why do firms use private equity to opt out of public...

48
Why Do Firms Use Private Equity to Opt Out of Public Markets? Sreedhar T. Bharath Ross School of Business, University of Michigan Amy K. Dittmar Ross School of Business, University of Michigan We investigate how firms weigh the costs and benefits of being public in the decision to opt out of the public market and go private. We draw on previous studies of going private and on the subsequent well-developed theoretical literature on why firms go public to develop our hypotheses. We employ a comprehensive sample of going-private transactions from 1980 to 2004 in the United States and examine how these firms differ over their public life (from IPO to going private) relative to a sample of firms that went and remained public. Our results provide strong support for the importance of information and liquidity consid- erations in being a public firm. These factors are evident at the IPO, on average thirteen years before the going-private decision. Access to capital and control considerations be- come increasingly important in the choice of going private over the public life of the firm. (JEL G34) In the 1980s, an unprecedented number of public companies went private in leveraged buyouts (LBOs), partly fueled by the development of the junk bond market. LBO activity increased from $1.4 billion in 1979 to $77 billion in 1988. The six years since 2000 saw a resurgence in going-private transactions but were fueled this time by the development of the private equity market. Private equity firms in the United States raised about $225 billion in 2006, up from $159 billion raised in 2005. 1 Given the size and growth of this market, The authors acknowledge the research support of the Ross School of Business, University of Michigan. They would like to thank Yakov Amihud, Ekkehart Boehmer, David Brophy, Martijn Cremers, Ted Fee, Laura Field, Ken French, Charlie Hadlock, Steve Kaplan, Han Kim, Josh Lerner, Mike Long, Stewart Myers, Amiyatosh Purnanandam, Lubos Pastor, Gordon Phillips, Rina Ray, Michael Roberts, Henri Servaes, Anil Shivdasani, Matt Speigel (the editor), Michael Stegemoller, Jeremy Stein, Anjan Thakor, David Yermack, Rebecca Zarutskie, an anonymous referee, and seminar participants at Bocconi University in Milan, Italy; Boston College; Dartmouth College; Erasmus University Rotterdam; Harvard Business School; IBS ISCTE Business School in Lisbon, Por- tugal; McGill University; Michigan State; New York University; Ohio State University; Ross School of Business, University of Michigan; University of Amsterdam; University of Washington; Yale University; UNC-Duke 2006 Corporate Finance Conference; 2007 European Finance Association Conference; 2008 American Finance Asso- ciation conference; and the 2008 Financial Intermediation Research Society Conference for helpful comments and Brad Bernatek for research assistance. The article won the LECG Award for Best Paper in Corporate Fi- nance at the EFA Conference 2007. An earlier version of the article was titled “To Be or Not to Be(Public).” The authors are responsible for all errors. Send correspondence to Sreedhar T. Bharath (Amy K. Dittmar), Department of Finance, Ross School of Business, University of Michigan, Suite R3332 (R4436), Ann Arbor, MI 48109; telephone: 734-763-0485 (734-764-3108). E-mail: sbharath(adittmar)@umich.edu. 1 Source: Business Week, November 7, 2006. c The Author 2010. Published by Oxford University Press on behalf of The Society for Financial Studies. All rights reserved. For Permissions, please e-mail: [email protected]. doi:10.1093/rfs/hhq016 Advance Access publication March 23, 2010 at Univ of RochesterLibrary on April 15, 2013 http://rfs.oxfordjournals.org/ Downloaded from

Upload: others

Post on 10-Jul-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Outof Public Markets?

Sreedhar T. BharathRoss School of Business, University of Michigan

Amy K. DittmarRoss School of Business, University of Michigan

We investigate how firms weigh the costs and benefits of being public in the decision to optout of the public market and go private. We draw on previous studies of going private andon the subsequent well-developed theoretical literature on why firms go public to developour hypotheses. We employ a comprehensive sample of going-private transactions from1980 to 2004 in the United States and examine how these firms differ over their public life(from IPO to going private) relative to a sample of firms that went and remained public.Our results provide strong support for the importance of information and liquidity consid-erations in being a public firm. These factors are evident at the IPO, on average thirteenyears before the going-private decision. Access to capital and control considerations be-come increasingly important in the choice of going private over the public life of the firm.(JEL G34)

In the 1980s, an unprecedented number of public companies went private inleveraged buyouts (LBOs), partly fueled by the development of the junk bondmarket. LBO activity increased from $1.4 billion in 1979 to $77 billion in1988. The six years since 2000 saw a resurgence in going-private transactionsbut were fueled this time by the development of the private equity market.Private equity firms in the United States raised about $225 billion in 2006, upfrom $159 billion raised in 2005.1 Given the size and growth of this market,

The authors acknowledge the research support of the Ross School of Business, University of Michigan. Theywould like to thank Yakov Amihud, Ekkehart Boehmer, David Brophy, Martijn Cremers, Ted Fee, Laura Field,Ken French, Charlie Hadlock, Steve Kaplan, Han Kim, Josh Lerner, Mike Long, Stewart Myers, AmiyatoshPurnanandam, Lubos Pastor, Gordon Phillips, Rina Ray, Michael Roberts, Henri Servaes, Anil Shivdasani, MattSpeigel (the editor), Michael Stegemoller, Jeremy Stein, Anjan Thakor, David Yermack, Rebecca Zarutskie, ananonymous referee, and seminar participants at Bocconi University in Milan, Italy; Boston College; DartmouthCollege; Erasmus University Rotterdam; Harvard Business School; IBS ISCTE Business School in Lisbon, Por-tugal; McGill University; Michigan State; New York University; Ohio State University; Ross School of Business,University of Michigan; University of Amsterdam; University of Washington; Yale University; UNC-Duke 2006Corporate Finance Conference; 2007 European Finance Association Conference; 2008 American Finance Asso-ciation conference; and the 2008 Financial Intermediation Research Society Conference for helpful commentsand Brad Bernatek for research assistance. The article won the LECG Award for Best Paper in Corporate Fi-nance at the EFA Conference 2007. An earlier version of the article was titled “To Be or Not to Be(Public).”The authors are responsible for all errors. Send correspondence to Sreedhar T. Bharath (Amy K. Dittmar),Department of Finance, Ross School of Business, University of Michigan, Suite R3332 (R4436), Ann Arbor,MI 48109; telephone: 734-763-0485 (734-764-3108). E-mail: sbharath(adittmar)@umich.edu.

1 Source: Business Week, November 7, 2006.

c© The Author 2010. Published by Oxford University Press on behalf of The Society for Financial Studies.All rights reserved. For Permissions, please e-mail: [email protected]:10.1093/rfs/hhq016 Advance Access publication March 23, 2010

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 2: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

it is important to understand the economic forces that drive a firm to go frompublic to private markets, which is the goal of our study.

This article is not the first to investigate going-private transactions.DeAngelo, DeAngelo, and Rice (1984a, b), Lehn and Poulsen (1989), andKaplan (1989a, b, 1991) study the gains from LBOs in the 1980s and showthat agency problems and tax benefits explain the value gain to these transac-tions.2 This article complements and expands on these studies by consideringa much broader set of factors, employing a longer sample period drawn fromseveral ways a firm can go private, and examining the entire public life of thefirms prior to going private. In order to fully understand the costs and benefitsof being a public firm, we draw on the insights from the more recent theo-retical literature developed after these studies were published. These studiesweigh the costs and benefits of being a public versus private firm to explainwhy firms go public.3

Of course, to draw on the theories of going public to get insight into goingprivate, it is important that the theories of going public are reversible. Many(but not all) of the costs and benefits modeled in these papers are reversible;they explain why firms go public along with providing insight into why firmsgo private. Each of these going-public theories emphasizes the trade-off be-tween different economic forces that confer costs and benefits of going andbeing public. We focus on the costs and benefits of being (rather than going)public, since these factors are reversible. For example, the trade-off betweeneconomic force X (say, liquidity benefits of being a public firm) and economicforce Y (say, the costs of losing control in decision making) could be the ten-sion in a model that generates predictions on the going-public decision. In thisexample, when the benefits of liquidity exceed the cost of having lesser controlin decision making, firms go public. We argue that since the nature of the the-ories of going public are trade-off type theories, the reverse is also true. Whenthe costs of having lesser control exceed the benefits of liquidity, firms willreverse their decision, that is, firms will exit the public markets and go private.Thus, by systematically reversing many of the predictions of the going-publictheories and examining firms’ decisions to go private, we can determine whichfactors drive the choice between being a private or a public firm and test theirvalidity empirically. We also supplement these factors with the free cash flow(agency) considerations emphasized by the earlier going-private studies in ouranalysis.

2 DeAngelo, DeAngelo, and Rice (1984a, b) were among the first to find that public shareholders gain about22% in going-private transactions. Lehn and Poulsen (1989) show that much of the shareholder gain in going-private transactions stems from mitigating agency conflicts associated with free cash flow. Kaplan (1989a, b,1991) examines the benefits of going private using a sample of LBOs and highlights the importance of tax andincentive improvements due to the high leverage in these transactions. Other studies on similar issues on goingprivate include Schipper and Smith (1991), Slovin, Shushka, and Bendeck (1991), Denis (1992), and Opler andTitman (1993).

3 We summarize these theories in Section 2.

1772

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 3: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

To examine the choice between being public or private, we employ a largedata set of firms that went private and compare them with firms that remainedpublic. We track firms over their entire public life, from the IPO to either theirgoing-private decision or, for the comparison firms, the year 2004 (the endof our sample period). Our initial going-private sample consists of 1,377 U.S.firms (called the “private sample” hereafter) that went private between 1980and 2004 and is an exhaustive sample of going-private firms. To determine thebeginning of a firm’s public life and also provide a control group, we mergethis sample with an extensive sample of IPO dates. We retain those firms in theIPO master list that did not go private for comparison (called the “comparisonsample” hereafter). This comparison sample consists of 6,464 IPO firms thatwent and remained public. To determine why firms go private, we examinefirm characteristics of the private and comparison samples at the IPO and overtheir public lives to determine how both inherent differences and the evolutionof firm characteristics impact the decision to go private.

To explore the importance of changes in firms’ characteristics over theirpublic life in predicting the going-private decision, we first compare the privatesample firms at the time of the IPO and the time they went private. We find thatsome of the firms’ characteristics (such as analyst following) evolve in waysthat should increase the benefits of being public, but others (such as liquidity)change in ways that make it more likely that the firm will go private; thus, thetrends do not provide clear predictions of why firms go private and illustrate theimportance of benchmarking these changes to those of the comparison sample.We therefore line up our comparison and going-private firms in event timefrom the IPO and compare them at the IPO and over their public lives. Despitethe fact that, on average, the private sample firms remain in the public marketfor over thirteen years, firms that ultimately go private are very different anddiscernable in information and liquidity considerations, relative to firms thatremain public, throughout their public life and even at the time of the IPO.We therefore examine how inherent differences at the time of the IPO and theevolution of firm characteristics drive a firm to go private.

In our primary analysis, we use a Cox proportional hazard model, whichdetermines the hazard (probability) a firm will go private given its initial firmcharacteristics and their evolution over time, relative to other firms. In doingso, we reveal many novel insights into why firms go private and shed lighton the relevant costs and benefits of being a public firm. We show that firmsare more likely to go private when it is more costly to produce information,and thus, they have less information available about them in the public mar-ket. Specifically, firms are more likely to go private if they have less analystcoverage and lower institutional ownership. We further confirm these resultsby showing that firms are more likely to go private if they have more concen-trated ownership or more mutual fund ownership. Additionally, we documentthe importance of liquidity to a public firm. We find that firms are more likelyto go private when they are less liquid relative to the comparison sample. We

1773

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 4: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

also find strong support for the importance of high levels of free cash flow indetermining who goes private, but only in the 1980s. We find evidence thatfinancially constrained firms are more likely to remain public (supporting theimportance of access to capital) and that firms that go private engage in feweracquisitions over their public life, suggesting support for the role of controlconsiderations in the public–private choice. Further, we explore the impor-tance of macroeconomic forces using the Cox model and find that our resultsare robust in controlling these market forces and that many macrovariables areimportant in a firm’s decision to go private.

The results from the hazard model show that information, liquidity, accessto capital, and control factors determine a firm’s hazard rate of going private.These results derive from both the importance of cross-sectional differencesbetween firms and the evolution of firm characteristics over time. As discussedearlier, many of these differences were apparent at the IPO. Thus, to better un-derstand the inherent differences in firms that go private and those that remainpublic, we examine the effect of the starting point in a firm’s public life. Todo this, we estimate a logit model using explanatory variables only at the yearfollowing the IPO to predict if a firm will ultimately go private. The resultsare striking. We find that firms that are more likely to ultimately go privatehave less analyst coverage, less institutional holdings, more concentrated own-ership, and more mutual fund ownership at the time of the IPO compared tofirms that remain public, supporting the importance of information consider-ations in the choice between being public or private. We also find that firmsthat go private are more illiquid and have less share turnover, supporting theimportance of liquidity issues. These results are surprisingly similar to thosefrom the hazard model even though we utilize only the initial conditions in thelogit model and not the entire time path of evolution of firm characteristics.However, the access to capital and control considerations are not significant inthe logit model. Since the Cox model utilizing the entire time path of evolutionof these factors showed them to be important determinants of the going-privatehazard, this result suggests that both of these factors become increasinglyimportant in the choice of being a public versus a private firm over a publicfirm’s life.

The logit results highlight the importance of the initial, intrinsic character-istics. However, it is not just the initial conditions that determine if a firm willultimately go private but also the evolution of these conditions relative to thefirms that remained public. To show this, we reestimate our hazard model usingalternative comparison samples. These comparison samples consist of firmsthat are in the same industry and went public in the same calendar year as thegoing-private firm. Further, we match the comparison sample by firm charac-teristics shown to be significant in the logit analysis (such as analyst coverage,market to book, market capitalization, and turnover). This enables us to matchthe starting conditions for both samples and examine the importance of therelative evolution of factors suggested by theory. The results are similar to the

1774

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 5: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

results from the original hazard model, which did not match firms on initialconditions. We conclude that in addition to the initial conditions, the evolutionof information, liquidity, access to capital, and control factors over time alsodetermines if a firm will opt out of the public markets.

Our analysis examines why firms go private. Since there are multiple ways afirm can go private, our sample includes transactions that are driven by privateequity firms, management, and private operating firms. In the main analysis,we treat these deals uniformly. However, it is possible that the factors drivingfirms to go private differ by the parties involved in the transaction. We thereforecollect data on the parties that drive each of these transactions and examinehow motives for going private differ (or not) by subgroup. We find that mostof the factors impact each subgroup similarly. Regardless of how a firm goesprivate, information, liquidity, and growth (measured by market-to-book ratio)considerations are important. The key differences between the subgroups arethat, for the private equity subgroup, firms are more likely to go private if theyhave high leverage. For the private equity and management subgroups, firmsare more likely to go private if they have low research and development (R&D)expenses or low capital expenditures.

The previous analysis shows the importance of several factors in determin-ing the choice between being a public or a private firm. To determine thestrength of these tests, we evaluate out of sample, the accuracy of our haz-ard and logit models predictions on which firms will go private. Specifically,we use the 1980–1990 sample period to estimate our hazard model and thenuse this model to predict who will go private in the following five years. Werepeat this analysis rolling forward by estimating the model during the 1980–1991 period and predicting the following five years and so on. Using a hazardrate greater than the median as our threshold to classify firms as forecasted bythe model to go private, we accurately predict about 83% of the going-privatedecisions. A similar analysis using the logit model accurately predicts whowill go private 82% of the time. We further assess the accuracy of each modelusing a receiver operating characteristic (ROC) curve analysis, which plots thetrue-positive rate against a false-positive rate threshold of the model, which isvaried. The area under this curve is a measure of the ability of the model to cor-rectly classify a firm as going private or not without requiring the researcherto specify a threshold; thus, this provides an objective measure of the model’sability to accurately predict if a firm will go private. Based on this measure, thehazard model has a 78% and the logit model a 71% accuracy. These percent-ages can be compared to a benchmark of 50% accuracy with a random guessand reflect substantial improvement over a naive model.

The primary contribution of this article is to examine how firms weigh thecosts and benefits of being public in the decision to go private. The sample andresearch design we use allows us to examine a broader set of the reasons why afirm may go private, thus testing many implications of the theories of why firmsgo public. We examine both a sample that chooses to be private and one that

1775

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 6: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

remained public and tracks the firms in similar event time. This research designconfers two crucial advantages: (i) ex ante data on the going-private firms andthe comparison sample are readily available due to the disclosure requirementsof public markets; and (ii) measuring ex ante the impact of market forces, suchas liquidity and information production, on the public–private decision is easybecause the firms are traded on a public exchange. These market forces arecentral to many theories but cannot be tested using data on private firms thatare considering going public because market forces are not measurable untilafter it is public. Thus, this article potentially contributes to both the literatureon why firms go private and on how firms weigh the costs and benefits of beingpublic versus private that are modeled in the large IPO literature.

The remainder of the article is organized as follows. We describe ourhypotheses in Section 1. Section 2 discusses data and sample selection. InSection 3, we present summary statistics, and in Section 4, we discuss methodsand present results. We conclude in Section 5.

1. Theoretical Predictions and Tests

The decision between being a public or a private firm is complex, with manyfactors to be considered. In this article, we draw on theories of why firms gopublic, as well as the literature on going private, to examine how firms weighthe costs and benefits of being a public firm in the decision to go private. Thetheories of why firms go public provide insight into why firms go private be-cause they examine many of the costs and benefits of being a public firm. Ofcourse, the costs of going public are also represented in these models. How-ever, these factors are not redeemable and thus, given the nature of our reverseresearch design, we do not relate to this study. Thus, this article does not dis-cuss the relative importance of these costs and benefits going public.

In this section, we summarize the theories of why firms go public, focus-ing on the costs and benefits of being a public firm and specifying the testableimplications as they relate to the decision to go private. We also discuss otherfactors that may influence a firm to go private that do not directly relate to thegoing-public decision. We draw on both of these literatures to formulate ourhypotheses. We categorize the various theories based on the driving force be-hind the models into five groups: (i) information considerations; (ii) access tocapital; (iii) control; (iv) liquidity; and (v) agency and other considerations. Inthis last group, we discuss factors of particular importance to firms decidingto go private. In addition to the theory, we discuss the empirical implicationsfrom these categories of models. As we do so, we refer to several empiricalproxies used in this study. It is important to note that many of the empiricalpredictions are not mutually exclusive among the various theories. The sourcefor this data and a detailed description of each variable are discussed in Sec-tion 3 and Appendix A. A summary of the (reverse) predictions and empirical

1776

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 7: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

proxies for the costs and benefits of being public (groups (i) through (iv)) isprovided in Appendix B.

1.1 Information considerationsWhen a firm is public and widely held, investors may be less informed thaninsiders, and information may be costly to obtain. Several theories highlightthe importance of the information environment and its effect on the costs andbenefits of being a public firm. In this subsection, we discuss several of thesetheories, dividing them by the key economic driver in the model.

1.1.1 Adverse selection. Investors are less informed than the issuers aboutthe true value of the companies going public; thus, the investors’ concern thatthey might be purchasing a “lemon” adversely affects the average quality of thecompanies seeking a new listing and the price at which their shares can be soldinitially and during the public life of the firm. Leland and Pyle (1977) showthat entrepreneurs can signal the quality of their projects by investing moreof their wealth into these projects; thus, private firms do not face as severeadverse selection costs. This adverse selection cost is a more serious obstacleto young, small companies that have low visibility (low analyst following andlow institutional ownership) and concentrated ownership; thus, these firms willbe more likely to go private to avoid the adverse selection costs.

1.1.2 Duplicative monitoring. Chemmanur and Fulghieri (1999) highlightthe costs of duplication of information production by a large number of in-vestors in public firms.4 These costs can be mitigated by the availability of apublic price that conveys information to all investors so that only a fractionof the investors will incur the information production cost. Ultimately, thesecosts have to be borne by the firm. Their model suggests that once a firm ispublic, if information production costs of outsiders increase or the stock priceis not able to aggregate information effectively due to lack of liquidity, thenmore firms would choose to go private, since the value of the firm does not

4 Ritter (1987) estimates that variable costs, which are yearly layouts on auditing, certification, disseminationof accounting information, stock exchange fees, etc., are about 7% of the gross proceeds of the IPO. Further,disclosure rules in public markets that force companies to part with private information necessary for their com-petitive advantage might be an important consideration in the going-public decision, as pointed out by Campbell(1979) and Yosha (1995). Maksimovic and Pichler (2001) develop a model of the going-public decision drivenby product market competition between innovative private firms in an industry. Raising capital in the equitymarket by going public allows a firm that is an industry leader to raise external capital at a cheaper rate thanprivate financing, thus allowing it to implement its project at its optimal scale. However, going public has thedisadvantage of releasing confidential information to competing firms, which can then compete more effectivelywith the firm going public. This theory and the importance of variable costs imply that the advent of SOX andthe associated disclosure requirements might also prompt firms to go private, since confidentiality is a deterrentfor obtaining funding in public markets.

1777

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 8: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

accurately reflect available information about the firm. We therefore predictthat firms where information is more costly or less available and firms withlow liquidity will be more likely to go private. The variables discussed abovethat exacerbate adverse selection costs also increase the cost of monitoring. Wewill further discuss proxies for liquidity in Section 1.3.

1.1.3 Serendipitous information production. Subrahmanyam and Titman(1999) discuss how the public market provides a trade-off between costs of du-plication of information and the benefits of serendipitous information (definedin their model as the information that stock market investors by chance comeacross in their day-to-day activities). Serendipitous information, though noisy,is likely to be diverse across market participants; thus, when aggregated acrossmany investors, it can provide a useful signal. They predict that firms preferto be public when the benefit of this signal outweighs the cost of duplication.This logic suggests that as the costs of generating serendipitous informationincrease, firms would choose to go private. They suggest that serendipitous in-formation is less available in high-tech firms, such as those with high R&Dexpenditures. We therefore predict that high R&D firms are more likely to goprivate.

1.1.4 Investor recognition. Merton (1987) provides an extension to CAPMthat relaxes the assumption of efficient information for all investors and showsthat expected returns decrease with the size of investor base, which he char-acterizes as the “degree of investor recognition.” Thus, the benefit of beingpublic is diminished for firms with high ownership concentration or lower in-vestor recognition, and these firms are more likely to opt to be private.

1.2 Access to capitalThe opportunity to tap public markets for equity capital is appealing for highgrowth firms with large current and future investments that may have limitedaccess to other financing alternatives due to high leverage or high transactionscosts and is a leading reason why firms go public (Kim and Weisbach 2008).Thus, firms that do not have large investments and future growth opportunitiesare more likely to go private.5

A related motivation for going and being a public firm is to minimize thecost of capital for the firm and thus maximize the value of the company. Thelower the cost of capital in the public versus the private market, the greater theincentive to be a public firm. This argument has been advanced by Modiglianiand Miller (1963) and Scott (1976). This argument suggests that as the cost of

5 Shah and Thakor (1988) also predict that firms with more favorable investment opportunities prefer to be public,but in their model, this is driven by risk sharing in public companies.

1778

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 9: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

capital for firms increases in public markets (e.g., due to increased informationproduction costs or lower liquidity), firms are more likely to go private.

The above discussion also suggests that financially constrained firms wouldprefer to be public to fund their investment opportunities since (i) visibilityin the public markets can be a prerequisite to raise further (even nonequity)capital, as suggested by Pickens (1987); and (ii) their public status allowsthem to enjoy competition among suppliers of finance and to negotiate privatecapital at better rates (Roell 1996) than they would be otherwise. We measurethe degree of financial constraint by whether or not a firm paid a dividend(non-dividend-paying firms are financially constrained) or by the KZ Indexdefined in Appendix A. We expect non-dividend-paying and high KZ Indexfirms (financially constrained) to be less likely to go private. The empiricalproxies for these theories are summarized in Appendix B.

1.3 LiquidityAmihud and Mendelson (1988) were the first to emphasize the importance ofliquidity considerations for a public firm. Bolton and von Thadden (1998) em-phasize the trade-off between liquidity and control in determining corporateownership structure. Boot, Gopalan, and Thakor (2007) also study a liquidity-control trade-off but identify a manager’s control benefit as the ability tomake investment decisions despite the disagreement of outside shareholders.Zingales (1995) and Mello and Parsons (1998) posit that the role of an IPO isto establish the market price/value for a firm and thus provide liquidity. Basedon the theoretical implications of these models, liquidity is therefore a bene-fit to being a public firm that is considered in the choice between public andprivate ownership. Further, share trading on an exchange is cheaper comparedto bilateral trades, and this liquidity benefit (which is an increasing function ofthe trading volume) leads companies to go public. We therefore predict that,as the liquidity benefit in the market deteriorates, firms are more likely to goprivate.

1.4 Control considerationsZingales (1995) argues that an IPO can serve as the first step toward selling acompany at an attractive price through a takeover. In going public, the initialowner sells a portion of his cash flow and control rights. Many of the importantaspects of this theory are difficult to capture in our cross-sectional analysis.However, it suggests that if the firm is not very active in the market for corpo-rate control, it will more likely go private. The spirit of the theory, that is, theimportance of value and control issues, is also highlighted in Brau, Francis, andKohers (2003) and Brau and Fawcett (2006), who suggest that an IPO createspublic shares for a firm that may be used as currency in acquiring other com-panies or in being acquired in a stock deal. This reasoning again suggests thatif the firm is not active in the market for corporate control using stock deals,

1779

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 10: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

it is likely to go private. In testing these predictions, we examine how activelya firm acquires other firms. Further, Zingales shows that the value of the cashflow rights is more sensitive to change in market conditions than the controlrights. Thus, we also examine a firm’s value as measured by the market-to-book ratio. In doing so, we examine the importance of control and the impactof the price on going private. Additionally, as firms may need capital to financean acquisition, this hypothesis is related to the access to capital hypothesis.

1.5 Agency considerationsEach subsection above discusses implications from theories of going public onthe decision to go private. Prior to much of this literature, researchers directlyexamined the value implications of the decision to go private. Given the se-quential nature of these research streams, the literature on going private doesnot address the theories of going public. Our goal in this article is both to testthe above-mentioned theories in order to understand how firms weigh the costsand benefits of being a public firm and to enhance our understanding of whyfirms go private. Thus, we also will consider some reasons why a firm may goprivate, specific to the going-private decision.

The literature on going private suggests that LBOs lead to efficiency gainsbecause of higher debt payments and alignment of the management incentivesby increased equity positions (Jensen 1986). Thus, one motive for going privateis to improve the incentive alignment and governance structure of the firm.This motivation is particularly important for high free cash flow firms, whichcan more easily destroy value by wasting accessible cash flow. This motive forgoing private is well supported in Lehn and Poulsen (1989). We proxy for theseconsiderations empirically, consistent with Lehn and Poulsen (1989), by usingfree cash flow to assets, firm leverage, cash to assets, and tangibility of assetsof the firm. Many of these variables also measure a firm’s need for capital andthus are relevant to the access to capital hypothesis as well.

In addition to the factors discussed above, macroeconomic factors may alsoinfluence a firm to go private. We therefore also examine the impact of theseinfluences. These factors are described in detail in Section 4.3.

2. Data and Sample Selection

To construct our sample, we rely on the legal definition of going private asdefined by the SEC. Based on this, “A ‘true’ going private transaction...is oneby which an individual or a group of individuals controlling a public corpora-tion...undertakes a corporate transaction in order to acquire...the entire equityinterest in the corporation” (Borden and Yunis 2007, pp. 1–3). The structuraltest to deem a transaction going private is defined by Rule 13e-3 promulgatedunder the Securities Exchange Act of 1934, which went into effect September7, 1979. According to this rule, the purchase must be by an affiliate of the com-

1780

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 11: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

pany as defined under the 1934 act (Koenig 2004; Borden and Yunis 2007).Affiliates include insiders and entities under the common control of the com-pany. In practice, the definition of an affiliate is quite broad and can be invokedeven if management does not participate in ownership, but there is a highdegree of continuity of management. This raises the question of what types oftransactions fit the legal definition of going private. The short answer is thatboth management buyout (MBO) and LBO by a financial sponsor in which aschedule 13e-3 is filed are technically going-private transactions. According toBorden and Yunis (2007, pp. 3–12): “In LBOs...management is almost invari-able invited to participate...(and) it has become practice to treat the transactionsas subject to the Rule (13e-3).” This rule also includes transactions where aprivate operating company acquires a public firm and a schedule 13e-3 is filed.

To construct our sample of going-private firms, we therefore search SECfilings for all 13e-3, including 13e-3, DEF13e-3, and PRE13e-3, submissionsfrom 1980 to 2003. Using the 13e-3 filing to designate going-private transac-tions has also been used by other papers, such as Engle, Hayes, and Wang(2004) and Leuz, Triantis, and Wang (2005). Examining the filing of theDEF13E3 and PRE13E3 is less common but important during the early partof our sample. These are form types that are no longer filed, which may bewhy they have not been included in other studies. The DEF13E3 and PRE13E3forms are filed as part of proxy statements but contain 13e-3 materials. AfterJanuary 24, 2000, some firms provide the Schedule 13e-3 information as partof Schedule TO, the tender offer statement, so we search these documents aswell. We also retrieve SEC Form 15 and Form 25 filings involving notificationof termination of security registration to make sure that the deal is completed.We supplement our sample with that used in Lehn and Poulsen (1989) to en-sure that we do not miss any early transactions used in previous studies.

After collecting going-private transactions, we screen all sample firms toverify that they are no longer trading and available on CRSP within twenty-four months of initial 13e-3 submission. In doing so, we exclude any firmfrom the private sample that was dropped from CRSP because it was moved totrading on an exchange not covered by CRSP. This includes those trading overthe counter or on pink sheets. Thus, our sample does not include those firmsthat have “gone dark,” as described in Leuz, Triantis, and Wang (2005). Theseauthors examine why firms choose to delist rather than go private, whereaswe examine why a firm goes private rather than remain public. Though thereis some overlap in the reasons a firm goes dark versus goes private, Leuz,Triantis, and Wang show that firms that go dark are significantly different fromfirms that go private. Thus, the questions of interest as well as the methodsemployed differ between our study and theirs. If a firm meets the criteria de-scribed above, we designate the firm as going private. The top half of the firstcolumn of panel A of table 1 details the breakdown from these data sources.

1781

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 12: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

Table 1Summary statistics, Panel A: Sample selection

Sample source/screen Number of firms13e-3 filings only

Edgar only 1,121Lehn and Poulsen (LP) only 92Edgar / LP Common 164Total base sample 1,377

Less: firms not on Compustat −152in years between IPO and going privateLess: firms with 6 or 49 SIC Code −176Less: firms where IPO date is after going-private date −26Total sample 1,023

Table 1Panel B: Industry distribution

S.No. Description Number of firms13e-3 filings only

1 Consumer nondurables 1512 Consumer durables 413 Manufacturing 1854 Energy 535 Chemicals 206 Business equipment 1137 Telecom 328 Utilities 09 Shops 19510 Healthcare 4611 Money 012 Other 187

Total 1023

Our base sample includes 1,377 firms that go private between 1980 and 2003.6

We focus on this full sample throughout most of the article but will discusssubsamples based on the type of transaction (LBO, MBO, or acquisition by aprivate operating company) in Section 4.7.

The sample described above follows the legal (SEC) definition of going pri-vate and is the primary sample used in our analysis. We rely on this sam-ple because the filing makes it unambiguous as to who is going private andthe legal definition is consistent with our question of interest: Why does afirm choose to go private? However, this sample may exclude some deals thatare commonly considered going-private transactions. For instance, the KKR-Nabisco LBO did not fall under rule 13e-3; thus, some LBOs and other (possi-bly more hostile) transactions involving private operating companies will notfall under 13e-3. Why do some deals fall under 13e-3 and others do not? Asmentioned above, a 13e-3 legal going-private transaction involves an affiliate.The rule broadly defines an affiliate as “a person that directly or indirectly

6 Note that this sample size is comparable to recent studies by Engle, Hayes, and Wang (2004), who have 343firms that went private between 1998 and 2003, and the going-private comparison group used in Leuz, Triantis,and Wang (2005), which has 406 firms that went private between 1998 and 2003. Of the 1,377 firms in our initialsample, 415 occur between 1998 and 2003. We further detail our sample by year in panel C of table 1, but thistable contains only firms that pass various screens and are therefore not comparable to the sample descriptionsprovided in these papers.

1782

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 13: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

Table 1Panel C: Time-series distribution

Year Going-private IPO year—going- IPO year—comparisonsample private sample sample

Before 1970 2171970 17 661971 35 851972 196 1951973 12 261974 3 31975 17 51976 13 291977 7 191978 12 251979 9 511980 21 24 731981 41 28 2091982 53 15 761983 49 52 3881984 78 29 1811985 69 21 1771986 72 43 3071987 58 26 2511988 80 9 1081989 53 11 1341990 42 11 1161991 16 21 2541992 14 13 3571993 11 27 4661994 20 24 4051995 26 23 4281996 15 30 6161997 30 39 3881998 35 13 2281999 54 19 3972000 54 6 3322001 48 1 692002 362003 372004 11Total 1023 1023 6464

Panel A of this table presents data on the derivation of the going-private sample for the period 1980–2004 from various data sources and used in the article. The second column consists of firms ob-tained from 13e-3 filings. Panel B reports the industry distribution of the going-private sample of firms.We use the twelve industry classification of Fama and French (available at Ken French’s Web sitehttp://mba.tuck.dartmouth.edu/pages/faculty/ken.french) to classify the firms. Panel C presents the number offirms in the sample that went private in each year in the first column using the 13e-3 filings sample. Resultsare very similar if we report using the sample that includes firms from the SDC database as well. The secondcolumn reports the time-series distribution of the same firms based on their IPO year. The third column reportsthe time-series distribution by IPO year of all other firms that did not go private until the end of our sampleperiod and is labeled the comparison sample.

through one or more intermediaries controls, is controlled by or is under com-mon control with the issuer” (Borden and Yunis 2007, pp. 10–9). In fact, anaffiliate can become an affiliate “through means other than equity ownershipas part of the overall sale.” For instance, retaining or increasing compensationof management may be viewed as recognition of their importance and deemthem an affiliate (pp. 10–3). Further, parties can claim an exemption fromfiling 13e-3. Two common exemptions under the rule include 13e-3g(1) and

1783

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 14: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

13e-3g(2). An exemption under g(1) relates to two-step clean-up transactionsand under g(2) relates to if securities are issued in partial payment for the trans-action. Additionally, other exceptions include those under g(3), g(4), and g(5),which relate to public utilities; g(10), which relates to transactions in class ac-tion litigations; and g(12), which allows the court to make exceptions to therule (Borden and Yunis 2007, pp. 10–9 to 10–30). However, as also explainedin Borden and Yunis, questions as to the effect of the rule normally relate towhether or not an affiliate is involved or if a transaction is one of a series.

Because some of these exceptions may hold for transactions that aresimilar in spirit to going-private deals in our sample, we supplement oursample using data from the SDC database in which SDC flags the transactionas “going private.” This adds 377 deals after applying data filters to the sample.We refer to this as our “broader sample” and repeat the entire analysis usingthis broader sample even though our preferred sample in terms of data qualityand reliability is based on filings with the SEC. These analysis are availableas a Web appendix and are qualitatively similar to the results presented in thisarticle. At the same time, our data investigations suggest that there are prob-lems for going-private studies that rely exclusively on SDC data. As an exam-ple, of the 1,377 deals in our primary sample, the SDC did not include 997%or 72% of the total deals. This means that many firms missed by the SDC filed13e-3 submissions and are clearly under the legal definition of going private.We believe, based on the results in this article, that the best empirical strategyin going-private studies is to use 13e-3 filings as the primary sample. One canalways supplement it with the SDC database for transactions that fall outsidethe ambit of 13e-3 but check for consistency of results between the legal andbroader samples.

We gather an IPO date for all firms that went private from three sources toensure that we do not lose sample observations. First, we use the SDC, ex-cluding those with offer prices under five dollars, unit offers, ADRs, or REITs.If the date is missing from this source, we use Jay Ritter’s database providedon his Web page (http://bear.cba.ufl.edu/ritter/ipodata.htm). If the date is notavailable from either of these sources, we use the first date the firm was listedon the CRSP, checking this with the Jovanovic and Rousseau (2001) data forthe pre-CRSP period. For robustness, we repeat all analysis excluding firms forwhich the IPO date is obtained only from Jovanovic and Rousseau (2001) orthe CRSP listing date and all of our results survive. In twenty-six cases in boththe primary and broader samples, the IPO date we obtain is after the going-private date, and we delete these firms from our sample. We also exclude 152cases where Compustat data are not available and 176 cases that have eithera one-digit SIC Code of 6 (financial institutions) or a two-digit SIC Code of49 (utilities) in the primary sample. This results in a primary sample of 1,023going-private transactions (“private sample”). This sample construction pro-cess is described in table 1, panel A.

1784

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 15: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

Panel B of table 1 describes the industries in which the sample firmsoperate, using the twelve industry classifications from Ken French’sWeb page (http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/datalibrary.html). The most prominent industries are retail and manufacturing,although no one industry dominates. However, though not presented here, theprevalence of manufacturing (business equipment) decreases (increases) overthe sample period. Panel C describes the year in which the primary samplefirms went private (column 1) and public (column 2). Those transactions thatoccur in 2004 were filed in 2003. There is an increase in firms going privateduring the mid-to-late 1980s and a resurgence after 1998.

We also construct a comparison sample using Compustat data. We follow thesame procedure we used to collect IPO dates for the private sample. Since allof our private sample firms went public before 2002, we limit our comparisonsample to IPOs occurring before 2002. To do so, we include all IPOs fromthe SDC, Jay Ritter’s Web site, and the Jovanovic and Rousseau (2001) data,excluding those in the broader going-private sample. Specifically, we includeall IPOs listed on the SDC with an offer date between 1980 and 2001, exceptthose with offer prices under five dollars, unit offers, ADRs, or REITs. Wesupplement this list with all IPOs not on the SDC from the 1970s from JayRitter’s Web site and the Jovanovic and Rousseau (2001) data. Similar to ourprivate sample, we also exclude financial institutions (one-digit SIC Code of6) or utilities (two-digit SIC Code of 49). This results in 6,464 comparisonIPOs.7 The third column in panel C of table 1 describes the year in which a setof comparison firms went public.

We chose our comparison group to provide a large sample with a clear start-ing point. However, we also consider two alternative comparison samples. Onealternative comparison sample is to include all firms in Compustat and not limitthe sample by IPO date. We chose to use the IPO comparison sample becausemany of our tests line up the two samples in IPO event time, and this com-parison is not possible for firms in Compustat for which we do not have anIPO date. However, it is possible to consider the first time a firm is listed onCRSP as the begin date. We therefore repeat the analysis in this article usingall firms on Compustat and assuming that the first date they appear on CRSP istheir begin date. The results are robust to the use of this alternative comparisonsample. For brevity, we do not present these results, but they are available fromthe authors upon request. A second alternative for the comparison firms is tocreate a matched sample at the time of the IPO based on firm characteristics.We examine the robustness of our results relative to several samples matched

7 Unlike the going-private sample, our comparison sample begins with IPO dates in 1970, because we use CRSPstart dates for the private sample to prevent losing sample firms but not for the comparison sample. All analysispresented in this article is qualitatively unchanged whether we exclude all firms that went public before 1970 orwe expand our comparison sample with the sample of firms for which Jay Ritter’s database provides CRSP startdates as IPO dates.

1785

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 16: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

Table 2Age distribution and comparison of firm age at IPO of going-private sample and comparable IPO firms

Age No. of firms Age at IPO Age at IPONo. of years Going-private sample Going-private sample (A) Comparison sample (B)1 3 40 3492 52 28 3593 74 24 5174 71 39 4505 65 18 3896 59 22 3327 42 29 3158 35 14 2609 41 21 21310 47 22 17511–19 332 106 95920–29 129 61 35330–39 30 50 14540–49 12 31 10350–59 19 18 8360–69 9 22 6070–79 2 15 5480–89 1 7 4790–99 5 34100>= 6 41Mean 13.5 21.8 13.8Median 11 12 7

t-stat for difference in means (B)–(A): −9.4∗∗∗z-stat for difference in medians (B)–(A) : −9.5∗∗∗This table presents the distribution of firm age. The first column presents the age of the firm in public markets(number of years the firm stayed public before becoming private) for the going-private sample. Column Apresents the distribution of firm age at the time of the IPO for the going-private sample. Column B presents thesame information for the comparison sample. t-stat (z-stat) reports the results of a test of difference in means(medians) between the (B) and (A) columns and is equal to zero. ∗∗∗,∗∗, and ∗ indicate significance at the 1%,5%, and 10% levels, respectively. Note that we do not have firm age information for all firms in our sample.

by IPO year, industry, and several firm characteristics and present these resultsin Section 4.5.

3. Summary Statistics

In table 2, we examine the age of the firm when they go public and the numberof years the firms are public before going private. The first column shows thenumber of years the private sample firms were in the public markets (i.e., num-ber of years from IPO to going private). On average, the firms in our privatesample were public for about 13.5 years. The median number of years the firmsare public is eleven years. The second column shows the age of the samplefirms at the time of the IPO for those firms for which these data were available.The average (median) sample firms was 21.8 (12) years old at the time of theIPO. We contrast this with the comparison sample. Column 3 shows that thesecomparison firms were much younger when they went public, with an average(median) age of 13.8 (7) years. As noted in the table, the age of the privatesample firms is statistically greater than that of the comparison firms.

1786

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 17: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

In table 3, we compare firm characteristics between our sample of going-private firms and the comparison sample in the year following the IPO date,labeled the IPO year. We divide this table by the hypotheses being tested. Theresults are striking: the going-private sample is significantly different from thecomparison firms at the IPO on several dimensions, despite this time being,on average, 13.5 years before the actual decision. The first section providessupport for the information consideration hypotheses: the going-private firmshave significantly less analyst following and less institutional ownership.8

These results indicate the importance of information asymmetry and infor-mation gathering costs to a firm’s choice between being private or public.The sample firms also are larger based on median sales and median assets butsmaller based on mean or median market values. The sample firms also havelower mean and median R&D expenditures.

The second panel of table 3 examines the characteristics related to the pre-dictions of the access to capital hypothesis. The statistics show that the privatesample firms have lower capital expenditures, indicating that they have fewerinvestments and less need for capital. Thus, it may not be surprising that theyultimately abandoned public life and went private. Additionally, a greater per-centage of the going-private firms pay a dividend, which may indicate that theywere less constrained. However, the constraint dummy variable based on theKZ Index is not significant.9

The third panel of table 3 shows that the sample firms are significantly lessliquid with less turnover at the time of the IPO, indicating the importance ofliquidity to the decision of being public. The last two sections present a numberof interesting statistics. The sample firms engage in a statistically similar num-ber of mergers and acquisitions as the comparable firms, in the year followingthe IPO, indicating that there is no difference in the control activities or needfor capital to finance acquisitions, initially. Consistent with Lehn and Poulsen(1989), the going-private firms have higher free cash flow than the compara-ble firms; thus, they may benefit from going private to reduce potential agencyproblems. The free cash flow results further support the importance of accessto capital on the decision to go private, in that these firms have less need forexternal funding. This panel also presents statistics indicating that the sam-ple firms have significantly lower market-to-book ratios, higher leverage, lesscash, and more tangible assets in contrast to the comparison sample of firms.

Next, we examine how the firm characteristics of the private sample evolveover time. We do this in two ways. In table 4, we examine several firm charac-teristics in the first year following the IPO date along with the year prior to the

8 In unreported results, we also verify that going-private firms have more concentrated institutional ownership,fewer funds holding the stock, and more informed trading as measured by PIN (Easley, O’Hara, and Hvidkjaer2004), consistent with less information production.

9 In untabulated results, we replace the KZ Index Dummy variable with a constraint dummy based on the Whitedand Wu (2006) financial constraint variable. The results are similar to those presented here with the KZ Index:the constraint dummy using the Whited and Wu measure is not significant.

1787

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 18: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

Table 3Comparison of firm characteristics at IPO—going-private sample vs. comparison IPO sample

Measure Private firms Comparison sample Differencet= IPO year t= IPO year t-stat (z-stat)

(A) (B) (B)–(A)Information variablesSales 246.72 248.96 0.1

91.41 40.07 −14.1∗∗∗Assets 227.85 303.93 0.9

72.79 51.35 −8.0∗∗∗Market Value 303.30 505.39 2.2∗∗

85.16 100.12 2.6∗∗∗Analyst Coverage 0.78 1.10 3.9∗∗∗

0.00 0.00 4.4∗∗∗Institutional Holdings 11.87 13.84 2.3∗∗

4.26 7.06 1.6R&D/Sales 0.12 1.34 1.9∗

0.00 0.00 15.7∗∗∗

Access to capitalCapex/Sales 0.21 0.82 1.9∗

0.04 0.06 8.3∗∗∗KZ Index Dummy 0.45 0.42 −1.3

0.00 0.00 −1.3Dividend Dummy 0.35 0.15 −14.5∗∗∗

0.00 0.00 −14.3∗∗∗

Liquidity variablesIlliq 0.03 0.01 −8.0∗∗∗

0.01 0.00 −13.0∗∗∗Turnover 0.05 0.08 6.4∗∗∗

0.03 0.05 11.9∗∗∗

Control considerationsMerger Count 0.25 0.30 1.1

0.00 0.00 −0.1Market-to-Book 2.11 3.27 7.0∗∗∗

1.55 2.22 14.4∗∗∗

Agency considerationsFree cash flow/Assets 0.05 −0.03 −9.0∗∗∗

0.07 0.05 −9.9∗∗∗Leverage 0.16 0.11 −8.0∗∗∗

0.12 0.02 −11.6∗∗∗Cash to Assets 0.18 0.35 17.2∗∗∗

0.10 0.29 16.0∗∗∗Net Fixed Assets/Assets 0.31 0.21 −12.2∗∗∗

0.26 0.13 −13.5∗∗∗

This table presents firm characteristics for the going-private sample (column A) and the comparison sample(column B) at the same point in time (i.e., at the time of the IPO). t-stat (z-stat) reports the results of a test ofdifference in means (medians) between the two columns and is greater than 0. ∗∗∗,∗∗, and ∗ indicate signifi-cance at the 1%, 5%, and 10% levels, respectively.

going-private date for the private sample. In doing so, we hope to add insightas to how the costs and benefits of being public changed over the firms’ publiclife. As before, we divide table 4 into panels by hypotheses. In figure 1, weshow how these characteristics change by presenting mean firm characteris-tics for the five years following the IPO and five years preceding going private(for only the firms that went private, since comparison firms do not have anend date). Because some firms were not public for more than a few years, we

1788

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 19: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

Table 4Going-private sample—comparison of firm characteristics at IPO and going-private years

Measure Private firms Private firms Differencet= IPO year t= Private year t-stat (z-stat)

(A) (B) (B)–(A)Information variablesSales 246.72 588.20 10.0∗∗∗

91.41 149.79 15.1∗∗∗Assets 227.85 554.00 8.6∗∗∗

72.79 114.92 14.7∗∗∗Market Value 303.30 395.37 0.4

85.16 60.85 −0.3Analyst Coverage 0.78 2.24 10.0∗∗∗

0.00 1.00 9.9∗∗∗Institutional Holdings 11.87 18.01 8.9∗∗∗

4.26 11.09 8.7∗∗∗R&D/Sales 0.12 0.11 −0.2

0.00 0.00 2.1∗∗

Access to capitalCapex/Sales 0.21 0.08 −3.4∗∗∗

0.04 0.03 −7.3∗∗∗KZ Index Dummy 0.45 0.31 −4.6∗∗∗

0.00 0.00 −4.5∗∗∗Dividend Dummy 0.35 0.38 −2.9∗∗∗

0.00 0.00 −2.2∗∗

Liquidity variablesIlliq 0.03 0.04 0.9

0.01 0.01 0.7Turnover 0.05 0.03 −4.7∗∗∗

0.03 0.01 −5.9∗∗∗

Control considerationsMerger Count 0.25 0.23 0.9

0.00 0.00 0.9Market-to-Book 2.11 1.28 −11.4∗∗∗

1.55 1.08 −14.3∗∗∗

Agency considerationsFree cash flow/Assets 0.05 0.04 −2.2∗∗

0.07 0.06 −4.9∗∗∗Leverage 0.16 0.20 5.9∗∗∗

0.12 0.17 5.0∗∗∗Cash to Assets 0.18 0.13 −7.4∗∗∗

0.10 0.07 −8.0∗∗∗Net Fixed Assets/Assets 0.31 0.33 3.1∗∗∗

0.26 0.27 3.3∗∗∗

This table presents firm characteristics for the going-private sample at two points in time: the time of the IPO(column A) and the time of going private (column B). t-stat (z-stat) reports the results of a test of difference inmeans (medians) between the two columns and is equal to 0. ∗∗∗,∗∗, and ∗ indicate significance at the 1%, 5%,and 10% levels, respectively.

only include sample firms in figure 1 that were public at least seven years.10

Specifically, the graphs show the first five years of both samples and the lastfive years for the private sample.

10 This analysis is somewhat similar to that in Mikkelson, Partch, and Shah (1997), who examine the operatingperformance and ownership of IPO firms for ten years post-IPO.

1789

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 20: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

0.00

1.00

2.00

3.00IP

O 0

IPO

1

IPO

2

IPO

3

IPO

4

Priv

ate

4

Priv

ate

3

Priv

ate

2

Priv

ate

1

Priv

ate

0

AAnalyst Following

Going Private Comparison

0.00

0.02

0.04

0.06

0.08

IPO

0

IPO

1

IPO

2

IPO

3

IPO

4

Priv

ate

4

Priv

ate

3

Priv

ate

2

Priv

ate

1

Priv

ate

0

BTurnover

Going Private Comparison

0.00

0.20

0.40

0.60

IPO

0

IPO

1

IPO

2

IPO

3

IPO

4

Priv

ate

4

Priv

ate

3

Priv

ate

2

Priv

ate

1

Priv

ate

0

CMerger Count

Going Private Comparison

1790

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 21: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

0.00

1.00

2.00

3.00

IPO

0

IPO

1

IPO

2

IPO

3

IPO

4

Priv

ate

4

Priv

ate

3

Priv

ate

2

Priv

ate

1

Priv

ate

0

DMarket to Book

Going Private Comparison

-0.14

-0.09

-0.04

0.01

0.06

IPO

0

IPO

1

IPO

2

IPO

3

IPO

4

Priv

ate

4

Priv

ate

3

Priv

ate

2

Priv

ate

1

Priv

ate

0

EFree Cash Flow

Going Private Comparison

Figure 1Trends in key variables in event timeThis figure displays the evolution of firm characteristics in event time for the going-private sample and thecomparison sample. We present the cross-sectional means of several variables for firms for five years from theIPO year (for both samples) and five years prior to going private (for the going-private sample).

The first panel of table 4 shows that firms are larger, more followed by an-alysts, and have greater institutional holdings when they go private relative towhen they went public. These results indicate that the information environmentis improving over time. Taken in conjunction with the results from table 3, thisindicates that though firms that go private find the information environment ofbeing public more costly than comparable firms at the time of the IPO, theychange such that relative to themselves they will gain more from being public.While this might indicate that firms are becoming less likely to go private basedon the predictions of the information consideration hypotheses, it is importantto see the changes relative to the comparison sample. We do this in figure 1and in the multivariate analysis that follows. Panel A of figure 1 illustratesthese comparisons and shows that, despite these changes over time, the num-

1791

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 22: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

ber of analysts for going-private firms never reaches the level of comparablefirms five years after the IPO.

The second panel of table 4 presents summary statistics related to the firms’need and access to capital. The results show that at the time of going private,the sample firms have lower capital expenditures and are less constrained,as indicated by the lower percentage of firms with the KZ Index constraintdummy equal to 1 and the greater percentage of firms that pay a dividend.Taken together, these results indicate that these firms have less need for theaccess to capital that a public firm status provides, when they decide to goprivate.

The third panel of table 4 presents two measures of liquidity. The statisticsshow that firms have lower turnover at the time they go private. These resultsare consistent with the hypothesis that firms go private when the relative valueof having a liquid stock price declines. However, the illiquidity measure doesnot differ across the two time periods. Panel B of figure 1 illustrates that thegoing-private sample firm’s turnover remains substantially lower than compa-rable firms over their life.

The last two panels of table 4 present the statistics related to control, valua-tion, and agency costs. The merger activity does not significantly change overthe private firm’s public life but remains below the comparison firm, depictedin panel C of figure 1. The results also show that the market-to-book ratio of thefirms is significantly and substantially lower at the time they go private. Thismay indicate declining growth opportunities or it may indicate a decreasingrelative valuation. As shown in panel D of figure 1, the market-to-book ratioof comparable firms declines as well, but the ratio for the going-private samplefirms remains below that of comparable firms. In the multivariate analysis, wewill include the adjusted stock return to further examine the impact of relativevaluation. Firms also have slightly less free cash flow at the time they go pri-vate. This evidence may seem inconsistent with the agency story supported byLehn and Poulsen (1989) that firms benefit from going private to reduce poten-tial wasting of free cash flow. However, as shown in table 3 and illustrated inpanel E of figure 1, it is how these firms fare relative to the comparison samplethat matters. These firms have considerably more free cash flow than compara-ble firms at the IPO, so the decrease does not appear to be material. This is il-lustrated in figure 1, panel E. Additionally, the results of table 4 show that firmsare more levered, with less cash and more tangible assets when they go private.

4. Why Do Firms Go Private?

The previous univariate analysis shows that firms that go private are systemat-ically different from firms that do not go private, on average, 13.5 years beforethey make this decision, that is, at the time of the IPO. This indicates that alot of what determines the choice between being private and public may beinherent and observable to the firm at least at the time of the IPO. The uni-

1792

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 23: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

variate analysis also shows that many of these firms, characteristics changedover the firms, public life. In this section, we will examine how both of thesedifferences impact the probability that the firm goes private.

4.1 MethodsTo examine how changes in firm characteristics impact the probability of goingpublic, we use a hazard model to investigate if and when a firm goes private.Specifically, we are interested in the length of time it takes to go private and theinfluence of various variables informed by theory on that duration, controllingfor the fact that our comparison firms may go private at some later unobserv-able time. While there are parametric models of duration that are simple, byimposing structure as much as they do, the models might distort the estimatedhazard rate. Since fewer restrictions can result in a better representation, weuse the Cox proportional hazard model, which is commonly employed by re-searchers (see, for example, Shumway 2001 to model bankruptcy prediction),using the Breslow method for ties.

The model to be estimated is

h(t, X (t)) = h(t, 0) exp(β ′ X (t)),

where h(t, X (t)) is the hazard rate at time t for a firm with covariates X (t). Itis important to note that the variables of interest vary with time and that thismodel will examine the effect of differences between firms, as well as changesover time. Second, some of our observations are right censored. That is, atthe end of the sample period, there are some firms that continue to remainin the public markets, even though there is a positive probability that theymay go private. The hazard model is flexible enough to handle both thesecomplications. The Cox regression estimates the coefficient vector β. TheCox proportional hazard model does not impose any restriction on h(t, 0),the baseline hazard. Cox’s partial likelihood estimator provides a way ofestimating β without estimating h(t, 0). A positive coefficient on variable xin the hazard model implies that a higher x is linked to higher hazard rateand thus a lower expected duration. The hazard ratio, which is simply exp(β),tells us how much the hazard (i.e., instantaneous risk) the going private eventincreases for a unit change in the independent variable.

As described in Section 3 and table 1, panel C, our comparison sample doesnot begin until 1970, whereas our private sample begins before 1970. In es-timating the hazard model, we limit our private sample to those with an IPOdate on or after 1970. We do this to make duration or the number of years afirm may be public comparable across the two samples since the hazard modelpredicts the time to going private.

1793

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 24: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

Table 5Cox proportional hazard models for time to go private

(13e-3 filings sample)(1) (2) (3) (4) (5)

Log(Sales) 0.09∗∗∗ 0.09∗∗∗ 0.09∗∗∗ 0.09∗∗∗ 0.17∗∗∗(0.03) (0.03) (0.03) (0.03) (0.03)

Analyst Coverage −0.04∗∗∗ −0.04∗∗∗ −0.04∗∗ −0.06∗∗∗(0.01) (0.01) (0.01) (0.02)

R&D/Sales 0.008 0.007 0.007 0.006 −0.004(0.006) (0.007) (0.007) (0.008) (0.03)

Capital Expenditure/Sales −0.21 −0.23 −0.22 −0.19 −0.15(0.14) (0.16) (0.16) (0.16) (0.18)

Dividend Dummy 0.25∗∗∗ 0.20∗∗ 0.18∗ 0.13(0.10) (0.10) (0.10) (0.10)

Turnover −13.47∗∗∗ −12.70∗∗∗ −12.26∗∗∗ −12.28∗∗∗ −18.06∗∗∗(1.63) (1.61) (1.59) (1.67) (1.83)

Market-to-Book −0.40∗∗∗ −0.40∗∗∗ −0.41∗∗∗ −0.46∗∗∗(0.06) (0.06) (0.06) (0.07)

Free Cash Flow/Assets 0.59∗∗(0.27)

Free Cash Flow post-1990s −0.10 −0.01 −0.25(0.25) (0.26) (0.26)

Free Cash Flow pre-1990s 4.24∗∗∗ 4.09∗∗∗ 4.27∗∗∗(0.57) (0.57) (0.63)

Leverage 0.09 0.11 0.13 0.08(0.21) (0.21) (0.21) (0.23)

Cash/Assets 1.04∗∗∗ 0.96∗∗∗ 0.95∗∗∗ 0.74∗∗∗(0.27) (0.27) (0.27) (0.29)

Net Fixed Assets/Assets 0.26 0.21 0.15 0.004 0.42∗(0.20) (0.20) (0.20) (0.22) (0.22)

Merger Count −0.12∗∗ −0.11∗ −0.10∗ −0.13∗∗ −0.20∗∗∗(0.06) (0.06) (0.06) (0.06) (0.07)

Inst. Ownership −0.009∗∗∗(0.002)

Annual Adjusted Return 0.08 −0.04(0.05) (0.06)

KZ Index −0.28∗∗∗(0.09)

Illiq 0.09(0.08)

Obs. 33979 33979 33979 30718 31258

This table reports the results of Cox proportional hazard models with time-varying covariates based on allvariables. The dependent variable is time to private, which measures the time between IPO of the firm andthe time they go private. All specifications include Fama–French Industry classification controls. The tablereports the coefficients and, in parentheses, the standard errors. Variables are defined in Appendix A. ∗∗∗,∗∗, and∗ indicate significance at the 1%, 5%, and 10% levels, respectively.

4.2 Hazard model resultsWe present the results of the hazard model estimation in table 5. All specifica-tions include dummy variables for each of the Fama–French twelve industriesdetailed in table 1, panel B. Column 1 shows that firms have a greater haz-ard rate of going private if they have less analyst coverage and lower liquidity,supporting the information and liquidity hypotheses. However, large firms alsohave a greater hazard rate of going private. Firms also have a greater hazardrate of going private if they have a low market-to-book ratio, which may in-dicate that the firm has fewer investment opportunities or a low value. Theseresults are robust across all specifications. Based on the evidence in column 1,

1794

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 25: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

firms have a greater hazard rate of going private if they pay a dividend, engagein fewer mergers and acquisitions, and have high free cash flow, supporting theagency, control, and access to capital hypotheses. However, these results arenot robust to all specifications.

Column 2 of table 5 interacts free cash flow with a variable equal to 1 ifthe year is less than 1990 and another if the year is greater than or equal to1990. We do this because the prior literature using data before 1990 foundmixed evidence on the applicability of the free cash flow hypothesis (FCFH) togoing-private transactions. Lehn and Poulsen (1989), Denis (1992), and Oplerand Titman (1993) provide evidence that is consistent with FCFH. In contrast,Maupin, Bidwell, and Ortegren (1984), Kieschnick (1989, 1998), and Servaes(1994) provide evidence that is inconsistent with FCFH. We therefore wantto test the FCFH with a sample consistent with that used in these studies, aswell as examine its impact over the full sample period. Thus, we contrast thepre-1990 and post-1990 results on the FCF variable in our sample. Consistentwith Lehn and Poulsen (1989) and other papers cited above, we show that inthe 1980s, firms with higher free cash flow have a greater hazard rate of goingprivate. However, in the later period, the coefficient on free cash flow is in-significant, indicating that the agency hypothesis is less important during thistime period.11 This change is consistent with the increasing importance of op-erational restructuring relative to governance restructuring in private equity. AsJensen et al. (2006) explain, “financial and governance engineering...(are) theessence of the story of how LBOs and private equity created value in the ’80s.Today, there is another piece that was less prevalent in the ’80s but has becomeincreasingly important. You might call this piece operational engineering.”

Column 3 of table 5 replaces analyst coverage with institutional ownership.The coefficient on institutional ownership is negative and significant, furthersupporting the information hypothesis.12 Column 4 includes the annualadjusted return to determine if stock market performance impacts the hazardrate of going private. We include this additional variable because Pastor andVeronesi (2005) predict that firms will rationally go public after a period ofhigh returns. If one reverses this argument, it may suggest that firms go privateafter periods of low returns. Though we see firms go private when they havelow market-to-book ratios and we see market-to-book ratios decline overthe firm’s public life (consistent with their argument), the coefficient on themarket-adjusted stock return is not statistically significant. Column 5 furtherexplores the access to capital and liquidity hypotheses, including a dummyvariable equal to 1 if the KZ Index indicates the firm is constrained and illiq-uidity measure (illiq) of Amihud (2002). Because the KZ Index is estimated

11 We test the stability of the free cash flow result to slight changes in the cut-off year and find that the result isstable in these variations.

12 In untabulated results, we repeat this analysis including both analyst coverage and institutional ownership. Co-efficients on both variables are negative and significant.

1795

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 26: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

.006

.008

.01

.012

.014

Sm

ooth

ed h

azar

d of

goi

ng p

rivat

e

5 10 15 20 25Time since IPO

Analyst Coverage=0 Analyst Coverage=8

A

0.0

05.0

1.0

15.0

2.0

25S

moo

thed

haz

ard

of g

oing

priv

ate

5 10 15 20 25Time since IPO

Turnover=0.004 Turnover=0.14

B

.005

.01

.015

.02

Sm

ooth

ed h

azar

d of

goi

ng p

rivat

e

5 10 15 20 25Time since IPO

Market to Book=0.81 Market to Book=4.21

C

1796

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 27: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

.006

.008

.01

.012

Sm

ooth

ed h

azar

d of

goi

ng p

rivat

e

5 10 15 20 25Time since IPO

Merger Count=0 Merger Count=1

D

.008

.01

.012

.014

.016

.018

Sm

ooth

ed h

azar

d of

goi

ng p

rivat

e

5 10 15 20 25Time since IPO

Free Cash Flow (pre)=-0.15 Free Cash Flow (pre)=0.09

E

.007

.008

.009

.01

.011

.012

Sm

ooth

ed h

azar

d of

goi

ng p

rivat

e

5 10 15 20 25Time since IPO

Free Cash Flow (post)=-0.19 Free Cash Flow (post)=0.14

F

1797

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 28: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

.007

.008

.009

.01

.011

.012

Sm

ooth

ed h

azar

d of

goi

ng p

rivat

e

5 10 15 20 25Time since IPO

G

Figure 2Economic significance of covariatesThis figure displays the effect of five covariates from figure 1 on the hazard of going private based on thehazard function estimated using the Cox model for table 5, specification 2. The hazard is evaluated at the tenthand ninetieth percentile for each specified variable. The hazard is evaluated at the mean values of the otherexplanatory variables. Panel G displays the Cox model hazard function for table 5, specification 2. The hazardis evaluated at the mean values of the explanatory variables. The analysis time is in years from the IPO date.

using several of the control variables, we exclude any control variable includedin this index estimation. The evidence supports the importance of the access tocapital hypothesis as the coefficient on the KZ Index is negative and significant.The coefficient on illiq is insignificant.

Figure 2 illustrates the economic significance of several variables in the haz-ard model analysis. Specifically, in figure 2, we estimate the hazard functionusing specification 2 of table 5. All covariates are held at their sample mean ex-cept those indicated. In all figures (panels A–F), we illustrate the difference ineach specified variable at the tenth and ninetieth percentiles. These figures canbe compared to panel G, which illustrates the estimated hazard function hold-ing all variables at their sample mean. The differences in the hazard functionsfor firms with high and low analyst coverage, turnover, free cash flow in thepre-1990 period, and market to book are substantial and reflect the economicimportance of these variables in determining the choice to go private versusremaining public. The differences in the hazard function of the high and lowpost-1990 free cash flow and merger count are less dramatic.

4.3 Effect of market and macroeconomic factorsIn table 6, we explore the importance of the evolution of market-wide andmacroeconomic forces in the timing of the going-private decision using theCox proportional hazard model after controlling for the firm and industry char-

1798

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 29: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

acteristics using specification 2 of table 5. We do this for two reasons. First,waves in going-private transactions (as detailed in panel C of table 1) sug-gest that exogenous factors may influence the number of firms going privateand thus impact a firm’s decision to go private. In this section, we thereforeexamine the importance of these exogenous factors. Second, as we describethroughout this article, the difference in firms that go private and remain pub-lic is evident at the IPO. One interpretation of this is that the marginal benefitis just above the marginal cost of being public for these firms at the IPO. Overtime, forces push the marginal costs over the marginal benefits, and thus, thesefirms go private. However, as in figure 1 and table 4, many of the firm’s char-acteristics do not change in ways that would increase the probability of goingprivate. Thus, it may be that exogenous market and macroeconomic forces alsoplay a role in pushing the firm over the threshold.

We find that the hazard rate of going private increases significantly in highsentiment (measured as in Baker and Wurgler 2006) (specifications 1 and 2 oftable 6) and hot private equity markets (specifications 5 and 6 of table 6) butdecreases in hot IPO markets (specification 5 of table 6). These results suggestthat fluctuations in the private equity, IPO, and overall market impact the de-cision to go private. The negative relation with the IPO market may indicatetime-varying fluctuations in the costs and benefits of being a public firm. Inspecification 3, we examine the impact of the term, default risk premiums onthe decision to go private. We find that firms have a lower hazard rate of goingprivate when the yield curve is steep and increasing, suggesting that the sup-ply of debt in the economy might be an important factor in the going-privatedecision.13 We also find that as default risk premium increases in the econ-omy, the likelihood of a firm going private increases, suggesting that costs ofbankruptcy for public firms might be substantial and an influencing factor. Inspecification 4, we examine the supply of bank loans as a proportion of GNP.Bank loans are computed as the total year 2000 real dollar amount in the syn-dicated loan market from the Dealscan database each year that includes thehighly leveraged loan market, which supplies debt to LBO transactions. Weshow that firms have a high hazard rate of going private when the supply ofbank loans is great, indicating the importance of this source of financing forprivate firms. In specifications 5 and 6, we include measures of IPO and privateequity market activity. The private equity market data are not available before1980, and thus, the number of observations drops in these two columns intable 6. Not surprisingly, specifications 5 and 6 show that firms have a higherhazard rate of going private when the private equity market is hot. We notethat the state of the private equity market is a valid control since the firms inour sample go private by operating or management buyouts in addition to pri-vate equity buyouts. In untabulated results, we repeat specifications 5 and 6,including the private equity and IPO variables independently, and find simi-

13 We thank Steve Kaplan for suggesting this analysis and interpretation.

1799

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 30: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

Table 6Cox proportional hazard models for time to go private—impact of macrovariables

(1) (2) (3) (4) (5) (6)Log(Sales) 0.08∗∗ 0.08∗∗ 0.10∗∗∗ 0.16∗∗∗ 0.10∗ 0.10∗

(0.03) (0.03) (0.03) (0.03) (0.05) (0.05)

Analyst Coverage −0.04∗∗∗ −0.04∗∗∗ −0.04∗∗∗ −0.05∗∗∗ −0.05∗∗ −0.05∗∗(0.01) (0.01) (0.01) (0.01) (0.02) (0.02)

R&D/Sales 0.009 0.008 0.009 0.009 0.007 0.007(0.006) (0.006) (0.006) (0.006) (0.008) (0.008)

Capital Expenditure/ −0.24 −0.23 −0.25 −0.23 −0.17 −0.17Sales (0.15) (0.15) (0.15) (0.15) (0.21) (0.21)

Dividend Dummy 0.23∗∗ 0.24∗∗ 0.11 0.02 −0.28 −0.29(0.10) (0.10) (0.10) (0.10) (0.20) (0.20)

Turnover −13.40∗∗∗ −13.12∗∗∗ −10.07∗∗∗ −8.72∗∗∗ −10.57∗∗∗ −10.52∗∗∗(1.62) (1.62) (1.55) (1.53) (2.06) (2.06)

Market-to-Book −0.39∗∗∗ −0.39∗∗∗ −0.37∗∗∗ −0.35∗∗∗ −0.37∗∗∗ −0.38∗∗∗(0.06) (0.06) (0.06) (0.06) (0.09) (0.09)

Free Cash Flow/Assets 0.61∗∗ 0.63∗∗ 0.61∗∗ 0.62∗∗ 0.42 0.44(0.27) (0.27) (0.28) (0.28) (0.38) (0.39)

Leverage 0.08 0.08 0.03 −0.05 0.09 0.10(0.21) (0.21) (0.22) (0.22) (0.33) (0.33)

Cash/Assets 1.02∗∗∗ 0.99∗∗∗ 1.04∗∗∗ 1.27∗∗∗ 0.55 0.58(0.27) (0.27) (0.27) (0.27) (0.43) (0.43)

Net Fixed Assets/Assets 0.26 0.25 0.26 0.31 0.08 0.08(0.20) (0.20) (0.20) (0.20) (0.32) (0.32)

Number of Mergers −0.12∗∗ −0.12∗∗ −0.08 −0.05 −0.03 −0.02(0.06) (0.06) (0.05) (0.05) (0.06) (0.05)

Sentiment 1 0.28∗∗∗(0.06)

Sentiment 2 0.28∗∗∗(0.08)

Term Premium −0.11∗∗∗(0.04)

Default Risk Premium 0.85∗∗∗(0.09)

Real $ Bank Loans 50.37∗∗∗(3.17)

Number IPO −0.01∗∗∗(0.005)

Private Equity Deals 0.0002∗∗∗(0.00005)

IPO Volume 0.94(2.23)

Real $ Private Equity 62.34∗∗∗(20.55)

Obs. 33979 33979 33979 33979 20244 20244

This table reports the results of Cox proportional hazard models with time-varying covariates based onmacrovariables. The dependent variable is time to private, which measures the time between IPO of the firmand the time they go private. All specifications include Fama–French industry classification controls. The tablereports the coefficients and, in parentheses, the standard errors. Variables are defined in Appendix A. ∗∗∗,∗∗,and ∗ indicate significance at the 1%, 5%, and 10% levels, respectively.

lar results. The analysis shows that most of the previously documented resultsare robust to including these macrofactors. The coefficients on the variablesmeasuring information and liquidity considerations remain significant in allspecifications. The coefficients on the variables measuring control and accessto capital considerations are subsumed by the macrovariables, suggesting thatthese factors themselves are related to the macroeconomic conditions, an argu-ment made by Shleifer and Vishny (2003). The results presented in table 6 also

1800

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 31: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

raise the related question of what drives private equity waves. However, thisquestion is beyond the scope of this article, and we leave it to future research.

4.4 Effect of the starting point of firms in public markets on thegoing-private decision

The hazard model analysis presented in table 5 illustrates the relative impor-tance of the firm characteristics, informed by theory, to significantly impact theprobability of going private. This analysis and the earlier summary statisticsshow that both the cross-sectional differences and the evolution of variablesover time impact the choice between being private versus public. However, theearlier summary statistics also highlight the possible importance of the start-ing point for these firm characteristics—at the IPO. We therefore use a logitmodel to examine how inherent differences at the time of the IPO determine afirm’s decision to go private. It is important to note that, on average, the IPO is13.5 years before the going-private decision.

We present these results in table 7, panel A, using the specifications thatmimic the hazard model analysis in table 5, except we do not include the prioryear annual adjusted return since this is not available at the IPO year. Thedependent variable is a dummy variable that equals 1 if the firm ultimately goesprivate and 0 otherwise. All data are examined on an annual basis, at the firstyear following the IPO date. These tests will help us assess the importance ofintrinsic characteristics of the firms (measured at the time of the IPO) that canbe used to successfully predict the going-private decision. Note that the logitanalysis does not incorporate the relevance of the path the firm takes and howits firm characteristics change and will therefore examine the stark differencebetween private and public firms at the time of the IPO highlighted both byfigure 1 and table 3. However, it ignores the truncation issues dealt with inthe hazard model. We therefore employ the logit model to provide insight intoand stress the importance of the characteristics at the starting point (IPO) butrecognize imperfections in the analysis.

Many of the results presented in table 7, panel A, are very similar in signand statistical significance to the hazard model presented in table 5. Specifi-cally, firms are more likely to ultimately go private if they have low analystcoverage and low turnover, supporting the predictions of the information andliquidity hypotheses. The liquidity hypothesis is further supported by the pos-itive and significant coefficient on illiq. The FCFH continues to get supportbut again only in the early pre-1990 sample period. Differences in the re-sults from these analyses include insignificant coefficients on market to book,merger count, and the dividend dummy, indicating that it is more the changes inthese variables over time than their inherent differences that matter. Given thatthe firm characteristics in the logit are measured, on average, 13.5 years priorto the going-private decision, the results presented in table 7 are remarkable.It appears that many of the factors that determine if a firm will go private—

1801

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 32: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

Table 7Logit models. Panel A: logistic regression models explaining the decision to go private using all variablesat the time of IPO

(13e-3 filings Sample)(1) (2) (3) (4)

Log(Sales) 0.08∗ 0.07∗ 0.06 0.22∗∗∗(0.05) (0.05) (0.04) (0.05)

Analyst Coverage −0.07∗∗ −0.07∗ −0.14∗∗∗(0.04) (0.04) (0.05)

R&D/Sales 0.0002 −0.0003 −0.0004 −0.003(0.009) (0.009) (0.01) (0.04)

Capital Expenditure/Sales −0.01 −0.01 −0.01 −0.03(0.03) (0.03) (0.03) (0.06)

Dividend Dummy −0.02 −0.02 −0.03(0.16) (0.16) (0.16)

Turnover −5.73∗∗∗ −5.63∗∗∗ −5.45∗∗∗ −5.74∗∗∗(1.31) (1.31) (1.33) (1.64)

Market-to-Book −0.0009 −0.0007 −0.004(0.02) (0.02) (0.02)

Free Cash Flow/Assets 0.61∗(0.35)

Free Cash Flow post-1990s 0.33 0.38(0.37) (0.37)

Free Cash Flow pre-1990s 1.47∗∗ 1.56∗∗(0.68) (0.69)

Leverage −0.16 −0.16 −0.13(0.34) (0.34) (0.34)

Cash/Assets −0.89∗∗ −0.91∗∗∗ −0.96∗∗∗(0.35) (0.35) (0.35)

Net Fixed Assets/Assets 0.52∗ 0.52∗ 0.49 1.07∗∗∗(0.31) (0.31) (0.31) (0.36)

Merger Count −0.08 −0.08 −0.10 −0.04(0.07) (0.07) (0.07) (0.08)

Inst. Ownership −0.005(0.004)

KZ Index −0.23(0.16)

Illiq 5.45∗∗(2.20)

Obs. 4171 4171 4171 3013Pseudo R-Sq 0.06 0.06 0.06 0.08

information considerations, liquidity, and free cash flow—are seen at the timeof the IPO itself. Thus, it is the inherent characteristics of the firm at the IPOin addition to the changes that happen to it when it is public that determine ifthe firm will ultimately go private.

In table 7, panel B, we present the economic impact of a one-half stan-dard deviation change in each variable using the coefficients from column 2of panel A on the probability of going private. If the variable is not includedin the column 2 specification, we use the available coefficient in specifications3 or 4. For comparison, note that, as presented in the last row of panel B, thepredicted probability of going private is 6.02%. A one-half standard deviationdecrease in the analyst coverage increases the probability of going private byalmost 1%, approximately one-eighth the predicted probability. Further, a one-half standard deviation decrease in turnover increases the probability of goingprivate by 3.1%, approximately one-half the predicted probability. A one-half

1802

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 33: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

Table 7Panel B—economic significance of forces that explain the decision to go private using variables at thetime of IPO

Variable Change in probability (%)Log(Sales) 0.88Analyst Coverage −0.79R&D/Sales −0.04Capital Expenditure/Sales −0.52Dividend Dummy −0.05Turnover −3.08Market-to-Book −0.02Leverage −0.17Cash/Assets −1.51Net Fixed Assets/Assets 0.63Merger Count −0.56Free Cash Flow/Assets Pre-1990s 0.95Free Cash Flow/Assets Post-1990s 0.48Inst. Ownership −0.53KZ Index −3.08Illiq 0.66E (Private| X)% 6.02

Panel A reports the results of logistic regressions that predict the propensity of a firm to go private based onall variables for which data available are measured at the time of the IPO year. The dependent variable is 1 if afirm goes private and 0 otherwise. All specifications include Fama–French industry classification controls. Thepanels report the coefficients, and in parentheses are the standard errors. Variables are defined in Appendix A.∗∗∗,∗∗, and ∗ indicate significance at the 1%, 5%, and 10%, levels respectively. Panel B reports the increasein probability of going private for a one-half standard deviation around the mean for each explanatory variable,holding all other variables at their mean. For dummy variables, the change is from 0 to 1. All the results are basedon specification 2 of panel A. The economic significance of the variables that are not included in specification2 is based on the specifications (3–4) in panel A in which they first appear.

standard deviation increase in illiq results in a 0.66% increase in the probabil-ity of going private. A one-half standard deviation increase in free cash flowin the pre-1990 period results in a 0.95% increase in the probability of goingprivate, respectively. Thus, almost all of the statistically significant variableshave economically meaningful effects.

In unreported analysis, we tabulate results from rolling forward five yearsand repeating the logit analysis in panel A of table 7 in each of the next fiveyears. Obviously, as we roll forward, we lose those firms that were only publicfor a few years, since some of these firms would not have data five years afterthe IPO. The results are quite similar to those at the time of the IPO fromtable 7, panel A.

The results at the IPO not only provide evidence of the choice between beingprivate or public but also indicate that it is possible to determine the relativecosts and benefits of being public early in the firm’s public life. Stated anotherway, it seems that at least in part there is something inherent to the firm atthe time of the IPO that determines if it will ultimately go private. Assumingthat the firm’s IPO decision was optimal, these results suggest that firms thatultimately go private may be just above this threshold, where benefits of beingpublic exceed the costs at the time of the IPO. Firms then reverse the going-public decision once they fall below the threshold. Alternatively, the resultsmay indicate that these are firms that should not have gone public and thus in

1803

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 34: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

time reverse this suboptimal decision. Regardless, the variables examined hereand informed by theory are able to successfully capture the relative costs andbenefits long before the actual decision is made.

4.5 Effect of changes over time relative to matched sampleThe analysis in table 5 shows the importance of both the cross-sectional dif-ferences and time-series changes in the variables on the hazard rate of goingprivate. Table 7 highlights the importance of the initial differences in liquidityand information. In this section, we focus on better understanding the impor-tance of the time-series changes in these variables. We do this by repeating theanalysis for column 2 of table 5 using several matched samples as the compar-ison group. In doing so, we both explore the robustness of the results to thesealternative comparison groups and focus our analysis on how these variableschange over time, since some of the samples are matched based on the firmcharacteristics important in the earlier hazard model and logit analysis.

For each going-private firm, we first match the set of firms in the sameFama–French industry classification that had an IPO in the same year asour going-private firm but remained public. We use this sample as our firstmatched sample. Next, we impose a screen and keep only those firms that wereclose (within a specified range) based on the initial condition, such as marketto book, analyst coverage, market value (equity market capitalization), andturnover. For continuous variables, such as market to book, market value, andturnover, we pick comparison firms for which the matching variable is within±10% of the going-private firm. (For robustness, we repeat our analysis withcomparison firms whose characteristics are within ±5%,±15%, or ±20% ofthe going-private firm, and the results remain qualitatively unchanged.) Foranalyst coverage, we pick comparison firms that are within ±1 analyst of thegoing-private firm and repeat it for ±2 analysts for robustness. Thus, all firmsin the comparison groups are similar to the going-private firms, but the numberof firms within this range may vary.

The results from table 8 are remarkably robust to these matched samples.The coefficients on all variables except size are consistently significant and ofthe same sign as those in table 5. Specifications 1 and 4, which match by in-dustry and equity market capitalization at the time of the IPO, show that all theresults emphasized in table 5 continue to be important. Of particular impor-tance are the results in columns 2, 3, and 5, as these match based on market-to-book ratio, analyst coverage, and turnover, respectively, characteristics thatwere found to be important in the earlier hazard analysis. Any significance inthese variables in the results presented in table 8 must obtain primarily fromthe time-series changes since the initial conditions of both samples have beencontrolled for in the matching technique. The results in columns 3 and 5 showthat it is not just the initial value of analyst coverage and turnover that impactthe hazard rate of going private but also how these characteristics evolve over

1804

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 35: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

Table 8Cox proportional hazard models for time to go private—matched initial conditions analysis

Industry MB Analyst MktVal TurnoverLog(Sales) 0.06∗ 0.05 0.05∗ 0.09∗∗∗ 0.03

(0.03) (0.03) (0.03) (0.03) (0.03)

Analyst Coverage −0.05∗∗∗ −0.04∗∗∗ −0.04∗∗ −0.03∗∗ −0.02∗∗(0.01) (0.01) (0.01) (0.01) (0.01)

R&D/Sales 0.006 0.006 0.006 0.006 0.009(0.008) (0.008) (0.008) (0.008) (0.008)

Capital Expenditure/Sales −0.26 −0.26 −0.27 −0.25 −0.40∗∗(0.17) (0.16) (0.17) (0.16) (0.19)

Dividend Dummy 0.26∗∗∗ 0.24∗∗ 0.28∗∗∗ 0.21∗∗ 0.12(0.10) (0.10) (0.10) (0.10) (0.10)

Turnover −9.62∗∗∗ −9.22∗∗∗ −9.69∗∗∗ −9.00∗∗∗ −7.69∗∗∗(1.46) (1.46) (1.47) (1.45) (1.41)

Market-to-Book −0.42∗∗∗ −0.38∗∗∗ −0.41∗∗∗ −0.42∗∗∗ −0.31∗∗∗(0.06) (0.06) (0.06) (0.06) (0.06)

Free Cash Flow post-1990s −0.07 −0.11 −0.13 −0.12 0.28(0.25) (0.23) (0.24) (0.24) (0.27)

Free Cash Flow pre-1990s 4.51∗∗∗ 4.39∗∗∗ 4.74∗∗∗ 3.67∗∗∗ 1.92∗∗∗(0.57) (0.57) (0.56) (0.58) (0.62)

Leverage 0.23 0.16 0.24 0.11 0.11(0.21) (0.22) (0.21) (0.22) (0.22)

Cash/Assets 0.81∗∗∗ 0.90∗∗∗ 0.68∗∗ 0.88∗∗∗ 0.46∗(0.27) (0.27) (0.27) (0.27) (0.28)

Net Fixed Assets/Assets 0.13 0.13 0.15 0.20 0.34(0.20) (0.20) (0.20) (0.20) (0.21)

Merger Count −0.12∗∗ −0.12∗∗ −0.13∗∗ −0.14∗∗ −0.10∗(0.06) (0.06) (0.06) (0.06) (0.06)

Obs. 30482 27202 29995 22358 17111

This table reports the results of the Cox proportional hazard models with time-varying covariates based on allvariables. The dependent variable is time to private, which measures the time between IPO of the firm and thetime they go private. All specifications include the going-private firms and a comparison sample matched byFama–French industry classification and the IPO year of the going-private firm. Thus, for each going-privatefirm, there could be multiple matches. Specification 1 is the industry and the IPO year matched sample. Speci-fications 2 through 5 match firms in addition to industry and the IPO year, other characteristics in the IPO year.These include market-to-book (MB, specification 2), Analyst Coverage (analyst, specification 3), Market Capi-talization (MktVal, specification 4), and Turnover (Turnover, specification 5). The table reports the coefficientsand, in parentheses, the standard errors. Variables are defined in Appendix A. ∗∗∗,∗∗, and ∗ indicate significanceat the 1%, 5%, and 10% levels, respectively.

time. Recall that the initial values of these characteristics matter, as evidencedby the significant coefficients in the logit analysis in table 7. The results incolumn 2 show that, similar to table 5, the coefficient on market to book isnegative and significant. The coefficient on this variable was not significantin the logit analysis in table 7. Thus, it may be suggested that the time-serieschanges in market to book, and perhaps not the initial conditions, impact thehazard rate of going private. Overall, these results highlight that it is both howthese firms compare to other firms at their starting point at the IPO and howthey evolve relative to other firms that determine if they go private.

4.6 Out-of-sample forecasts and ROC analysis resultsIn this section, we test out of sample our model’s ability to forecast which firmswill go private. We do this to gauge the accuracy of our analysis. To determinethe accuracy of our hazard model, we estimate specification 2 from table 5

1805

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 36: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

Table 9Out-of-sample forecasts

Time period: 1991–200430,617 firm years, 407 going-private transactions.

Decile Cox Logit1–5 83.3 81.71–3 60.5 61.71 26.7 30.1

This table reports on the success of forecasting by sorting firms each year by the Cox proportional hazards modelforecast (or the corresponding logit model) and counting the fraction of going-private firms over the next fiveyears that correspond with each decile of the model-predicted hazard of going private. We use the first ten yearsof the sample (1980–1990) to develop the Cox proportional hazards (logit) model. Thus, the table only considersfirms going from 1991 to 2004, and it includes the fitted values of the hazard or logit model (specification 2from tables 5 and 7, panel A, respectively) as predictors. The Cox (logit) model is then reestimated on a rollingbasis each year as we move forward in time, using all available data till each year from 1980, and the resultingcoefficients are used to form the predictors.

using data from before 1991. We then use this model to predict who will goprivate in the next five years, 1991–1995.14 We sort the private sample intodeciles by the model’s predicted hazard rate and count how many firms in eachdecile actually went private during this five-year period. If our model providesan accurate forecast, we expect that a greater percentage of the going-privatesample will fall in the high hazard rate deciles relative to the low hazard ratedeciles. We repeat this analysis rolling forward each year after 1991, such thatwe next use the pre-1992 period as the estimation period and the 1992–1996as the prediction period and then the pre-1993 period as the estimation and1993–1997 as the prediction period, so on and so forth, through using after2002 as the prediction period. We then take an average of the accuracy rateover each of these periods. Of the 407 firms that go private in our sample(during 1991–2004), 83.3% have a hazard rate greater than the median (topfive deciles), 60.5% fall in the top three deciles, and 26.7% fall in the topdecile. These statistics are presented in table 9 and demonstrate that our modelis quite accurate in predicting who will go private. For comparison, we repeatthis test using a naive hazard model where the market value of equity is theonly explanatory variable. This test yields a 60.4% accuracy for deciles 1–5and 11.5% accuracy for the top decile. Thus, the naive model provides onlya modest improvement over a random guess of 50% (deciles 1–5) and 10%(for the top decile), while employing variables predicted by theory produces asubstantial improvement in prediction accuracy.

We repeat this analysis using the logit model and find similarly accurateresults. Specifically, we find that 81.7% of the going-private firms have a pre-dicted probability of going private greater than the median (top five deciles),61.7% fall in the top three deciles, and 30.1% fall in the top decile. Given thatthese predictions only use data from the year following the IPO, the accuracy

14 We choose a five-year period since we need a reasonable number of going-private transactions in the assessmentperiod. Results are similar, if we use a three-year prospective period.

1806

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 37: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

of this model indicates the importance of the inherent characteristics of thefirm.

Our second method of assessment is using the ROC curve analysis. A ROCcurve for our classification model (hazard or logit) plots the true-positive rateof the model (y-axis) against its false-positive rate (x-axis). The true-positiverate, also known as sensitivity, is the proportion of actual going-private trans-actions correctly classified by the model. The false-positive rate, known as1-specificity, is the proportion of not going-private transactions incorrectlyclassified as going-private transactions by the model. Each point on the ROCcurve is a (sensitivity, 1-specificity) combination representing a prediction re-sult. The best possible prediction method would yield a point in the upper leftcorner or coordinate (0,1) of the ROC space. This represents 100% sensitiv-ity, which means that all true positives are found, and 100% specificity, whichmeans no false positives are found. A completely random guess would give apoint along a diagonal line, the line of no-discrimination from the left bottomto the top right corners. By changing the cutoff or the threshold to classify pre-dictions of the model into going private or not going private group, we obtaina (sensitivity, 1-specificity) trade-off or a point on the ROC curve. Repeatingthis exercise with different cutoffs produces the ROC curve. The area under theROC curve thus generated can be interpreted as the accuracy of the diagnosticmodel. One advantage of this method is that it does not require the researcherto specify a threshold at which the prediction is correct. While the applica-tion of this idea for the logit model is well known, we follow the method ofHeagerty and Zheng (2005) to obtain ROC curves for Cox proportional hazardmodels. ROC curves have been used in the default risk literature to assess thequality of bankruptcy prediction models (Sobehart, Keenan, and Stein 2000).

We present these curves in figure 3, with the hazard (logit) model at the top(bottom). Again, the results provide strong support for our model’s accuracy.The area under the ROC curve indicates that the hazard model is 77.9% accu-rate and the logit is 70.6% accurate. In other words, by varying the cutoff thatpredicts who will go private, on average, our hazard (logit) model will accu-rately predict the firms that go private 78% (71%) of the time. This is relativeto the diagonal line, which would indicate a random guess or a 50/50 chance ofguessing correctly. Alternatively, one can compare this to a ROC curve from anaive hazard (logit) model where market value is the only explanatory variable,which yields a 57.7% (52.3%) accuracy. Again, this test formally and stronglysupports the accuracy and importance of our results.

4.7 Subsample analysisAs explained in Section 2, our sample consists of firms going private throughvarious methods. We break down the sample firms that went private into threecategories: a buyout by a private operating firm, a buyout by a private equityfirm, or a buyout by management. We source the classifications for these deals

1807

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 38: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

1.00

0.75

0.50

0.25

0.00

0.00 0.25 0.50 0.75 1.001 – Specificity

0.00 0.25 0.50 0.75 1.001 – Specificity

Sen

sitiv

ity1.

000.

750.

500.

250.

00S

ensi

tivity

Area under ROC curve = 0.7793

Area under ROC curve = 0.7062

A

B

90

Figure 3Prediction accuracy and ROC curvesThis figure displays the ROC curves to assess the accuracy of the hazard (top panel) and logit (bottom panel)models to predict going-private transactions (specification 2 in tables 5 and 7, panel A, respectively). In the y-axis, we plot the true-positive rate (sensitivity) (i.e., the proportion of actual going-private transactions correctlyclassified by the model). In the x-axis, we plot the false-positive rate (1-specificity) (i.e., the proportion of notgoing-private transactions), incorrectly classified as going-private transactions by the model. Points above thediagonal (random guess) indicate good classification results. The area under the curve measures the accuracy ofthe model.

1808

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 39: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

PrivateOperating Privatefirms Equity

firms

Management buyout

13 E3 filings sample (1023 firms)

458 880

0

9369

164

Figure 4Breakdown of sample firmsThis figure displays the breakdown of the sample firms that went private into three categories: a buyout bya private operating firm, and a buyout by a private equity firm, a buyout by the management. We source theclassifications for these deals by reading newspaper reports from Factiva.

by reading newspaper reports from Factiva around the announcement of thedeal. For the sample of firms from the SDC database, we use the classificationin that database. We note that the same deal could be classified in multiple cat-egories and provide this information for both the primary and broad samplesin figure 4. As shown in the diagram, the majority of transactions involve aprivate operating firm, and many transactions involve multiple players. In thissection, we ask if the factor driving the hazard rate for going private differsbetween these groups. We therefore reestimate specification 2 of table 5 sepa-rately for each subgroup. Because much of the prior research on LBOs, citedin the introduction, groups management and private equity buyouts together,we also estimate treating both types as one group. One disadvantage of theanalysis by subgroup is a loss in power as our going-private sample in eachestimation will significantly decrease in size.

The results from the subgroup analysis are presented in table 10. The resultsare largely consistent across the subgroups and with those in table 5. Thisanalysis therefore suggests that, regardless of who chooses to take the firmprivate, the costs and benefits of being public as identified by theory drive thisdecision and are universally important across deal types. The exceptions arethat firms have a lower hazard rate of going private with a private equity groupor management if they have high R&D or capital expenditures. Further, firmshave a higher hazard rate of going private with a private equity firm if they havemore debt, perhaps indicating the need for a private equity restructuring. Also,the coefficient on turnover is insignificant but of the same sign in the privateequity only group. All other results are broadly consistent across subgroupsand table 5.

1809

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 40: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

Table 10Cox proportional hazard models for time to go private—subsample analysis

(13e-3 filings sample)PE and Mgmt PE Mgmt Operating

Log(Sales) 0.08 0.14∗ 0.07 0.14∗∗∗(0.05) (0.08) (0.06) (0.04)

Analyst Coverage −0.10∗∗∗ −0.06∗ −0.11∗∗∗ −0.03†

(0.03) (0.03) (0.04) (0.02)

R&D/Sales −4.20∗∗ −6.55∗ −3.86∗∗ 0.005(1.78) (3.46) (1.97) (0.006)

Capital Expenditure/Sales −1.33∗∗ −0.42 −2.35∗∗ −0.12(0.63) (0.63) (0.96) (0.15)

Dividend Dummy 0.34∗∗ 0.42∗ 0.29∗ 0.14(0.15) (0.23) (0.17) (0.13)

Turnover −5.36∗∗∗ −1.57 −7.06∗∗∗ −18.81∗∗∗(2.08) (2.48) (2.61) (2.38)

Market-to-Book −0.81∗∗∗ −0.63∗∗∗ −0.93∗∗∗ −0.26∗∗∗(0.13) (0.18) (0.16) (0.07)

Free Cash Flow post-1990s 0.79 0.98 0.87 −0.26(0.57) (0.97) (0.64) (0.27)

Free Cash Flow pre-1990s 6.88∗∗∗ 8.05∗∗∗ 6.73∗∗∗ 3.99∗∗∗(0.68) (1.02) (0.76) (0.76)

Leverage 0.69∗∗ 1.18∗∗ 0.48 −0.22(0.33) (0.46) (0.40) (0.29)

Cash/Assets 2.27∗∗∗ 2.12∗∗∗ 2.48∗∗∗ 0.92∗∗(0.41) (0.70) (0.45) (0.36)

Net Fixed Assets/Assets 0.45 0.27 0.60 0.34(0.33) (0.50) (0.38) (0.27)

Merger Count −0.01 0.01 −0.09 −0.11(0.06) (0.07) (0.11) (0.07)

Obs. 31879 30946 31530 32315

This table reports the results of Cox proportional hazard models with time-varying covariates based on all thevariables. The dependent variable is time to private, which measures the time between the IPO of the firmand the time it goes private. All specifications include Fama–French industry classification controls. The tablereports the coefficients and, in parentheses, the standard errors. Variables are defined in Appendix A. ∗∗∗,∗∗,∗,and † indicate significance at the 1%, 5%, 10%, and 12% levels, respectively.

5. Conclusion

In this article, we investigate how firms weigh the costs and benefits of beinga public firm in the decision to opt out of the public market and go private. Indoing so, we draw on previous studies from the 1980s on the value implicationsof the decision to go private and on the well-developed theoretical literature onwhy firms go public. We test the relative importance of each of these factors byexamining the public life cycle of firms from IPO to the going-private decisionand compare these firms to a set of firms that had IPOs during a similar periodbut did not go private.

Our findings are two fold. First, we find support for the importance of manyof the costs and benefits of being a public firm. Particularly, we find strongsupport for the importance of theories that stress information and liquidityconsiderations. We also find support for the role of free cash flow, primarilyin the pre-1990s period and support for control and access to capital consider-ations. We show that our hazard model is quite accurate, correctly predictingwho will go private 83.3% of the time. Second, we find that despite the fact

1810

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 41: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

that, on average, our sample firms remain public for over thirteen years, onecan reliably predict that they will go private at the time of the IPO. Specifically,many of the factors that significantly determine the duration of public life alsosignificantly predict who will go private using data in the year following theIPO. Using only data at the time of the IPO, we are able to correctly predictwho will go private 81.7% of the time. This result implies that it is not only thepath that the firm takes but also factors inherent and observable about the firmat the time of going public that determine if it eventually will go private.

A secondary contribution of our study is to shed light on the going-publicdecisions of firms. Going public is one of the most important events in a privatecompany’s development; thus, the IPO is often seen as a rite of passage in thelife cycle of a young, successful firm.15 A large theoretical literature discussedearlier developed in the last decade, weighing the costs and benefits of beingpublic versus private to explain why firms go public. However, empirical anal-ysis investigating the relative importance of these costs and benefits remainsscarce because, ideally, we need data on private firms that choose to go publicand data on firms that continue to remain private to provide a direct compari-son of the choice between being public or private.16 Data on private firms arenot readily available. Our research design circumvents this data challenge byinvestigating the reversible predictions of the going-public theories, thus in-forming us of the important economic forces among the many highlighted bytheory.

15 See Jenkinson and Ljungquist (2001) and Ritter (2003) for surveys of the IPO literature.

16 Pagano, Panetta, and Zingales (1998), Boehmer and Ljungvist (2004), Helwege and Packer (2004), Kim andWeisbach (2008), Rosen, Smart, and Zutter (2005), Poulsen and Stegemoller (2006), and Chemmanur, He, andNandy (2007) are notable examples of studies on the going-public decisions of firms.

1811

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 42: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The

Review

ofFinancialStudies

/v23

n5

2010

Appendix A. Description of variablesThis appendix describes the list of variables constructed from various databases and used as proxies for testing the various theories of going public in the article.It also lists the time period of their availability. Compustat variables are available from 1950 to 2004.

Variable name Description SourceSales Real net sales in year 2000 million dollars (Data12). Compustat

Assets Real assets in year 2000 million dollars (Data6). Compustat

Market Value Market capitalization of equity in year 2000 million dollars Compustat(Data24∗Data25).

Age Age of firm at the IPO. Loughran and Ritter (2004)Jovanovic and Rousseau (2001)

Analyst Coverage Number of analysts following the firm in any fiscal year IBES(set equal to zero if missing). 1974–2004

Institutional Holdings Expressed as percentage of total shares outstanding for each firm. CDA SPECTRUM1983–2004

Capex/Sales Ratio of capital expenditure to sales (Data128/Data12). Compustat

R&D/Sales Ratio of research and development expenditure to sales Compustat(Data46/Data12) (set to 0 if missing).

KZ Index Dummy −1.001909CFit + 3.139193TLTDit − 39.36780TDIVit Compustat−1.314759CASHit + 0.2826389Qit,CF—cash flow to total assets; TLTD—long-term debt to total assets.CASH—liquid assets to total assets; TDIV—total dividends to assetsQ—market-to-book ratio. As in Lamont, Polk, and Saa-Requejo (2001),firms in the highest KZ Index tercile (constrained most) arecoded 1 for this variable and the rest as 0.

Dividend Dummy Equals 1 if a firm paid out dividends during the fiscal year and Compustat0 otherwise.

1812

at Univ of RochesterLibrary on April 15, 2013 http://rfs.oxfordjournals.org/ Downloaded from

Page 43: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why

Do

Firm

sU

seP

rivateE

quityto

OptO

utofPublic

Markets?

Illiq Illiquidity measure computed as in Amihud (2002) CRSPScaled by 100,000. 1963–2004.

Turnover Ratio of daily turnover volume over the past 12-month CRSPperiod divided by number of shares o/s. Scaled by 1000. 1963–2004.

Merger Count Number of acquisitions in which the company is an SDCacquirer during the year. 1980–2004.

Market-to-Book (Data6−Data60+Data24∗Data25)/Data6 CRSPFree cash flow/Assets (Data13−Data16−Data15−Data21−Data19)/Data6 CompustatLeverage Data9/Data6 CompustatCash to Assets Data1/Data6 CompustatNet Fixed Assets/Assets Data8/Data6 CompustatIPO Volume Real 2000 dollars value of IPOs in each calendar year SDC

(standardized by real GNP).Number IPO Number of IPOs in each calendar year SDCSentiment 1 sf1 measure of Baker and Wurgler (2006). Wurgler’s Web siteSentiment 2 sf2 measure of Baker and Wurgler (2006). Wurgler’s Web siteAnnual Adj Ret Stock’s market adjusted return CRSP

over CRSP Value-Weighted Index in the previous fiscal year.Term Yield Spread differential between 10 years. FRED

1-year Treasury bonds.Default Yield spread differential between BBB FRED

and AAA corporate bonds.Bank Real 2000 dollars value of bank loans FRED

(standardized by real GNP) each year. FREDPrivate Equity Real 2000 dollars value of Moneytree report of

private equity deals (standardized by real GNP). PricewaterhousecoopersPrivate Equity Deals Number of Moneytree report of

private equity deals. Pricewaterhousecoopers1813

at Univ of RochesterLibrary on April 15, 2013 http://rfs.oxfordjournals.org/ Downloaded from

Page 44: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The

Review

ofFinancialStudies

/v23

n5

2010

Appendix B. Summary of empirical implications of theories that consider the choice between being a public or private firmThis appendix summarizes the list of variables constructed from various databases and used as proxies for testing the various theories described in Section 2. Eachof these theories suggests that firms are more likely to reverse the going-public decision if information is costly or difficult to obtain. Many of the theories discussedabove examine related but subtly different aspects of information considerations. In the theories, these differences are important. Unfortunately, it is difficult to proxyfor these important but subtle differences. In our empirical analysis,we group this set of theories as those that highlight the importance of information considerationsin the decision to go private. The empirical proxies that we use in our analysis are summarized below.

Driving Force Models Effect on the Empirical proxy(ies)going-private decision and predictions

a. Adverse Selection Leland and Pyle (1977) As information Firms are more likely toasymmetry increases, go private if they :more firms go private. –are smaller, younger

–have lower institutionalb. Duplicative Monitoring Chemmanur and Fulghieri (1999) If information gathering

costs increase, holdings (and greater concn) andmore firms go private. analyst coverage.

–have higher R&D intensity.c. Serendipitous Titman If serendipitous information

Information Production Subrahmanyam and Titman (1999) gathering costs increase,more firms go private

Summary and empirical proxies for access to capital considerations:Driving Force Models Effect on the Empirical proxy(ies)

going-private decision and predictionsa. Cost of Capital Modigliani and Miller (1963) As cost of capital Firms are more likely to

Scott (1976) increases in public markets, go private if they:firms go private. –are dividend payers.

–have low Capex and M/B.b. Overcoming Lack of large investments –are not financially

financial (Capex), not constrained constrained (KZ Index).constraints in public markets lead

more firms to go private.

1814

at Univ of RochesterLibrary on April 15, 2013 http://rfs.oxfordjournals.org/ Downloaded from

Page 45: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why

Do

Firm

sU

seP

rivateE

quityto

OptO

utofPublic

Markets?

Summary and empirical proxies for control considerations:Driving Force Models Effect on the Empirical proxy(ies)

going-private decision and predictionsBenefit of Zingales (1995) Facilitate control transfers Firms are more likely toCorporate control Mello and Parsons (1998) increase in public markets, go private if they:

Brau, Francis, and Kohers (2003) Stock as a currency for future, –engage in fewer mergers.acquisitions. –have lower market to book.

Summary and empirical proxies for liquidity considerations:Driving Force Models Effect on the Empirical proxy(ies)

going-private decision and predictionsBenefit of Zingales (1995) Establish market price Firms are more likely toliquidity Mello and Parsons (1998) of the firm. go private if they have:

Bolton and von Thadden (1998) Liquidity benefit. –high illiquidity (illiq).Boot et al. (2007) Managerial autonomy. –low share turnover.

1815

at Univ of RochesterLibrary on April 15, 2013 http://rfs.oxfordjournals.org/ Downloaded from

Page 46: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

ReferencesAmihud, Y. 2002. Illiquidity and Stock Returns: Cross-Section and Time Series Effects. Journal of FinancialMarkets 5:31–56.

Amihud, Y., and H. Mendelson. 1988. Liquidity and Asset Prices: Financial Management Implications. Finan-cial Management 17:5–15.

Baker, M., and J. Wurgler. 2006. Investor Sentiment and the Cross-Section of Stock Returns. Journal of Finance61:1645–80.

Boehmer, E., and A. Ljungvist. 2004. On the Decision to Go Public: Evidence from Privately Held Firms.Working Paper, SSRN.

Bolton, P., and E. L. von Thadden. 1998. Liquidity and Control: A Dynamic Theory of Corporate OwnershipStructure. Journal of Institutional and Theoretical Economics 154:177–209.

Boot, A. W. A., R. Gopalan, and A. V. Thakor. 2007. Market Liquidity, Investor Participation, and ManagerialAutonomy: Why Do Firms Go Private? Journal of Finance 63(4):2013–59.

Borden, A., and J. Yunis. 2007. Going Private. New York, NY: Law Journal Press.

Brau, J. C., and S. E. Fawcett. 2006. Initial Public Offerings: An Analysis of Theory and Practice. WorkingPaper, SSRN.

Brau, J., B. Francis, and N. Kohers. 2003. The Choice of IPO Versus Takeover: Empirical Evidence. Journal ofBusiness 76:583–612.

Campbell, T. 1979. Optimal Investment Financing Decisions and the Value of Confidentiality. Journal of Finan-cial and Quantitative Analysis 14:913–24.

Chemmanur, T. J., and P. Fulghieri. 1999. A Theory of the Going-Public Decision. The Review of FinancialStudies 12:249–79.

Chemmanur, T. J., S. He, and D. Nandy. 2007. The Going-Public Decision and the Product Market. WorkingPaper, Boston College.

DeAngelo, H., L. DeAngelo, and E. Rice. 1984a. Going Private: The Effects of a Change in Corporate OwnershipStructure. Midland Corporate Finance Journal (Summer):35–43.

DeAngelo, H., L. DeAngelo, and E. Rice. 1984b. Going Private: Minority Freezeouts and Stockholder Wealth.Journal of Law and Economics 27:367–402.

Denis, D. J. 1992. Corporate Investment Decisons and Corporate Control: Evidence from Going-Private Trans-actions. Financial Management (Autumn):80–92.

Easley, D., M. O’Hara, and S. Hvidkjaer. 2004. Factoring Information Into Returns. Working Paper, CornellUniversity.

Engle, E. R., M. Hayes, and X. Wang. 2004. The Sarbanes–Oxley and Firms’ Going Private Decisions. WorkingPaper, University of Chicago.

Heagerty, P., and Y. Zheng. 2005. Survival Model Predictive Accuracy and ROC Curves. Biometrics 61:92–105.

Helwege, J., and F. Packer. 2004. The Decision to Go Public: Evidence from Mandatory SEC Filings of PrivateFirms. Working Paper, Ohio State University.

Jenkinson, T., and A. Ljungquist. 2001. Going Public: The Theory and Evidence of How Companies RaiseEquity Finance, 2nd ed. Oxford: Oxford University Press.

Jensen, M. C. 1986. Agency Costs of Free-Cash-Flow, Corporate Finance, and Takeovers. American EconomicReview 76:323–29.

Jensen, M., S. Kaplan, C. Ferenbach, M. Feldberg, J. Moon, B. Hoesterey, C. Davis, and A. Jones. 2006.Morgan Stanley Roundtable on Private Equity and Its Import for Public Companies. Journal of AppliedCorporate Finance 18(3):8–27.

1816

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 47: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

Why Do Firms Use Private Equity to Opt Out of Public Markets?

Jovanovic, B., and P. Rousseau. 2001. Why Wait? A Century of Life Before IPO. American Economic Review101:336–41.

Kaplan, S. N. 1989a. The Effects of Management Buyouts on Operating Performance and Value. Journal ofFinancial Economics 24:217–54.

———. 1989b. Management Buyouts: Evidence on Taxes as a Source of Value. Journal of Finance 44(3):611–32.

———. 1991. The Staying Power of Leveraged Buyouts. Journal of Financial Economics 29:287–314.

Kieschnick, R. 1989. Management Buy Outs of Public Corporations: An Analysis of Prior Characteristics.In Y. Amihud (ed.), Leveraged Management Buyouts: Causes and Consequences. pp. 35-68. Homewood, IL:Dow Jones-Irwin.

———. 1998. Free Cash Flow and Stock Holder Gains in Going Private Transactions Revisited. Journal ofBusiness Finance and Accounting 25:187–202.

Kim, W., and M. S. Weisbach. 2008. Motivations for Public Equity Offers: An International Perspective. Journalof Financial Economics 87:281–307.

Koenig, J. 2004. Survey: A Brief Roadmap to Going Private. Columbia Business Law Review 2004: 505.

Lamont, O. A., C. K. Polk, and J. Saa-Requejo. 2001. Financial Constraints and Stock Returns. Review ofFinancial Studies 14:529–54.

Lehn, K., and A. Poulsen. 1989. Free Cash Flow and Stockholder Gains in Going Private Transactions. Journalof Finance 44:771–87.

Leland, H., and D. Pyle. 1977. Information Asymmetries, Financial Structure, and Financial Intermediation.Journal of Finance 32:371-87.

Leuz, C., A. Triantis, and T. Wang. 2005. Why Do Firms Go Dark? Causes and Economic Consequences ofVoluntary SEC Deregistrations. Working Paper, University of Pennsylvania.

Loughran, T., and J. Ritter. 2004. Why Has IPO Underpricing Changed Over Time? Financial Management33:5–37.

Maksimovic, V., and P. Pichler. 2001. Technological Innovation and Initial Public Offerings. The Review ofFinancial Studies 14:459–94.

Maupin, R., C. Bidwell, and A. Ortegren. 1984. An Empirical Investigation of the Characteristics of PubliclyHeld Corporations That Change to Private Ownership Via Management Buy Outs. Journal of Business Financeand Accounting 11:435–50.

Mello, A., and J. Parsons. 1998. Going Public and the Ownership Structure of the Firm. Journal of FinancialEconomics 49:79–109.

Merton, R. C. 1987. A Simple Model of Capital Market Equilibrium with Incomplete Information, Journal ofFinance 42(3):483–510.

Mikkelson, W. H., M. M. Partch, and K. Shah. 1997. Ownership and Operating Performance of Companies ThatGo Public. Journal of Financial Economics 44:281–307.

Modigliani, F., and M. Miller. 1963. Corporate Income Taxes and the Cost of Capital: A Correction. AmericanEconomic Review 53:433–43.

Opler, T., and S. Titman. 1993. The Determinants of Leveraged Buyout Activity: Free Cash Flow vs. FinancialDistress Costs. Journal of Finance 48(5):1985–99.

Pagano, M., F. Panetta, and L. Zingales. 1998. Why Do Companies Go Public? An Empirical Analysis. Journalof Finance 53:27–64.

Pastor, L., and P. Veronesi. 2005. Rational IPO Waves. Journal of Finance 60(4):1713–57.

1817

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from

Page 48: Why Do Firms Use Private Equity to Opt Out of Public Markets?schwert.ssb.rochester.edu/f423/RFS10_BD.pdf · reverse their decision, that is, firms will exit the public markets and

The Review of Financial Studies / v 23 n 5 2010

Pickens, T. B. 1987. Boone. Boston, MA: Houghton Mifflin.

Poulsen, A., and M. Stegemoller. 2006. Moving from Private to Public Ownership: Selling Out to Public FirmsVersus IPOs. Working Paper, University of Georgia.

Ritter, R. J. 1987. The Costs of Going Public. Journal of Financial Economics 19(2):269–81.

Ritter, R. J. 2003. Investment Banking and Securities Issuance. In G. Constantinides, M. Harris, and R. Stulz(eds.) Chapter 5 of Handbook of the Economics of Finance, pp. 255–306. Amsterdam: North-Holland.

Roell, A. 1996. The Decision to Go Public: An Overview. European Economic Review 40(3):1071–81.

Rosen, R. J., S. B. Smart, and C. J. Zutter. 2005. Why Do Firms Go Public? Evidence from the Banking Industry.Working Paper, SSRN.

Servaes, H. 1994. Do Takeover Targets Overinvest? Review of Financial Studies 7:253–78.

Schipper, K., and A. Smith. 1991. Effects of Management Buyouts on Corporate Interest and Depreciation TaxDeductions. Journal of Law and Economics 34(2):295–341.

Scott, J. H. 1976. A Theory of Optimal Capital Structure. Bell Journal of Economics 7:33–54.

Shah, S., and A. V. Thakor. 1988. Private vs. Public Ownership: Investment, Ownership Distribution, andOptimality. Journal of Finance 43:41–59.

Shleifer, A., and R. W. Vishny. 2003. Stock Market Driven Acquisitions. Journal of Financial Economics70(3):295–311.

Shumway, T. 2001. Forecasting Bankrutcy More Accurately: A Simple Hazard Model. Journal of Business101–24.

Slovin, M. B., M. E. Shushka, and Y. M. Bendeck. 1991. The Intra-industry Effects of Going Private Transac-tions. Journal of Finance 46(4):1537–50.

Sobehart, J. R., S. C. Keenan, and R. Stein. 2000. Validation Methodologies for Default Risk Models. Credit(May):51–56.

Subrahmanyam, A., and S. Titman. 1999. The Going-Public Decision and the Development of FinancialMarkets. Journal of Finance 54:1045–82.

Whited, T., and G. Wu. 2006. Financial Constraint Risk. Review of Financial Studies 19:531–59.

Yosha, O. 1995. Information Disclosure Costs and the Choice of Financing Source. Journal of FinancialIntermediation 4:3–24.

Zingales, L. 1995. Insider Ownership and the Decision to Go Public. Review of Economic Studies 62:425–48.

1818

at Univ of R

ochesterLibrary on A

pril 15, 2013http://rfs.oxfordjournals.org/

Dow

nloaded from