Federal Judge Ideology: A New Measure of Ex-Ante Litigation Risk

Drawing on the political theory of judicial decision making, our paper proposes a new and parsimonious ex ante litigation risk measure: federal judge ideology. We find that judge ideology complements existing measures of litigation risk based on industry membership and firm characteristics. Firms in liberal circuits (the third quartile in ideology) are 33.5% more likely to be sued in securities class action lawsuits than those in conservative circuits (the first quartile in ideology). This result is stronger after the U.S. Supreme Court’s ruling in the Tellabs case. We next show that the effect of judge ideology on litigation risk is greater for firms with more sophisticated shareholders and with higher expected litigation costs. Furthermore, judicial appointments affect litigation risk and the value of firms in the circuit, highlighting the economic consequences of political appointments of judges. Finally, using our new measure, we document that litigation risk deters managers from providing long-term earnings guidance, a result that existing measures of litigation risk cannot show.


Introduction
Litigation is costly to firms. The question of how the risk of securities class action lawsuits affects corporate decisions has attracted a great deal of research interest in the accounting and finance literature. 1, 2 Prior studies generally use industry membership (e.g., Schipper [1994a, 1994b]), either alone or in conjunction with firm characteristics, such as size, stock turnover, and returns, as measures of ex ante litigation risk (e.g., Johnson, Kasznik, and Nelson [2001], Field, Lowry, and Shu [2005], Rogers and Stocken [2005]). However, these measures are also likely to capture industry and firm characteristics unrelated to ex ante litigation risk. Consequently, such measures can induce spurious relations, which affect their results (Kim and Skinner [2012]). Our study proposes a new measure that better captures ex ante litigation risk by exploiting firms' external litigation environment: federal judge ideology.
It is documented in both legal and political science studies that ideology is among the most important of judges' personal attributes influencing civil liberties and economic lawsuit outcomes (Johnston [1976], Tate [1981], Segal and Cover [1989], Staudt, Epstein, and Wiedenbeck [2006]).
Specifically, political theory of judicial decision making argues that the materials available to judges do not usually provide sufficient clarity for resolving disputes (Grundfest and Pritchard [2002]). As such, judges' individual characteristics can influence legal outcomes. Judges, like other government officials, take advantage of their ability to shape case outcomes and project their views of justice onto society (Richards and Kritzer [2002], Cross [2007]). Through judicial votes, judges advance personal ideological preferences, which generally fall along the conventional liberal-to-conservative continuum 1 Federal securities class action lawsuits generally involve violations of the Securities Exchange Act of 1934 (commonly known as the Exchange Act). While shareholders can file class action and derivative lawsuits in state courts, these generally cover a narrow range of misbehavior, that is, almost entirely limited to two contextsacquisitions and self-dealing transactions-and focus largely on the duties and liabilities of directors, not officers; they also carry less severe penalties and are of diminishing importance (Thompson and Sale [2003]; Thompson and Thomas [2004]). In addition, state derivative lawsuits typically follow the filing of a federal securities class action suit; they do not uncover new facts and often result in poor quality settlements, and thus add little significant value beyond federal securities class actions (Choi, Erickson, and Pritchard [2017]). Due to the importance of federal securities class action lawsuits, we focus on the risk of such lawsuits throughout this paper. Please refer to Section 3.2 for a detailed discussion. 2 For example, prior studies examine litigation risk in financial reporting and disclosure decisions (e.g., Skinner [1994Skinner [ , 1997, Johnson, Kasznik, and Nelson [2000], Field, Lowry, and Shu [2005], Rogers and Van Buskirk [2009]), cash-holding and investment decisions (Arena and Julio [2015]), executive compensation (Peng and Röell [2008], Laux [2010]), IPO underpricing (Lowry and Shu [2002]), institutional monitoring (Cheng et al. [2010]), and auditor resignation (Shu [2000]).
in U.S. politics (see George [1998] for a review).
The most widely adopted measure of judge ideology is the appointing president's political affiliation (Goldman [1999], Pinello [1999]). Presidents almost always appoint judges whose ideology reflects that of their political party (Dorsen [2006], Federal Judicial Center [2006). Judges appointed by Democratic presidents are more liberal on the bench than those appointed by Republican presidents (Goldman [1975], Sunstein, Schkade, and Ellman [2004], Cross [2007]). For example, using presidential party affiliation as the measure, Tate [1981] finds evidence that judges' personal attributes explain 72% to 87% of their voting behavior, with ideology having the largest influence. In economic cases, studies show that a liberal (Democratic) ideology is more likely to result in outcomes that are anti-business or pro-economic underdog, that is, that favor the "have-nots" over the "haves," such as labor unions and employees over firms, and government regulations over the free market (Goldman [1966], Ryan and Tate [1975], Segal and Spaeth [1996]). In the political science literature, researchers apply the "liberal versus conservative" coding protocol to securities class action lawsuits (brought by investors or shareholders who suffered economic injury as a result of securities fraud). Findings indicate that liberal judges are more likely to vote in favor of investors (plaintiffs), while conservative judges are more likely to vote in favor of firms (defendants); as such, liberal judges pose a higher litigation risk to firms than conservative judges (Grundfest and Pritchard [2002], Sullivan and Thompson [2004], Spaeth [2006], Fedderke and Ventoruzzo [2016]). 3 Our study also shows that the political appointments of judges have economic consequences for firms. Prior research has documented how the characteristics and quality of securities laws and their enforcements affect firms (see review by La Porta, Lopez-de-Silanes, and Shleifer [2008]). However, the outcomes of securities law enforcement depend heavily on the judicial system (Hay and Shleifer [1998]). In the U.S., courts handle private enforcement of securities laws, and public enforcements such as those handled by the Securities and Exchange Commission (SEC) are often challenged in courts or, if settled, need approval from judges (Macchiarola [2012]). Investigating judges' influence in securities cases thus provides a more comprehensive understanding of the economic consequences of the legal system.
Empirically, we measure judge ideology at the circuit court level. Circuit and district court judges are usually the final arbiters of securities class action lawsuits, and circuit court decisions have binding constraints over district courts within their jurisdictions. Hence, the ideology of a circuit court has the greatest influence on expected lawsuit outcomes (refer to Section 3.2 for a detailed discussion) (Bowie and Songer [2009], Choi, Gulati, andPosner [2012]). Each case in a circuit court is assigned to a panel composed of three randomly selected judges from the circuit. We follow prior studies (e.g., Sunstein, Schkade, and Ellman [2004]) and measure the circuit's judge ideology as the probability that the panel is dominated by appointees of Democratic presidents, that is, the probability that the threejudge panel comprises at least two Democratic appointees. 4 To estimate a firm's litigation risk, we use the judge ideology of the circuit court whose jurisdiction covers the firm's headquarter (hereafter, the home circuit), as the headquarter is the usually requested filing location in civil procedures (28 U.S. Code §1391 (b); 1404). In our sample, 87% of securities class action lawsuits are filed in the firm's home circuit. 5 To the extent that plaintiffs understand how judge ideology affects case outcomes (de Figueiredo [2005]), we expect they are more likely to file securities class action lawsuits against firms when there are more Democratic judges in the circuit. 6 As a measure of litigation risk, judge ideology differs from industry membership and firm characteristics in several ways. First, measures based on industry and firm characteristics are estimated using filed lawsuits, with a purpose to predict lawsuit occurrences. They thus capture not only the ex 4 See Section 3.1 for a detailed discussion of how securities class action lawsuits are filed in the federal judicial system. 5 Cox, Thomas, and Bai [2009] also show that the majority of securities class action lawsuits are filed in firms' home circuits. 6 For example, Bernstein Litowitz Berger & Grossman, the largest plaintiff law firm in 2017, often discusses the impact of judge assignment on securities lawsuits in their analysis of the latest securities law issues (Zayenchik [2017], Eisinger [2018]).
Electronic copy available at: https://ssrn.com/abstract=3170842 ante litigation risk (i.e., the likelihood of being sued given that there is misconduct), but also firms' misconduct behaviors that lead to lawsuits. In contrast, our measure of judge ideology is based on the composition of the judges on the circuit court with jurisdiction over the state where a firm is located.
Compared with industry membership and firm characteristics, judge composition is less endogenous to omitted correlated variables that are also likely to affect firm operations or managerial decisions, such as risk and growth (Kim and Skinner [2012]), and can thus better serve the goal of isolating ex ante litigation risk. Second, judge ideology is measured based on the circuit's judge composition, which changes whenever a judge retires or a new judge is appointed (Figure 1). The interaction of exogenous cross-sectional and time-series variations in judge ideology makes it suitable to studies that require strong identification. Third, industry and firm characteristics capture litigation risks originating from the firms themselves, including their business models and internal operations. Our measure takes a different perspective by focusing on a set of critical players outside the firm, judges, who are central to the interpretation of legal doctrines and whose ideology is a key determinant of both the expected and realized lawsuit outcomes (Cross and Tiller [1998]). As such, our measure is especially useful for those studies interested in litigation risk stemming from the legal environment. In summary, we expect that judge ideology, as a parsimonious ex ante litigation risk measure, can be applied across a wide array of studies that require an exogenous measure for identification.
[Please insert Figure 1 here] To validate judge ideology as a measure of ex ante litigation risk, we first investigate whether it can predict lawsuit occurrence beyond existing measures based on industry membership and firm characteristics (Johnson, Kasznik, and Nelson [2001], Rogers and Stocken [2005], Kim and Skinner [2012]). Specifically, we test whether firms located in circuits with judges that are more liberal are more likely to be sued. We find strong evidence in both univariate and regression analyses to confirm our prediction. Our regression analyses, which include state economy and demographic variables, in addition to circuit and year fixed effects to control for economic conditions and political climate, indicate that judge ideology not only adds incrementally to existing litigation risk prediction models, but also captures variations in ex ante litigation risk with significant economic magnitudes. An increase Electronic copy available at: https://ssrn.com/abstract=3170842 in judge ideology (i.e., more liberal) from the first to the third quartile implies an increase in the ex ante probability of being sued by 33.5% in relative terms (from 2.45% to 3.27% in absolute terms).
Next, we conduct two sets of tests to explore variations in judge ideology as a measure of ex ante litigation risk and to provide further validations: 1) when judges have more discretion in their decisions, and 2) when plaintiffs are more capable or have more incentives to consider judge ideology in filing decisions. First, we identify the Supreme Court's decision on Tellabs, Inc. v. Makor Issues & Rights, Ltd. (551 U.S. 308, 2007) as an exogenous shock that may have affected judge ideology's role in predicting lawsuit occurrence. Because the Tellabs decision allows judges a higher degree of discretion in deciding a motion to dismiss (Miller [2009]) and provides further room for presiding judges' individual perspectives to shape case outcomes (Cox, Thomas, and Bai [2009]), we predict the effect of judge ideology to be stronger after the Tellabs decision. Our findings are consistent with this prediction. The effect of judge ideology is 1.4 times stronger after the U.S. Supreme Court's ruling in the Tellabs case compared to the period before the ruling. Second, we conduct a series of cross-sectional tests to identify subgroups of firms for which judge ideology plays a stronger role in predicting litigation risk. Because judge ideology affects the odds of lawsuit filings only when plaintiffs understand its effect on case outcomes, we base these tests on plaintiffs' ability and incentives to consider judge ideology in their filing decisions. 7 Specifically, we expect judge ideology to have a stronger effect when shareholders are more sophisticated or when the expected payoffs from the lawsuits are higher. We find consistent results using the percentage of institutional ownership to proxy for shareholder sophistication, and firm size and predicted litigation risk to proxy for expected lawsuit payoffs.
Third, we use judicial appointments to explore how time-series variations in judge ideology affect firms' litigation risks. Consistent with the appointment of liberal (conservative) judges increasing (decreasing) firms' litigation risk, we find that liberal (conservative) judicial appointments increase (decrease) the occurrence of lawsuit filings. Furthermore, given that securities class action lawsuits are costly to firms (Romano [1991]), we predict that the market reacts negatively when a liberal judge is appointed, and vice versa, and find supporting evidence. A liberal judicial appointment leads to a -0.26% three-day abnormal return for firms in the circuit, while a conservative judicial appointment boosts the stock prices of firms in the circuit by 0.27%. Finally, we find that the market reacts more strongly for the subsample of firms in industries with higher securities litigation risk, suggesting that market reactions are more likely to be driven by expected cost changes in securities class action lawsuits rather than other types of litigation. In summary, our identification tests support judge ideology as a measure of ex ante litigation risk that provides economically meaningful improvements to measures currently used in the literature. The relation between judge ideology and lawsuit occurrence also suggests that the political appointment of judges shapes firms' litigation risks and has sizable economic consequences.
Fourth, we investigate the relation between judge ideology and lawsuit outcomes, that is, whether a lawsuit is dismissed and, if not, its settlement amount, to gain further insights into the economic consequences of judicial appointments. Consistent with the traditional notion that liberal judges tend to favor economic underdogs, we find cross-sectional evidence that lawsuits filed in more liberal circuits have significantly lower dismissal rates and larger settlements. An increase in judge ideology from the first to the third quartile leads to a 14.6% decrease in the odds of cases being dismissed, and for cases that are not dismissed, an average increase of $907,274 in settlements. These findings suggest that judge ideology is a critical driver of judicial decision making and litigation costs.
Lastly, we use our litigation risk measure to shed new light to a particular research problem: how ex ante litigation risk affect a firm's voluntary disclosure decisions. Prior studies show that when facing bad earnings news, firms either preemptively warn investors, known as the "preemption effect" (Kasznik and Lev [1995], Field, Lowry, and Shu [2005]), or abstain from disclosure, or bundle bad news with other news (Francis, Philbrick and Schipper [1994b], Skinner [1997], Bliss, Partnoy andFurchtgott [2018]). Litigation risk, however, may also discourage firms from disclosing forwardlooking information. This is known as the "chilling effect," because the legal system cannot effectively distinguish between unexpected forecast errors due to chance and those due to deliberate management bias (Johnson, Kasznik, and Nelson [2001], Baginski, Hassell, and Kimbrough [2002], Rogers and Van Electronic copy available at: https://ssrn.com/abstract=3170842 Buskirk [2009]). We find results consistent with theoretical predictions; that is, firms located in more liberal circuits are, in general, more likely to issue short-term earnings forecasts to preempt upcoming negative earnings news and also more reluctant to release positive long-horizon earnings forecasts.
These results suggest that managers are aware of judge ideology's effect on their firms' litigation risk and that they incorporate this information in their decision making. Most importantly, we find that although measures based on industry membership and firm characteristics provide some support for the preemption effect, they fail to provide evidence of the chilling effect. 8 In a placebo test, we find no significant relation between judge ideology and firms' mandatory disclosures, measured using firms' mandatory 8-K filings (Cooper, He, and Plumlee [2016]).
Our study makes the following contributions to the literature. First, we contribute to empirical research on litigation risk and firm behavior in the accounting, finance, and economics literature. We introduce a new measure of ex ante litigation risk, federal judge ideology. Measured via the party affiliation of judges' appointing presidents, judge ideology is an intuitive measure that is well-grounded in legal and political science studies. Our findings based on the measure broaden the understanding of litigation risks to external factors beyond firm and industry characteristics currently used in the literature.
Empirically, our new measure adds explanatory power in predicting litigation occurrences beyond existing measures and provides economically significant effects. As a result, our measure will enable researchers to draw a more definitive conclusion regarding the impact of ex ante litigation risk on firm decisions.
Second, we provide evidence for the economic consequences of judicial appointments.
Research has shown how legal rules and their enforcements affect firms (La Porta et al. [1997], Hail andLeuz [2006], La Porta, Lopez-de-Silanes, and Shleifer [2006]). However, there is little evidence for the economic impact of the judicial system, which is heavily relied upon to enforce securities laws (Ford [2005], Pritchard [2011]). Our findings suggest that with the discretion allowed in the judicial doctrines 8 Prior studies find support for the chilling effect using regulation changes (Johnson, Kasznik, and Nelson [2001]; Huang, Shen, and Zang [2018]), actual lawsuit filings (Rogers and Van Buskirk [2009]), and cross-country samples (Baginski, Hassell, and Kimbrough [2002]). Our study is the first to document the chilling effect in a large sample of panel data.
(e.g., the Supreme Court's Tellabs Ruling), judge ideology is economically meaningful in predicting litigation occurrence and its outcomes. These results underline a key consequence of the political appointments of judges, which has not been documented in the literature. Thus, our study provides important insights, not only for academics but also for investors, regulators, and lawmakers.
Third, the judge ideology measure is especially relevant to the study of accounting regulations, as it enables researchers to examine litigation risk stemming from the legal system (e.g., Johnson, Kasznik, and Nelson [2001], Heflin, Subramanyam, and Zhang [2003]). Judge ideology measures how judges' preferences shape their interpretations of legal doctrines, which are crucial to regulatory enforcement. Evidence from studies based on judge ideology thus provides valuable insights, both to regulators for evaluating the effectiveness of regulations and to firms for determining how to cope with constantly evolving legal requirements. As an example, using the judge ideology measure, our study draws inferences relevant to the voluntary disclosure literature in ways that research using measures based on industry membership and firm characteristics cannot. In using judge ideology as a measure of litigation risk, it should be noted that circuit court judges are constrained by legal precedents in their own circuits and the Supreme Court. The effect of judge ideology is thus likely shaped by court decisions. Indeed, our empirical results show that judge ideology's effects became stronger after Supreme Court's Tellabs decision in 2007.
Our study also contributes to a second literature stream-legal and political science-by showing new evidence of judge ideology's effects in securities class action lawsuits. First, unlike prior legal studies, which examine how judges who preside over lawsuits affect case outcomes (e.g., Tate [1981], Sunstein, Schkade, and Ellman [2004]), our study is the first to show that judge ideology affects the occurrences of securities class action lawsuits. Second, the literature yields mixed results on whether judge ideology affects business and finance cases (Choi and Pritchard [2012], Choi, Gulati, andPosner [2013]). Our paper sheds light on this issue by highlighting the role of circuit court judge ideology and its dominance over district courts in securities class action lawsuits. Taken together, our results contribute to the theory of litigation by improving our understanding of judge ideology's effects in the judicial system.

Securities class action lawsuits and existing measures of litigation risk
Securities class action lawsuits generally involve violations of the Exchange Act, which gives shareholders the right to bring private actions in federal courts to recover damages sustained as a result of securities frauds. 9 These lawsuits are very costly for firms and their managers, who face the possibility of severe financial, reputational, and even criminal consequences if charged with misreporting (Romano [1991]). 10 Due to their importance, securities class action lawsuits have drawn attention not only from scholars, but also from practitioners, such as consulting firms, law firms, and auditors. For example, Cornerstone Research, NERA Economic Consulting, Berstein Litowitz Berger & Grossmann, and PwC each regularly put out publications summarizing trends in litigation filings and outcomes, analyzing recent court decisions, upcoming court cases, and judicial appointments. 11 Academic researchers are particularly interested in how securities litigation influences corporate decisions, such as financial reporting (e.g., Frankel, McNichols, and Wilson [1995], Skinner [1994,1997], Field, Lowry, and Shu [2005]). 12 The literature commonly uses industry membership and firm characteristics to proxy for litigation risk. These measures originate with Schipper's [1994a, 1994b] analyses, which are based on a sample of firms drawn from four industries with a high incidence of litigation (i.e., biotechnology, computers, electronics, and retail). Subsequent papers have used an indicator variable for whether a firm belongs to these industries as a proxy for high litigation risk (e.g., Matsumoto [2002], Jayaraman andMilbourn [2009], Brown andTucker [2011], Donelson et al. [2012]) or limited their analyses to firms in these industries (e.g., Ali and Kallapur 9 Most securities fraud allegations are brought pursuant to Section 10(b) of the Exchange Act, and the SEC Rule 10b-5. 10 Studies in accounting, finance, and law consistently find that in the post-PSLRA period, merits matter for both lawsuit incidence and outcomes. Specifically, firms are more likely to be sued when there are fraud indicators, such as intentional restatements, abnormal accruals, abnormal insider selling, and SEC enforcement action, and lawsuits with these fraud indicators are less likely to be dismissed and have larger settlements (Cox, Thomas and Kiku [2003], DuCharme, Malatesta and Sefcik [2004], Johnson, Nelson and Pritchard [2007], Cox, Thomas and Bai [2008], Hennes, Leone andMiller [2008], Klausner, Hegland, andGoforth [2013]). 11 In addition, popular press, such as newspapers and magazines, extensively discuss the impact of judicial appointments, including those to lower federal courts, on various topics and refer to these appointments as a President's "judicial legacy" (Toobin [2014], Scheindlin [2017], Berenson [2018]). 12 Other papers in the literature measure litigation risk using actual securities class action lawsuits filed against a firm (e.g., Rogers and Van Buskirk [2009]), legal regime changes, such as the passage of the PSLRA (e.g., Johnson, Kasznik, and Nelson [2001]), differences in legal environments across countries (Baginski, Hassell, and Kimbrough [2002]), and director and officer insurance premiums (Cao and Narayanamoorthy [2011]). Such measures rely on either a limited set of firms or a specific time period, and as such, their results may not be generalizable to broader samples.
The measures used in these studies, based on industry and firm characteristics, capture not only ex ante litigation risk, but also firms' underlying misconduct. This makes it difficult to attribute empirical findings to ex ante litigation risk. Industry and firm characteristics, such as information asymmetries, operational or financial riskiness, proprietary costs, and investor clientele, are likely to independently affect firm decisions. Using these variables to study how litigation risk affects corporate decisions thus creates a correlated omitted variable problem (Kim and Skinner [2012]). For example, although stock market variables provide significant explanatory power in predicting litigation (e.g., Rogers and Stocken [2005]), they are also related to firms' disclosure decisions (Ke, Huddart, and Petroni [2003]). Consequently, associations between a litigation risk measure constructed using stock market variables and voluntary disclosure cannot be interpreted as showing that litigation risk affects voluntary disclosure. Indeed, prompted by litigation risk's importance to firm decision making, recent studies have called for further exploration of ex ante measures of litigation risk (e.g., Kim and Skinner [2012]).

3.
Judge ideology and litigation risk

Judge ideology and how it affects judicial votes
Judge ideology denotes the extent to which judges are liberal or conservative (Goldman [1966], Segal and Spaeth [1996], Spaeth [2006]). In this section, we summarize the literature addressing the influence of judge ideology on judicial decisions and discuss how securities class action lawsuits are filed in the federal judicial system.
Judges in all federal courts are appointed by United States presidents with the approval of the U.S. Senate. Presidents almost always appoint judges with an ideology similar to their own party's; as such, judicial outcomes tend to be aligned with a president's own policy preferences (Goldman [1999], Scherer [2005], Dorsen [2006], Federal Judicial Center [2006). 13 Once appointed, federal judges almost always hold office for as long as they wish and their salaries are guaranteed by the U.S.
Constitution (Federal Judicial Center [2006]). Lifetime tenure and salary protection are meant to ensure an independent judiciary but also give judges significant latitude to vote according to their ideological preference.
It is well documented in the political science literature that judge ideology influences judicial votes in federal courts, with judges appointed by Democratic presidents being more liberal than those appointed by Republican presidents. Although legal doctrine plays an important role in deciding case outcomes, judges are more likely to obey (disobey) legal doctrine when such doctrine supports (does not support) their own partisan or ideological policy preferences (Cross and Tiller [1998]). For example, Cross and Tiller [1998] show consistent evidence that judge panels controlled by Democratic appointees are more likely to produce liberal decisions than those controlled by Republican appointees. In a metaanalysis of 84 studies in the legal literature between 1959 and 1998, Pinello [1999] concludes that judge partisanship contributes to almost one half of the variance in judicial actions. Similarly, Sunstein, Schkade, and Ellman [2004] document that circuit court judges' votes are explained by the political party of their appointing presidents. Specifically, they find that a panel composed of only Democratic appointees is 80% more likely to issue a liberal ruling than a panel composed of only Republican appointees. In economic cases, liberal ideology leads to outcomes that are anti-business or proeconomic underdog (Goldman [1966], Segal and Spaeth [1996], Spaeth [2006], Fedderke and Ventoruzzo [2016]). 14 In another study, Tate [1981] finds that an appointing president's party affiliation is the most important factor in explaining Supreme Court justices' votes on economic issues in 13 Note that while a president's judicial nominations may be weakened by a Senate controlled by the opposite party or by a "lame duck" status, he or she remains the most powerful player in the judge selection process (Segal [1987], Ruckman [1993], Guliuzza, Reagan, and Barrett [1994]). If the president must compromise when filling a vacated seat and nominates someone whose ideology is closer to the center of the ideological spectrum, this may introduce noise into our measure and bias against finding significant results. In Section 5.4, we use a measure of judge ideology that incorporates the political composition of the Senate at the time of judicial appointment and find similar results. 14 Liberal outcomes in economic cases may also be pro-competition, pro-indigent, pro-small business vis-à-vis large business, pro-debtor, pro-bankrupt, pro-Native American, pro-environmental protection, and pro-consumer, and, for issues pertaining to federal taxation, pro-United States (Spaeth [2006]).
Given how important judge ideology is to case outcomes, it is not surprising that plaintiffs take judge ideology into consideration when deciding whether to file a lawsuit. Using a sample of Federal Communications Commission (FCC) orders and trials, de Figueiredo [2005] finds that firms are more likely to appeal their cases when the circuit judges' ideology is distant from that of the FCC executives, that is, when circuit judges are more likely to overturn the FCC's decisions. This evidence suggests that the expectation of an ideological effect on judicial voting impacts the incidence of lawsuit filings, consistent with plaintiffs attempting to take advantage of a court's ideological makeup to achieve a desired ruling.
Although the literature provides ample theoretical and empirical research into the role of judge ideology in civil rights or liberties cases, it is less clear whether judge ideology plays a significant role in the business and finance subset of economic cases. 16 Prior studies recognize a divergence between how political preferences affect civil rights cases versus business and finance cases, and debate whether judges exhibit political or ideological preferences when it comes to the latter (Richards [2001]). 17 For example, Schneider [2001] argues that even if judges have political preferences in such cases, they are sufficiently weak such that other factors, such as the legal process, may work to neutralize them. This view is supported by the mixed empirical evidence on the influence of judge ideology in individual income tax and bankruptcy cases (Staudt, Epstein, and Wiedenbeck [2006], Nash and Pardo [2012]).
Similarly, previous research finds mixed evidence for judge ideology's role in securities class action lawsuit rulings. For example, Choi and Pritchard [2012] find no difference between Republican and Democratic district court judges' dismissal rates prior to the Supreme Court decision in the Tellabs case; 15 Personal attributes examined in Tate [1981] include birth, upbringing, education, experience, elective office, age, tenure, and partisanship. 16 Economic cases include civil liberties (e.g., labor and government regulation) and business and finance cases (e.g., securities and anti-trust) (Sullivan and Thompson [2004]). 17 We do note, however, that U.S. presidents usually express a consistent attitude toward social liberalism and economic liberalism. For example, Segal, Timpone, and Howard [2000] find correlations of 0.89 (0.93) between economic (social) liberalism and the presidential party. however, after the Tellabs decision, Republican judges were found to dismiss cases on scienter grounds at a higher rate than their Democratic counterparts. Choi, Gulati, and Posner [2013], in contrast, find that district court judge ideology does not affect case dismissal decisions. By coding a selected group of 48 Supreme Court's securities regulation decisions from 1936 through 2011, Fedderke and Ventoruzzo [2016] find that conservative justices are more likely to vote pro-business (58% of cases) than liberal justices (40% of cases). Within the context of this literature, our study provides additional insights by documenting a large sample of evidence for judge ideology's effects in a specific type of business and finance case, securities class action lawsuits.

How securities class action lawsuits are filed in the federal judicial system
In the U.S. judicial system, securities class action lawsuits filed in the federal court system may go through three levels: district courts (trial courts), circuit courts (courts of appeals), and the Supreme Court (the final level of appeals). 18 The plaintiff first files a complaint against the defendant in one of 94 district courts. After a case is filed, the district court randomly assigns it to one of its judges, such that judges in the same court have roughly equal caseloads (Federal Judicial Center [2014]). There are three potential outcomes for securities class action lawsuits in the district court: dismissal, out-of-court settlement, or trial. The vast majority of cases are either dismissed or settled out of court. For instance, Choi, Gulati, and Posner [2013] show that in their sample of securities class action lawsuits filed between 2003 and mid-2007, roughly half were dismissed (with voluntary dismissals at 11% and dismissals with prejudice at 38%), while the remaining half were settled out of court. Only eight lawsuits (1.6% of their sample) resulted in a trial verdict or a summary judgment.
If a judge in a district court dismisses a securities class action with prejudice, the plaintiff can 18 Immediately following the enactment of the PSLRA in 1995, plaintiffs' lawyers attempted to circumvent the heightened pleading requirement in the federal court by filing state securities class action lawsuits or parallel federal and state securities class action lawsuits (Walker, Levine, and Pritchard [1997], Martin et al. [1999]). The U.S. Congress responded by passing the Securities Litigation Uniform Standards Act (SLUSA) of 1998 (signed into law on November 3, 1998), thereby closing this loophole (Levine and Pritchard [1998]). Johnson [2012] confirms that after SLUSA was enacted, fewer securities class action lawsuits alleging 10b-5 violations were filed in state courts.
appeal the case to a circuit court. 19 There are 12 circuits in the U.S. that have jurisdiction over securities class action lawsuits, comprising 11 numbered circuits (from the 1 st to the 11 th Circuit) and the D.C.
Circuit. These 12 circuits are geographically defined (see Figure 2 for details). They have from 6 (1 st Circuit) to 29 (9 th Circuit) authorized judges, depending on the total area and population within their jurisdiction. After a case is appealed to a circuit court, it is assigned to a panel of three randomly selected judges. The panel then decides the case based on a majority opinion (i.e., at least two of the three judges on the panel must agree with the decision) and either affirms or reverses the lower court decision. In the set of circuit court cases that are terminated on their merits during the period from 2001 to 2017 (excluding cases voluntarily withdrawn or settled), 13% of the "Other Private Civil" cases, which include securities class action lawsuits, were reversed (data from the Administrative Office of the U.S.

Courts).
[Please insert Figure 2 here] After the circuit court decides a case, the losing party may file a petition requesting the Supreme Court to review it. Unlike the circuit court, however, the Supreme Court does not have to hear every case it is asked to review. Typically, it will agree to hear a case only when the case involves an unusually important legal principle or when two or more circuit courts have interpreted a law differently.
Of the three levels of federal court (i.e., the Supreme Court, circuit courts, and district courts), we choose to measure the ideology of circuit courts for three reasons. First, the Supreme Court, the highest federal court, makes only discretionary and infrequent reviews, that is, fewer than 1% of the more than 10,000 review requests that the Supreme Court receives from circuit courts each year (Bowie and Songer [2009]). In fact, research suggests that for securities class action lawsuits, monitoring by the Supreme Court is close to nonexistent (Pritchard [2011], Choi and Pritchard [2012]). Therefore, circuit courts and district courts generally serve as the final arbiters for the vast majority of, if not all, securities class action lawsuits, and thus have the greatest effect on case outcomes (Howard Jr. [1981], Songer, Segal, and Cameron [1994], Klein [2002], Choi and Pritchard [2012]). 20 Second, district court judge decisions are subject to mandatory and routine reviews by circuit courts. During a review, a circuit court can rule that a district court judge's decision is incorrect and reverse it. Following a reversal, the district court judge must hear new motions and conduct a new trial, thus increasing his or her workload. Reversals embarrass district court judges, damage their reputation, and reduce their chances of being appointed to a higher court. Such concerns are limited for circuit court judges, however, as they lack opportunities for further promotion (Savchak, Hansford, and Songer [2006]). Because district court judges must consider the probability that their decisions will be reviewed and reversed by circuit courts, they have been found to take the political preference of circuit courts into consideration when making decisions (Schanzenbach and Tiller [2007], Knight and Gulati [2010]).
Indeed, consistent with this argument, prior studies show weak or no evidence that district court judges' ideology affects their dismissal decisions for securities class action lawsuits (Choi and Pritchard [2012], Choi, Gulati, and Posner [2013]). In fact, in civil liberty and economic cases, district court judge rulings reflect the ideological preferences of the circuit court judges (Randazzo [2008], Choi, Gulati, and Posner [2012]).
Third, our primary objective is to measure ex ante litigation risk, which is affected both directly and indirectly by the circuit courts. Circuit courts directly influence the cases that they handle, which are 22% of our sample cases. 21  process, both parties consider the probability that a case will be dismissed and appealed to the circuit court. Thus, it is likely that circuit court judge ideology plays an indirect but important role in these settled cases. Consistent with this argument, we find that if securities class action lawsuits filed in more liberal circuits are settled, they tend to be settled for larger sums than those filed in less liberal circuits (see details in Section 5.8). Thus, when rational shareholders and plaintiff lawyers make filing decisions, they are likely to assess the potential influence of the circuit judge ideology, that is, the likelihood that the case will reach the circuit court and how friendly its judges would be if that were to happen.
Note that shareholders may file lawsuits in state courts alleging breaches of state fiduciary duty by corporate directors and officers (Badawi [2013]). However, we focus our study at the federal level for several reasons. First, lawsuits filed at the state level are comprised almost entirely of lawsuits challenging an acquisition or alleging self-dealing transactions, largely focusing on the duties and liabilities of directors and not those of officers. Indeed, Thompson and Thomas [2004] find that 88% or 808 out of 916 state-level cases filed in Delaware in 1999 and 2000 allege a fiduciary duty breach resulting from an agreement to sell the company for too low of a price. They also carry less severe penalties and are of diminishing importance (Thompson and Sale [2003], Thompson and Thomas [2004]). For example, nearly every acquisition in 2011 (95%) was accompanied by a state-filed lawsuit, the vast majority of which resulted in only additional disclosure about the merger to shareholders and a payment of a relatively small fee to the plaintiffs' lawyers (Cain and Davidoff [2012], Davidoff, Fisch, and Griffth [2015]).
Similarly, state derivatives lawsuits have been increasingly limited by a variety of procedural and substantive restrictions. For example, the prevalence of independent directors following NYSE and NASDAQ requirements makes it harder to establish conflicts of interest (Thompson and Thomas [2004] the political affiliation of the appointing president (Goldman [1999], Pinello [1999], Sunstein, Schkade, and Ellman [2004]). This measure is both objective and easy to implement. To identify each judge's appointing president, we obtain biographical data of circuit court judges from the Federal Judicial Center's website. Specifically, judge ideology in a circuit is calculated as the probability that Democratic presidents' appointees dominate a panel of three judges randomly selected from the circuit ( ), calculated using the following formula: To measure ex ante litigation risk at the firm-year level, we assign each firm-year observation to a circuit-month. The assignment is based on the location of the firm's headquarters at the beginning of the year, as civil procedure usually requires securities class action lawsuits to be filed in the circuit where the firm's headquarters are located (hereafter, the home circuit). 24 Because Compustat reports only the current headquarters of firms, we use PERL to extract the locations of firms' historical 22 In a sensitivity test, we use the percentage of judges appointed by Democratic presidents in a given circuit to measure the circuit court ideology. Our results are similar using this alternative proxy. 23 We include both active and senior judges in the calculation of judge ideology. For details on the difference between active and senior judges, please refer to footnote 26. 24 As discussed in Cox, Thomas, and Bai [2009] page 428, "it is usually required to file in the home circuit of the defendant firm because of the well-established doctrine of forum non conveniens (now subsumed in the statutory change of venue provision in 28 U.S.C. § 1404) available to defendants who believe the plaintiff's initial choice of forum poses substantial burdens on them." headquarters from their 10-K filings, downloaded from the SEC's Edgar database. 25 4.2 Descriptive statistics of the judge ideology measure Figure 2 displays the geographic boundaries of the 12 circuits. We report the judge ideology ( ) for each circuit and year in Table 1. As an illustration of the cross-sectional and timeseries variations in judge ideology, Figure 3 plots for the two most liberal circuits (the 9 th and 2 nd Circuits), the two most conservative circuits (the 7 th and 8 th Circuits), and the 4 th Circuit over time. Judge ideology varies greatly across circuits. In the 9 th Circuit, the most liberal circuit during the sample period, from 1996 to 2014, an average of 56.6% of judges were appointed by Democratic presidents; this translates into a mean of 0.599, that is, a 59.9% chance that a three-judge panel randomly drawn from the 9 th Circuit is dominated by liberal judges. At the other extreme, only 29.2% of judges in the 7 th Circuit, the most conservative circuit during the sample period, were appointed by Democratic presidents, with a mean of 0.187. There are also substantial time-series variations in judge ideology due to judge appointments and departures. A president can appoint a judge to a court when there is a vacant seat, which is created when an active judge dies, resigns, or retires. 26 A court becomes more liberal when Republican appointed judges, whether active or senior, leave the court (i.e., dies, resigns or fully retires), or when Democratic Presidents appoints new judges to fill vacant seats, and vice versa. Furthermore, because judge departures and appointments occur at different points of time across circuits, the circuits' judge ideology levels do not always move in the same magnitude or even in the same direction. For example, Figure 3 shows that during 2006 to 2012, the 7 th and 8 th Circuits became more conservative while the 1 st , 2 nd , and 9 th Circuits became more liberal.
The interaction of exogenous cross-sectional and time-series variations in judge ideology makes it 25 We assume that firms' choice of headquarters locations in the year are exogenous to judge ideology. Our findings are robust if we limit our sample to firms that have never changed their headquarters location. 26 Circuit and district court judges are eligible to retire after they are 65 years old and the combination of their age and years of service totals at least 80 (the Rule of 80, 28 U.S.C § 371). These judges can either fully retire and leave the bench, or take senior status and continue to serve in the courts. Judges usually take senior status soon after they become eligible (the median and average wait time is one and two years respectively) (Burbank, Plager, and Ablavsky [2012]). Most court vacancies are created when active judges take senior status. Anecdotal evidence suggests that judge's retirement choices are unlikely to be driven by the expected ideology of their replacements (Tables 26 and 28 Figure 2).

The effect of judge ideology on securities class action lawsuits
We conduct a number of empirical tests to validate judge ideology as a measure of litigation risk to examine its economic significance and to compare it with existing proxies of litigation risk.
Specifically, we investigate whether firms located in more liberal circuits are more likely to be sued, how the Supreme Court's Tellabs decision influences judge ideology's effects, and the cross-sectional variation in such effects. year observations. 27 The unconditional probability that a firm-year in our testing sample is in a securities class action lawsuit's class period is 4.56% (4,183 / 91,698).

Univariate analysis
[Please insert Table 2 here] Table 2 Panel B tabulates the number of lawsuits filed in each circuit. The largest circuit by population and area, the 9 th Circuit, covers the West Coast of the United States and contains 544 of the total 1,973 lawsuits in our sample period. In contrast, the smallest circuit by population and area, the D.C. Circuit, contains only 12 lawsuits over our sample period. We also report the firm-year distribution of class periods across circuits in Panel B. Firms located in the 9 th Circuit, which has the highest of all circuits (0.563), are 41% more likely to be sued (5.68% of its firm-years are involved in a lawsuit) than firms located in the 7 th Circuit (4.04% chance of being sued), which has the lowest of all circuits (0.186). This analysis provides a preliminary indication that the more liberal a circuit's judges are, the more likely firms located in that circuit are to be sued in securities class action lawsuits, which is consistent with judge ideology as a measure of firms' litigation risk.
To perform univariate tests of the relation between judge ideology and lawsuit occurrence, we sort our sample into five quintiles based on and report lawsuit occurrences across the quintiles in Table 3 Panel A. We find that firm-years in the top quintile of are 64% more likely to be sued (5.77%) than those in the bottom quintile (3.51%), a difference that is statistically significant at the 1% level.
[Please insert Table 3 here] Prior research establishes that firms in certain industries (e.g., biotechnology, computers, electronics, and retail, hereafter collectively referred to as FPS industries) have higher litigation risks than firms in other industries (Francis, Philbrick, and Schipper [1994b], Kim and Skinner [2012]). We investigate whether can improve our ability to predict litigation beyond predictions based on FPS industry membership. We report the results in Table 3 Panel B. First, our results confirm findings from prior research that firms in FPS industries have significantly higher litigation risk than those in non-FPS industries. Reported in the first column ("All Quintiles"), the probability that a firm- year observation in FPS (non-FPS) industries was involved in a securities class action lawsuit is 6.55% (3.81%). Next, we independently sort firm-year observations from FPS and non-FPS industries into five quintiles based on . If judge ideology measures an aspect of litigation risk that is incremental to that measured by , then, conditional on industry membership, we expect firm-years with a higher to have a higher litigation risk than those with a lower . This prediction is strongly supported by our results. Specifically, in non-FPS industries, the probability that a firm in the most liberal quintile is sued is 1.47% higher than its counterpart in the most conservative quintile (4.39% and 2.92%, respectively, difference significant at the 1% level). In FPS industries, the difference is even larger: a firm located in the most liberal quintile is 2.56% more likely to be sued than a firm located in the most conservative quintile (7.88% versus 5.32%, respectively, significant at the 1% level). This univariate analysis provides initial evidence that judge ideology measures a different aspect of litigation risk than is captured by industry membership.
Because prior studies also find that other firm characteristics predict securities class action lawsuits (Johnson, Kasznik, and Nelson [2001], Rogers and Stocken [2005] We define high (low) litigation firm-years as those with predicted litigation risk above (below) the sample median. Next, we independently sort firm-year observations in the high and low predicted litigation risk samples into five quintiles, each based on , and report the results in Table   3 Panel C. In short, we find results similar to those in Table 3 Panel B. In both the high and low predicted litigation risk samples, the probability that a firm in the most liberal circuits is sued is significantly higher than its counterpart in the most conservative circuits, suggesting that judge ideology measures an aspect of litigation risk that is not captured by firm and industry characteristics.

The effect of judge ideology on lawsuit occurrence
In this section, we further validate judge ideology as a measure of litigation risk by conducting a regression analysis. Following Johnson, Kasznik, and Nelson [2001], Rogers and Stocken [2005], and in each circuit changes only when there is a judge retirement or appointment, it is possible that the effect we observe for is driven by variations across circuits; this would suggest that the association between judge ideology and lawsuit filings may be driven by some correlated omitted variables at the circuit level. To mitigate this concern, we include circuit fixed effects in the regression.
We also control for other macroeconomic events and lawsuit filing trends using year fixed effects. Zstatistics (t-statistics) are based on standard errors clustered by states, as the primary variable of interest is based on geographic location. 28 [Please insert Table 4 here] We report the descriptive statistics of the variables in Table 4 Panel A. The mean (median) firm-year observation in our sample has a total market value of $1,918.6 ($227.2) million, a sales growth of 10.3% (3.4%) of total assets, and a market-adjusted return of 3.4% (-5.6%) in the prior year. Table 4 Panel B reports the Pearson and Spearman correlations among the variables. These results show that is significantly and positively correlated with both (Pearson correlation of 0.06) and (0.04), which is consistent with industry membership and judge ideology, respectively, predicting lawsuits. is also positively correlated with firm characteristics that prior studies document as significantly correlated with litigation risk, namely, size (0.14), sales growth (0.08), past returns (0.05), return volatility (0.05), and turnover (0.16) (Kim and Skinner [2012]).
[Please insert Table 5 here] In Table 5, we explore the relation between judge ideology and a set of variables, including firm characteristics that prior studies use to construct litigation risk measures, in addition to industry and market conditions and demographic variables. We report the regression results without fixed effects in Column (1)  [Please insert Table 6 here]  (1), we see that the estimated coefficient for is positive and significant at the two-tailed 1% level, which is consistent with plaintiffs taking judge ideology into account when making filing decisions. The effect of judge ideology is economically significant after controlling for firm, industry, market, and demographic characteristics. Reported at the bottom of the table, results further show that an increase in from Q1 to Q3 (from 0.247 to 0.500) implies an increase in the predicted odds of being sued by 33.5% in relative terms (from 2.45% to 3.27% in absolute terms). In comparison, an increase in from 0 to 1 leads to a 13.9% increase in the predicted odds of being sued (from 2.73% to 3.11%).
To illustrate the economic effect in a practical example, judge retirements, followed by President Obama's appointments of six judges to the 4 th Circuit, increased the chance of being sued for firms located in the circuit from 2.69% to 4.06% (or 50.9% in relative terms) between 2006 and 2014. All of our findings remain consistent when we replace circuit fixed effects with state fixed effects in column (2), and use a linear probability model in columns (3) and (4). Together with the results presented in 29 Note that during the Internet Bubble, the percentage of firms in California and Washington that were in FPS industries increased from 36.8% to 50.9%, three times the increase compared with other states (from 20.3% to 25.8%). During the same period, the judge ideology of the 9 th Circuit also increased twice as fast as other circuits (from 0.414 to 0.649 in the 9th Circuit and from 0.273 to 0.401 in other circuits). When we include in our main regression, we find robust results (tabulated in Internet Appendix Table IA8 Column 7). Table 3, the regression analysis shows that judge ideology identifies an aspect of litigation risk that is incremental to that explained by risk measures currently used in the literature.
The estimated coefficients of the control variables are generally consistent with findings in prior studies (e.g., Kim and Skinner [2012]). In particular, larger firms and firms with higher sales growth, higher beta, more volatile returns, higher share turnover, more institutional holdings, and larger negative minimum returns are more likely to be subjects of lawsuits. Firms in FPS industries and industries with more negative and volatile returns also have higher litigation risks. Finally, state GDP growth is significantly positive, suggesting that firms located in higher growth states are more likely to be sued.

Additional analysis on the economic impact of judge ideology on litigation risk
To further understand the economic significance of for firms' litigation risk, we re-estimate the logit model in Table 6, replacing circuit fixed effects with firm fixed effects to control for time-invariant firm characteristics (tabulated in column (1) of Internet Appendix Table IA5). Note that we exclude in this regression because of the inclusion of firm fixed effects. The estimated coefficient on (1.816, significant at the 5% level) shows that even after controlling for time-invariant firm characteristics, judge ideology still has a sizeable influence on firms' litigation risk.
For an average firm in this sample, the probability of being sued increases by a relative magnitude of 53.9% (from 12.5% to 19.3% in absolute terms) when increases from Q1 to Q3. 30, 31 The economic magnitude detected in this test is larger than for specifications with circuit and state fixed effects, suggesting that judge ideology has greater predictive power for firms already more likely to be subject to securities class action lawsuits.
Next, we investigate how much statistical explanatory power of adds to existing ex ante litigation risk models based on firm and industry characteristics (Johnson, Kasznik, and Nelson [2001], Rogers and Stocken [2005], Kim and Skinner [2012]) by comparing with that of the 30 This sample has a much higher unconditional probability of being sued, as it is comprised only of firms that have been sued during the sample period. 31 After controlling for firm fixed effects, we find that judge turnovers in the 4 th Circuit from 2006 to 2014 increased the chance of being sued for firms located there by 61.9% in relative terms (from 14.0% to 22.7%).
variable. 32 We use as a comparison benchmark for because both measures are parsimonious and more exogenous than other firm characteristics, which make them more desirable for studies requiring proxies of ex ante litigation risk. 33 To compare the performance of the prediction models, we calculate the models' area under the receiver operating characteristic (ROC) curve, or AUC, which measures the models' ability to discriminate (Hobson, Mayew, and Venkatachalam [2012], Kim and Skinner [2012], Larcker and Zakolyukina [2012]). In Internet Appendix Table IA6, the results in columns (1) to (3) show that both and increase the model's AUC (significant at the 1% and the 5% levels, respectively), but that the incremental information provided by is greater than that of ; that is, the AUC of column (2) is larger than that of column (3) (statistically significant at the 10% level). Columns (3) and (4) (1) and (2)). This suggests that including does not seem to affect judge ideology's improvement of the litigation prediction model, which is consistent with judge ideology as measuring an aspect of litigation risk that is orthogonal to the one measured by industry membership. 34 In summary, our results from these two sets of analyses show, first, that judge ideology provides incremental information that improves the ex ante litigation prediction models currently used in the literature, and second, that variation in ex ante litigation risk captured by judge ideology is economically significant, indicating the importance of political appointments for firms.

Robustness checks of the effect of judge ideology
We run a battery of sensitivity tests to ensure the robustness of our main results. First, we ensure that our results in Table 6 are not driven by certain subsamples. We report the tests in Internet Appendix 32 Note that this specification has fewer variables than the main specification in Table 5 because we want to show that judge ideology improves upon ex ante litigation risk measures used in prior studies, which usually do not include demographic variables, such as economic growth, unemployment rate, and political ideology. 33 Firm characteristics, such as firm size and stock return volatility, may be directly associated with operating decisions (Kim and Skinner [2012]). This creates a correlated omitted variable problem for studies requiring proxies for ex ante litigation risk. See Section 2 for a detailed discussion. 34 Our conclusion is unchanged if we replace the logit regression with a linear probability model and test the difference in adjusted R 2 s across specifications (reported in Internet Appendix Table IA7). Table IA8. We start by examining whether our results are driven by a particular circuit. Specifically, we focus on the 9 th and 2 nd Circuits, as the former includes 22% of the lawsuit firm-years in the sample and is the most liberal circuit in our sample, while the latter houses most of the IPO and foreign firmsrelated cases in our sample. In columns (1) and (2), we repeat our main test, but exclude the 9 th and 2 nd Circuits, respectively, and find that the estimated coefficients for are not affected. Next, we confirm that our main findings are consistent if we remove firm-year observations from finance industries (tabulated in column (3)) or cases involving IPO allocations, underwriters, analysts, or mutual funds (tabulated in column (4)). Finally, we run our analyses again, now excluding observations that occurred prior to the passage of the SLUSA on November 3, 1998, when plaintiffs had the ability to file state securities class action lawsuits (Walker, Levine and Pritchard [1997], Martin et al. [1999]).
Our results tabulated in column (5), remain similar.
Second, given that all federal judges must be confirmed by the Senate before being appointed, we consider the U.S. Senate's effect on the ideology of federal judge appointees. In this sensitivity test, we incorporate the partisan makeup of the Senate to estimate judge ideology and find consistent results (tabulated in Internet Appendix Table IA9).
Third, we explore the effect of district court judge ideology. As discussed in Section 4, we expect the effect of a district court's judge ideology on litigation risk to be dominated by that of the circuit court: district court judge ideology may matter, but only if it is consistent with that of the circuit court judge. Empirically, we measure district court judge ideology ( ) using the percentage of district court judges appointed by Democratic presidents, as district courts only assign one judge to each case.
To execute this test, we first re-estimate Eq.
(2), retaining all of the control variables but replacing with . The results, tabulated in Internet Appendix Table IA10 column (1), show a significantly positive coefficient for . After we include in our estimation, however, the coefficient for is no longer significant (column (2)). This suggests that district court judge ideology's effects are dominated by the judge Electronic copy available at: https://ssrn.com/abstract=3170842 ideology of the circuit court. 35 As a further confirmation, we partition the sample into two subsets based on whether the ideology of district court judges conforms to that of their circuit court judges (i.e., both are above or below their respective sample median). The results tabulated in Internet Appendix Table   IA11 show that is significantly positive only when the district court ideology conforms to the circuit court ideology. Thus, consistent with prior studies that examine other types of lawsuits (Randazzo [2008] making. In addition, the Tellabs ruling prescribes that circuit courts must assess "all the allegations holistically," conduct a "comparative assessment of plausible inferences," and consider all "competing inferences" when determining whether a given complaint "gives rise to a strong inference." Judges are 35 The coefficient of is also no longer significant if we include circuit and year fixed effects (tabulated in column (3) of Table IA10). 36 In Internet Appendix Table IA12, we examine whether judge ideology's effect varies with firms' political donations (Tahoun [2012]). Our main findings remain robust after including the firms' political donations and its interaction with . We also find that judge ideology's effect on litigation occurrence does not vary with firms' political donations. thus required to investigate all relevant information pertaining to the inference, from both within and outside of defendants' complaints, and to rely on their own judgment when evaluating competing inferences (Cox, Thomas, and Bai [2009]). As a result, the Tellabs ruling may allow judges a higher degree of discretion when deciding whether to issue a motion to dismiss (Miller [2009]); this, in turn, provides further room for presiding judges' individual perspectives to shape case outcomes (Cox, Thomas, and Bai [2009]). For these reasons, we expect the effect of ideology on securities class action lawsuits to be greater after the Tellabs ruling.
[Please insert Table 7 here] To test the effect of Tellabs on securities class action lawsuits, we re-run our analyses adding an indicator variable, , to denote whether the lawsuit filing date is after the Tellabs decision, (i.e., whether the plaintiffs have knowledge of the ruling). We then test its interaction with in the regression of the lawsuit occurrence test (Eq. (2)). We match each non-sued firm- year with a sued firm-year in the same industry and year, and with the closest predicted firm-specific litigation risk ( _ ). For sued firm-years, we use the actual lawsuit filing date to define . Following Rogers and Van Buskrik [2009], for non-sued firm-years, we use matched sued firm-years' lawsuit filing dates as pseudo filing dates to define . 37 Our findings, presented in Table 7 column (1) (2)), if we use linear probability 37 We include year fixed effects in the regression to control for macroeconomic events and lawsuit filing trends.
Year fixed effects are defined based on the fiscal year of the firm-year observation, while is defined based on the actual or pseudo lawsuit filing dates. In an un-tabulated sensitivity test, we exclude year fixed effects and find similar results. 38 For brevity, we report the coefficient estimates of the control variables in the Internet Appendix Table IA13. models (in columns (3) and (4)), or if we include interactions of with our control variables (Internet Appendix Table IA13). 39

The effect of judge ideology-cross-sectional tests
To investigate circumstances where judge ideology is more important, we conduct three crosssectional tests. Because the effect of judge ideology on plaintiffs' filing decisions depends on the plaintiffs' ability and incentive to consider the implications of judge ideology for lawsuit outcomes, we expect judge ideology to have a stronger effect when shareholders are more sophisticated or when the expected payoffs from the lawsuits are higher.
We first measure a potential plaintiff's sophistication using the proportion of institutional investors. The PSLRA's lead plaintiff provision dictates that the most adequate plaintiff, that is, the one with the largest financial interest in the case, shall be the lead plaintiff and will select and retain counsel to represent the class (Choi [2011]). Institutional lead plaintiffs are sophisticated players in securities class action litigations. They have resources to supervise a case more closely (Weiss and Beckerman [1995], Perino [2003Perino [ , 2012). Cox, Thomas, and Kiku [2006] and Cheng et al. [2010], for example, find that cases with an institutional lead plaintiff are less likely to be dismissed and have a significantly larger settlement after controlling for a variety of factors. With greater sophistication, more experience in securities class action lawsuits, and access to resources, institutional investors are likely better able to incorporate judge ideology information into their decision-making process than individual investors.
Therefore, we expect judge ideology to exhibit a stronger influence on litigation occurrence for firms with higher institutional ownership.
We next use the expected payoffs from a lawsuit to measure plaintiffs' incentives to take judge ideology into consideration. Here, we expect that the higher the expected payoffs, the more likely plaintiffs are to consider judge ideology in their filing decisions. As proxies for the expected payoffs 39 We also use one-to-one matched sample, in which each sued firm-year is matched to a non-sued firm year, to test the effect of the Tellabs decision. The results are consistent with circuit court judge ideology having a stronger effect after the Tellabs case (reported in Internet Appendix Table IA14). Note that using this sample, the main effect of is positive but not statistically significant, probably owing to the small sample size (less than 10% of the full sample), which reduces the power of the test. from the lawsuits, we use firm size and predicted litigation risk, measured using firm characteristics.
First, because larger firms are better able to pay a settlement, they are more likely to settle a case, which increases lawsuits' expected payoffs (House Conference Report [1995], Choi and Pritchard [2012]). This is consistent with the "deep-pocket strategy" utilized by plaintiff lawyers, where lawsuits, even completely meritless ones, are routinely filed against firms with deep pockets whenever there is a significant drop in their stock prices (Kasznik and Lev [1995], Skinner [1997]). Given the higher expected payoffs of litigation against larger companies, we expect plaintiffs will be more likely to assess judge ideology when deciding whether to file lawsuits against these firms. In addition to firm size, we use a composite measure of predicted firm-specific litigation risk as a proxy for lawsuits' expected payoff ( _ ). Previous research shows that shareholders expect to receive higher payoffs from lawsuits against firms with higher predicted litigation risk (Finnerty and Pushner [2002]). Similar to firm size, we expect that shareholders are more likely to consider judge ideology for these firms.
[Please insert Table 8 here] We estimate Eq.
(2) separately for subsamples based on the median value of these three measures and report the results in Table 8. In the first cross-sectional test, we partition the sample firms by their institutional holdings. We expect judge ideology to have a larger effect for firms with higher institutional holdings. The results show that the coefficient for is higher for high institutional ownership firms (column (1)) than for low institutional ownership firms (column (2)) and is significant at the 5% level. Next, we partition firms based on firm size, and expect the effect of judge ideology to be stronger for larger firms. The results show that is positive and significant (1.419, z-stat of 4.84) for larger firms (column (3)) but not significant (0.282, z-stat of 0.80) for smaller ones (column (4)). The Chi-square test result shows that the difference is significant at the 1% level.
Lastly, we partition our sample firms based on high and low predicted firm-specific litigation risk. We expect the effect of judge ideology to be stronger for firms with high litigation risk due to the higher expected payoffs from these firms. The results from this test show that the estimated coefficient for is positive and significant only for firms with high litigation risk (column (5)), but not for those with low litigation risk (column (6)), with the difference between columns significant at the 1% level. Taken together, the cross-sectional tests presented in Table 8 highlight the conditions under which judge ideology has a stronger effect, providing further support for its ability to measure ex ante litigation risk.

Judicial appointment test
To further validate judge ideology as a measure of litigation risk and to demonstrate the effect of political appointments, we explore how judicial appointments affect the likelihood of lawsuit filings and whether investors understand its implications. To test this empirically, we focus on firms located in circuits that experience new judicial appointments. The average time between the departure of one judge and the confirmation of a new judge is about two and a half years (Jennewine [2016]). The official judicial appointment is thus the event that marks the final confirmation of the change in judge ideology. Table 9 Panel A shows the descriptive statistics for judicial appointments over time and across circuits.
Altogether, there were 141 judicial appointments in our sample period from 1996 to 2014, including 82 Democratic and 59 Republican appointees.
[Please insert Table 9 here] First, to verify the effect of judicial appointments on lawsuit filings, we compare the likelihood that lawsuit filings occur before and after judicial appointments. We expect an increase in lawsuit filings in a given circuit after a Democratic judicial appointment and a decrease after a Republican judge is appointed. Specifically, we compare the percentages of firms sued in the two years before and the two years after the judicial appointment dates, respectively. The results in Table 9 Panel B show that a firm's lawsuit filing rate increases by 0.81% after a Democratic appointee arrives, translating to a 59.6% increase in the unconditional probability of being sued. In contrast, a firm's likelihood of being sued decreases by 0.77% following a Republican judge's appointment, a 36.3% decrease in the unconditional probability of being sued. Both changes are statistically significant at the 1% level.
Next, as securities class action lawsuits are costly to firms (Romano [1991]), we examine market reactions surrounding judicial appointment dates and expect that the market will react positively (negatively) to a Republican (Democratic) judicial appointments. Table 9 Panel C reports the mean and median three-day market-adjusted returns ( ) surrounding judicial appointment dates for firms located in the corresponding circuit. The results show that stock returns are significantly negative (mean and median of -0.261% and -0.263%, respectively) for firms affected by a Democratic appointment and significantly positive (mean and median of 0.272% and 0.045%, respectively) for those affected by a Republican appointment. This is consistent with the market pricing the influence of judge ideology on litigation risk.
Because judicial appointments can affect litigation risks not only for securities class action lawsuits but also for other types of lawsuits, we conduct an additional test to ensure that market reactions following judicial appointment dates are driven mainly by changes in securities class action lawsuit risk.
To do so, we regress on an indicator variable for whether the new judge is appointed by a Democratic President ( ) and its interaction with , controlling for size, book-to-market, beta, and momentum (Larcker, Ormazabal, and Taylor [2011]). We expect our findings to be stronger for firms in high litigation risk industries (Johnson, Kasznik, and Nelson [2000]). Our findings (tabulated in Internet Appendix Table IA15) show that although firms experience a negative market reaction to Democratic judge appointments in general (columns (1) and (3)), the magnitude is substantially larger for firms in high litigation risk industries ( • significant at the 5% levels). This suggests that our earlier findings are driven specifically by the effect of judge ideology on securities class action lawsuits.

The effect of judge ideology on lawsuit outcome
We next test the association between circuit court judge ideology and securities class action lawsuits' outcomes for two reasons. First, in proposing judge ideology as a measure of litigation risk, we draw on the political science literature to argue that judge ideology influences securities class action lawsuit rulings. In this section, we provide empirical evidence to corroborate our argument. Second, the test measures judge ideology's effect on the costs of litigation. Compared with dismissed lawsuits, settled lawsuits can cost firms much more due to lengthy discovery periods, expensive legal fees (Klausner, Hegland, and Goforth [2013]), increased director and officer insurance premiums (Donelson and Yust [2017]), and, of course, the settlement payout itself.
Our findings, discussed in detail in Internet Appendix I and tabulated in Internet Appendix Table IA16, indicate that securities class action lawsuits filed in more liberal circuits are less likely to be dismissed and, for those not dismissed, to be settled for larger amounts. The effect is both statistically and economically significant. An increase in from Q1 to Q3 results in a 14.6% decrease in the odds of cases being dismissed (from 34.8% to 29.7% in absolute terms), and for cases that are not dismissed, an average increase of $907,274 in the settlement amount. The significant magnitude not only confirms that judge ideology influences judicial decisions, but also suggests that it serves as an important driver of litigation costs. 40

Litigation risk and firm disclosures
We now test the effect of judge ideology in a specific economic problem: how ex ante litigation risk affects firms' voluntary disclosure decisions. On the one hand, prior studies have conflicting views on the preemption effect of litigation risk, that is, litigation risk motivates firms to disclose bad news promptly. Skinner [1994] and Marinovic and Varas [2016], for example, argue that firms have an incentive to warn investors of bad news to reduce the likelihood of litigation and the amount of potential settlements. Several other studies document evidence consistent with this argument (Kasznik and Lev [1995], Field, Lowry, and Shu [2005]). However, voluntary disclosure of bad news can also contribute to shareholder lawsuits (Francis, Philbrick and Schipper [1994b], Skinner [1997], Bliss, Partnoy, and Furchtgott [2018]). Consistent with the latter view, Johnson, Kasznik, and Nelson (2001) find an increase in earnings warnings following the passage of the PSLRA in 1995, which reduced firms' litigation exposure.
On the other hand, the literature also argues that litigation risk has a chilling effect on disclosure, that is, litigation risk may deter firms, in general, from making voluntary disclosures because plaintiffs in a potential lawsuit can use information provided in the disclosure as a rationale for bringing suit. 40 Note that this magnitude may understate judge ideology's effect on actual lawsuit outcomes for two reasons. First, the lawsuit outcome tests use all securities class action lawsuits as samples, including cases that ended in district courts and did not reach circuit courts, for which circuit court judge ideology only has an indirect effect. Second, because we are interested in whether ex ante litigation risk affects case outcomes, we use , the ex ante expected judge ideology in our tests, which may be different from actual judge ideology after the case is filed to the circuit. As such, is a noisy measure of actual judge ideology and may understate judge ideology's true impact. Rogers and Van Buskirk [2009], for example, find that firms reduce voluntary disclosures after being the target of actual litigation. They interpret this as an indication that firms perceive plaintiffs to use disclosures as a pretext for litigation. To examine whether managers understand judge ideology's effect on litigation risk, and to apply our measure of litigation risk to the voluntary disclosure literature, we estimate the following regression: where is the natural log of one plus the number of earnings guidance. We use earnings guidance as a measure of voluntary disclosure because we can accurately measure its timeliness, that is, the future fiscal period to which the disclosure is referring. In addition to controlling for firm characteristics and demographic variables, we control for earnings news ( ), earnings decreases ( ), EPS change ( ℎ ), earnings to price ratio ( ), loss ( ), and past earnings guidance ( ). Following prior disclosure studies (e.g., Ajinkya, Bhojraj, and Sengupta To study the preemption effect, we test whether firms facing negative earnings news in more liberal circuits are more likely to disclose quarterly earnings warnings before the forthcoming earnings announcements. 42 We present the results in Panel A of Table 10. 43 In specifications with circuit and firm fixed effects (Columns (1) to (4)), we find that the coefficients for are positive and significant. This is consistent with firms in more liberal circuits facing higher litigation risk and being 41 Our finding is similar if we replace circuit fixed effects with state fixed effects (tabulated in Internet Appendix Table IA17). 42 Compared with annual forecasts, quarterly forecasts have shorter horizons, and are thus more likely to be used for the "preemption" of negative earnings news (Skinner [1994], Baginski, Hassell, and Kimbrough [2002], Rogers and Stocken [2005]). 43 For brevity, we report the coefficient estimates of the control variables in Internet Appendix Table IA17. more likely to issue earnings warnings to pre-empt an adverse earnings outcome. However, while the coefficients for and _ are positive and significant when circuit fixed effects are included (in Columns (1) and (2)), as is consistent with prior findings (Francis, Philbrick, and Schipper [1994b], Skinner [1997], Field, Lowry, and Shu [2005]), both turn insignificant when firm fixed effects are included (in Columns (3) and (4), and (7) and (8)). Lastly, while is positively significant in Columns (1) to (4), when year fixed effects are included, it is also no longer statistically significant as a predictor of earnings warnings. This finding is consistent with the ambiguity of the net legal cost of earnings warnings, that is, whether it increases or decreases the expected litigation costs (Leuz and Wysocki [2016]). Another possibility is that firms bundle earnings warnings with positive information or noise to lower expected litigation costs (Bliss, Partnoy, and Furchogott [2018]). Further research may be needed to disentangle the various effects.
[Please insert Table 10 here] To examine whether our results support a chilling effect, we measure a firm's tendency to issue upward annual forecasts relative to the analyst consensus prior to the fourth quarter and perform the test on the same firm-year sample as the preemption test. In this case, we expect the chilling effect to be more pronounced for positive, longer-horizon forecasts, as these are less likely to be used to preempt forthcoming negative earnings surprises, and managers still bear litigation costs for issuing biased forecasts (Rogers and Stocken [2005]). Consistent with this prediction, Baginski, Hassell, and Kimbrough [2002] show that Canadian firms, which face a less litigious environment than U.S. firms, issue more annual forecasts than U.S. firms but not more interim forecasts. Similarly, Huang, Shen, and Zang [2018] find that when firms reduce their litigation risk by increasing disclosure of meaningful cautionary statements, they are more likely to issue annual earnings forecasts. Our results of negative coefficient for in Table 10 Panel B suggest that litigation risk reduces managers' incentive to issue positive long-horizon forecasts. The results also indicate that managers are aware of judge ideology's effect on litigation risk in making decisions.
However, unlike the results in the preemption test, we find that in this test, _ is not significant and is significant in the direction opposite to the chilling effect (in Columns (1) and (5) when firm fixed effects are not included). This suggests that in this setting, judge ideology is a superior measure of litigation risk compared to existing measures. This result is intuitive, as litigation risk associated with the chilling effect is more closely related to the legal system, which can be captured by judge ideology. In particular, the PSLRA, which was designed to curb frivolous lawsuits, provides companies with a safe harbor for forward-looking statements made in good faith and accompanied by "meaningful cautionary statements." However, the PSLRA does not clearly specify what qualifies as a meaningful cautionary statement. This ambiguity leaves room for judges to exert discretion in their judicial decisions. 44 Within this context, the presence of a liberal judge increases litigation risk, as he or she will be more likely to take a stricter view of whether a cautionary statement constitutes a "meaningful cautionary statements" as defined in the PSLRA, and of whether such statements can protect a firm from disclosure-related securities class action lawsuits. Thus, compared with firm and industry characteristics, judge ideology is a more direct and powerful measure of litigation risk for the chilling effect.
[Please insert Table 11 here] Finally, as a placebo test, we investigate the relation between judge ideology and firms' mandatory disclosures. Because issuing mandatory disclosures is not a choice for firms, we do not expect it to be related to judge ideology. We separately examine firms' positive and negative mandatory disclosures such that the results are comparable with the management guidance test. 45 To measure the number of positive and negative mandatory disclosures that a firm makes in a year, we count their 8-K filings that do not contain voluntary items 2.02, 7.01, and 8.01, and with positive or negative market reactions, respectively (Cooper, He, and Plumlee [2016]). As expected, the results in Table 11 Panels A and B show no significant relation between judge ideology and mandatory disclosures, regardless of earnings' directionality.
44 Judge Easterbrook points out in Asher v. Baxter that "the fundamental problem is that the statutory requirement of 'meaningful cautionary statements' is not itself meaningful" (Easterbrook [2004]). 45 In a sensitivity test, we use firms' overall mandatory disclosure and find similar results.
Taken together, the results for our tests of management forecasts and mandatory disclosures suggest not only that judge ideology is a valid measure of ex ante litigation risk and that enables researchers to draw inferences beyond existing measures based on industry membership and firm characteristics; it also indicates the importance of political appointments to corporate decision making.

Conclusions
In this study, we introduce a new ex ante litigation risk measure, federal judge ideology, and examine its economic consequences for firms. Ideology is well documented in the political science literature as one of the most important determinants of judges' voting behavior. We measure judge ideology using the partisanship of a judge's appointing president. In securities class action litigation, judges appointed by Democratic presidents are more liberal and tend to side with shareholders, while judges appointed by Republican presidents are more conservative and tend to favor firms in their decisions. Thus, firms have a higher litigation risk if they are located in a circuit with a greater proportion of judges appointed by Democratic presidents.
We conduct three sets of empirical tests to validate the measure and demonstrate its economic consequences. First, we test whether judge ideology predicts the occurrence of lawsuits. We find strong evidence that judge ideology is economically meaningful in predicting securities class action lawsuit filings and that it improves upon the ex ante litigation prediction models currently used in the literature.
Firms in more liberal circuits are more likely to be sued. We also find that the effect of judge ideology is stronger after the Supreme Court's Tellabs ruling, which is consistent with this precedent providing more room for judges to exercise discretion in their rulings. Second, as evidenced in the economic consequence of judicial appointments, we find that the lawsuit filing rate increases (decreases) after the appointments of liberal (conservative) judges and that the market reacts accordingly. Third, we find that lawsuits filed in more liberal circuits are less likely to be dismissed and more likely to settle in larger amounts. Lastly, we use managers' voluntary disclosure decisions as a specific economic problem and show that our new measure can provide new insights that litigation risk measures based on firm and industry characteristic cannot.
Electronic copy available at: https://ssrn.com/abstract=3170842 Our study has implications for empirical research related to ex ante litigation risk. The judge ideology measure will enable researchers to isolate the effects of ex ante litigation risk attributable to the legal system and to study how this affects firms' economic decisions. This will also be useful for both regulators and managers by improving their understanding of regulations' practical implications.
In addition, our evidence for the occurrence of private securities class action lawsuits provides insights into the implications of the political theory of decision making for securities class action lawsuits, and highlights the economic consequences of political appointments for firms.
We acknowledge two caveats of the paper. First, judge ideology is based on geographic location of the firm's headquarter and is affected by shifting trends in the political environment. As such, depending on the research questions, future studies using the measure may need to control for these environmental variables in the tests. Second, judge ideology in our study is based on federal circuit courts. As such, our measure of litigation risk is limited to lawsuits alleging violations of federal laws and cannot provide insights to issues in state courts' jurisdiction.
This appendix reports the logit regression results on the relation between the securities class action lawsuits occurrence and industry and firm characteristics. We estimate the logistic model = (  ,  ,  ,  ,  ,  ,

Judge ideology measure
The probability that a three-judge panel randomly selected from a circuit court has at least two judges appointed by Democratic presidents, that is, is a binomial coefficient indicating the number of possible combinations of objects from a set of distinct objects, is the number of Democratic appointees in the circuit, and is the total number of judges in the circuit. We measure at the end of each month and assign each firm-year observation to a circuit court-month based on the firm's headquarters at the beginning of the year. Historical headquarters information is extracted from firms' 10-K filings. Circuit court judges' appointing presidents are obtained from the Federal Judicial Center's website;

Variables used in litigation risk validity tests
an indicator variable that equals one if the firm-year overlaps with the class period of a securities class action lawsuit, and zero otherwise. Securities class action lawsuits are obtained from the SCAC's website; _ the predicted firm-specific litigation risk from the logit regression of = ( , , , an indicator variable that equals one if the firm's headquarters state favors a Democratic candidate in the most recent presidential election prior to year t, and zero otherwise; an indicator variable that equals one if the actual or pseudo lawsuit filing date of a firm-year is after the Supreme Court's ruling on the Tellabs case, that is, June 21 st , 2007, and zero otherwise. Each non-sued firm-year is matched to a sued firm-year in the same industry and year, and with the closest _ ; for a sued firm-year, we use its actual lawsuit filing date; for a non-sued firm-year, we use the actual lawsuit filing date of its matched sued firm-year as its pseudo lawsuit filing date; the three-day cumulative market-adjusted returns surrounding the judicial appointment date in percentage; market-adjusted returns are return minus the value-weighted market return;      This panel reports the number and percentage of firm-years that overlap with class periods of lawsuits across quintiles of within high and low predicted litigation risk firm-years. Firm-years are sorted into the high and low predicted litigation risk groups based on whether their _ is above or below the sample median, respectively. *, **, and *** indicate significance at the 0.   Table 5 Relation between judge ideology and firm characteristics and geographic variables Table 6 Relation between judge ideology and lawsuit occurrence This table reports the regression results on the relation between securities class action lawsuit occurrences and circuit court judge ideology. We estimate the model = (  ,  ,  _  ) + ;  _  includes  ,  ,  ,  , ,

Variables used in earnings guidance and mandatory disclosure tests
Columns (1) and (2) report the logit regression results and columns (3) and (4) report the OLS regression results. z-stats (t-stats) based on standard errors estimated clustered by state are reported in parentheses below the coefficients. *, **, and *** indicate significance at the two-tailed 0.1, 0.05, and 0.01 levels, respectively. Variable definitions are in Appendix B. (1) (2) (3) Columns (1) and (2) report the logit regression results and columns (3) and (4) report the OLS regression results. z-stats (t-stats) based on standard errors clustered by state are reported in parentheses below the coefficients. *, **, and *** indicate significance at the 0.1, 0.05, and 0.01 levels, respectively. Variable definitions are in Appendix B. (1) (2)   Table 9 Consequence of circuit court judicial appointments 67 Pane B: Lawsuit filing rates before and after judicial appointments This panel reports the likelihood of lawsuit filings in the two years before and the two years after judicial appointment dates for firms located in the affected circuits. *, **, and *** indicate significance at the 0.1, 0.05, and 0.01 levels, respectively, with standard errors clustered by states. This panel reports the market reaction to judicial appointments for firms located in the circuit affected. The market reaction ( ) is the three-day cumulative market-adjusted returns surrounding the judicial appointment date. The statistical significance of mean is based on the t-test of the difference between the and zero, with standard errors clustered by states. The statistical significance of the median is based on the non-parametric sign test of the difference between the and zero. *, **, and *** indicate significance at the 0.1, 0.05, and 0.01 levels, respectively.  This panel reports the OLS regression results on the relation between the number of negative short-horizon earnings guidance and litigation risk. The sample consists of 35,890 firm-year observations with negative earnings news during the 1995-2013 period. In columns (1), (3), (5) and (7), we estimate the OLS regression _ ℎ = ( , ,