The (out)performance of zeroleverage firms in recessions

Transcription

1 Master thesis Finance The (out)performance of zeroleverage firms in recessions And its implications on dominant capital structure theories Faculty: Tilburg School of Economics and Management Department: Finance Author: Jonathan van den Hoek Student number: Thesis supervisor: Dr. F. Braggion Defense date:

2 2

3 3 Abstract Recently several studies have focused on explaining the zero-leverage (ZL) phenomenon. This thesis provides an approach different to those studies, in an attempt to discover why increasing numbers of U.S. listed firms eschew debt in their capital structures. Using dominant capital structure theories and the hypotheses of financial constraints and financial flexibility as a theoretical framework, performance of ZL firms in times of recession and growth periods is analyzed. Unlike other studies, here ZL behavior is not regarded as a consequence of performance. Rather, being unlevered is considered a strategic decision of dividend-paying ZL firms. The resulting financial flexibility should lead to outperformance in recessions. For non-dividend firms, being ZL is the consequence of a restriction in the supply side of debt which is expected to lead to underperformance in financially adverse times. Empirical evidence reinforces the expectation that dividend-paying ZL firms seek financial flexibility. Consequently these firms outperform comparable levered firms in recessions. Non-dividend ZL firms show signs of financial constraints and high growth. Contrary to expectations they do not clearly underperform to levered firms in recessions. Moreover, there is no evidence that ZL firms capital structure decision making is inconsistent with the static tradeoff and pecking order theories when the hypotheses of financial flexibility and financial constraints are accepted alongside these models.

4 4 Table of Contents Introduction Literature review Modigliani-Miller s capital market assumptions Static tradeoff theory Tax benefits of debt Bankruptcy costs Agency costs Free cash flow benefits Challenge to the static tradeoff theory Pecking order theory Challenge to the pecking order theory Market timing theory Potential hypotheses for the low leverage and zero leverage anomalies Financial constraints hypothesis Underinvestment and Financial flexibility hypotheses Current state of literature on ZL firms Devos, Dhillon, Jagannathan, Krishnamurthy (2012) Strebulaev and Yang (2013) Bessler, Drobetz, Haller, Meier (2013) Dang (2013) Macroeconomic conditions Implications of current literature to this study Hypothesis Data Recession periods Definition of zero leverage Definition of performance variables Regression dependent variables Performance variables in summary statistics Summary statistics Summary statistics of ZL firms Summary statistics of dividend-paying and non-dividend ZL firms... 25

5 5 5.3 Summary statistics compared to control group Regression analysis The regression model Estimation results Fixed effects models Endogeneity and Instrumental Variable regression Endogeneity The instrument Limitations to IV regression for the estimation problem of interest Instrumental Variable regression model Testing for endogeneity and regression results Conclusions, limitations and recommendations for future research Conclusions Results on Hypothesis Results on Hypothesis Limitations and Recommendations for future research References Appendix... 48

6 6 Introduction One of the persistent debates in modern financial research is about the capital structure decision of firms. In the seminal 1984 paper The Capital Structure Puzzle, Stewart C. Myers starts by declaring: How do firms choose their capital structures?" [ ] The answer is, "We don't know (p.575). Thirty years later, this statement can still be considered to be true. Myers himself introduced the pecking order theory of corporate investment financing, which is recognized as one of the dominant capital structure theories. This theory suggests firms have a predetermined pecking order for financing investments, based on the costs of the source of financing. Internal financing is cheapest and therefore most preferable. If this is unavailable, debt is considered. Only as a last resort equity will be issued. The main counterpart of this theory is the static tradeoff theory. Here the bottom line is that the firm s capital structure is the optimal composition of debt, equity and internal funds for that firm. This is mainly focused at the benefits and costs of debt. Although in many situations firms behave in line with capital structure theory, these frameworks are far from perfect. In the last decades a growing fraction of firms refrains from incorporating debt into their capital structures. This trend is a challenge to theory since debt is widely accepted as a viable source of financing. The phenomenon is first referred to by Korteweg (2010) as the zero-leverage (ZL) puzzle. According to Korteweg (2010) the presence of ZL firms explains the low leverage puzzle ; the fact that in practice firms are on average lower levered than deemed optimal by capital structure theory. He finds that excluding ZL firms, most firms are close to their optimal leverage levels. Since Korteweg (2010) made his observation of the ZL puzzle, several studies investigated the phenomenon. Although progress has been made, existing literature still has many gaps. So far, being a ZL firm was regarded a result of firm performance, or external factors. But not as a strategic decision. Hence the research of the ZL phenomenon focused on the factors that were expected to cause firms to be unlevered. Simultaneously, researches observed an upward trend in the number of ZL firms (e.g. Devos et al., 2012; Strebulaev & Yang, 2013). These rising numbers could indicate that rather than being forced by past performance or external factors, some firms choose to become ZL as well. If this is the case, these firms must believe that becoming ZL leads to higher performance. As a result, this should be observable by comparing the performance of ZL firms with levered firms. Dang (2013) concluded that the group of ZL firms is not homogeneous. He argues that some firms strategically eschew debt to achieve financial flexibility. By saving debt capacity, firms are more flexible to respond to profitable investment opportunities. According to Dang (2013), another group of ZL firms is forced by financial constraints. These firms would like to use debt, but are limited by the supply side (banks and investors). Hence their performance is expected to be lower than comparable levered firms. Moreover in financially adverse times (i.e. recessions), these effects are expected to be clearer. Since firm performance is under pressure in recessions, firms able to flexibly finance investments have a competitive advantage. On the other hand firms with limited debt capacity have fewer abilities to finance investments compared to levered firms. Therefore they are expected to underperform. In this thesis, both these expectations are investigated. Also an attempt is made to explain the ZL phenomenon in the light of the dominant capital structure theories. Research is done on a sample of U.S. listed firms between 1962 and This is largely the same as other studies on ZL firms, so that the outcomes can be compared. The results of summary statistics and regression analysis show that ZL firms seeking financial flexibility outperform comparable levered firms. Moreover in recessions their performance deteriorates less. Hence

7 7 their strategy seems successful. Contrary to the conjecture, ZL firms expected to be financially constrained do not unambiguously underperform to comparable levered firms. Also in recessions, not all measured performance indicators worsen more than those of levered peers. Next to these findings, no evidence can be found that ZL firms purposely act in contrast with dominant capital structure theories. The behavior of these firms is very reasonable when taking into account the financial flexibility and financial constraints hypotheses. This finding brings the ZL puzzle a step closer to being solved. Even so, there is still work to be done. While the presence of omitted variables or reverse causality cannot be ruled out, instrumental variable analysis in this study presents puzzling results. The remainder of this thesis is structured as follows. In Section 2 a brief overview is given of the theoretical framework of dominant capital structure theories and the financial flexibility and financial constraints hypotheses. Also the current state of literature on the field of ZL firms is discussed. In Section 3 the hypotheses are formally introduced. Section 4 contains the description of the data, and an explanation of the selected performance variables. In Section 5 summary statistics are described. Regression analyses are discussed in Section 6. The possibility of endogeneity in the model and Instrumental Variable analysis are presented in Section 7. Finally in Section 8, conclusions, limitations and recommendations for further research are formulated.

8 8 2. Literature review The financial conservatism of firms remains a challenge for dominant capital structure theories. Since the seminal paper of Modigliani and Miller (1958) much research has been done on firm s capital structures. In modern literature, the two main capital structure theories are the static trade-off theory and the pecking order theory. In recent years also the market timing theory received renewed attention. In this section the basis for the capital structure theories in the work of Modigliani and Miller will be outlined. Also the basic idea of the aforementioned theories will be explained, and it will be made clear why these theories do a fairly poor job in explaining the observation that many firms follow low leverage policies. Subsequently the financial flexibility and financial constraints theories will be discussed. Finally, the current state of literature on zero-leverage firms and the relevance of studying these firms in periods of diverse macroeconomic conditions will be described. 2.1 Modigliani-Miller s capital market assumptions The assumptions on perfect capital markets introduced by Modigliani and Miller in their 1958 paper are the starting point for most capital structure theories. Modigliani and Miller s proposition I stated that in perfect capital markets, the capital structure decision does not affect firm value. A capital market can be called perfect under the assumptions that there are no corporate or personal taxes, no financial distress costs, no transaction or issuance costs, no information asymmetries, and firm investment decisions are pre-determined. These assumptions provide an economic framework for researchers to depart from. In real capital markets Modigliani and Miller s assumptions do not hold because of the presence of market frictions. To create testable models that better predict real capital market behavior, researchers developed capital structure theories by relaxing different perfect capital market assumptions. 2.2 Static tradeoff theory One of the first and foremost of these theories, the static trade off theory, sets off relaxing the no taxes and no distress costs assumptions. In real capital markets, these frictions are present and influence the capital structure decision. The presence of taxes generates benefits for using debt in the firm s capital structure, since the interest payments of debt are tax deductible. Thus, financing the firm with debt will reduce the amount of taxes to be paid. On the other hand, distress costs pose a threat to the use of debt. Leverage comes with principal and interest payments that have to be fulfilled by firms. This is obligatory of nature. If a firm is unable to comply with this obligation, the ownership of the firm s assets is legally transferred to the debt holders. These could file for bankruptcy, destroying firm value. This tradeoff between benefits and costs of debt is the basis of the static tradeoff theory. In the course of the years more classes of costs and benefits were added to the theory by researchers, relaxing other perfect capital market assumptions as well. The static tradeoff theory predicts that a firm s capital structure is the optimal composition between the various benefits and costs of debt that are briefly introduced below Tax benefits of debt Modigliani and Miller showed in their tax correction article in 1963 that if a firm would always be able to meet their debt obligation, the market value of a firm is a linear function of the amount of debt used in

9 9 the capital structure. Kraus and Litzenberger (1973) incorporated bankruptcy costs into this function in their conceptual State-preference model of optimal financial leverage, showing that if debt benefits exceed the present value of distress costs, a levered firm is valued higher than an unlevered firm. Despite these theoretical models, empirical evidence for the tax benefits of debt took researchers almost two more decades. In 1990, MacKie-Mason showed that firms which are close to tax-exhaustion through deductions other than debt tax shields (e.g. loss carryforwards) or have a high probability of a zero tax rate, are less likely to finance with debt. This suggests that at the margin, taxes influence capital structure decisions. Graham (2000) was the first to quantify tax benefits for firms. He estimated a series of marginal tax rates on a sample of firms by simulating tax rates as if firms spent 0 to 800 percent of their real interest expenses. Graham (2000) found that less than half of the sample firms use debt up to the point that marginal tax rates decline. He estimated tax benefits of firms to be about 10 percent of firm value. However firms could obtain additional gross tax benefits of about 15 percent by levering up to the point where the interest deduction function begins to decline. Finally Graham (2000) signaled that growth firms use debt conservatively, as well as large, liquid and profitable firms. He inferred conservative debt policy is persistent, and indicated this as an interesting field of future research. Korteweg (2010) uses market values of corporate debt and betas of equity to identify the net benefits of debt in a 1994 to 2004 panel data sample. He finds that the slope of net benefits of debt is upward for low-levered firms, but downward for firms with very high leverage levels. This implies the existence of an optimal capital structure. Korteweg (2010) finds net benefits of debt as high as 5.5 percent of firm value for the median sample firm. Like Graham (2000), he indicates that on average firms are slightly underlevered but attributes this to a group of firms that persistently keeps zero leverage. The subset of firms that does lever up comes close to optimal debt levels. According to Korteweg (2010) the behavior of zero leverage firms, as if they face high friction costs to re-levering, is puzzling Bankruptcy costs A disadvantage of using debt is the risk of bankruptcy in case a firm cannot comply with its debt obligations. The first study on direct bankruptcy costs of large, publicly listed firms is that of Warner (1977). He found that the direct costs of bankruptcy of the firms he studied cumulated to an average 5.3 percent of firm value in the month bankruptcy was filed. Weiss (1990) found that direct costs of bankruptcy in his sample were on average 3.1 percent of book value of debt plus market value of equity, in the year prior to the filing. Now direct costs include only administrative and legal costs. The potential indirect costs of bankruptcy, e.g. the distortion of business, could be much higher. However these indirect costs are hard to quantify and researchers that tried (e.g. Andrade & Kaplan, 1998) came up with rather ambiguous results. Moreover, Miller discusses bankruptcy costs in the light of the static trade off theory in his 1977 paper, concluding the supposed trade-off between tax gains and bankruptcy costs looks suspiciously like the recipe for the fabled horse-and-rabbit stew - one horse and one rabbit (Miller, 1977, p.264). Hence Miller (1977) implies the tax gains of debt are much larger than the potential costs of bankruptcy Agency costs Whereas the early static tradeoff theory focused on the tradeoff between tax benefits and financial distress costs of debt as described above, Jensen and Meckling (1976) concluded that these factors are not the only ones influencing capital structure decisions. The way modern firms are structured leads to conflicts of interest between ownership and control, since the firm s owners (principal) delegate control of the firm s business to management (agent). Both parties by nature act to maximize their own interests,

10 10 but these interests are not always aligned. They called this conflict the agency problem, which consequently can lead to agency costs. The issuance of debt leads to conflicting interests between debt holders and the firm s owners, the equity holders. The latter have limited liability and will hence favor more risky investments with a large upside. Debt holders receive a fixed return on their investment and will favor safer, less profitable investments. Jensen and Meckling (1976) define this conflict of interest as the incentive effect. Debt holders can mitigate this effect by asking a premium on the debt issued, or monitoring management in an attempt to control their behavior. In both cases, debt holders will transfer the agency costs to the firm by asking a premium on the debt issued. Hence these agency costs have a negative influence on the firm s decision to lever up. The issuance of outside equity by the ownership of the firm ultimately leads to the same agency costs as debt issuance. If ownership of a firm lies fully with the management, the perks the management enjoys at the expense of the firm also fully come at its own cost. When ownership is shared with outside equity holders, these holders will want to monitor management to ensure these perks do not come at the expense of their ownership in the firm. So both the issuance of debt and sharing of ownership will lead to agency costs in the form of monitoring or bonding costs that come at the expense of the firm. These costs inevitably have an impact on capital structure decision making Free cash flow benefits Whereas Jensen and Meckling point out the disadvantages of debt in a firm s capital structure in their 1976 paper, in a later study in 1986 Jensen argues there are benefits to levering up as well. He calls this the control hypothesis for debt creation. Jensen (1986) argues that in firms with large free cash flows and little fertile investment opportunities, management can repurchase shares or announce a permanent increase in dividends. However these promises are weak, and capital markets subsequently punish dividend cuts severely. To accredit their claim and simultaneously reduce monitoring and bonding costs, management can substitute equity for debt (without retention of the principal payment). By doing so, the free cash flow of the firm will be limited and management effectively restricted in their control over the firm s free cash flows Challenge to the static tradeoff theory The static tradeoff theory suggests that a firm s capital structure decision is a function of the benefits and costs of debt and equity. In this function, the tax benefits of debt are accredited to play a dominant role. However Graham (2000) concluded that on average, firms are heavily under levered. Korteweg (2010) discovered that this is largely due to a group of firms that eschew debt completely. Strebulaev and Yang (2013) found that this fraction consists of 10.2% of their sample of U.S. publicly listed firms. So the zero leverage puzzle, as Korteweg (2010) calls it, is persistent and goes against the basic principle of the static tradeoff theory.

11 Pecking order theory In his 1984 paper The Capital Structure Puzzle, Myers is the first to effectively express the pecking order theory that had been floating around in descriptive literature for a while already. The cornerstone of this theory is the relaxation of the perfect information assumption of Modigliani and Miller s perfect capital market. Although Myers (1984) starts by pointing out we do not know what drives firms capital structure decisions, he argues that the pecking order theory works at least as well as static tradeoff models in explaining what is known about capital structure decision making. He also points out the omission of adjustment costs in the static tradeoff models. If firms face costs for adjusting their debt-to-value ratios to the optimum, this will cause lags. Large adjustment costs could potentially explain why many firms are not levered to their optimum debt-to-value ratios. The fact that this is rarely mentioned in static tradeoff literature is criticized by Myers (1984). While Myers is critical of the static tradeoff theory, until his 1984 paper he dismisses the pecking order idea since he sees no proficient theoretical explanation for the preference of leverage over equity. He denies issue costs as a catalyst for the pecking order since the magnitude of these costs is too small to override the benefits and costs of debt argument of the static tradeoff theory. However in literature on the asymmetric information idea, Myers finds predictions in line with the pecking order. Potential costs of asymmetric information could be of such a magnitude that internal financing becomes most attractive, and external equity financing the least. The rationale is as follows. A firm needs financing for a positive net present value (NPV) investment opportunity. If the investment goes through, the firm will be worth more than when the opportunity is passed. The firm s manager exactly knows how much the firm is worth after investing, and when the investment is passed by. However to investors this is unknown, they observe a value in the range of the low and high bound. This information asymmetry is taken as given. If the manager issues stock for a total value of the investment, the benefit will be the investment s positive NPV value. But there can also be a cost. If the firm s current stock is undervalued, the manager will need to issue more stocks to collect the financing need. Only the manager knows if the stock is undervalued, investors do not. Here Myers (1984) makes an assumption about the behavior of the manager. He regards it most reasonable that the manager serves the existing shareholders primarily. Therefore the manager is expected to maximize the intrinsic value of the existing shares in the firm. This could lead to passing by a positive NPV investment if the difference between the real stock value and the undervaluation is larger than the benefits of the investment. And since investors expect the manager to be on the side of the existing shareholders, they will adjust the price they are willing to pay downwards, increasing undervaluation. Thus, Myers and Majluf (1984) argue asymmetric information leads to a cost of relying on external capital, which is typically the indirect cost to a firm of passing up a positive NPV investment. This can be prevented by retaining enough internal funds to finance positive NPV projects and so in the pecking order, internal financing is the most preferable stream of funds. The other key point defines the pecking order between debt and equity in the case that the firm does seek external financing. Myers (1984) states the general rule: Issue safe securities before risky ones (p. 584). The safer security here is the one that changes least in value when the manager s exclusive information is revealed to the market. This is the security that has the smallest difference between the true value and the value observed by investors. Myers (1984) argues cases in which this is reasonably always debt. For example when the firm is able to issue default-free debt, the security s value will always be equal to the true value and hence debt always more favorable. Also if default risk is allowed, this is

12 12 mostly the case if the firm is undervalued. However the firm could also be overvalued by investors. If this is the case and inside information is unfavorable, managers could try to exploit investors by trying to increase overvaluation leading to the rule "Issue debt when investors undervalue the firm, and equity, or some other risky security, when they overvalue it" (Myers, 1984; p. 585). But Myers argues that since investors know managers will reason this way, they will not buy equity until a firm s debt capacity is fully used. This effectively forces the pecking order upon the firm. Myers (1984) stresses these are extreme situations that in practice are not very likely to occur, but they do outline how asymmetric information can lead to the pecking order theory. In short, the pecking order theory uses the asymmetric information principle to explain why external financing can lead to costs for the firm. Because of these information asymmetries, the firm has a risk of passing by profitable investments. Therefore internal financing should always be preferred. If this is unavailable, the rule for issuing external financing is to issue safe securities before risky ones. Securities are safer if their value changes less when exclusive information becomes know to the market. Considering this, debt is generally safer than equity. Therefore in the pecking order debt comes before equity, which will only be issued when debt capacities are fully used Challenge to the pecking order theory The existence and persistence of zero leverage firms, as shown by Strebulaev and Yang (2013), is a direct challenge to the rationale of the pecking order theory. While the theory suggests firms will use external equity financing only as a last resort, in practice a large group of firms prefers external equity financing over debt financing. These firms issue equity while they still have debt capacity available, and according to Strebulaev and Yang (2013) also pay high dividends and retain high cash balances. The pecking order theory offers no explanation as to why firms might prefer the least favorable means of financing while internal equity and debt capacity are available as well. 2.4 Market timing theory Myers (1984) outlined the concept of what he called timing of security issues (p. 586). Myers (1984) found empirical evidence in literature that one of the main drivers for firm equity issuance is past stock price movements. This would imply that firms favor the source of external financing that is cheapest at that point in time. This fact goes against the rationale of both the static trade-off and pecking order theory. Baker and Wurgler (2002) investigate the link between market timing and firm capital structures. Their conclusion is that firm s capital structures are a combination of the issuance of the most favorable sources of financing through time. Instead of a pecking order or an assessment of the costs and benefits of external finance, managers in need of external financing will choose the source that is most favorable at that moment. Although Modigliani and Miller (1958) show that in integrated and efficient capital markets the costs of different forms of capital are equal, Baker and Wurgler (2002) argue that markets are in fact inefficient and segmented. Therefore timing the market can benefit ongoing shareholders. If managers favor ongoing shareholders, they can attempt to time the market to exploit these inefficiencies. In their study, Baker and Wurgler (2002) moreover find evidence that low-leverage firms are those that issued equity in a favorable climate. On the contrary, high leverage firms issued their equity when valuations were low. This concept is interesting in the light of this thesis and will be taken into account in the further analysis.

13 Potential hypotheses for the low leverage and zero leverage anomalies Financial constraints hypothesis One possible explanation for the observed fact that an increasingly large fraction of firms is unlevered can be found in the supply side of debt. In the presence of market frictions, firms could be unable to obtain debt financing for positive net present value projects. For example Stiglitz and Weiss (1981) propose that information asymmetry amongst banks can lead to financial constraints for firms. They argue that banks have perfect information about their own clients, but not about clients of other banks. If a bank X with excess credit supply offers a loan to a profitable client of bank Y against better conditions, bank Y will offer the same conditions as bank X and retain the client. If the client is in fact unprofitable, bank Y refrains from offering and bank X acquires a bad client. If banks know this, they will stop offering to other banks clients since they will only acquire the bad clients. So banks are left with excess credit supply but do not acquire new clients or lower interest rates. This leaves less profitable firms financially constrained. Faulkender and Petersen (2006) also focus on the impact of market frictions in the supply side of credit. Taking debt ratings as a proxy for firms access to public bond markets, they find that firms with a debt rating have 35 percent more debt than firms without, even after controlling for firm characteristics and possible endogeneity of having a debt rating. This evidence suggests financial constraints from the supply side of credit could be a main driver for zero leverage firm behavior. The conclusions of Faulkender and Petersen (2006) are backed by Lemmon and Zender (2010). By integrating debt capacity in tests of theories of capital structure, they find that in the absence of debt capacity restrictions leverage is preferred as external financing. When debt becomes restricted, for example for small, high-growth firms, external equity is frequently issued. This reconciles the pecking order theory with equity issuance by financially constrained firms. Frank and Goyal (2009) indicate that dividend-paying status, size and growth opportunities are proxies for financial constraints. Smaller firms with high growth opportunities, who do not pay dividends, are typically financially constrained. Hadlock and Pierce (2010) find that financially constrained firms are typically smaller and younger Underinvestment and Financial flexibility hypotheses Next to being forced to remain unlevered by restriction from the supply side of debt, firms could be ZL from strategic viewpoints. The underinvestment and financial flexibility hypotheses are seen as the main drivers for firms propensity to remain unlevered. The underinvestment hypothesis has roots in Myers 1977 paper on Determinants of Corporate Borrowing. Myers argues that the maturity of outstanding debt influences corporate investment. If a firm has risky (long maturity) debt outstanding, it could refrain from investing in profitable new projects if enough of the benefits of the projects would flow to the debt holders. This is called the debt overhang problem. Following this reasoning, firms should issue short maturity debt to reduce the underinvestment problem (Johnson, 2003) or eschew debt ex ante profitable investment opportunities. The latter could be an explanation for firms ZL behavior. The financial flexibility hypothesis stems from Myers and Majluf s (1984) adverse selection problem. Firms will favor internal or debt financing to fund new projects, and will therefore hoard cash balances and save debt capacity to mitigate financing distortions. Another argument for this behavior is the presence of adjustment costs of capital structures shown by Leary and Roberts (2005). If these costs are large enough, firms that use debt capacity for investments will choose conservative debt levels prior to investment.

14 14 Especially growth firms that rely heavily on their capacity to finance risky investments could strategically refrain from using their debt capacity to remain financially flexible, and thus be ZL firms. 2.6 Current state of literature on ZL firms The notion of Korteweg (2010) that the average debt conservatism of firms is mainly attributable to ZL firms has not gone unnoticed in financial research. In 2012 and 2013, at least four credible papers have been written on this subject. These papers will be briefly reviewed to present a clear picture of the current state of literature on ZL firms Devos, Dhillon, Jagannathan, Krishnamurthy (2012) Aside from the capital structure theories reviewed above, another branch of research focuses on management characteristics and corporate governance to explain capital structure decision making. Following this research, Devos et al. (2012) examine whether ZL firms are managed by entrenched managers that shun leverage. They also investigate financial constraints as a possible explanation. Finally they explore the role of these factors in firms decisions to abandon their ZL policies. The sample comprises of all US non-financial Compustat and CRSP firms between 1990 and Devos et al. (2012) find no evidence that managerial entrenchment plays a role in the decision of firms to remain unlevered. The summary statistics reported on board size, insider ownership, outside director percentages, takeover activities, and 5% blockholder percentages show no major deviations from the control group. Furthermore in the reported multivariate logistic regression analysis with a ZL dummy as dependent variable (1 for ZL firms and 0 otherwise), almost all included managerial entrenchment variables are statistically insignificant. However the variables used as a proxy for financially constrained firms are statistically significant. These suggest ZL firms are younger, have less collateral to finance debt and are more restricted in investment activities. Next Devos et al. (2012) look at firm characteristics around debt initiation. The results are robust with their previous findings. Governance characteristics do not change around debt initiation. However in the debt initiating year, firms increase investments by over 4 percent of assets, cash ratios and Tobin s Q decline and net fixed asset ratios increase. This is all consistent with the financial constraints hypothesis and shows that firms for which constraints are lifted converge to the average levered firm s characteristics Strebulaev and Yang (2013) The research of Strebulaev and Yang (2013) is contemporaneous to that of Devos et al. (2012) and interestingly, both studies investigate governance characteristics. Furthermore Strebulaev and Yang (2013) consider the potential tax benefits of ZL firms extending Graham s (2000) research. Finally they consider the relationship between ZL behavior and industry and size, and the persistence of ZL behavior. The sample used is a Compustat and CRSP dataset of US listed non-financial firms in the period 1962 to The data is largely filtered using the same criteria as Devos et al. (2012), although encompassing a much larger period of time. Strebulaev and Yang (2013) find that ZL firms on average leave 7.6 percent of their market value on the table compared to the proxy group of firms they defined. Separating on dividend-paying status this percentage is over 15 for dividend-paying ZL firms and less than 3 for non-dividend firms, indicating a difference in motives between these groups of firms. They also find that industry and size cannot fully explain ZL behavior and ZL dividend-paying firms are more profitable, and pay higher dividends and taxes.

15 15 Like Devos et al. (2012) a multivariate logistic regression on the ZL dummy is reported. The independent variables consist of several CEO and board characteristics, as well as a large set of firm characteristics variables as controls. The outcomes on the governance variables are almost all highly statistically significant, in contrast to Devos et al. (2012). However there are a number of differences between the estimations. Strebulaev and Yang (2013) define ZL as all firms with market leverage ratio below 5 percent instead of zero, lag all independent variables one year and use year fixed effects. They conclude that firms with large CEO-ownership and CEO-friendly boards, as well as family firms, are more likely to be ZL. A conclusion that is drastically different from Devos et al. (2012). But as indicated, the results are established using different econometrics. Also a large set of control variables is included that could potentially spur endogeneity Bessler, Drobetz, Haller, Meier (2013) This study presents some new insights in the ZL phenomenon, mainly along three dimensions. First, Bessler et al. (2013) try to explain the upward trend in the percentage of ZL firms. Second, instead of investigating only US firms, they present international evidence. Third, both the demand and supply side for leverage of ZL firms are investigated. The database used is a 1988 to 2011 Compustat North America and Compustat Global database that includes publicly traded firms from 20 developed countries. Bessler et al. (2013) find that the upward trend of ZL firms is partly due to IPO waves in the sample period. Many small, young firms that go public have little to no debt in their capital structures. Moreover, they report an industry effect in the data due to changes in industry composition. A shift towards more debt conservative industries is observed. Unlike other studies so far, Bessler et al. (2013) show that the ZL phenomenon is international. They report a higher propensity for firms to remain ZL in countries with a common law system. The credit supply-side analysis of ZL firms reveals most firms are credit restricted and therefore unlevered by financial constraints. These firms are smaller, younger, less profitable and riskier. They are also the most active equity issuers. Bessler et al. (2013) also report a small group of financially unconstrained firms that pay high dividends and are more profitable. This is robust to the other studies we have seen so far. The econometrics used in the study is by construction the same as those of Devos et al. (2012) and Strebulaev and Yang (2013). A multivariate logistic model is used, with the ZL dummy as dependent variable. Again a large set of firm performance characteristics is used as control variables. Almost all the coefficients are highly statistically significant Dang (2013) Two potential explanations for the ZL behavior of firms are the main focus in this paper. First, Dang (2013) investigates whether firms are unlevered due to financial constraints, or for strategic reasons. Second, he attempts to reveal if firms ZL policies are affected by macroeconomic variables rather than firm-level specifics. Dang s (2013) sample is different than the other studies, which mainly retrieve data from Compustat databases. This study focuses on data of U.K. firms in the period collected from Worldscope. Like the other three studies discussed, Dang (2013) reports multivariate logistic regressions with the ZL dummy as dependent variable. He studies largely the same variables and comes to similar conclusions. ZL firms consist of dividend-paying and non-dividend firms that have different motives for their behavior. The first group contains largely small and young firms with poor performance. This group is ZL mainly because of a lack of credit available to them. The group of dividend-payers is different. They are more

16 16 mature and profitable and face weaker credit restrictions. Hence they are suspected to be ZL for strategic reasons, as predicted by the underinvestment and financial flexibility hypotheses Macroeconomic conditions Although the conclusions and methods of Dang (2013) so far are in line with what was already concluded by other papers, the inclusion of macroeconomic conditions is new. To investigate the effects of the macro environment, Dang (2013) includes real GDP growth and term structure as variables in his regressions. The coefficients of these variables are almost all significant, indicating these conditions have effects on firms propensity to eschew debt. Dang (2013) concludes that adverse macroeconomic conditions lead to more firms becoming ZL, which is consistent with Bernanke and Gertler s (1989) balance sheet channel theory that leverage is pro-cyclical. While Dang shows that macroeconomic conditions have effects on the propensity of firms to be unlevered, his method is unable to indicate whether this is for financial flexibility or financial constraints reasons. He conjectures that if this is for flexibility reasons, firms seeking ZL policies should be able to maintain investment and outperform in adverse macroeconomic times. He therefore recommends studying ZL firm performance in periods of financial crisis Implications of current literature to this study The conclusions of the four studies vary. Across the literature there is evidence that a fraction of ZL firms is in fact financially constraint, while another part is seeking financial flexibility. The division can be done on the basis of dividend-paying status. Overall ZL firms are young, small, more risky and less profitable. The role of governance characteristics leads to contradictive conclusions in the papers of Devos et al. (2012) and Strebulaev and Yang (2013). The ZL phenomenon is furthermore shown to be international. Macroeconomic conditions influence firms propensity to be unlevered, although it is unclear to what magnitude these conditions affect performance. So far, all studies performed on ZL firm behavior approach the subject from roughly the same angle. All try to explain why firms are unlevered against dominant capital structure theories. Multivariate logistic regressions are used, with the ZL dummy as dependent variable. Different variables are included to investigate the roles of different aspects of economic theory (e.g. governance variables in Devos et al. (2012) and Strebulaev & Yang, 2013 or macroeconomic variables in Dang, 2013). Firm characteristics are used to try explaining why firms are unlevered, but the performance of ZL firms has not yet been deeply assessed. As a result, the drivers for ZL behavior and characteristics of ZL firms have become more visible, but the effect of being a ZL firm on performance is unclear. Moreover the possible role of endogeneity in the models is largely ignored by researchers. By approaching the ZL phenomenon from a different angle and with a different regression model, I hope to present new insights concerning the reasons for firms to pursue ZL policies. This will be discussed further in the Hypothesis section.

17 17 3. Hypothesis In the literature review section, the shortcomings of the static tradeoff and pecking order theory in explaining the ZL phenomenon were outlined. Furthermore, hypotheses that could support ZL behavior were outlined and an image was given on the current state of literature on the ZL phenomenon. Overall it is clear that this literature still has gaps that should be further explored. I intend to add to existing literature along two dimensions. Existing studies investigate the factors that lead firms to become ZL. However this could also be a decision rather than a consequence. By comparing firm performance of ZL firms to similar firms through summary statistics and regression analysis, more should become clear about the effectiveness of the capital structures of these firms. Interesting in this dimension is the group of dividend paying firms that strategically shun leverage. Considering these firms are not restricted in their choice of capital structure and if the static tradeoff theory holds, being ZL is perceived the optimal capital structure by these firms. Hence these firms are expected to have a strategic reason for their ZL policy. This should result in superior performance of dividend-paying ZL firms against similar levered firms. Following the same line of reasoning, ZL non-dividend firms are expected to be financially restricted and therefore unable to move to the desired capital structure according to the static tradeoff as well as the pecking order theory. Since credit restrictions limit their financing abilities, these firms are expected to underperform in comparison to similar firms that do have access to debt. The second dimension stems from a combination of empirical evidence by Dang (2013), and the financial flexibility and financial constraints hypotheses. Dang found that macroeconomic variables (real GDP growth and term structure) influence firms propensity to become unlevered. However his research made no implications on the performance of ZL firms in diverse macroeconomic situations. In his conclusions Dang (2013) makes the following recommendation to future research: Firms that eschew debt and accumulate cash to preserve financial flexibility should be well positioned to meet financing needs and maintain investment in the future, even under adverse economic conditions characterized by a lack of credit. Hence, it would be equally interesting to examine how zero-leverage firms fare in periods of financial crisis. (p. 200) Dang (2013) conjectures firms that are ZL for strategic reasons will outperform in financial crises. Moreover according to Gertler and Gilchrist (1993), the impact of business cycle variables varies with the degree of constraints faced by firms. Taking size as a proxy for constraints, Gertler and Gilchrist (1993) show that smaller (constrained) firms face credit rationing in recessions with a monetary contraction. Subsequently the performance of these firms is expected to decline compared to unconstrained firms. Hence both dimensions lead to the same expectation of ZL firm performance in times of recession and growth. Considering dividend-paying status as a good proxy for the degree of credit restriction of firms, the following hypotheses can be formulated: H1: The decision of dividend-paying firms to become zero-leverage leads to higher performance in recessions due to financial flexibility benefits and the absence of the negative effects of debt. H2: Non-dividend zero-leverage firms are financially constrained and therefore underperform to levered firms in recessions.

18 18 4. Data Following Devos et al. (2012) and Strebulaev and Yang (2013), the data comes from the merged annual CRSP (Center for Research in Security Prices) and Compustat database. In this database, information on all U.S. listed firms is contained. The sample period is , since data before 1962 in the Compustat database are biased towards large firms and 2012 is the most recent complete year in the database. I choose the largest possible sample period so that multiple recessions fall into its scope. This will result into more robust outcomes. In the same manner as Strebulaev and Yang (2013) a number of filters are applied to the data set in order to attain to the desired sample. Financial Services firms and Utilities (SIC codes and respectively) are excluded, as well as non U.S. based companies and nonpublicly traded firms and subsidiaries. In line with Strebulaev and Yang (2013), I exclude firms with inflation-adjusted book assets below $10 million. This is done by adjusting the total book value of assets (AT in the CRSP-Compustat database) for inflation. Using the Consumer Price Index (CPI) figures from the U.S. Bureau of Labor Statistics, all firm-year observations are adjusted and observations with adjusted total book assets below $10 million are excluded. The rationale for excluding firms below $10 million in book assets is the presence of noise in the accounting data for small firms reported by Strebulaev and Yang (2013, p. 3). The main variable of interest in this paper will be based on the book leverage ratio of firms. This ratio is constructed in the same manner as in Strebulaev and Yang (2013), Bessler et al. (2013), and capital structure papers such as Lemmon, Roberts, Zender (2008): Book leverage it = DLTT it + DLC it AT it DLTT and DLC are the Compustat items for total long term debt and short term debt respectively, while AT is the total book value of assets. The ratio is calculated for firm i in year t. Accordingly, the market leverage ratio is also constructed: DLTT it + DLC it Market leverage it = DLTT it + DLC it + CSHO it PRCC_F it Here CSHO is the number of common share outstanding, which is multiplied by the annual fiscal year-end stock price (PRCC_F) of firm i in period t. All observations in the data set with a missing book leverage or market leverage ratio are excluded, since these ratios form the cornerstone of this research. With these restrictions applied, there are 166,545 firm-year observations across 14,730 firms in the data set. 4.1 Recession periods Recessions in the U.S.A. are identified by the National Bureau of Economic Research (NBER). The bureau divides years into business cycles of contraction and expansion. The NBER report the following statement on their website concerning the identification of a recession period:

19 19 The NBER does not define a recession in terms of two consecutive quarters of decline in real GDP. Rather, a recession is a significant decline in economic activity spread across the economy, lasting more than a few months, normally visible in real GDP, real income, employment, industrial production, and wholesale-retail sales. (nber.org/cycles) The Bureau indicates the period of a business cycle from peak to trough. All the months in between are months of contraction (i.e. recession), while the months from a trough to the next peak are months of expansion. The interest is in the contraction months in the period (see Appendix A). The NBER identified seven troughs in this period. However this is monthly data, while the dataset used contains annual information. To adjust this, a recession year is identified when in a year 6 or more months are reported in a trough period by the NBER. This way nine recession years can be identified: 1970, 1974, 1980, 1981, 1982, 1990, 2001, 2008 and All other years in the data are years of expansion. These years are considered normal business years in the rest of this thesis and will be indicated as growth years. 4.2 Definition of zero leverage Strictly, a firm i is zero-leverage in year t if both long term debt and short term debt equal zero. Out of 166,545 observations in the data set, 18,562 (11.1%) observations have zero book leverage. Strebulaev and Yang (2013) add to their research a category of almost zero-leverage (AZL) firms. This includes all firms with a book leverage ratio of less than The rationale behind this mainly lies in ambiguity in accounting conventions. This could lead to the exclusion of firms that are in reality ZL. Arguably, for the purpose of this paper being zero-levered or having a book leverage ratio below 5% makes little difference. Fact is that a leverage ratio smaller than 5% is still very far from the optimal ratio according to capital structure theory. Also it seems unlikely that these firms use all their debt capacity, which retains the financial flexibility argument. In preliminary analysis of the results of this thesis, both ZL and AZL groups were tested. However a comparison of the outcomes learned that the qualitative results of the analyses were not significantly different. Therefore, for the sake of clarity only the ZL analyses are reported in this thesis. 4.3 Definition of performance variables Above, the variables for book leverage, market leverage and recession have been introduced. The influence of these variables and their interactions on firm performance is the main interest of this thesis. Below, all variables included in either summary statistics or regression analysis, or both, are shortly introduced. As the identifier for the ZL dummy that is the variable of interest in the regressions, book leverage is chosen. Hence the dummy will be 1 if a firm has zero book leverage, and zero otherwise. This is done since book leverage is expected to be less affected by recessions, the variable with which the ZL dummy will be interacted. Market leverage is the ratio of total debt and total debt plus the market value of equity. Thus it will rise in recessions, since recessions are marked by a sharp decline in the market values of equity. The book leverage ratio on the other hand has book assets in the denominator, which will typically adjust much slower to macroeconomic circumstances and is therefore more reliable when analyzing recession periods.