會(huì)計(jì)欺詐外文翻譯文獻(xiàn)

會(huì)計(jì)欺詐外及翻譯文獻(xiàn)（文檔含中英文對(duì)照即英文原文和中文翻譯）譯文：會(huì)計(jì)欺詐和機(jī)構(gòu)投資者查得-R?拉森介紹美國(guó)資本市場(chǎng)依賴財(cái)務(wù)報(bào)告系統(tǒng)來幫助有效分配資本。最近的財(cái)務(wù)報(bào)告過程中故障在許多高調(diào)公司新的人員的監(jiān)管機(jī)構(gòu)，會(huì)計(jì)欺詐和市場(chǎng)參與者的興趣。兩個(gè)重要的經(jīng)驗(yàn)規(guī)律的文獻(xiàn)記錄極端操縱收益的決定因素和后果。首先，股票市場(chǎng)反應(yīng)的啟示會(huì)計(jì)處理顯著負(fù)面。估計(jì)下降公告后的市場(chǎng)價(jià)值會(huì)計(jì)操作范圍從20-40%（Palmrose.理查德森和2003李，和馬丁2007）。第二，會(huì)計(jì)操作是可預(yù)測(cè)的。文獻(xiàn)會(huì)計(jì)操作的文檔可以預(yù)測(cè)的措施準(zhǔn)確的質(zhì)量、會(huì)計(jì)性能、非金融變量聲明和股市變量（如,Beneish1999;Dechow,通用電氣,拉爾森和斯隆2007）。雖然會(huì)計(jì)操作導(dǎo)致重大投資損失和與公司相關(guān)的特點(diǎn)和性能，幾乎沒有證據(jù)表明存在成熟的投資者是否能夠避免損失與會(huì)計(jì)欺詐。機(jī)構(gòu)投資者已成為市場(chǎng)的重要力量在過去的幾十年。上世紀(jì)八十年代初到九十年代末，機(jī)構(gòu)投資者所有權(quán)翻倍，股市50%以上（龔帕斯和Metric時(shí),2001）。機(jī)構(gòu)投資者在美國(guó)市場(chǎng)存在的上升有意義的促使文獻(xiàn)調(diào)查他們是否執(zhí)行是有利可圖的交易。文獻(xiàn)的結(jié)果是喜憂參半。幾項(xiàng)研究文檔積極變化之間的相互關(guān)系，這些機(jī)構(gòu)投資者的資產(chǎn)和未來的收益和回報(bào)，這表明機(jī)構(gòu)通知交易員（如，柯和2005;阿里列弗，確認(rèn)我也承認(rèn)泰德?克里斯坦森的指導(dǎo)，沒有它我就不會(huì)了博士學(xué)位的挑戰(zhàn)。最后，我感謝我的家人的支持。沒有這篇論文的完成并不意味著幾乎一樣多。介紹美國(guó)資本市場(chǎng)依賴財(cái)務(wù)報(bào)告系統(tǒng)來幫助有效分配資本。最近的財(cái)務(wù)報(bào)告過程中故障在許多高調(diào)公司新的人員監(jiān)管機(jī)構(gòu)，會(huì)計(jì)欺詐和市場(chǎng)參與者的興趣。兩個(gè)重要的經(jīng)驗(yàn)規(guī)律的文獻(xiàn)記錄極端操縱收益的決定因素和后果。首先，股票市場(chǎng)反應(yīng)的啟示，會(huì)計(jì)處理顯著負(fù)面。估計(jì)下降公告后的市場(chǎng)價(jià)值會(huì)計(jì)操作范圍從20-40%（Palmrose、理查德森和朔爾茨2003;Karpo_,李，和馬丁2007）。第二，會(huì)計(jì)操作是可預(yù)測(cè)的。文獻(xiàn)會(huì)計(jì)操作的文檔可以預(yù)測(cè)的措施合格的質(zhì)量、會(huì)計(jì)性能、非金融變量聲明和股市變量（如,Beneish1999;Dechow,通用電氣,拉爾森和斯隆2007）。雖然會(huì)計(jì)操作導(dǎo)致重大投資損失和與公司相關(guān)的特點(diǎn)和性能，幾乎沒有證據(jù)表明存在成熟的投資者是否能夠避免損失與會(huì)計(jì)欺詐。機(jī)構(gòu)投資者已成為市場(chǎng)的重要力量在過去的幾十年。從八十年代初到九十年代末，機(jī)構(gòu)投資者所有權(quán)翻倍，股市50%以上（。機(jī)構(gòu)投資者在美國(guó)市場(chǎng)存在的上升促使文獻(xiàn)調(diào)查他們是否執(zhí)行有利可圖的交易。文獻(xiàn)的結(jié)果是喜憂參半。幾項(xiàng)研究文檔積極變化之間的相互關(guān)系，這些機(jī)構(gòu)投資者的資產(chǎn)和未來的收益和回報(bào)，這表明機(jī)構(gòu)通知交易員（如。、柯和Ramalingegowda2005;柯和Petron2004;阿里Dutch列弗,Trembles2004）。另一方面,一些文獻(xiàn)表明，知情交易可能更有限,發(fā)現(xiàn)性能優(yōu)越的共同基金很少持續(xù)（Carat1997;布朗一個(gè)曼1995年）和交易考慮通知可能只是動(dòng)量交易的結(jié)果（Bushee和古德曼,2007）。會(huì)計(jì)欺詐和大市場(chǎng)的可預(yù)測(cè)性與會(huì)計(jì)欺詐相關(guān)的損失表明，它是一個(gè)理想的設(shè)定檢查成熟的機(jī)構(gòu)投資者。如果機(jī)構(gòu)投資者擁有優(yōu)越的信息和復(fù)雜的會(huì)計(jì)信息的使用者對(duì)會(huì)計(jì)欺詐，他們應(yīng)該在欺詐上市公司公開披露前欺詐。我的主要研究問題是機(jī)構(gòu)投資者預(yù)期會(huì)計(jì)欺詐的啟示和剝離的股票欺詐公司公開披露之前騙子。作為一個(gè)次要的研究問題，我檢查是否機(jī)構(gòu)作為有效的公司監(jiān)控預(yù)防欺詐。我使用的會(huì)計(jì)、審計(jì)和執(zhí)行版本（AAER）涉及欺詐和會(huì)計(jì)操作作為一個(gè)代理第一新聞文章Factiva提及會(huì)計(jì)違規(guī)的公眾披露欺詐。我檢查機(jī)構(gòu)交易模式在322年企業(yè)，美國(guó)證券交易委員會(huì)（SEC）中標(biāo)識(shí)執(zhí)行行動(dòng)從1982年到2005年有操縱會(huì)計(jì)收益。我的分析是在兩個(gè)階段進(jìn)行。第一階段是本文分析聚合機(jī)構(gòu)公司級(jí)和欺詐檢查他們的交易行為。第二階段是一個(gè)階段分析，利用機(jī)構(gòu)投資者之間的異構(gòu)性，欺詐行為檢查他們的交易行為。公司的分析，我遵循Bushee（2001）組織機(jī)構(gòu)分為三類根據(jù)自己的投資風(fēng)格:瞬態(tài)和專用。多元化的投資組合和較低的投資組合營(yíng)業(yè)額機(jī)構(gòu)的特點(diǎn)。多樣化的投資組合投資組合交易描述瞬態(tài)高機(jī)構(gòu)，和高度集中的投資組合和較低的投資組合營(yíng)業(yè)額專門機(jī)構(gòu)的特點(diǎn)。符合文學(xué)之前,我希望發(fā)現(xiàn)瞬態(tài)機(jī)構(gòu)最有可能發(fā)起有利交易欺詐啟示的預(yù)期和機(jī)構(gòu)不大可能發(fā)起有利益可得交易欺詐的預(yù)期啟示我沒有強(qiáng)烈的專門機(jī)構(gòu)研究預(yù)測(cè)通常找到貿(mào)易基于即將到來的未來事件。然而，欺詐是一個(gè)獨(dú)特的設(shè)置可能導(dǎo)致專門機(jī)構(gòu)剝離他們的位置。如果專門機(jī)構(gòu)投資公司基于他們的信心的和意愿完整性管理、檢測(cè)欺詐行為會(huì)引起專門機(jī)構(gòu)剝離他們的股份。此外，由于專門機(jī)構(gòu)的特點(diǎn)是高度的投資組合,他們可能會(huì)有更大比例的投資組合風(fēng)險(xiǎn)欺詐是否顯示。因此，他們可能會(huì)有最強(qiáng)的激勵(lì)預(yù)測(cè)欺詐和詐騙剝離他們的股票。文獻(xiàn)表明，機(jī)構(gòu)投資者作為公司監(jiān)控（如果是這樣的話,那么有可能是詐騙公司低水平的機(jī)構(gòu)投資欺詐行為，因?yàn)樗麄冎叭狈π时O(jiān)控。因此，我的第一組測(cè)試檢查機(jī)構(gòu)所有權(quán)水平是否欺詐公司立即釋放之前第一次欺詐收益報(bào)告不同于人口控制的公司。在不變的分析中，我發(fā)現(xiàn)欺詐公司實(shí)際上有更高水平的總機(jī)構(gòu)所有權(quán)，所有權(quán)和瞬態(tài)比公司所有權(quán)制度。專門機(jī)構(gòu)所有權(quán)沒有顯著區(qū)別樣本公司。接下來，我將進(jìn)行回歸分析，控制公司的特點(diǎn)。我立即發(fā)現(xiàn)欺詐的開始之前，機(jī)構(gòu)所有權(quán)欺詐公司的總體水平高于制度所有權(quán)控制公司的一個(gè)示例。然而，我發(fā)現(xiàn)更高層次的機(jī)構(gòu)主要是所有權(quán)的結(jié)果表明斜面更高層次的瞬態(tài)機(jī)構(gòu)所有權(quán)，專門機(jī)構(gòu)所有權(quán)控制后幾乎是相同的形式特征。大學(xué)和回歸結(jié)果表明，機(jī)構(gòu)所有權(quán)水平不作為一個(gè)伶俐的監(jiān)控裝置在欺詐的預(yù)防。我的下一套測(cè)試提供相關(guān)的證據(jù)，我的主要研究問題。我第一次檢查機(jī)構(gòu)所有權(quán)水平變化在欺詐公司在此期間公司提交欺詐。季度發(fā)行前的第一次欺詐收益報(bào)告，直到季度會(huì)計(jì)欺詐的公開披露之前，我發(fā)現(xiàn)機(jī)構(gòu)所有權(quán)欺詐公司增加了近14%，代表一家欺詐公司已發(fā)行股份的3.9%。2因?yàn)槠墼p公司經(jīng)驗(yàn)股價(jià)下跌約35%一旦發(fā)現(xiàn)欺詐，欺詐的機(jī)構(gòu)增加3.9%所有權(quán)不是微不足道的。事實(shí)上，計(jì)算表明，總機(jī)構(gòu)損失322年我的樣品是詐騙公司的1380億美元。增加3.9%機(jī)構(gòu)所有權(quán)欺詐時(shí)期代表約200億美元的損失。之前的研究已經(jīng)檢查機(jī)構(gòu)是否能預(yù)測(cè)即將發(fā)生的事件在短窗口（Briber,詹金斯和王,2006）。因此，在我的下一組測(cè)試，我觀察機(jī)構(gòu)所有權(quán)的變化后的季度馬上前,公開揭露欺詐。我在季度立即欺詐曝光之前，所有權(quán)制度降低了大約一個(gè)半欺詐公司已發(fā)行股份的百分比。我發(fā)現(xiàn)能有效的降低瞬態(tài)機(jī)構(gòu)持有,而專門的機(jī)構(gòu)持有的變化無關(guān)緊要的不會(huì)。我也在季后立即找到有效減少欺詐啟示。這些結(jié)果是強(qiáng)勁的幾個(gè)控制變量包括現(xiàn)在和過去的股票回報(bào)，意想不到的收益,以及分享營(yíng)業(yè)額的變化。雖然我找到證據(jù)表明瞬態(tài)機(jī)構(gòu)能夠預(yù)測(cè)欺詐前一個(gè)時(shí)期它的啟示這些證據(jù)必須解釋的證據(jù)我之前測(cè)試。一個(gè)半百分比下降機(jī)構(gòu)所有權(quán)欺詐曝光之前雖然顯著，稍微減輕相當(dāng)大的機(jī)構(gòu)投資者的損失。機(jī)構(gòu)更異于三類我受聘于企業(yè)層面分析。因此，我進(jìn)行第二次分析進(jìn)一步利用機(jī)構(gòu)投資者之間的異質(zhì)性。我創(chuàng)建代理機(jī)構(gòu)的信息環(huán)境和機(jī)構(gòu)的激勵(lì)機(jī)制，以避免負(fù)面的市場(chǎng)后果與會(huì)計(jì)欺詐的啟示。條件擁有欺詐公司股票欺詐開始之前，我測(cè)試是否與機(jī)構(gòu)的所有權(quán)的變化相關(guān)聯(lián)的代理是欺詐公司前會(huì)計(jì)欺詐的啟示。結(jié)果提供一些證據(jù)表明與最強(qiáng)的激勵(lì)制度，以避免會(huì)計(jì)欺詐和最高的質(zhì)量信息環(huán)境剝離前股票欺詐公司會(huì)計(jì)欺詐的啟示。盡管數(shù)據(jù)符合資產(chǎn)剝離率的增加在這些機(jī)構(gòu)中，我無法確定這些關(guān)系的結(jié)果通知交易或自然的所有權(quán)水平均值回歸。這項(xiàng)研究應(yīng)該感興趣的機(jī)構(gòu)投資者和研究者。研究結(jié)果表明,機(jī)構(gòu)投資者失去錢的重要性通過投資公司提交會(huì)計(jì)欺詐。進(jìn)一步的證據(jù)有助于我的研究文獻(xiàn)記錄復(fù)雜的機(jī)構(gòu)投資者。至少在這個(gè)特殊的背景下，大多數(shù)機(jī)構(gòu)似乎沒有復(fù)雜的會(huì)計(jì)信息用戶;然而，我確實(shí)提供了有限的證據(jù)前立即通知本季度交易欺詐機(jī)構(gòu)之間的一個(gè)子集的啟示。這些投資的標(biāo)準(zhǔn)可能會(huì)導(dǎo)致這些機(jī)構(gòu)傾斜特征，更有可能來證明他們的投資組合的審慎投資。例如發(fā)現(xiàn)高的銀行向企業(yè)傾斜投資組合標(biāo)準(zhǔn)普爾股票評(píng)級(jí)。Bushee和古德曼（2007）,這是一個(gè)指示符變量等于一個(gè)如果一個(gè)機(jī)構(gòu)的市場(chǎng)價(jià)值的股票投資組合的五等分頂層，否則所有機(jī)構(gòu)在一個(gè)特定的季度和0。因?yàn)榇蠖鄶?shù)機(jī)構(gòu)擁有更多的資源，我希望ISIZE是一個(gè)機(jī)構(gòu)的代理獲取和處理信息的能力。因此，我認(rèn)為,大型機(jī)構(gòu)將更有可能出售公司的股票有欺詐行為。我發(fā)現(xiàn)了兩個(gè)額外的措施,代理機(jī)構(gòu)的私人信息和激勵(lì)措施,以避免會(huì)計(jì)欺詐。選擇，第一個(gè)是一個(gè)變量，措施的相對(duì)大小的股權(quán)機(jī)構(gòu)風(fēng)險(xiǎn)在一個(gè)特定的公司。打賭測(cè)量作為應(yīng)聲股本旗下機(jī)構(gòu)j公司我在一季度t擴(kuò)展機(jī)構(gòu)j的總市場(chǎng)價(jià)值的投資組合在季度t。我希望押注是負(fù)相關(guān)的制度變遷在欺詐公司的所有權(quán)。與更高水平的機(jī)構(gòu)選擇相對(duì)比水平較低的企業(yè)風(fēng)險(xiǎn)價(jià)值選擇，因此，這些機(jī)構(gòu)有更大的激勵(lì)來收集私人信息，避免投資公司有欺詐行為。我最后的機(jī)構(gòu)公司水平變量，這是一個(gè)指示符變量等于一個(gè)如果一個(gè)機(jī)構(gòu)持有的流通股總量的百分比在公司五等分頂層的制度，公司所有權(quán)和零。我希望塊與私人信息優(yōu)勢(shì)，因?yàn)檫@些機(jī)構(gòu)更有可能獲得私人信息和更愿意承擔(dān)私人信息采集和處理的成本。因此，我希望阻止將負(fù)相關(guān)的機(jī)構(gòu)持有的詐騙公司的變化。Bushee和古德曼(2007)是第一個(gè)采用這兩種措施指出是一個(gè)很好的衡量去激勵(lì)收集公司信息。原文：AccountingFraudandInstitutionalInvestorsByChadR.LarsonIalsoacknowledgethementorshipofTedChristensenwithoutwhichIwouldhavenevertakenonthechallengeofadoctorate.Lastly,Iamgratefulforthesupportofmyfamily.Withoutthemthecompletionofthisdissertationwouldnotmeannearlyasmuch.IntroductionU.S.capitalmarketsrelyonfinancialreportingsystemstohelpeffectivelyallocatecapital.Therecentbreakdownsinthefinancialreportingprocessatmanyhighprofilecompanieshaverenewedresearchers',andmarketparticipants'interestinaccountingfraud.Twoimportantempiricalregularitiesemergefromthebodyofliteraturedocumentingthedeterminantsandconsequencesofextremeearningsmanipulations.First,stockmarketreactionstotherevelationofaccountingmanipulationsaresignificantlynegative.Estimateddeclinesinmarketvaluefollowingthepublicannouncementofaccountingmanipulationsrangefrom20to40percent(Palmore,Richardson,andScholz2003;Karpo_,Lee,andMartin2007).Second,accountingmanipulationsarepredictable.Abodyofliteraturedocumentsthataccountingmanipulationscanbepredictedwithmeasuresofaccuratequality,accountingperformance,non-financialstatementvariables,andstockmarketvariables(e.g.,Beneish1999;Dechow,Ge,Larson,andSloan2007).Althoughaccountingmanipulationsresultinsignificantinvestorlossesandareassociatedwithfirmcharacteristicsandperformance,littleevidenceexistsonwhethersophisticatedinvestorsareabletoavoidlossesassociatedwithaccountingfraud..Institutionalinvestorshavebecomeasignificantmarketforceoverthelastseveraldecades.Fromtheearly1980stothelate1990s,institutionalinvestorsdoubledtheirownershipintheequitymarketstoover50percent(GompersandMetrick,2001).TherisingpresenceofinstitutionalinvestorsintheU.S.marketshasspurredasigni_cantbodyofliteratureinvestigatingwhethertheyexecutepro_tabletrades.Theresultsoftheliteraturearemixed.Severalstudiesdocumentpositiveassociationsbetweenchangesininstitutionalinvestors'holdingsandfutureearningsandreturns,suggestingthatinstitutionsareinformedtraders(e.g.,KeandRamalingegowda2005;KeandPetroni2004;Ali,Durtschi,Lev,andThrombley2004).Ontheotherhand,someliteraturesuggeststhatinformedtradingmightbemorelimited,findingthatsuperiormutualfundperformanceisrarelypersistent(Carhart1997;BrownanGoetzmann1995)andtradingpatternspreviouslyconsideredinformedmightsimplybetheresultofmomentumtrading(BusheeandGoodman,2007).Thepredictabilityofaccountingfraudandthelargemarketlossesassociatedwithaccountingfraudsuggestthatitisanidealsettingtoexaminethesophisticationofinstitutionalinvestors.Ifinstitutionalinvestorspossesssuperiorinformationandaresophisticatedusersofaccountinginformationwithrespecttoaccountingfraud,theyshouldsellsharesinfraudfirmspriortopublicrevelationsoffraud.Myprimaryresearchquestioniswhetherinstitutionalinvestorsanticipateaccountingfraudrevelationsanddivestsharesinfraudfirmspriortothepublicrevelationoffrauds.Asasecondaryresearchquestion,Iexaminewhetherinstitutionsactaseffectivefirmmonitorsinthepreventionoffraud.IuseAccounting,Auditing,andEnforcementReleases(AAER)involvingaccountingmanipulationsasaproxyforfraudandthefirstpressarticleinFactivamentioninganaccountingirregularityasthepublicrevelationoffraud.1Iexamineinstitutionaltradingpatternsin322firmsthattheSecuritiesandExchangeCommission(SEC)identifiedinenforcementactionsfrom1982through2005ashavingmanipulatedtheiraccountingearnings.Myanalysisisconductedintwostages.Thefirststageisafirm-levelanalysisthataggregatesinstitutionsatthefirm-levelandexaminestheirtradingbehaviorinfraud_rms.Thesecondstageisaninstitution-levelanalysisthatexploitstheheterogeneityamonginstitutionalinvestorsandexaminestheirtradingbehaviorinfraud_rms.Formyfirm-levelanalysis,IfollowBushee(2001)bygroupinginstitutionsintothreecategoriesbasedontheirinvestmentstyles:quasi-indexer,transient,anddedicated.Diversifiedportfoliosandlowportfolioturnovercharacterizequasi-indexerinstitutions.Diversifiedportfoliosandhighportfolioturnovercharacterizetransientinstitutions,andhighlyconcentratedportfoliosandlowportfolioturnovercharacterizededicatedinstitutions.Consistentwithpriorliterature,Iexpecttofindthattransientinstitutionsarethemostlikelytoinitiateprofitabletradesinanticipationofafraudrevelationandquasi-indexerinstitutionsareunlikelytoinitiateprofitabletradesinanticipationofafraudrevelation(e.g.,KeandRamalingegowda2005;Hribar,Jenkins,andWang2006).Imakenostrongpredictionsfordedicatedinstitutionsasresearchtypicallyfindsthattheydonottradebasedonimpendingfutureevents.However,fraudisauniquesettingthatmayleaddedicatedinstitutionstodivesttheirpositions.Ifdedicatedinstitutionsinvestinfirmsbasedontheirconfidenceinthevisionandintegrityofmanagement,detectingafraudmightleaddedicatedinstitutionstodivesttheirshares.Inaddition,sincededicatedinstitutionsarecharacterizedbyhighly-concentratedportfolios,theyarelikelytohavealargerpercentageoftheirportfoliosatriskiffraudisrevealed.Therefore,theyarelikelytohavethestrongestincentivestoanticipatefraudanddivesttheirsharesinfraud_rms.Abodyofliteraturesuggeststhatinstitutionalinvestorsactasfirmmonitors(e.g.Chung,Firth,andKim2002;Chen,Harford,andLi2007).Ifthisisthecase,thenitispossiblethatfraudfirmshavelowlevelsofinstitutionalinvestmentpriortocommittingfraudbecausetheylackefficiencymonitoring.Therefore,myfirstsetoftestsexamineswhetherinstitutionalownershiplevelsinfraudfirmsimmediatelypriortothereleaseofafirstfraudulentearningsreportdifferfromapopulationofcontrolfirms.Inunvariedanalysis,Ifindthatfraudfirmsactuallyhavehigherlevelsoftotalinstitutionalownership,quasi-indexerownership,andtransientinstitutionalownershipthannon-fraudfirms.Dedicatedinstitutionalownershipisnotsignificantlydifferentfromthenon-fraudsampleoffirms.Next,Iconductregressionanalysisthatcontrolsforfirmcharacteristics.Ifindthatimmediatelypriortothebeginningofafraud,fraudfirms'totallevelofinstitutionalownershipishigherthaninstitutionalownershipforasampleofcontrolfirms.However,Ifindthatthehigherlevelofinstitutionalownershipisprimarilytheresultofasigncanthigherleveloftransientinstitutionalownership,whilequasi-indexeranddedicatedinstitutionalownershipisnearlyidenticalaftercontrollingformcharacteristics.TheuniversityandregressionresultssuggestthatthelevelofinstitutionalownershipdoesnotactasasapientmonitoringdeviceinthePreventionoffraud.Mynextsetsoftestsprovideevidencerelatingtomyprimaryresearchquestion.Ifirstexaminechangesininstitutionalownershiplevelsinfraudfirmsovertheperiodfirmscommitfraud.Fromthequarterpriortotheissuanceofafirstfraudulentearningsreportuntilthequarterpriortothepublicrevelationofanaccountingfraud,Ifindthatinstitutionalownershipinfraudfirmsincreasesbyalmost14percent,representing3.9percentofafraudfirm'soutstandingstock.2Becausefraudfirmsexperiencestockpricedeclinesofapproximately35percentoncethefraudisrevealed,the3.9percentincreaseininstitutionalownershipoverthefraudperiodisnottrivial.Infact,calculationssuggestthattotalinstitutionallossesforthe322fraudfirmsinmysampleareinorderof$138billion.The3.9percentincreaseininstitutionalownershipoverthefraudperiodrepresentsapproximately$20billionofthoselosses.Priorresearchhasexaminedwhetherinstitutionscanpredictimpendingeventsovershortwindows(Hribar,Jenkins,andWang,2006).Accordingly,inmynextsetoftests,Iobservechangesininstitutionalownershipinthequartersimmediatelypriortoandfollowingthepublicrevelationoffraud.Iamthatinthequarterimmediatelypriortoafraudrevelation,institutionalownershipdecreasesbyapproximatelyoneandahalfpercentofafraudfirm'soutstandingstock.Ifindsignificantdecreasesfortransientinstitutionalholdings,whilechangesinquasi-indexeranddedicatedinstitutionalholdingsareinsignificant.Ialsofindsignificantdecreasesinthequarterimmediatelyfollowingthefraudrevelation.Theseresultsarerobusttoseveralcontrolvariablesincludingcurrentandpaststockreturns,unexpectedearnings,andchangesinshareturnover.AlthoughIfindcomeevidencethattransientinstitutionsareabletoanticipatefraudoneperiodpriortoitsrevelation,thisevidencemustbeinterpretedinlightofevidencefrommyprevioustests.Theoneandahalfpercentdecreaseininstitutionalownershippriortofraudrevelations,thoughstatisticallysignificant,onlyslightlymitigatessubstantiallossesforinstitutionalinvestors.InstitutionsaremoreheterogeneousthanthethreecategoriesIemployinmyfirm-levelanalysis.Therefore,Iconductasecondanalysisattheinstitution-levelthatfurtherexploitstheheterogeneityamonginstitutionalinvestors.Icreateproxiesforinstitutions'informationenvironmentsandinstitutions'incentivestoavoidthenegativemarketconsequencesassociatedwiththerevelationofaccountingfraud.Conditionalonowningfraudfirmsharespriortoafraudbeginning,Itestwhethertheproxiesareassociatedwithinstitutions'ownershipchangesinfraudfirmspriortotherevelationofanaccountingfraud.Theresultsprovidesomeevidencethatinstitutionswiththestrongestincentivestoavoidaccountingfraudandwiththehighestqualityinformationenvironmentsdivestsharesinfraudfirmspriortotherevelationofaccountingfraud.Althoughthedataareconsistentwithanincreasedrateofdivestituresamongtheseinstitutions,Iamunabletoestablishwhethertheserelationsarearesultofinformedtradingornaturalmeanreversioninownershiplevels.Thisstudyshouldbeofinteresttobothinstitutionalinvestorsandresearchers.Theresultssuggestthatinstitutionalinvestorslosesignificantamountsofmoneybyinvestinginfirmsthatcommitaccountingfraud.Mystudycontributesfurtherevidencetothebodyofliteraturedocumentingthesophisticationlevelofinstitutionalinvestors.Atleastforthisparticularcontext,mostinstitutionsdonotappeartobesophisticatedusersofaccountinginformation;however,Idoprovidelimitedevidenceofinformedtradinginthequarterimmediatelypriortofraudrevelationsamongasubsetofinstitutions.Theremainderofmydissertationproceedsasfollows.Chapter2examinespriorliteratureandoutlinesmyempiricalpredictions.Chapter3outlinesmyresearchdesign.Chapter4describesmysampleselectionprocessandprovidesdescriptivestatistics.Chapter5detailsmytestsandpresentsresultsandchapter6concludes.PriorLiteratureandEmpiricalPredictionsMydissertationbuildsontwostreamsofpriorliterature.Thefirststreamofliteratureexaminesthedeterminantsandconsequencesofaccountingmanipulations.Thesecondstreamofliteratureexaminesthetradingbehaviorofinstitutionalinvestors.AccountingManipulationsPriorresearchhasidentifiedcharacteristicsoffirmsthatmanipulatetheirfinancialstatements.Dechow,Ge,Larson,andSloan(2007)investigateacomprehensivesampleofall895firmssubjecttoAccounting,Auditing,andEnforcementReleases(AAER)from1982throughJuly2005.Theyexaminetheuseofseveralfinancialstatementvariables,o_-balancesheetandnon-financialvariables,andmarket-relatedvariablestopredictaccountingmanipulations.TheyhadthatfirmsaccusedbytheSECofmanipulatingtheir_financialperformancetendtohavehadstrongperformancepriortomanipulations.Theyalsoandthatmanipulationsappeartobemotivatedbymanagers'desiretoobfuscatedeterioratingfinancialperformance.Duringmanipulationyears,theyfindthatcashprofitmarginsandreturnonassetsaredecliningwhileaccrualsareincreasing.Theyalsofindthatfirmsmanipulatingfinancialreportingaremorelikelytoissuedebtandequity.Rankingfirmsbasedonthepredictedlikelihoodofaccountingmanipulationsfromalogisticmodel,theyclassifyalmost50percentofmanipulationfirmsinthetop20percentoftheirmanipulationindexand65percentofmanipulationfirmsinthetop40percentoftheirindex.Beneish(1999)createsafraudpredictionmodelbasedonasampleof74firmsthatmanipulatedearningsandasampleof2,332matchedfirms.Estimatingprobitymodelsofaccountingmanipulationsasafunctionofeightaccountingbasedvariables(indexedday'ssalesinreceivables,grossmargin,assetquality,salesgrowth,depreciation,sales,generalandadministrativeexpenses,leverage,andaccrualstototalassets)heisabletocorrectlyclassifyapproximately50to75percentoffraudfirms,whileincorrectlyclassifyingonly10to20percentofmatchedfirms.Severalotherstudiesdocumentrelationsbetweenearningsmanipulationfirmsandfirmcharacteristics.TwoothernotablestudiesincludeDechow,Sloan,andSweeney(1996)andBrazel,Jones,andZimbelman(2006).Dechowetal,(1996)examineasampleof92firmswithanAAERfrom1982to1992.TheydocumentthatAAERsareassociatedwithexternalfinancingneedsandpoorcorporategovernance.TheyalsoshowthatAAERfirmsexperiencesignifiescantincreasesintheircostofcapitalaftertherevelationsofaccountingmanipulations.Brazeletffal,(2006)alsoandthatseveralnon-financialmeasurescanbeusefulinpredictingaccountingmanipulations.AlthoughthenumberofTypeIerrorsinfraudpredictionmodelsisrelativelyhigh,therelativecostofTypeItoTypeIIerrorsforinstitutionalinvestorsislikelyextremelylow.Severalstudieshaveestimatedinvestmentlosseswhenaccountingmanipulationsarerevealed.ThelatestlargesampleevidencesuggeststhatthecostofTypeIIerrorsaverageapproximately40percentofaninstitution'sinvestmentinafraudfirm(Karpo,Lee,andMartin,2007).Ontheotherhand,thecostofaTypeIerrorisextremelylowinamarketwithmanysubstituteassetsasinvestorscansimplychoosenottoholdfirmswithahighprobabilityoffraud.Investorsmayalsobeprivytoprivateinformationregardingfirmperformanceandaccountingmanipulations.Totheextentthatinvestorspossessprivateinformationandchoosetouseotherqualitativeinformation,theymaybeabletosignificantlyreducethenumberofTypeIandIIerrorsincurredwhenattemptingtoidentifyaccountingfrauds.ThehighnumberofTypeIerrorsassociatedwithusingearningsmanipulationpredictionmodelsmightalsosuggestthatinvestorswouldbewillingtolivewiththenegativereturnsassociatedwithfraud_firmsifthenegativereturnsarebalancedoutwithsuficientlypositivereturnsfromnon-fraud_firmswithstrongsignalsoffraud.Inaconcurrentworkingpaper,BeneishandNichols(2007),showthatthisisnotthecase.Theirresultsrevealthat_firmswithahighprobabilityofmanipulatedearnhavelowerfutureearningsandreturns.Theyalsoshowthatatradingstrategybasedontheprobabilityofearningsmanipulationyieldsanabnormalhedgereturnof13.9percent.Throughadditionalteststheyconcludethatthereturns,whichareconcentratedontheshortside,arenotaresultofasymmetricarbitragecosts,butratheraresultofasymmetricerrorsinmarketexpectations.BeneishandNichols(2007)donotprovidedirectevidenceon_firmsthatactuallymanipulateearn,rathertheyexamineportfoliosoffirmswithahighprobabilityofmanipulation.Theyfindthatinstitutionalinvestorsincreasetheirholdingsinfirmswithahighprobabilityofmanipulation.Mystudyfocusesontheactualincidenceoffraud.Iamabletoprovidemoredetailedanddirectevidenceonthetradingbehaviorofinstitutionsinactualfraud_firmsbefore,during,andaftertheperiodinwhichfirmscommitfraudandthefraudsbecomepublic.2.2InstitutionalInvestorsFrom1980to1996,institutionalinvestorsdoubledtheirshareofthemarketandnowcontroloverhalfoftheU.S.equitymarket(GompersandMetrick,2001).Theincreasedimportanceandperceivedsophisticationofinstitutionalinvestorshasspawnedalargebodyofliterature.Onebranchoftheliteratureexamineswhetherinstitutionalinvestorsactasmonitorsandinfluencemanagements'decisions.Theevidencesuggeststhatthelevelofinstitutionalownershipandthecompositionofafirm'sinstitutionalownershipbasematterswhendeterminingwhetherinstitutionalownersarelikelytoactaseffectivemonitors.Bushee(1998)findsthatmanagersarelesslikelytocutresearchanddevelopmentexpenseswhenfacinganearningsshortfallifinstitutionalownershipishigh.Buthealsofindsthatlargeproportionsofownershipbyinstitutionsthattradebasedonmomentumandhavehighportfolioturnoverincreasethelikelihoodthatafirmwillcutresearchanddevelopmenttomeetanearningsshortfall.Chung,Firth,andKim(2002)findthatlargeinstitutionalshareholdingsinafirmreducethelikelihoodofearningsmanagementusingaccruals.Chen,Harford,andLi(2007),usingacquisitiondecisionstorevealmonitoring,findthatinstitutionswithlong-terminvestmentsspecializeinmonitoringwhileotherinstitutionsdonotmonitor.Bushee(2001)findsthathighlevelsofshort-terminvestorsareassociatedwithanover-weightingofnear-termexpectedearningsandunder-weightingoflong-termexpectedearnings.Inlightofthiscombinedevidence,myfirstpredictionisthatfraudfirms,priortotheissuanceoftheirfirstfraudulentearningsreport,arelikelytohavelowlevelsofinstitutionalownership.Ialsoexpectthatfraudfirmswillhavehigherlevelsofshort-term,transient,institutionalownershipandlowerlevelsoflong-term,dedicated,institutionalownership.Muchoftheaccountingresearchoninstitutionalinvestors'tradingbehaviorsuggeststhatinstitutionalinvestorsaresophisticatedusersofaccountinginformation.Forexample,previousliteraturehasdocumentedthatthehigherthelevelofinstitutionalownership,thesmallerthemarketreactionsurroundingearningsannouncements(El-Gazzar,1998).Balsam,Bartov,andMarquardt(2002)findthatthevaluationimplicationsoflargediscretionaryaccrualsareincorporatedintostockpricesmorequicklyfor_rmswithlargeinstitutionalinvestorbases.Thepresenceofinstitutionalinvestorsisalsopositivelyassociatedwiththeextentthatpricesleadearnings(Jiambalvo,RajGopal,andVenkatachalam,2002).Studieshavealsoshownthatinstitutionalinvestorsexploitaccountingbasedstockpriceanomaliessuchasthepost-earningsannouncementdrift(KeandRamalingegowda,2005)andtheaccrualsanomaly(Collins,Gong,andHribar,2003).LevandNissim(2006)alsoshowthattheaccrualsanomalyisexploitedbysomeinstitutionalinvestors,butthemagnitudeofthisaccruals-relatedtradingisrathersmall.Theyshowthatthecontinuedpersistenceoftheaccrualsanomalyisnotexplainedbyalackofunderstandingamonginstitutions,butratheraninstitutionaldistasteforextreme-accrualsfirmsthataretypicallysmall,unprofitable,andrisky.KeandRamalingegowda(2005)findthatinstitutionsalsopossessinformationthatallowsthemtoavoidnegativestockpriceshocksassociatedwithabreakinastringofconsecutiveearningsincreases.Althoughmuchoftheliteratureoninstitutionalinvestorssuggeststhattheyaresophisticatedusersoffinancialinformation,thisliteraturestandsincontrasttoevidencethatquestionswhetherinstitutionsprofitfrominformationaladvantages.Forexample,muchoftheliteratureonmutualfundperformancesuggeststhatsuperiorperformanceisnotpersistent(e.g.,BrownandGoetzmann1995).Additionally,O'BrienandBhushan(1990)_findthatinstitutionsareattractedtofirmswithmoreanalystfollowing.Similarly,BusheeandNoe(2000)_findthatinstitutionsareattractedtofirmswithhigh-qualitydisclosureregimes.Therefore,ifpublicandprivateinformationaresubstitutes,institutionsshouldhavefeweropportunitiestobenefitfrominformationaladvantages.Ifinstitutionalinvestorspossesssuperiorprivateinformationorinformationprocessingabilities,Iexpecttofindsupportformysecondpredictionthatinstitutionalinvestorsdivestsharesinfirmsthatarecommittingaccountingfraud.Alackofevidencethatinstitutionsdivestsharesinfraudfirmspriortopublicrevelationsoffraudwouldsuggestthateitherinvestorsareunabletouseprivateinformationtoanticipatepublicannouncementsoffraudorthecostofanticipatingthepublicannouncementsoffraudaretoogreatrelativetothebenefits.Institutionalinvestorsexhibitheterogeneityintheirinvestmentstyles.Priorliteraturehasshownthatthelikelihoodofinformedtradingvarieswithinstitutionalinvestors'characteristics(e.g.,HribarandJenkins2004;KeandRamalingegowda2005).MuchofthepriorliteraturehasreliedonamethodologyproposedbyBushee(1998).Inthismethodology,institutionsarefirstclassifiedintooneofthreeinvestmentstrategies(quasi-indexer,transient,anddedicatedinstitutionalinvestors)basedonportfolioturnoverandstakesizes.Theinstitutionsarethenaggregatedatthefirmlevel.Thebodyofevidencethatusesthismethodologytypicallyfindsthatprofitabletradinginanticipationoffutureeventsisonlyidentityableforthetransientinvestorcategory.Therefore,Iexpectanyevidencethatinstitutionalinvestorspredictaccountingfraudwillbeconcentratedamongtransientinstitutionalinvestors.Becausededicatedinstitutionalownershavethelargestportionoftheirportfoliosatstakewhenafraudisrevealed,Ialsoanticipatethepossibilitythattheymaydivestsharesinanticipationoffraudrevelations.Inarecentpaper,BusheeandGoodman(2007)exploittheheterogeneityamonginstitutionalinvestorsandthepositionstheyholdbyconductingananalysisthatincludesnotonlyinstitution-levelvariablessuchasportfoliosizeandtradingstrategybutalsoinstitutionfirm-levelcharacteristicssuchasthesizeofapositioninaparticularfirmandthesizeofth