1 and negotiates a quick settlement, obtains no benets of opportunism. For simplicity suppose that an entrepreneur, who does not expect to negotiate a quick settle- ment with shareholders at t = 1, invests half of the total funding K between t = 0 and t = 1, and the remaining half between t = 1 and t = 2. So at t = 1, such an entrepreneur has K=2 left in cash. The assumption establishes a baseline; it provides no additional insight nor does it alter the implications of the model. For example, shareholders never negotiate a quick settlement with a good entrepreneur, so she invests K=2 between t = 0 and t = 1, and K=2 between t = 1 and t = 2. She holds K=2 in cash at t = 1. When conditions are such that shareholders never negotiate a quick settlement with either an unlucky or a bad entrepreneur, then that entrepreneur also investsK=2 betweent = 0 andt = 1, andK=2 betweent = 1 andt = 2. She or he holdsK=2 81 in cash att = 1. Here, unlucky and bad entrepreneurs simply replicate the cash out ows of good entrepreneurs. But obviously good entrepreneurs put the cash out ows to better use: this is why good entrepreneurs always report high cash ows X H at t = 2. Given Problem 2, the following proposition characterizes the equilibria that may arise at t = 1: Proposition 1 The following equilibria may arise at t = 1: 1. If Problem 2 has no feasible solution Q , then a pooling equilibrium arises in which an unlucky and a bad entrepreneur both hold K=2 in cash at t = 1. A bad entrepreneur chooses a level of opportunism w from Equation 2.4. 2. If Problem 2 has a feasible solution Q and Constraint 2.6 binds rst, then a pooling equi- librium arises in which an unlucky and a bad entrepreneur both hold K=2 in cash at t = 1. A bad entrepreneur chooses a level of opportunism w from Equation 2.4. 3. If Problem 2 has a feasible solution Q and Constraint 2.7 binds rst, then if [b(Q )[(X(w )X H d)s(1)(X H X L (w ))]] [X(w max )X(w )](1)[(X H X L (w max ))s(X H X L (w ))] p 1, a pooling equilibrium arises in which an unlucky and a bad entrepreneur both hold Q +B = X H + (1 )X L (w )+(1) [hs (X H X L (w ))B]+B in cash att = 1. A bad entrepreneur chooses a level of opportunism w from Equation 2.4. if 0 p < [b(Q )[(X(w )X H d)s(1)(X H X L (w ))]] [X(w max )X(w )](1)[(X H X L (w max ))s(X H X L (w ))] , a separating equilib- rium arises in which an unlucky entrepreneur holds Q +B =X H + (1)X L (w ) + (1) [hs (X H X L (w ))B] +B, and a bad entrepreneur holds K=2 in cash at t = 1. A bad entrepreneur chooses a level of opportunism w max . p denotes a bad entrepreneur's belief about the probability that shareholders le a lawsuit at t = 1. Details behind Proposition 1 are in the Appendix. 82 Item 1 in Proposition 1 asserts that when no feasible solution exists to Problem 2, unlucky and bad entrepreneurs simply follow the baseline: they replicate the cash out ows of good en- trepreneurs and invest K=2 between t = 0 and t = 1, and K=2 between t = 1 and t = 2. So at t = 1 they have a cash level of K=2. Note that a feasible solution does not exist when an unlucky entrepreneur's benet functionb(Q) is such that Constraint 2.6 does not hold. For instance, in the presence of shareholder litigation insurance, b(Q) is likely to be low or zero. Shareholder litiga- tion insurance obviates the need to save cash in precaution of a lawsuit ling. When shareholder litigation insurance is unavailable or does not suciently cover the risk of shareholder litigation, then b(Q) becomes signicant and leads to a feasible solution to Problem 2. Item 2 in Proposition 1 asserts that when a feasible solution exists to Problem 2 and the solution is such that Constraint 2.6 binds rst, unlucky and bad entrepreneurs again follow the baseline: they replicate the cash out ows of good entrepreneurs and invest K=2 between t = 0 andt = 1, andK=2 betweent = 1 andt = 2. So at t = 1 they have a cash level of K=2. Unlucky and bad entrepreneurs follow the baseline, because they have no incentive to save cash for a quick settlement. Specically, because Constraint 2.7 does not bind, shareholders cannot benet from a quick settlement. Item 3 in Proposition 1 asserts that when a feasible solution exists to Problem 2 and the solution is such that Constraint 2.7 binds rst, one of two equilibria arises. The rst is a pooling equilibrium in which unlucky and bad entrepreneurs both hold Q +B in cash at t = 1. If a lawsuit ling occurs at t = 1, B is used to pay for the ling; thereafter, shareholders obtain Q . An unlucky entrepreneur directly benets from the quick settlement, because Constraint 2.6 does not bind. The probabilityp that shareholders le a lawsuit att = 1 dictates whether or not a bad entrepreneur can benet from mimicking an unlucky entrepreneur. Under a pooling equilibrium, a bad entrepreneur benets from mimicking an unlucky entrepreneur. So shareholders cannot distinguish between an unlucky and a bad entrepreneur. 83 The second is a separating equilibrium in which unlucky entrepreneurs hold Q +B in cash at t = 1, while bad entrepreneurs follow the baseline and hold K=2 in cash at t = 1. That is, bad entrepreneurs invest K=2 between t = 0 and t = 1, and K=2 between t = 1 and t = 2. Separation occurs because p is such that a bad entrepreneur cannot benet from mimicking an unlucky entrepreneur. When a bad entrepreneur separates, shareholders can distinguish between an unlucky and a bad entrepreneur. In turn, courts nd a bad entrepreneur guilty with probability one. A bad entrepreneur, who is always found guilty in the low state, pushes his opportunistic choices to the corner solution w max . Throughout the chapter, I assume shareholders and entrepreneurs to be risk-neutral. Had I assumed shareholders or entrepreneurs to be risk-averse, the assumption would accentuate saving att = 1. Shareholders would have more incentive to replace an unlucky entrepreneur at t = 1; an unlucky entrepreneur would have more incentive to retain cash at t = 1 in order to preclude the replacement. Also, I assume L to be a xed value for all entrepreneurs. L, however, may come from some distribution that places a non-zero probability on the value X H + (1)X L (w ) + (1)[h s (X H X L (w ))B]. Each entrepreneur would know the distribution of L, but not the exact value of L that applies to him or her. An entrepreneur's savings at t = 1 could depend on the distribution of L. 2.5 Debt Thus far, shareholders oer each entrepreneur full funding K att = 0. At t = 1, an entrepreneur who provides negative news is at risk of a lawsuit ling. As a result, an entrepreneur who provides negative news at t = 1 may save cash as a precautionary measure against a premature lawsuit ling. When she obtains full fundingK, debt nancing need not be used as a precautionary mea- sure. Cash satises the entire precautionary need. Besides, debt nancing carries the additional 84 risk of nancial distress; so an entrepreneur who provides negative news at t = 1 resorts to the additional risk only if cash is limited. In this section, I demonstrate debt to be an alternative to cash as a precautionary measure against a premature lawsuit ling. I proceed in a manner similar to Townsend (1979) and Gale and Hellwig (1985). Suppose there exist two regions: a region V where assets are seized, and a complement regionV 0 where assets are not seized. Assets are seized when an entrepreneur reports cash ows below a specic threshold. At or above the threshold, assets are not seized. A standard debt security stipulates payments that establish regions V and V 0 . Debt gives a lender the right to seize assets when a borrower does not pay the threshold level of cash ows. Shareholders do not have the right to seize assets. When a lender seizes assets, nothing is left for an entrepreneur to manage. So the control rent evaporates. By expanding the set of states in which an entrepreneur obtains no control rent, debt imposes an additional cost on an entrepreneur who provides negative news at t = 1. The additional cost distorts the level of opportunistic behavior. The distortion resembles that of overinvestment under asset substitution (Jensen and Meckling, 1976). By distorting the level of opportunistic behavior, debt helps courts identify a bad entrepreneur. That is, the probability of nding a bad entrepreneur guilty of wrongdoing with debt exceeds the probability of nding a bad entrepreneur guilty of wrongdoing without debt. When cash is limited, each entrepreneur must resort to debt nancing in order to obtain the required funding K. Thereafter, if a lender seizes assets, an entrepreneur loses her or his control rent. So debt imparts an additional cost on an entrepreneur: it increases the likelihood of losing the control rent. A good entrepreneur is unaected by the additional cost, because she always reports high cash ows X H and obtains the full control rent C. An unlucky entrepreneur has no control over outcomes and must accept the additional cost: she obtains the control rent C with probability and loses everything with probability 1. A bad entrepreneur accounts for the 85 additional cost by adjusting the level of opportunistic behavior. So with debt, the problem of a bad entrepreneur becomes Problem 3 Find a level of opportunism w (or alternatively a level of investment I) that maxi- mizes [C + (X(w)X H )] + (1) [s(w) (Cd (X H X L (w)))] (2.9) subject to KwI: (2.10) The solution w implies an optimal level of opportunism that satises X 0 (w ) =(1) [s 0 (w ) [Cd (X H X L (w ))] +s(w )X 0 L (w )]: (2.11) The following lemma compares solution w to solution w in Equation 2.4. Lemma 2 Comparing Equations 2.4 and 2.11, the levels of opportunistic behavior are such that w >w . Debt increases the level of opportunistic behavior. Details behind Lemma 2 can be found in the Appendix. Lemma 2 demonstrates that the use of debt can increase opportunism. An increase in oppor- tunism allows a court to more easily detect the opportunism. When a court detects opportunism, shareholders and the lender obtain a benet in the form of damages. Shareholders force an entrepreneur to resort to debt nancing by withholding funding att = 0. An entrepreneur seeks the remainder of funding from a lender at t = 1. Every lender expects zero prots. Also, because no lender wastes resources at t = 1 to distinguish among the three entrepreneur types, all three types obtain the same debt contract. Debt must be risky; otherwise its use is trivial. As a result, a bad entrepreneur must not be able to save some amount of cash g in order to deceive a lender about the true region at t = 2. 86 That is, if the true region isV in which a lender seizes assets, a bad entrepreneur cannot saveg to misrepresent the region asV 0 in which a lender does not seize assets. The use of debt is incentive compatible if and only if either [C +X(w;g)X H ] ... + (1) [Cds(w;g)(X H X L (w;g))]C; (2.12) or (1)[1s(w )](Cd) ... + (1) [s(w )[X H X L (w )]s(w;g)[X H X L (w;g)]] ... [X(w )X(w;g)]: (2.13) Assume X(w;g) to be decreasing in g, and s(w;g) to be nondecreasing in g. Condition 2.12 stipulates that a bad entrepreneur who saves g obtains a lower expected value than an unlucky entrepreneur who does not saveg. Condition 2.13 stipulates that a bad entrepreneur who saves g obtains a lower expected value than a bad entrepreneur who does not save g. Proposition 2 A contract (F;w ) is incentive compatible if and only if 1. A lender obtains zero-prots 2. A bad entrepreneur does not misrepresent. F denotes the face value of debt. Details of Proposition 2 can be found in the Appendix. Given the use of debt, shareholders solve the following problem Problem 4 Find a constant F that maximizes (X H F ) + (1)[hs(w ) (X H BF )] (2.14) 87 subject to F + (1) [hs(w )F + [1hs(w )] (X L (w )B)] =D (2.15) F >X L (w )B > 0; in region V 0 (2.16) X L (w )B > 0; in region V (2.17) (F;w ) is incentive compatible: (2.18) Constraint 2.15 is the zero-prot condition. Constraints 2.16 and 2.17 are non-negativity con- straints. Given Lemmas 1 and 2, the following proposition characterizes the optimum for Prob- lem 4: Proposition 3 Lenders demand an F =X L (w ) +g , where g is such that Condition 2.13 of incentive compatibility binds. Details of Proposition 3 can be found in the Appendix. GivenF , a lender provides each entrepreneur withD att = 1. Att = 0 each entrepreneur knows the debt nancing terms that will arise att = 1. Consequently, a bad entrepreneur expects to choose a level of opportunistic behavior equal to w . By Proposition 2, a bad entrepreneur cannot do better than w . Shareholders provide the remainder of funding (KD ) at t = 0. (KD ) must satisfy KD f(X H F ) ... + (1f) [(X H F ) + (1)hs(w ) (X H BF )]: (2.19) The above \staged capital commitment" resembles the discussion in Bolton and Scharfstein (1990). Finally, note that debt can be an alternative to cash as a precautionary measure against a premature lawsuit ling or a premature entrepreneur replacement. Specically, the following 88 lemma characterizes the simple condition for debt to preclude a premature lawsuit ling or a premature entrepreneur replacement after the arrival of negative news at t = 1. Lemma 3 As long as F > X L (w )B (i.e. debt is risky), shareholders have no incentive to prematurely replace or le a lawsuit against an entrepreneur at t = 1, after he or she provides negative news. Details behind Lemma 3 can be found in the Appendix. 2.6 Conclusion In this chapter, I provide a model to help explain the in uence of shareholder litigation on a rm's nancial policies. Shareholder litigation provides a mechanism for capital market disci- pline. Litigation disciplines both entrepreneurs (or managers) and shareholders (or investors): entrepreneurs obtain a safeguard against discontent shareholders, and shareholders obtain a safe- guard against unscrupulous entrepreneurs. As a result, both entrepreneurs and shareholders are willing to engage in the investment process. Shareholder litigation entails high costs { both direct and indirect. If shareholder litigation insurance does not fully cover an entrepreneur's costs of litigation, then litigation may in uence the nancial policy choices of an entrepreneur. Before a potential lawsuit ling, an entrepreneur saves cash as a precautionary measure. The savings allow an entrepreneur to reach a quick settlement with shareholders and to continue managing the rm. If an entrepreneur saves cash and a ling does not occur, then she invests all of her cash in hopes of superior future performance. But if future performance wanes notwithstanding, shareholders then le a lawsuit against their entrepreneur. Cash alone can act as a precautionary measure against a potential lawsuit ling. But when cash is limited, debt acts as an alternative precautionary measure against a potential lawsuit ling. The use of debt, however, imposes an additional cost on entrepreneurs: in states of nancial 89 distress, liquidation evaporates an entrepreneur's control rent. To compensate for the additional cost, an entrepreneur who misbehaves increases the level of misbehavior so as to recuperate value in states of no nancial distress. Because debt distorts the level of misbehavior, courts can more easily identify an entrepreneur who misbehaves and nd him guilty. Shareholder litigation has two primary shortfalls. First, shareholders cannot dierentiate between an entrepreneur who misbehaves (a bad entrepreneur like Jim in the introduction) and an entrepreneur who does not misbehave but has poor luck (an unlucky entrepreneur like Jen in the introduction). As a result, shareholders oversue, ling lawsuits against both types and bearing unnecessary legal costs. Second, courts cannot detect every instance of misbehavior. Even if a court could detect every instance of misbehavior, it may not capture the full extent of the misappropriation. So shareholders may not recuperate the full losses that result from their entrepreneur's misbehavior. Shareholder litigation plays an important role when little information exists about an en- trepreneur and her project. That is, instances entailing a high risk of asymmetric information, such as an initial public oering or an early stage nancing. Shareholder litigation oers a means for equity investors to obtain an expected return in accord with the risk of asymmetric infor- mation, thus circumventing a potentially inecient replacement of an entrepreneur. Shareholder litigation also improves funding availability. Equity investors as well as lenders are more likely to nance the project of a new entrepreneur, knowing that legal channels exist to provide recompense should the entrepreneur's actions turn out to be unscrupulous. 90 2.7 Figures Figure 2.1: Payo from a hypothetical put writing strategy. Defense attorneys enable the writing of puts on behalf of defendants: the ocers or directors of a rm. S denotes equity value without a purported misrepresentation. S 0 denotes equity value after the purported mis- representation has been revealed. The misrepresentation decreases equity value. B denotes the cost of a lawsuit ling, e.g., immediate payment to defense attorneys. Courts, e.g., arbitrators or judges determine the strike price. The strike price here is SB. The supposed damages are SS 0 . If the settlement is SS 0 , then it covers the full extent of the damages. 6 - prot equity value 0 B B ( SS 0 ) (S 0 B) ( SB) B cost of a lawsuit ling ( SS 0 ) damages 91 Figure 2.2: Payo from buying a security which has a hypothetical put baked into its price. Plaintis' attorneys enable the buying of puts on behalf of plaintis: the shareholders of a rm. The top panel demonstrates the payo from the put. S,S 0 , andB are dened in Figure 2.1. Courts, e.g., arbitrators or judges determine the strike price. The strike price here is SB. The supposed damages are SS 0 . If the settlement is SS 0 , then it covers the full extent of the damages. Because shareholders purchase a security that has the hypothetical put baked into its price, their overall prot actually resembles that of a protective put, a.k.a., synthetic long call. The bottom panel demonstrates shareholders' prot from the protective put or synthetic long call. 6 - payo equity value ( SS 0 ) q q q q q q q 0 B (S 0 B) S 0 ( SB) S @ @ @ @ @ @ @ @ @ @ @ @ @ ( SS 0 ) damages 6 - prot equity value 0 (S 0 B) S 0 ( SB) S B ( SS 0 ) q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q 92 Figure 2.3: The setup of the model with the arrival of news at t = 1 and with no shareholder litigation. Shareholders obtain news about their entrepreneur at t = 1. The news arrives privately to shareholders. The news is either positive or negative. f denotes the t = 0 probability that an entrepreneur is good; 1f denotes thet = 0 probability that an entrepreneur is bad. e denotes the probability that an initially good entrepreneur has poor luck and provides negative news at t = 1, thus becoming unlucky. When positive news arrives, shareholders learn that their entrepreneur is good. A good entrepreneur always reports cash ows X H at t = 2. So given positive news, shareholders' expected value at t = 1 is E p 1 =X H . When negative news arrives, shareholders learn that their entrepreneur is either unlucky or bad. Without assistance, shareholders cannot distinguish between an unlucky and a bad entrepreneur: both types report cash ows X H with probability and cash ows X L with probability 1, where X H > X L . Shareholders can replace their entrepreneur at t = 1 to obtain a net present value of L. So given negative news, shareholders' expected value at t = 1 is E n 1 = maxfL;X H + (1)X L g. Note, the gure only contains information on shareholders' cash ows; it does not contain information on entrepreneurs' control rents. t = 0 t = 1 t = 2 t = 3 K f 1f e good POSITIVE NEWS E p 1 =X H unlucky bad NEGATIVE NEWS E n 1 = maxfL;X H + (1)X L g XH 1 XL XH XL 93 Figure 2.4: The setup of the model with the arrival of news at t = 1 and with share- holder litigation. With shareholder litigation (specically, given Denition 1), shareholders have no incentive to replace their entrepreneur at t = 1 after negative news arrives. So given negative news, shareholders' expected value att = 1 becomesE n 1 =X H +(1)[X L hs(X H X L )B]. When low cash ows X L are reported at t = 2, shareholders le a lawsuit against their en- trepreneur. B denotes the cost of a lawsuit ling. h denotes the portion of entrepreneurs who are perceived to be bad after negative news arrives at t = 1. s denotes the probability that courts nd a bad entrepreneur guilty of wrongdoing and that shareholders obtain damages in the amount ofX H X L att = 3. By assumption courts always dismiss a lawsuit against an unlucky entrepreneur. All remaining notation is described in Figure 2.3. Note, the gure only contains information on shareholders' cash ows; it does not contain information on entrepreneurs' control rents. t = 0 t = 1 t = 2 t = 3 K f 1f e good POSITIVE NEWS E p 1 =X H unlucky bad NEGATIVE NEWS E n 1 =XH + (1)XL ... +(1)[hs(XH XL)B] XH 1 XL XH hs XH B (1hs) XL B 94 2.8 Tables Table 2.1: Summary of probability related notation. This table contains all notation, used throughout the chapter, that species a probability value. Notation that may aect the probabilities is not contained in this table. f Initial (i.e. t = 0) probability of an entrepreneur being good. 1f Initial (i.e. t = 0) probability of an entrepreneur being bad. f p Conditional probability of an entrepreneur being good given positive news att = 1. f n Conditional probability of an entrepreneur being good given negative news at t = 1. e Probability that a good entrepreneur is unlucky and provides negative news at t = 1. h The probability of an entrepreneur being bad after negative news arrives at t = 1. Probability of the high cash ow X H at t = 2. 1 Probability of the low cash ow X L at t = 2. s Probability that courts nd a bad entrepreneur guilty of wrongdoing and that shareholders obtain damages at t = 3. 1s Probability that courts do not nd a bad entrepreneur guilty of wrongdoing and that shareholders do not obtain damages at t = 3. p A bad entrepreneur's belief about the probability that shareholders le a lawsuit at t = 1. 1p A bad entrepreneur's belief about the probability that shareholders do not le a lawsuit at t = 1. 95 Table 2.2: Summary of cash related notation. This table contains all notation, used throughout the chapter, that has a dollar-value associated with it. K The total amount of funding that an entrepreneur requires to fund her or his project. C An entrepreneur's control rent. w Funding spent on opportunistic behavior. X(w) Cash ow that a bad entrepreneur attains in the high state att = 2, after engaging in opportunism. X H The reported cash ow in the high state at t = 2. X L (w) The reported cash ow in the low state att = 2, when some entrepreneurs engage in opportunism. L Shareholders' net present value of replacing an entrepreneur at t = 1. Q The quick settlement that shareholders can obtain after ling a lawsuit at t = 1. b(Q) An entrepreneur's benet from a quick settlement at t = 1. B Shareholders' cost of ling a lawsuit. d An entrepreneur's unavoidable cost of undergoing a lawsuit. g Savings that a bad entrepreneur uses to deceive a lender at t = 2. D Financing from a lender at t = 1. F The face value of debt at t = 2. 96 Table 2.3: Summary of remaining notation. This table contains all remaining notation, used throughout the chapter, that does not belong in the prior two tables. N The number of investors who become shareholders at t = 0 after investing in an entrepreneur's project. V The region in which the lender seizes assets. V 0 The region in which the lender does not seize assets: complement to V . 97 Bibliography Appel, I., 2016. Governance by litigation, Working Paper. Arena, M., Julio, B., 2015. The eects of securities class action litigation on corporate liquidity and investment policy. The Journal of Financial and Quantitative Analysis 50, 251{275. Bates, T. W., Kahle, K. M., Stulz, R. M., 2009. Why do us rms hold so much more cash than they used to? The Journal of Finance 64, 1985{2021. Brochet, F., Srinivasan, S., 2014. Accountability of independent directors: Evidence from rms subject to securities litigation. Journal of Financial Economics 111, 430{449. Chalmers, J. M., Dann, L. Y., Harford, J., 2002. Managerial opportunism? evidence from direc- tors' and ocers' insurance purchases. The Journal of Finance 57, 609{636. Diamond, D. W., 1984. Financial intermediation and delegated monitoring. The Review of Eco- nomic Studies 51, 393{414. Diamond, D. W., 1991. Debt maturity structure and liquidity risk. The Quarterly Journal of Economics 106, 709{737. Diamond, D. W., 2004. Committing to commit: Short-term debt when enforcement is costly. The Journal of Finance 59, 1447{1479. Drake, P. D., Vetsuypens, M. R., 1993. IPO underpricing and insurance against legal liability. Financial Management 22, 64{73. 98 DuCharme, L. L., Malatesta, P. H., Sefcik, S. E., 2004. Earnings management, stock issues, and shareholder lawsuits. Journal of Financial Economics 71, 27{49. Engelmann, K., Cornell, B., 1988. Measuring the cost of corporate litigation: Five case studies. The Journal of Legal Studies 17, 377{399. Ferris, S. P., Jandik, T., Lawless, R. M., Makhija, A., 2007. Derivative lawsuits as a corporate governance mechanism: Empirical evidence on board changes surrounding lings. The Journal of Financial and Quantitative Analysis 42, 143{165. Fich, E. M., Shivdasani, A., 2007. Financial fraud, director reputation, and shareholder wealth. Journal of Financial Economics 86, 306{336. Froot, K. A., Stein, J. C., 1998. Risk management, capital budgeting, and capital structure policy for nancial institutions: An integrated approach. Journal of Financial Economics 47, 55{82. Gale, D., Hellwig, M., 1985. Incentive-compatible debt contracts: The one-period problem. The Review of Economic Studies 52, 647{663. Gande, A., Lewis, C. M., 2009. Shareholder-initiated class action lawsuits: Shareholder wealth eects and industry spillovers. The Journal of Financial and Quantitative Analysis 44, 823{850. Hanley, K. W., Hoberg, G., 2012. Litigation risk, strategic disclosure and the underpricing of initial public oerings. Journal of Financial Economics 65, 309{335. Jensen, M. C., 1986. Agency costs of free cash ow, corporate nance, and takeovers. The Amer- ican Economic Review 76, 323{329. Jensen, M. C., Meckling, W. H., 1976. Theory of the rm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3, 305{360. Karpo, J. M., Lee, D. S., Martin, G. S., 2008a. The consequences to managers for nancial misrepresentation. Journal of Financial Economics 88, 193{215. 99 Karpo, J. M., Lee, D. S., Martin, G. S., 2008b. The cost to rms of cooking the books. The Journal of Financial and Quantitative Analysis 43, 581{611. Keynes, J. M., 1936. The general theory of employment, interest, & money. London: Macmillan . Kim, C.-S., Mauer, D. C., Sherman, A. E., 1998. The determinants of corporate liquidity: Theory and evidence. The Journal of Financial and Quantitative Analysis 33, 335{359. Klausner, M., Hegland, J., 2010. When are securities class actions dismissed, when do they settle, and for how much? { Part II. Professional Liability Underwriting Society Journal 23, 1{5. Klausner, M., Hegland, J., 2013. When are securities class actions dismissed, when do they settle, and for how much? an update. Professional Liability Underwriting Society Journal 26, 1{5. Lin, C., Ocer, M. S., Wang, R., Zou, H., 2013. Directors' and ocers' liability insurance and loan spreads. Journal of Financial Economics 110, 37{60. Lin, C., Ocer, M. S., Zou, H., 2011. Directors' and ocers' liability insurance and acquisition outcomes. Journal of Financial Economics 102, 507{525. Lowry, M., Shu, S., 2002. Litigation risk and IPO underpricing. Journal of Financial Economics 65, 309{335. McTier, B. C., Wald, J. K., 2011. The causes and consequences of securities class action litigation. Journal of Corporate Finance 17, 649{665. Opler, T., Pinkowitz, L., Stulz, R. M., Williamson, R., 1999. The determinants and implications of corporate cash holdings. Journal of Financial Economics 52, 3{46. Seru, A., 2014. Firm boundaries matter: Evidence from conglomerates and R&D activity. Journal of Financial Economics 111, 381{405. Townsend, R. M., 1979. Optimal contracts and competitive markets with costly state verication. Journal of Economic Theory 21, 265{293. 100 Appendix Proof of Lemma 1 Given the existence of shareholder litigation, shareholders choose to retain their entrepreneur after negative news arrives at t = 1. In turn, shareholders' expected value is X H + (1)X L (1)B +h(1)[s(X H X L )]: Recall, h denotes the portion of entrepreneurs who are perceived to be bad after negative news arrives. At the same time, the expected value of an entrepreneur who provides negative news at t = 1 is C + (1)(Cd)h(1)[s(X H X L )]: Combining the two expressions yields [X H + (1)X L ] +C (1)(B +d): X H + (1)X L represents shareholders' expected value without shareholder litigation. With shareholder litigation, expected total value rises only if C > (1)(B +d). 101 Proof of Proposition 1 Problem 2 seeks to maximize Q subject to three constraints. If the three constraints cannot be met, then a feasible solution does not exist. In turn, unlucky and bad entrepreneurs pool and operate at the baseline: invest K=2 between t = 0 and t = 1, and K=2 between t = 1 and t = 2. So at t = 1 they hold K=2 in cash. A bad entrepreneur chooses a level of opportunistic behavior equal to w from Equation 2.4. If the three constraints can be met and Constraint 2.6 binds rst, then unlucky and bad entrepreneurs are indierent between saving cash to negotiate a quick settlement and operating at the baseline. But because Constraint 2.7 does not bind, shareholders have no incentive to le a premature lawsuit and negotiate a quick settlement. So unlucky and bad entrepreneurs again pool and operate at the baseline: invest K=2 between t = 0 and t = 1, and K=2 between t = 1 andt = 2. So att = 1 they holdK=2 in cash. A bad entrepreneur chooses a level of opportunistic behavior equal to w from Equation 2.4. If the three constraints can be met and Constraint 2.7 binds rst, then a bad entrepreneur's belief about the probability that shareholders le a premature lawsuit at t = 1 determines the equilibrium. Suppose p to denote a bad entrepreneur's belief about the probability that share- holders le a premature lawsuit at t = 1. A bad entrepreneur mimics an unlucky entrepreneur when p[Cd +b(Q )] ... + (1p) [ (C +X(w )X H ) + (1) (Cds (X H X L (w )))] ... (C +X(w max )X H ) + (1) [Cd (X H X L (w max ))]: Q denotes the solution to Problem 2. The left-hand side of the inequality denotes a bad en- trepreneur's expected value from mimicking an unlucky entrepreneur. With probability p, a bad 102 entrepreneur obtains Cd +b(Q ), and with probability 1p, a bad entrepreneur obtains the expected value from the solution in Equation 2.4. When a premature lawsuit ling occurs at t = 1, a bad entrepreneur obtains no benets of opportunism after negotiating a settlement Q . The right-hand side of the inequality denotes a bad entrepreneur's expected value from separa- tion. When a bad entrepreneur separates, shareholders can distinguish between an unlucky and a bad entrepreneur. In turn, courts nd a bad entrepreneur guilty with probability one. A bad entrepreneur, who is always found guilty in the low state 1, pushes the level of opportunistic behavior to the corner solutionw max . A bad entrepreneur's expected value at the corner solution is his expected value from separation. Rearranging the above inequality yields [b(Q ) [(X(w )X H d)s (1)(X H X L (w ))]] [X(w max )X(w )] (1)[(X H X L (w max ))s (X H X L (w ))] p 1; Becausep is a probability, the benet functionb(Q) at the feasible solutionQ must be such that the left-hand side is greater than or equal to zero. If the above inequalities hold, unlucky and bad entrepreneurs pool and save Q +B at t = 1. B is used to pay for the lawsuit ling; thereafter, shareholders obtainQ . A bad entrepreneur chooses a level of opportunistic behavior equal tow from Equation 2.4. Alternatively, if the above inequality were to be 0p< [b(Q ) [(X(w )X H d)s (1)(X H X L (w ))]] [X(w max )X(w )] (1)[(X H X L (w max ))s (X H X L (w ))] : then unlucky and bad entrepreneurs would separate. An unlucky entrepreneur would saveQ +B att = 1. A bad entrepreneur would operate at the baseline: invest K=2 betweent = 0 andt = 1, and K=2 between t = 1 and t = 2. At t = 1 a bad entrepreneur would hold K=2 in cash. Finally, a bad entrepreneur would choose a level of opportunistic behavior equal to w max . 103 Proof of Lemma 2 Equation 2.4 states that X 0 (w ) = (1) [s 0 (w ) (X H X L (w ))s(w )X 0 L (w )]: while Equation 2.11 states that X 0 (w ) =(1) [s 0 (w ) [Cd (X H X L (w ))] +s(w )X 0 L (w )]: Rearranging Equation 2.11 yields X 0 (w ) = (1) [s 0 (w ) (X H X L (w ))s(w )X 0 L (w )s 0 (w )(Cd)]: The term s 0 (w )(Cd) in the rearranged Equation 2.11 is strictly positive. Given the assumed properties of X(w) and s(w), inputting w from Equation 2.4 into the rearranged Equation 2.11 yields X 0 (w )> (1) [s 0 (w ) (X H X L (w ))s(w )X 0 L (w )s 0 (w )(Cd)]: w >w to turn the above inequality into an equality. 104 Proof of Proposition 2 Item 1 holds by assumption. So long as the zero-prot condition holds, a lender provides nancing. Item 2 follows from Conditions 2.12 and 2.13. To obtain Condition 2.12, begin with [C +X(w;g)X H ] ... + (1) [s(w;g) [Cd (X H X L (w;g))] + (1s(w;g)) (Cd)] ... <C; where the left-hand side of the inequality denotes the expected value of a bad entrepreneur who misrepresents while the right-hand side of the inequality denotes the concurrent expected value of an unlucky entrepreneur. Combining terms yields Condition 2.12. To obtain Condition 2.13, begin with [C +X(w;g)X H ] ... + (1) [s(w;g) [Cd (X H X L (w;g))] + (1s(w;g)) (Cd)] ... <[C +X(w )X H ] ... + (1) [s(w ) [Cd (X H X L (w ))]]; where the left-hand side of the inequality denotes the expected value of a bad entrepreneur who misrepresents while the right-hand side of the inequality denotes the concurrent expected value of a bad entrepreneur who does not misrepresent. Rearranging and combining terms yields Con- dition 2.13. The above conditions state that if misrepresentation is feasible, then the expected value of mis- representation must be less than either the expected value of acting like an unlucky entrepreneur 105 (i.e. Equation 2.12) or the expected value of not committing the misrepresentation (i.e. Equa- tion 2.13). As a result, a bad entrepreneur has no incentive to misrepresent. 106 Proof of Proposition 3 Between Conditions 2.12 and 2.13, Condition 2.13 binds rst. Suppose to the contrary that a g 0 exists that maximizes Problem 4 and allows Condition 2.13 to be slack. I establish that g 0 does not maximize Problem 4. g 0 implies the following face value of debt: F 0 =X L (w ) +g 0 . In turn, shareholders who face a bad entrepreneur obtain an expected value of (X H F 0 ) + (1)[hs(w ) (X H BF 0 )]. Since Condition 2.13 is slack, one can always decrease g 0 by some innitesimal amount and maintain incentive compatibility. The face value of debt then becomes F 0 =X L (w ) +g 0 , and shareholders who face a bad entrepreneur obtain an expected value of ( XF 0 +) + (1 )[s(w )( XBF 0 +)]. Clearly, (X H F 0 +) + (1)[s(w )(X H BF 0 +)] ... >(X H F 0 ) + (1)[s(w )(X H BF 0 )]: The above inequality contradicts the assertion that g 0 maximizes Problem 4. So Condition 2.13 must bind. 107 Proof of Lemma 3 Given negative news, a rm's total value at t = 1 is X H + (1) [hs(w )X H + (1hs(w ))X L (w )B]: The above total value sums the expected value of shareholders and the lender at t = 1. Recall, if shareholders attempt to replace or le a lawsuit against an entrepreneur who provides negative news at t = 1, at most they can obtain the rm's total value shown above. At the same time, however, shareholders will be required to pay the face value of debt F to the lender. Subtracting out the payment of face value from the rm's total value yields (X H F ) + (1) [hs(w ) (X H BF ) + (1hs(w )) (X L (w )BF )]: So when F > X L (w )B, the term (1hs(w )) (X L (w )BF ) is strictly negative, which implies that the expected value to shareholders from replacing or ling a lawsuit against an entrepreneur who provides negative news at t = 1 is strictly less than the expected value to shareholders of waiting for the resolution of uncertainty at t = 2. Recall that the expected value to shareholders of waiting for the resolution of uncertainty at t = 2 is (X H F ) + (1) [hs(w ) (X H BF )]: 108
Abstract (if available)
Abstract
Shareholder litigation risk varies across time and across firms. When shareholder litigation risk is high, it can increase (decrease) a firm's cash and investment before (after) a lawsuit filing. When shareholder litigation risk is low, little to no impact occurs. A quasi-natural experiment using two legal shocks, In re IPO and CAFA, supports a causal interpretation. Shareholder litigation risk can also impact a firm's debt around the time of a lawsuit filing, but the empirical results are less clear-cut. In addition to the empirical results, I offer a theoretical framework to help explain the changes in cash, investment, and debt around the time of a lawsuit filing. The framework posits that an entrepreneur, at risk of a lawsuit filing, may save cash as a precautionary measure. When the cash accumulates and a lawsuit filing does not occur, an entrepreneur increases investment in hopes of superior future performance. But if future performance wanes notwithstanding, shareholders then file a lawsuit against their entrepreneur. When cash is limited, debt may act as an alternative precautionary measure against a lawsuit filing. Finally, the theoretical framework emphasizes a benefit of shareholder litigation—the option to file a lawsuit against an entrepreneur encourages shareholders to fund projects and to retain entrepreneurs.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Executive compensation: the trend toward one size fits all
PDF
Slashing liquidity through asset purchases: evidence from collective bargaining
PDF
Essays on information and financial economics
PDF
Incentives and relative-wealth concerns: theory and evidence
PDF
Three essays on board structure
PDF
Share repurchases: how important is market timing?
PDF
Investment behavior of mutual fund managers
PDF
Essays in corporate finance
PDF
Dividend policy, earnings announcements, and analysts' earnings forecasts
PDF
The structure of strategic communication: theory, measurement, and effects
PDF
Essays in financial intermediation
PDF
Credit risk of a leveraged firm in a controlled optimal stopping framework
PDF
Evolution of returns to scale and investor flows during the life cycle of active asset management
PDF
Implied equity duration: international evidence
PDF
Does mandatory adoption of international financial reporting standards decrease the voting premium for dual-class shares: theory and evidence
PDF
Essays in behavioral and entrepreneurial finance
PDF
Creating destruction: the political economy of zombie firms
PDF
RGBD camera based wearable indoor navigation system for the visually impaired
PDF
Essays on understanding consumer contribution behaviors in the context of crowdfunding
PDF
Efficient policies and mechanisms for online platforms
Asset Metadata
Creator
Talijan, Vuk
(author)
Core Title
Shareholder litigation as a disciplining device: evidence from firms' financial policies
School
Marshall School of Business
Degree
Doctor of Philosophy
Degree Program
Business Administration
Publication Date
02/23/2018
Defense Date
02/14/2018
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
corporate finance,OAI-PMH Harvest,shareholder litigation
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Murphy, Kevin (
committee chair
), Hoberg, Gerard (
committee member
), Lee, Yoon-Ho Alex (
committee member
), Ozbas, Oguzhan (
committee member
)
Creator Email
talijan@usc.edu,vtalijan@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c40-478454
Unique identifier
UC11268218
Identifier
etd-TalijanVuk-6062.pdf (filename),usctheses-c40-478454 (legacy record id)
Legacy Identifier
etd-TalijanVuk-6062.pdf
Dmrecord
478454
Document Type
Dissertation
Rights
Talijan, Vuk
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
corporate finance
shareholder litigation