Equity Issuance of Distressed Firms: Debt Overhang or Agency Problem?

Transcription

1 Equity Issuance of Distressed Firms: Debt Overhang or Agency Problem? May 25th, 2017 James L. Park * Abstract This study empirically tests whether firms are more likely to issue equity when distressed. Contrary to Myers (1977) debt overhang predictions, I find that all else being equal, distressed firms issue more equity compared to less-distressed firms, and this disparity is greater for firms with weaker governance. Moreover, I find distressed equity issuing firms yield lower delisting and CEO turnover rates in the short-horizon but eventually increase to similar levels of comparable distressed firm. Distressed issuers also exhibit significant negative equity returns, lower Tobin s Q, deteriorating profitability, lower investments, and excess cash holdings following issuance, which indicate that managers subsidize losing operations rather than invest in positive NPV projects. These findings collectively suggest agency problems in distressed equity issuance. Keywords: Equity Issuance, Distress, Debt Overhang, Agency Problem JEL Classification: G12, G14, G31, G32, G33 * Korea University, Business School Main Building A504, 145 Anam-ro, Seongbuk-gu, Seoul, , South Korea. jpark1@korea.ac.kr. Telephone: This paper is based on the first part of my dissertation at the Wharton School. I would like to thank my advisor, Joao ˆ Gomes, Craig MacKinlay, Luke Taylor, and Amir Yaron, for their invaluable insight and comments

2 1. Introduction One of the central questions in capital structure literature is how firms finance in distress. The prevalent view of Myers (1977) predicts that managers are hesitant to issue equity under distress because a substantial portion of the benefits reaped from additional equity issuance is transferred from shareholders to junior creditors, creating inefficient underinvestment, i.e., the debtoverhang problem. According to the debt overhang problem, a key assumption of such reluctance to issue equity is that managers act in accordance with their shareholders best interest. A challenge to this assumption originates from literature on agency theory, which assumes that conflicts of interest exist between managers and their shareholders. According to the agency theory, risk-averse managers have strong private incentives 1 to finance even negative net present value (NPV) projects during distress because self-interest motivates them to continue their own employment by delaying default and financing ongoing projects, regardless of the best interest of shareholders. 2 Thus, the agency theory predicts that managers have greater incentives to issue equity in distress, which contrasts with the debt-overhang theory. In this study, the agency problem will refer to this specific type of shareholder-manager conflict. 3 This specific type of manager-shareholder agency problem should also be differentiated from other managerial decisions that would simply 1 Grossman and Hart (1982), Gilson (1989), Gilson and Vetsuypens (1993), Eckbo, Thorburn, and Wang (2016) show that the loss of benefits, specialized human capital, and reputation, can be large for distressed firm managers and can play an important private incentive for managers to continue funding deteriorating operations. 2 Aghion and Bolton (1992), Boot (1992), and Dewatripont and Tirole (1994) provide descriptions of self-interested managers extending negative NPV projects against the shareholders best interests. 3 The term agency problem in agency theory literature is used more broadly and could refer to a variety of conflicts among company agents, including managers and equity/debt holders. Even within the manager-shareholder conflict, it refers to the misaligned interests between self-interested, career-concerned, and risk-averse managers and risk-seeking shareholders, although I acknowledge that there are other shareholder-manager conflicts in the literature (e.g., empire-building, myopic investments, herding, quiet life, and overconfidence). See Stein (2003) for a survey of papers. 2

3 benefit either creditors or shareholders at the expense of the other in the proximity of distress (e.g., risk-shifting and asset substitution) that we test but do not find evidence for in this paper. To substantiate the hypothesis that the shareholder-manager agency problem better predicts distressed equity issuance of firms, contrary to the predictions of the debt-overhang problem, I explore equity issuance data in the broad cross-section of distress. To test the above hypothesis, this study examines whether firms issue more equity when distressed, and whether the firms with weaker governance (measured by institutional ownership, insider ownership, and governance index) do so to a greater degree. Subsequently, this study analyzes delisting rates, (forced) CEO turnovers and equity returns to study the impact of distressed issuance on the company distress levels, CEO tenure, and shareholder value, respectively..finally, this study examines various accounting ratios on post-equity issuance valuation, profitability, investment, and risk to distinguish itself from competing hypotheses. To examine the issuance behavior of firms in distress, I conduct pooled panel regression analyses to predict the following years equity issuance using the distress measure of Campbell, Hilscher, and Szilagyi (2008) and various control variables. 4 The study finds a strong positive relation between distress and equity issuance, robust against the use of direct distress measures or relative degrees of distress sorted by the cross-section of firms. In economic magnitudes, I find that a 1% increase in the 12-month-ahead probability of failure rate predicts an increase in quarterly equity issuance rate of 1.11%. I also show that the results are robust to using alternative distress measures as Altman s Z-score (1968) and Ohlson s O-score (1980). This strong positive relation between distress and equity issuance is contrary to the predictions of the debt overhang problem. I then test whether managers of distressed firms issue more equity in firms with weaker 4 Although I control for various firm characteristics and fixed effects throughout this study, this relationship is neither causal nor free of possible omitted variable bias. 3

4 governance. I use three different measures to proxy alternative channels that impact the governance of firms as follows: (1) institutional ownership as a proxy for outside shareholder monitoring, (2) managerial ownership as a proxy for shareholder-manager alignment, and (3) the Governance Index (G-index) developed by Gompers, Ishii, and Metrick (2003) as a proxy for the external threat of hostile takeover. This study finds that firms with weaker corporate governance (i.e., lower institutional/managerial ownership and higher G-index) issue more equity when distressed. 5 When governance is strong, there is no measurable effect of distress on equity issuance suggesting that debt-overahng theory may be appropriate when we can assume strong governance. Together with the previous result of strong positive relation between distress and equity issuance, however, we can infer that the shareholder-manager conflict of interest in the U.S. is strong enough to drive an overallpositive relation between distress and equity issuance, and thus, it strongly supports the agency problem hypothesis to better explain distressed equity issuance patterns in the U.S. To further substantiate the ex post effect of the equity issuance on distress and private benefits for managers, I study the post-issuance delisting rates and CEO turnover rate following the equity issuance. I find that the delisting rates of equity issuing firms are initially lower in the short-run (1 year) but evenetually increase to a level similar to other non-equity issuing distressed firms in longer horizons (4 year). The longer horizon results are driven by significantly higher performance related delistings of distressed firms that do not issue equity. CEO turnover data also exhibit similar patterns: the rate is initially lower but increases in the longer horizons. Further looking into the nature of CEO turnover indicates that CEOs that issue equity are increasingly forced out in the longer horizons, 5 The study assumes that managers can sell equity to outside sophisticated investors by issuing equity at a discount [e.g., Brophy, Ouimet, and Sialm (2009) and Chaplinsky and Haushalter (2010)] while shareholders of companies with weak governance have difficulty preventing self-interested managers from issuing equity. 4

5 even when firms do not experience eventual delistings. These results together suggest that the distressed equity issuance provides only a temporary solution to distress rather than a permanent solution. Despite these results for the firm, it is important to note that managers still enjoy a longer tenure by delaying default and financing ongoing projects, providing strong empirical evidence of the incentives for managers to issue equity in distress as suggested by the agency theory literature. Next, I analyze the optimality of equity issuance for stockholders and its impact on share holder value by predicting the following year s equity returns. I find that distressed firms that issue equity have an abnormal quarterly equity return of 3.084% following equity issuance, relative to non-equity issuing distressed firms. This negative abnormal return suggests that distressed equity issuers have negative long-run effects on equity value compared to their non-equity issuing peers. These results augment the agency view that distressed equity issuance is not optimal for equity holders, despite potential benefits (e.g., buying time to prevent immediate bankruptcy and reorganization). Finally, this study examines accounting ratios to test whether other company attributes corroborate the agency view of distressed equity issuance. I find that distressed firms that issue the most equity have lower Tobin s Q, decreasing profitability, non-increasing investments, and excess cash holdings in the year immediately following such issuance. These results, when combined, suggest that the primary motivation behind distressed equity issuance is not to initiate new investments in sufficiently positive NPV projects, but instead to prolong existing negative NPV projects by subsidizing them through cash holdings from equity issuance. This result is consistent with the agency problem depicted in this study. This study implements several additional tests to exclude alternative interpretations of its 5

6 empirical results. 6 (1) To address concerns that leverage rebalancing (i.e., optimal capital structure) or risk shifting (i.e., asset substitution) motivations better explain equity issuance patterns than the agency problem, this study tests changes in risk (i.e., leverage, cash flow volatility, equity beta, and asset beta) of distressed issuers, but finds insignificant results. (2) To address concerns that distressed firms issuing equity are simply more distressed than other distressed firms, or that equity financing is a substitute for debt financing (i.e., the pecking order theory), this study tests differences in ex ante distress levels, debt covenant violations, and debt issuance of equity issuers, and finds insignificant differences in such ex ante distress characteristics. 7 (3) To address the possibility that lower returns and profitability of distressed equity issuers result from their managers earnings management, this study tests changes in discretionary (current) accruals following issuance, and finds insignificant results. (4) Finally, to address concerns that the results of this study are driven by the unique characteristics of the distress measure of Campbell, Hilscher, and Szilagyi (2008), the study shows that the main results are robust to the use of alternative distress measures of Altman s (1968) Z-score and Ohlson s (1980) O-score. The findings of this study collectively provide corroborating evidence for the agency problem perspective of distressed equity issuance, and thus, make three significant contributions to financial literature. First, this study provides strong support for a specific shareholder-manager agency problem in distress. Prior empirical literature, which discusses the significance of the debt overhang problem, mainly focuses on estimating the magnitude of investment distortions and wealth transfer between shareholders and creditors. As wealth transfer and investment distortions are not directly 6 See Sections 4.4 and 4.5 for further discussions. 7 Notice that the insignificant differences in debt covenant violations also suggest that creditors are not major players in distressed equity issuance. As creditors can significantly benefit from issuing additional equity, they may force managers to issue such equity against shareholders best interests when their influence is stronger (i.e., debt covenant violations). However, empirical evidence of this conjecture is unfounded. 6

7 observable, the literature heavily utilizes structural models and their parameter estimates and finds mixed results. 8 Unlike these studies, which assume benevolent managers and use structural models for estimation, this study tests the possibility of shareholder-manager conflict by taking a more empirical approach. This study tests directly observable equity issuance patterns predicted by the debt overhang theory against opposite patterns predicted by the agency theory, where both problems are amplified under distress. The paper rejects the debt overhang theory by showing that debt overhang theory s assumption of benevolent manager is violated enough to affect the overall equity issuance pattern in distress. Thus, the major concern for the distressed firm financing is not to find ways to encourage managers to invest in positive NPV projects, but rather to deter them from financing negative ones. To the best of my knowledge, this is the first paper to argue that agency problem drives a strong positive distressed equity issuance pattern in the cross-section of all publicly traded U.S. firms. 9 Further, this paper adds to the literature on empirical agency theory, which suggests that managers would act against shareholders best interests when facing distress related risks. DeAngelo, DeAngelo, and Wruck (2002) argue a strong agency motivation by showing how a distressed firm (i.e., L.A. Gear) manager could prolong the life of a company by financing deteriorating operations through the sale of liquid assets. Gormley and Matsa (2011) explain that risk-averse managers respond to carcinogen liability risk by acquiring unrelated cash-rich businesses with high operating cash flows to provide liquidity to the firm, a behavior which is 8 Some studies find insignificant results while others find significant results. Mello and Parsons (1992) and Parrino and Weisbach (1999) are examples of the former. Hennessy (2004) and Moyen (2007) are examples of studies that find significant results. 9 Franks and Sanzhar (2006) and Jostarndt (2009) are the only two papers we are aware of that focus on distressed equity issuance to test the debt-overhang problem. Both papers, however, focus on the debt-equity conflict, rather than manager-shaerholder agency problem, in the U.K. and Germany where bankruptcy codes are much different from those in the U.S., making their results difficult to generalize. 7

8 11 against the risk-seeking equity holders best interest. Despite such studies that use a limited group of firms that finance in a specialized manner, large sample evidence of such shareholder-manager conflict has been difficult to document. 10 It is still debatable whether managers of distressed firms engage in projects benefiting stockholders, or act more conservatively in self-interest in the crosssection. 11 Contrarily, this study shows consistent evidence across the board of all public firms through a more common financing tool of equity issuance in distress. Finally, this paper contributes to the empirical literature on equity issuance. Equity issuance has commonly been argued to be motivated by managers selling over-priced equity [see Loughran and Ritter (1995), and Baker and Wurgler s (2002) market timing hypothesis]. Other possible motivations for equity issuance include financing of growth options for young firms [see Carlson, Fisher, and Giammarino (2006))] and near-term cash needs [see DeAngelo, DeAngelo, and Stulz (2010)]. Contrary to studies that assume benevolent managers, this study focuses on a specific situation where the conflict of interest between shareholders and managers is more apparent. Consequently, this study finds that managers of firms facing financial distress issue equity to extend their careers at the expense of current shareholders, which offers stronger evidence for a distress-agency motivation for equity issuance. 12 The remainder of this paper is organized as follows. Section 2 reviews the relevant theories and their predictions, Section 3 describes the distress measure and equity issuance datasets used for the analyses, Section 4 presents the main empirical results and discusses alternative interpretations of the results, and Section 5 concludes. 10, 11 See page 7 of Ayotte, Hotchkiss, and Thorburn (2013) 12 Empirical literature on shareholder-manager agency conflict as a motivation for equity issuance has been sparse. Jung, Kim, and Stulz (1996) also find support for a managerial agency model by showing that many firms without valuable investment opportunities still issue public equity. This paper differs from Jung et al. (1996) in that it focuses on a specific type of agency problem that occurs specifically only in distress situations. where the Jung et al. (1996) studies managerial agency model in general situations.. 8

9 2. Review of Theories and Development of Hypotheses This section reviews the debt overhang and agency problems and their predictions of distress and equity issuance Debt overhang problem Myers (1977) debt overhang problem suggests that when firms are in distress, the majority of additional value created by investing in new projects is secured by existing junior creditors. This value transfer to creditors deters additional financing, even for positive NPV projects, which creates an under-investment problem. Managers will refrain from issuing equity under these circumstances, as issuance will force shareholders to bear losses while creditors reap the benefits [see Myers (2003)]. Benevolent managers would only issue equity when the benefits to shareholders are sufficiently large to offset the value transfer to creditors. This study empirically tests the validity of the debt overhang problem by focusing on its prediction on distress and equity issuance, a prediction largely ignored in the literature. The existing literature on the debt overhang problem focuses on estimating the magnitude of value transfer and/or investment distortions, assuming managers are benevolent. However, this study takes a more empirical approach and tests the possibility of shareholder-manager conflict by focusing on the observable equity issuance predictions of debt overhang against the opposite predictions of the agency problem, where both problems are argued to be more prevalent under distress. I prepare the following three predictions for the debt overhang problem. Prediction DO1: The debt overhang problem predicts that distress and equity issuance are negatively correlated and that governance does not predict distressed equity issuance. All else being equal, the debt overhang theory suggests that distressed firms would cut back on 9

10 equity-financed investment [see Myers (2003)]. Although it is arguable if the magnitude of the debt overhang problem is significant, at minimum, the debt overhang problem predicts that wealth transfer would increase and increasingly deter equity issuance, as firms become more distressed. Therefore, the debt overhang problem predicts a decreasing, or at best, a nonincreasing relationship between distress and equity issuance after controlling for other motivations of equity issuance (e.g., life-cycle, size, investment opportunities, and market timing hypothesis). Moreover, the debt overhang theory assumes benevolent managers. Under the debt overhang theory, the levels of company governance should not significantly influence the equity issuance decisions of distressed firms. Thus, the debt overhang theory predicts that benevolent managers will decide whether to issue equity, regardless of the degree of governance, as they always consider the best interests of their shareholders. Overall, the debt overhang theory predicts a negative relationship between distress and equity issuance, along with a weak interaction between corporate governance proxies. Prediction DO2: The debt overhang problem predicts that distressed firms only issue equity when they can reduce default rates and/or increase shareholder value. Although the predictions of DO1 provide strong support for the debt overhang theory, they do not speak to the optimality of the investment decision on shareholder value. Distressed firms could still issue more equity than stable firms if they find highly profitable investment opportunities more frequently (i.e., positive NPV projects), which outweigh any wealth transfer to creditors. Profitable investment opportunities also include actions that reduce default risk and secure time to restructure company operations and eventually increase shareholder value. 13 However, accurately estimating 13 Notice that Jensen and Meckling s (1976) risk shifting theory predicts the opposite: shareholders would not want to reduce the company s risk, but rather increase it to benefit from the upside potential, at the expense of creditors bearing the downside risk. 10

11 and comparing the magnitude of wealth transfer and value of investment opportunities has proven to be difficult. This study instead evaluates the optimality of such decisions by examining delisting rates and equity returns that follow equity issuance. If the purpose of equity issuance is to decrease distress costs and restructure operations, firms should exhibit lower delisting rates after equity issuance, both in the short- and long-horizons. Importantly, if such issuance is in the best interests of shareholders, regardless of purpose, it should increase value for the equity holders. Therefore, if the proceeds from equity issuance are used efficiently, such firms should experience positive, or at least non-negative, abnormal returns compared to non-equity issuing distressed firms. Prediction DO3: The debt overhang problem predicts that distressed firms will have improved accounting ratios following equity issuance. Although the debt overhang theory suggests that distressed firms usually do not want to issue equity (DO1), any equity issuance could still be justified for investment opportunities that create sufficient value to justify any wealth transfer to creditors. In addition to the optimality of such decisions for shareholder value (DO2), ex post accounting ratios should also show corroborating evidence of improvements in the firm s operations. Equity issuance should lead to increased investments in positive NPV projects. Therefore, we expect that investment (e.g., capital investment, research and development, and acquisitions) will increase after equity issuance. Further, we expect profitability of the distressed firm to increase, as the investments would target positive NPV projects that are sufficient enough to overcome any wealth transfer to shareholders, which would result in improved profitability. Subsequently, the distressed firms that issue equity should have better valuation ratios following these investments and improved operating profits Agency Theory 11

12 In contrast to the debt overhang problem, the agency theory predicts that self-interested managers would issue equity even in cases where the investment opportunity does not maximize shareholder value [see Jensen and Meckling (1976) and Jensen (1989)]. Risk-averse managers have strong private incentives to continue funding negative NPV projects under distress to continue their employment 14 rather than optimally liquidate projects (or liquidate the firm as a whole) and return capital to the shareholders [see Aghion and Bolton (1992), Boot (1992), Dewatripont and Tirole (1994), and Baker (2000)]. It is important to note that the agency theory does not interfere with the wealth transfer (from shareholders to creditors) part of debt overhang. The agency theory argues that managers may still issue equity to prolong negative NPV projects, while debt overhang negates such behavior, as wealth transfer to creditors magnifies the cost to shareholders when distress levels increase. The main difference between the two theories is that the debt overhang theory does not consider the possibility that managers would pursue self-interest, while the agency theory predicts that managers would be incentivized to issue equity when in distress. For clarification, the agency problem, as described in this study is particular to the shareholdermanager conflict in distress situations, where the major concern is of managers not wanting to terminate unprofitable projects. Other shareholder-manager conflicts in the literature include motivations such as tendency to empire-build, conduct myopic investments, tendency to herd, preference for quiet life, and overconfidence. Any debt-equity conflict, in general, could also be referred to as an agency problem in the literature, but not in this study. See Stein (2003) for detailed references of these conflicts. 14 Grossman and Hart (1982), Gilson (1989), and Gilson and Vetsuypens (1993) argue that the loss of benefits, specialized human capital, and reputation, are important private incentives for managers to avoid default. The literature also documents how managers are risk averse, and how risk-averse managers may affect the firm s financing decisions [e.g., Lewellen (2006)]. 12

13 I prepare the following three predictions of the agency theory in parallel with the predictions of the debt overhang problem. Prediction AT1: The agency theory predicts that distress and equity issuance have a strong positive relationship, and more so for firms with weaker governance. The crux of the agency problem is that it is difficult to prevent managers from subsidizing existing negative NPV projects. To subsidize deteriorating projects, managers require additional financing and equity issuance as a mean to meet this goal. 15 Self-interested managers have a strong incentive to issue equity under high distress levels, whereas stockholders would not want to issue equity because the debt overhang would result in higher wealth transfers to creditors. If agency problems dictate equity issuance more than debt overhang problems in the cross-section of distress, we should observe that distressed firms issue more equity on average. Further, I test whether firms with weaker governance issue more equity when distressed. If governance is behind the positive relationship between distress and equity issuance, firms with stronger governance should show less of this behavior while firms with weaker governance should show more. As governance is not directly observable, this study uses three major proxies to measure different aspects of corporate governance. I use institutional ownership as a proxy for outside shareholder monitoring [e.g., Shleifer and Vishny (1986)], managerial ownership as a proxy for inside shareholder-manager alignment [e.g., Jensen (1993)], and anti-takeover provisions in the G-index as a proxy for weaker takeover threat [see Zwiebel (1996) and Gompers, Ishii, and Metrick (2003)]. When these proxies interact with distress, I predict that the managers of distressed firms with less institutional ownership, less managerial ownership, and higher G-index, issue more equity under the agency theory assumption. 15 Asset sales (DeAngelo, DeAngelo, and Wruck, 2002) and acquisitions (Gormley and Matsa, 2011) can provide an alternative means of financing for managers, in response to distress risk. 13

14 Prediction AT2: The agency theory predicts that equity issuance by distressed firms not only fails to deter default and CEO turnover in the long-run, but also results in negative equity returns following the issuance. Similar to DO2, I test the impact of distress equity issuance on a firm s ex post distress level, CEO Turnover, and shareholder value. Since the agency theory argues that equity issuance is motivated by a manager s private incentive to prolong his/her career, equity issuance is viewed as a simple technique to provide liquidity to postpone bankruptcy rather than improve operations. Without restructuring the firm, such actions only delay default rather than turn the company around [see case of L.A. Gear in DeAngelo, DeAngelo, and Wruck (2002)]. To measure the extent of delay in bankruptcy, I compare the delisting rates of distressed firms that issue equity with non-equity issuing distressed firms for different periods. The agency theory predicts that a firm s delisting rate would not improve in the long-horizon despite possible improvements in the short term. To substantiate the private incentives for managers we study the CEO turnover rate following the distressed equity issuance. Since agency theory argues that the major private incentive for managers to issuing equity in this specific agency problem is to prolong the current employment, we expect a relatively lower turnover rate following the equity issuance. Without benefiting shareholders in this process, however, we expect CEO turnover to increase back to its benchmark levels longer horizons, suggesting only a temporary solution to the problem. The fraction of Forced CEO turnovers are also expected to increase as shareholders realize that the distressed equity issuance did not benefit shareholders, driving the long-run turnovers of distressed equity issuers. More generally, as the agency problem suggests that managers of distressed firms would issue equity regardless of their shareholders best interests, we expect a negative impact on the value of equity following issuance. I measure the impact of distressed equity issuance by examining their post-issuance one-year abnormal equity returns. Gompers, Ishii, and Metrick (2003) and Giroud 14

15 and Mueller (2011) are among several studies, which contend that firms with weaker governance have lower equity returns. 16 Therefore, contrary to the predictions of debt overhang, the agency theory expects that equity issuing distressed firms to have greater negative abnormal returns than comparably distressed firms that do not issue equity. Prediction AT3: The agency theory predicts that managers of equity issuing distressed firms subsidize deteriorating operations. The Agency theory also predicts that funds obtained through equity issuance are used to finance ongoing negative NPV projects, rather than in investment or restructuring. Deteriorating project subsidies can be measured by observing fluctuations in accounting ratios of distressed companies. Thus, the agency theory predicts that such firms would not make new investments but instead increase cash holdings after equity issuance to prolong their existing operations, rather than invest in new positive NPV projects. Such actions eventually worsen operations and result in deteriorating profitability (ROA), which would consequently be observed in the lower valuations of these firms. Finally, beyond these predictions of the debt overhang and agency theory, in Section 4.5 this study examines interpretations and evidence for alternative theories in the literature for robustness; however, it finds insignificant results. For instance, it examines risk changes after equity issuance to find evidence of Jensen and Meckling s (1976) risk-shifting theory (i.e., asset substitution) but finds insignificant results. The study also examines changes in leverage, ex ante distress levels, debt issuance, and covenant violations for distressed equity issuers to reveal possible trade-off and pecking order behavior; however, it finds insignificant results. 16 As argued in the literature, the impact of agency costs may affect stock returns in longer horizons. Initially, investors may underreact to the agency problem, as it is not apparent how the proceeds are used when equity issuance is announced. If equity issuance is followed by inefficient investments or sustained losing operations, the stock prices ultimately reflect these agency costs in the long-run (possibly after the market observes deteriorating profitability). Thus, one would observe a delayed impact of distressed equity issuance in stock prices (i.e., after announcement day). The delay in CEO turnovers in Table IV also provides supporting evidence of the delayed response to equity issuance. 15

16 3. Data I use four main data sources for the analysis in this study. For stock market data, I use the CRSP monthly database, which I also use to find market information and extract net issuance. For accounting data, I use the Compustat (CRSP/Compustat Merged) quarterly database, where I use permno to match firm observations with the CRSP database. The Compustat quarterly database is used to obtain accounting data, construct the Campbell, Hilscher, and Szilagyi (2008) distress measure, and to construct distress bins. In addition, I use Thomson Reuter and the IRRC database to collect corporate governance data and ExecuComp data to collect CEO turnover data Distress Measure The distress measure used in this study is from Campbell, Hilscher, and Szilagyi (2008) (CHS measure). The CHS measure is a 12-month-ahead probability of failure estimated by a logit model. Failure is defined as delisting for performance-related reasons, receiving a D rating from a rating agency, or filing Chapter 7 or Chapter 11. The Appendix contains detailed equations and derivations of each variable that compose the distress measure. The study uses the CHS distress measure for several reasons. First, this measure is the most recent distress measure based on both accounting and financial information, which is known to outperform other distress measures such as Altman s (1968) Z-score, Ohlson s (1980) O-score, among others. Second, the estimation of the measure is based on periods from 1981 to 2003, which largely overlaps with my sample period. Later in Section 4.4, I show that results with alternative distress measures are robust Equity Issuance 16

17 ex Net equity issuance is calculated using CRSP monthly returns excluding dividends ( R ), stock, price ( P ) and the number of shares outstanding ( it, N ) of firm i at both month t and t 1. To it, adj adj calculate the split-adjusted net equity issuance ( N, / N, ), I use the fact that CRSP monthly returns, 1 ( 1 R ex,, / adj, 1, / adj it Pit Pit Pit Pit, 1 it adj adj + = = ), and market value, ( Pit, Nit, = Pit, N ), are indifferent to it, using split-adjusted price and shares outstanding. 17 Then, net equity issuance can be calculated by 1 P N it ex it, it, adj adj adj adj 1+ Rit, Pit, 1 Pit, Nit, Nit, it, = = adj adj adj adj Pit, 1 Nit, 1 Pit, Pit, 1 Nit, 1 Nit, 1 Net Equity Issuance = it Net equity issuance for each firm can be calculated each month if we have price, shares outstanding, and return data. However, the CRSP database does not necessarily observe the number of share increases each month. In most cases, CRSP updates the number of shares at the end of each calendar quarter; therefore, a lag of one or two months is possible in the equity issuance data compared to the actual equity issuance. Therefore, I accumulate the quarterly net issuance for each firm as one would compound stock returns. As the CRSP net issuance includes all transactions that increase outstanding shares, the database could be regarded as representative of the equity issuance population, making it possible to analyze equity issuance in the broad crosssection of all traded firms The sample Initally, the sample is formed by using observations that include accounting and market information. The following information is also required for each observation to be included in the sample: CHS measure, Tobin s Q, profitability (ROA), leverage (TLMTA), cash holdings 17 Pontiff and Woodgate (2008) and Daniel and Titman (2006) are examples of studies that use changes in shares outstanding or sale of equity to measure net issuance. As a robustness check, I use the quarterly Compustat database to confirm my results. First, I use cash flows from the sale and repurchase of common and preferred stock, adjusted for sale and repurchases of preferred stock, to calculate net issuance. Then I divide net issuance by total market value to calculate the net issuance rate. In unreported results, I find similar strong positive correlations between equity issuance and distress using the Compustat database. 17

18 (CASHMTA), and past returns at the year-ends of each year during the period of in the period 1990 to 2011 (year t 1 ). In addition, I require that firms have benchmark returns (henceforth, DGTW 18 ) for return adjustments (Daniel, Kent, Grinblatt, Titman, and Wermers, 1997). For predictive panel regressions, I measure distress by forming relative distress bins according to the convention of Fama and French (1993) by lagging the accounting variables for at least six months to ensure that there is sufficient time for data to be publicly available at the date of portfolio formation in July. To follow the screening procedures of Campbell, Hilscher, and Szilagyi (2008), I require sample observations to be common stocks traded on the NASDAQ, NYSE, and Amex exchanges excluding financial firms, while firms with share prices less than $1 as of July in year t are excluded. The sample span is from July 1991 to June Additionally, I use Thomson Reuter s database to collect institutional and insider holdings data for proxies of corporate governance. Institutional holding is calculated by dividing institutional holdings by the number of shares outstanding. Managerial ownership is the total number of shares held by the top five officers divided by outstanding shares. Observations are required to include ownership information in year t 1, which yields 76,401 firm-year observations. From the Investor Responsibility Research Center (IRRC) database, I collect G-index scores from 1990 to 2006, which are provided every three years. I follow the convention of filling in the missing G-index variables using the last observation lagged up to two years until I also scale the G-index dividing it by 24 (the highest G-index that can be achieved) so that the maximum value is equal to 1. To prevent sampling bias, I do not restrict the full sample observations to have the G-index, as the G-index is only available for larger firms drawn from the S&P The number of firm-year observations decreases from 76,401 to 27,737 for the 18 The DGTW benchmarks are available at 18

19 subsample with G-index information. Additionally, CEO turnover information is collected from the ExecuComp database, which is only available for 27,602 firm-year observations. Table I contains the summary statistics with mean, median, and standard deviation of both the full sample and the highest distress bin subsample. The Appendix contains the detailed description and derivation of each variable. The summary statistics show that the firms in the highest distress bin exhibit characteristics of a distressed firm. Firms are younger and smaller with lower Tobin s Q. Compared to the full sample, distressed firms have lower profitability (ROA) and higher leverage, while cash holdings and capital investments are similar. For proxies of governance, distressed firms exhibit mixed results. They have lower institutional ownership, but higher insider ownership, and lower G-index. 4. Results To test the debt overhang theory against the agency theory, I investigate which theory s predictions are more consistent with the data analyses in this study. First, the study investigates the relationship between equity issuance, distress, and corporate governance (predictions DO1 and AT1). Second, it investigates the relationship between distressed firms that issue equity and their delisting rates and equity returns (predictions DO2 and AT2). Finally, it investigates accounting ratios of distressed firms that issue equity for additional evidence (DO3 and AT3) Equity Issuance, Distress, and Corporate Governance I document the relationship between the degree of equity issuance, distress, and corporate governance using pooled panel regression analysis to study how distressed firms issue equity. The main variables are as follows: I use quarterly log net issuance rate as the dependent variable to 19

20 mitigate the concern of non-normality. 19 I then adjust the above net issuance by subtracting median past net issuance for the same two-digit SIC industry within the same size and book-to-market (three-by-three sorting), where past net issuance is the average quarterly log net issuance rate measured from January to December of year t 1. I predict industry-adjusted quarterly net issuance in percentages from July of year t to June of year t + 1 using explanatory variables from the end of the year t 1. The main explanatory variables are various proxies of distress that are measured by the distress measure of CHS, and corporate governance, which is proxied by institutional and insider ownership, and the G-index. For other explanatory variables in this test, I include past net issuance, and lagged control variables throughout the study to mitigate potential endogeneity and omitted variable problems in the sample. Modeling distress, equity issuance, and governance simultaneously can be problematic if equity issuance endogenously affects past distress levels and firm governance. Prior theories, however, suggest that the causal relationship, in general, goes from governance and distress to equity issuance, rather than vice versa. Without a clear instrumental variable that could be used for the broad cross-section of firms, this study follows the literature on capital structure by examining how lagged distress and governance characteristics predict future equity issuance by including end of last year s equity issuance and control variables. 20 I also control for various firm characteristics, which the literature argues to motivate equity issuance, to mitigate omitted variable issues in this study, although the dependent variables are 19 In unreported results, I plot the distribution of quarterly and monthly net issuance and log net issuance to address the concern that net issuance may lead to non-normal distribution. I observe an improvement in distribution in the quarterly log net issuance, and thus, use the results as the main measure of equity issuance in this study. The quarterly rate is also consistent with the CRSP database, which updates the number of shares outstanding at the end of each calendar quarter for most firms. 20 See Harford, Mansi, and Maxwell (2008) for a discussion on including lagged variables in pooled panel regressions for predicting various accounting ratios on corporate governance proxies. 20

21 already adjusted by industry, size, and book-to-market median values. I use control variables following DeAngelo, DeAngelo, and Stulz (2010) who discuss different motivations for equity issuance. The logarithm of age and assets is used to control for lifecycle motivations, Tobin s Q to control for investment opportunities or valuation of the stock, return on asset (ROA) to control for firm profitability, cash holdings to control for current cash reserves, leverage to control for debt capacity, and past returns to control for market timing motivations of equity issuance. Accounting ratios are winsorized below and above at the 0.5% level so that extreme values do not drive the results. Definitions and detailed derivations of each variable can be found in the Appendix. The regressions also include industry and year fixed effects. The choice of industry fixed effects, instead of firm fixed effects, is due to the insufficient within-variations of firms corporate governance, a concern voiced by Gompers, Ishii, and Metrick (2003) and Giroud and Mueller (2011). However, the results in the following section are qualitatively similar with firm fixed effects. I follow Petersen (2009) and report the t-statistics for pooled results using standard errors corrected for firm level clustering Distress and equity issuance I first test the relationship between distress and the following year net equity issuance. I use variations of the distress measure including the lagged distress measure (CHS) directly and the failure probability, which is the logistic transformation of the distress measure [1 / (1+ exp( CHS )]. I also show results for indicator functions of the top 20%, 10%, 5%, and it, 1 1% distress firms to show robustness and also variation in equity issuance as per different levels of distress. Notice that the indicator function of the top 10% distress level is used later as the baseline to indicate distress in this study. Table II reports the regression results. The first two regressions that use the CHS measure and 21

22 the failure probability to measure distress present statistically significant coefficients of (tstat = 6.57) and (t-stat = 6.94), respectively. These coefficients have positive signs, confirming that distressed firms issue more equity. The economic magnitude of this coefficient for failure probability can be interpreted as follows: a 1% increase in the 12-month-ahead failure probability predicts a 1.11% higher quarterly equity issuance. Regressions (3) through (6) present results of indicator functions for different levels of distress: (3) indicator function of the top 20% distress firms, (4) top 10% distress firms, (5) top 5%, and (6) for top 1% distressed firms. All coefficients are positive and statistically significant. Moreover, the magnitudes of the coefficients are monotonically increasing as the functions indicate higher levels of distress. This pattern suggests that the positive relationship found in regression (1) and (2) is due to neither a mere local fluctuation in the distribution of firms sorted by distress, nor a result driven by non-distressed firms. The positive relationship between distress and equity issuance is strongly driven by the most distressed firms each year. The results also reveal that control variables predict equity issuance significantly in directions expected from those of various motivations of equity issuance. Younger and smaller firms tend to issue more equity following the life-cycle motivation of equity issuance. Firms with higher Tobin s Q generally issue more equity, suggesting that firms with better investment opportunities issue more equity. Firms with less cash issue more equity, consistent with the cash demand motivation of equity issuance. Less profitable firms (ROA) with less debt capacity (high leverage) issue more equity, consistent with the predictions of the pecking-order theory. Finally, firms with higher past returns issue more equity, consistent with the market timing hypothesis. Excluding any combination of these control variables does not change my main results. 22

23 For robustness, instead of using net equity issuance, using indicator functions of the firms with net issuance above a certain threshold, or those in the top decile or quintile of equity issuance yield similar results. In addition, using monthly or annual instead of quarterly panel regressions yield similar results. When running variables that are used to construct the CHS measure as right-hand side variables to predict equity issuance, I find that seven out of eight variables accurately predict higher issuance in the same direction that they predict distress. 21 This result suggests that my results are not driven by a particular characteristic of the distress measure. Consequently, any distress measure that uses any combination of these variables would also find positive results. I also present results using Altman s (1980) Z-score and Ohlson s O-score distress measures in regressions (7) and (9), respectively, and use these distress measures to construct an indicator function of the top 10% ditress in regressions (8) and (10) to predict equity issuance.. All results show statistically significant positive results Distressed equity issuance and corporate governance This section further tests whether corporate governance is responsible for the positive relation between distress and equity issuance. Broadly, this section tests whether the assumption of benevolent or self-interested managers better explains the patterns of equity issuance in the crosssection of distress. As theories such as debt overhang, trade-off, and pecking order assume benevolent managers, the results of this test can be critical. To implement this analysis, I add lagged corporate governance proxies and their interaction terms with indicator functions of distress to other control variables included in Table II. Indicator 21 The only exception is past returns (EXRETAVG). Past return predicts distress negatively, suggesting that past losers are more distressed, while it predicts equity issuance positively, consistent with the market timing hypothesis. This opposite prediction result helps exclude market-timing motivation as an alternative interpretation of my results. 23

24 functions of II tttttt 10% and II tttttt 20% are used to indicate distress in regressions (1) through (3), and (4) through (6), respectively. 22 The governance proxies include institutional and insider ownership for all regressions, and the G-index is further included in regressions (3) and (6). Table III presents the results. First, looking at regression (1), we observe that distress significantly predicts future quarterly net issuance, even after controlling for lagged firm characteristics as shown earlier. These results suggest that governance is important for the decision to issue equity. Specifically, institutional and insider ownership have negative coefficients, suggesting that firms with weaker governance issue less equity as per the agency theory. Regression (2) shows the main result of the table, which includes the interaction term of distress with institutional and insider ownership. The interaction terms are (t-stat = 6.33) and (t-stat = 1.92), respectively. These two negative statistically significant coefficients suggest that firms issue less equity when they are distressed but have more institutional and insider ownership. Further, notice that the coefficient for intuitional ownership is now insignificant, suggesting that the negative relationship between institutional ownership and equity issuance in regression (1) is mainly driven by distressed firms. This means that monitoring of equity issuance by institutional investors is particularly important for distressed firms. In addition, the absolute magnitude of the coefficients of the interaction term is larger for institutional ownership, and similar for insider ownership, when compared to that of the indicator function of distress (1.067 [t -stat = 7.83]). As the sum of the coefficients of distress and distress interacted with ownership are negative or close to zero, these coefficients suggest that firms with 22 These indicator functions are utilized instead of directly using the CHS measure or a continuous variable of the distress decile bin ranking to make the interpretation of coefficients easier, and to avoid potential multicollinearity problems with the control variables. Using continuous variables as a proxy for distress render similar or stronger results, as seen in the previous section. 24