The International Zero-Leverage Phenomenon

Transcription

1 The International Zero-Leverage Phenomenon Wolfgang Bessler a, Wolfgang Drobetz b, Rebekka Haller c, and Iwan Meier d * This draft: August 2012 Abstract Extreme debt conservatism is an international phenomenon which has increased over time. While only 5% of the firms in our international sample maintained a zero-leverage policy in 1989, this fraction rose to roughly 15% by The increasing prevalence of zero-leverage firms is related to the different IPO waves and the accompanying changes in industry composition. In addition, we attribute the higher propensity to maintain a zero-leverage policy throughout all size and age groups to the increasing asset volatility during our sample period. Country-specific factors also affect extreme debt conservatism. Countries with a market-oriented financial system, higher creditor protection, and a classical tax system exhibit the highest percentage of zero-leverage firms. Analyzing the supplyside of firms financing choices, we show that only a small number of profitable firms with high payout ratios deliberately maintain zero-leverage. In contrast, most zero-leverage firms are constrained by their debt capacity; they tend to be smaller, riskier, and less profitable, and they are the most active equity issuers. With respect to the demand-side of financing choices, we document that firms which follow a zero-leverage policy only for a short period of time seek financial flexibility. After abandoning zero-leverage, these mostly unconstrained firms jump to higher leverage ratios, make higher investments, and reduce their cash holdings by a larger amount compared to constrained zero-leverage firms, which remain debt-free for longer periods of time. Keywords: Capital structure, debt conservatism, financial constraints, financial flexibility JEL classification codes: G32 a Wolfgang Bessler, Center for Banking and Finance, Justus-Liebig-University Giessen, Licher Straße 74, Giessen, Germany. Mail: wolfgang.bessler@wirtschaft.uni-giessen.de. b Wolfgang Drobetz, Institute of Finance, University of Hamburg, Von-Melle-Park 5, Hamburg, Germany. Tel.: ; fax: ; mail: wolfgang.drobetz@wiso.uni-hamburg.de. Corresponding author. c Rebekka Haller, Institute of Finance, University of Hamburg, Von-Melle-Park 5, Hamburg, Germany. Mail: rebekka.haller@wiso.uni-hamburg.de. d Iwan Meier, HEC Montréal, 3000 Chemin de la Côte-Sainte-Catherine, Montréal (Québec), Canada, H3T 2A7. Mail: iwan.meier@hec.ca. * We would like to thank Yakov Amihud, András Danis, Dave Denis, Espen Eckbo, Miguel Ferreira, Dimitrios Gounopolos, Ulrich Hofbaur, Philipp Immenkötter, Peter Limbach, and participants at a research seminar at the University of Hannover, the 2011 Financial Management Association (FMA) Europe Conference in Porto, the 2011 Swiss Finance Association Conference in Zurich, the 2011 Symposium for Banking, Finance, and Insurance in Karlsruhe, the 2012 WHU Campus for Finance Conference, and the 2012 European Financial Management Association (EFMA) Conference in Barcelona for helpful comments.

2 The International Zero-Leverage Phenomenon This draft: August 2012 Abstract Extreme debt conservatism is an international phenomenon which has increased over time. While only 5% of the firms in our international sample maintained a zero-leverage policy in 1989, this fraction rose to roughly 15% by The increasing prevalence of zero-leverage firms is related to the different IPO waves and the accompanying changes in industry composition. In addition, we attribute the higher propensity to maintain a zero-leverage policy throughout all size and age groups to the increasing asset volatility during the sample period. Country-specific factors also affect extreme debt conservatism. Countries with a market-oriented financial system, higher creditor protection, and a classical tax system exhibit the highest percentage of zero-leverage firms. Analyzing the supplyside of firms financing choices, we show that only a small number of profitable firms with high payout ratios deliberately maintain zero-leverage. In contrast, most zero-leverage firms are constrained by their debt capacity; they tend to be smaller, riskier, and less profitable, and they are the most active equity issuers. With respect to the demand-side of financing choices, we document that firms which follow a zero-leverage policy only for a short period of time are seeking financial flexibility. After abandoning zero-leverage, these mostly unconstrained firms jump to higher leverage ratios, make higher investments, and reduce their cash holdings by a larger amount compared to constrained zero-leverage firms, which remain debt-free for longer periods of time. Keywords: Capital structure, debt conservatism, financial constraints, financial flexibility JEL classification codes: G32

3 2 1. Introduction Major listed firms in the Standard & Poor s 500 index, such as Apple, Yahoo, Texas Instruments, Bed Bath & Beyond or Urban Outfitters, all have something in common: they are debtfree. These firms are examples for the stylized fact in corporate finance that the proportion of zero-leverage firms has increased over time. The zero-leverage puzzle is not only a phenomenon in the United States, but it is also observable on other major financial markets. While only 5.14% of all firms in our international sample renounced the use of debt in 1989, the percentage of zero-leverage firms rose to 14.97% by In the presence of market frictions, there are two main theories of capital structure: the trade-off theory (Kraus and Litzenberger, 1973) and the pecking order theory (Myers and Majluf, 1984). Both theories advocate the use of debt because of either tax benefits or lower costs of asymmetric information. The empirical literature is undecided which theory better describes firms financing decisions (Frank and Goyal, 2008). 1 Perhaps most troubling, neither the static trade-off theory nor the pecking order theory is able to explain why so many firms in our international sample follow a zero-leverage policy. This study contributes to our understanding of a zero-leverage policy in three ways. First, we offer explanations for the upward trend in the fraction of zero-leverage firms. On the one hand, the increasing prevalence of zero-leverage is related to the different IPO waves in our sample countries. A related explanation is that there are industry effects due to changes in industry composition toward sectors where extreme debt conservatism is more commonly adopted. On the other hand, the zero-leverage phenomenon is driven by a vintage effect, where newly listed firms in the later years of the sample period are more likely to a pursue a zero-leverage policy. We attribute this increasing propensity to maintain a zero-leverage policy throughout all size and age groups to firms increasing asset volatility (or business risk) over the sample period. Second, the zero-leverage puzzle extends to major financial markets outside the United States, although there are differences across countries. Our country-level analysis reveals that countries with capital-market-oriented financial systems, common law origins, high creditor protection, and classical tax systems exhibit the highest percentage of zero-leverage firms. A third contribution of our study is that we look at both demand- and supply-side explanations for zero-leverage. Most zero-leverage firms in our sample are financially constrained and have 1 Most empirical studies either focus on identifying factors that affect firms capital structures (Rajan and Zingales, 1995; Frank and Goyal, 2009; Gungoraydinoglu and Öztekin, 2011; Fan et al., 2012) or directly test theories of capital structure (Shyam-Sunder and Myers, 1999; Frank and Goyal, 2003; Bessler et al., 2011).

4 3 no other option than renouncing the use of debt (supply-side effect). These firms are smaller, younger, riskier, and less profitable; they are the most active equity issuers and hoard the largest cash holdings of all sample firms. Constrained firms do not possess sufficient debt capacity and thus (have to) maintain a zero-leverage policy for longer periods of time, on average. In contrast, there is a small subsample of firms that deliberately choose to adopt a zero-leverage policy. These financially unconstrained firms are more profitable, distribute higher dividends, and are older as well as larger than their constrained zero-leverage peers. When we restrict our international sample to established firms with a minimum of 15 years of historical data in order to analyze the time-series variation in leverage, we show evidence for the importance of flexibility in capital structure decisions (Graham and Harvey, 2001). Firms that maintain a zero-leverage policy only for a short period of time preserve their debt capacity for use in the near future (demand-side effect). 2 After abandoning zero-leverage, these unconstrained zero-leverage firms jump to higher leverage ratios, make higher investments, and reduce cash holdings by a larger amount compared to their constrained zero-leverage peers. Unconstrained zero-leverage firms have the flexibility to issue larger amounts of debt and use it together with their cash holdings to invest in future periods when good opportunities arise. In contrast, constrained firms lack sufficient debt capacity and are unable to issue as much debt; they follow a zero-leverage policy for longer periods of time, are more dependent on internal funds and thus are less flexible in their investments. The zero-leverage puzzle is neither compatible with the standard capital structure theories nor with more recent variations of them. While not all capital structure theories predict an optimal leverage ratio, none is able to explain why firms abstain from using debt. For example, based on the static trade-off theory, Leland (1994) forecasts an average debt ratio of approximately 60%. More recent simulation studies based on dynamic trade-off theories use contingent claim analysis and derive minimum leverage ratios as low as 10% (Morellec, 2003; Ju et al., 2005). Goldstein et al. (2001) and Hennessy and Whited (2005) also assume a dynamic framework. These models imply that firms, in the presence of adjustment costs or adverse selection costs, become debt-free in order to avoid either forgoing future investment or financing them with new risky securities. The model of DeAngelo et al. (2011) predicts that ex ante optimal finan- 2 Debt capacity is usually defined as a sufficiently high debt ratio so that the costs of financial distress curtail further debt issues (Shyam-Sunder and Myers 1999; Chirinko and Singha 2000).

5 4 cial policies preserve the ability to access the capital markets ex post in the event of shocks to investment opportunities or after unexpected earnings shortfalls. The option to issue debt later at comparable terms is valuable, and the opportunity cost of raising debt implies that target capital structures tend to be more conservative than those produced by standard trade-off models. DeAngelo and Roll (2012) analyze leverage stability and document that it arises only infrequently and then largely when firms have low leverage; nevertheless, very low leverage is almost always temporary. Leverage peaks and troughs are not exogenous shocks that induce rebalancing to some target leverage ratio. Instead, managers chose to pay down debt when attaining a leverage trough and then issue debt to fund expansion. In contrast to variants of the trade-off theory, the pecking order theory does not imply a welldefined target leverage. Myers (1984) argues that a firm s capital structure reflects the accumulation of its past financial requirements. When information asymmetry is temporarily low, firms with sufficient internal funds have less incentive to use external financing (Autore and Kovacs, 2009; Bessler et al., 2011). However, even a dynamic pecking order theory is unable to explain why firms with little or no debt tend to rely heavily on equity and are not willing to exhaust all internal funds (including large cash balances) prior to obtaining external financing. Our study follows along a growing literature on debt conservatism and zero-leverage. Graham (2000) develops interest-benefit deduction functions by estimating marginal tax rates and reports that large and profitable firms with high cash holding and low expected distress costs could significantly increase their value if they used the optimal amount of debt. Minton and Wruck (2001) analyze the low leverage phenomenon and show that financial conservatism is not an industry-specific phenomenon, even though conservative firms are more prone to industries related to high financial distress. DeAngelo and Roll (2012) document that the incidence of low leverage firms in recent years is due to a surge in listings by young growth firms that have little or no debt. A few recent studies explicitly analyze the prevalence of zero-leverage firms. One strand of this literature investigates firm-level fundamentals. Byoun et al. (2011), Devos et al. (2012), Strebulaev and Yang (2012), and Dang (2011) report that zero-leverage firms tend to be smaller, accumulate higher cash reserves, and exhibit higher market-to-book ratios as well as higher payout ratios. These findings are closely related to studies on cash holding that document a negative relationship between leverage and corporate liquidity (Opler et al., 1999; Drobetz et al., 2010). However, taken together, it is difficult to reconcile the ob-

6 5 served characteristics of zero-leverage firms with standard capital structure theories. Another strand of this literature explores the quality of firm-level corporate governance mechanisms and managerial entrenchment as a potential explanation for zero-leverage. Using data from the United States, Strebulaev and Yang (2012) suggest that managerial and governance characteristics are important determinants for the decision to maintain a zero-leverage policy. In contrast, Devos et al. (2012) and Byoun et al. (2011) report that zero-leverage firms do not suffer from weaker internal or external governance mechanisms. As an example, Devos et al. (2012) document that the debt initiation decisions of zero-leverage firms are not enhanced by shocks to their entrenchment, such as takeover threats or activist block holders. While poor firm-level corporate governance does not explain the zero-leverage puzzle, their results indicate that financial constraints contribute to extreme debt conservatism. In another approach, Lambrecht and Pawlina (2012) attribute the zero-leverage puzzle to human capital intensive industries. They develop a theoretical model in which the asymmetry between physical capital and transferable human capital induces negative net debt. 3 Only a few studies analyze low-leverage policies as a part of the intertemporal investment and financing decisions. Minton and Wruck (2001) show that firms with conservative debt policies increase leverage when they are faced with lower internal funds and higher investments. More recently, Marchica and Mura (2010) analyze the dynamics of low-leverage policies by using a sample from the United Kingdom. Based on DeAngelo s et al. (2011) arguments, they document that firms incur higher capital expenditures and higher abnormal investments following a period of low leverage, and that these investments are likely to be financed through the issuance of debt. Most important, there is a measurable effect from financial flexibility in the form of untapped borrowing reserves. Similarly, De Jong et al. (2012) document that US firms with high unused debt capacities and thus more financial flexibility make higher future investments than firms with less financial flexibility. Furthermore, more flexible firms are in a position to reduce investment distortions in constrained times because they have better access to the debt markets in these times. De Jong et al. (2012) conclude that firms save debt capacity for more constrained periods in the future. 3 Instead of assessing explanations for the prevalence of zero-leverage, Lee and Moon (2011) test the long-run equity performance of zero-leverage firms. Based on calendar-time portfolio regressions, they show that zeroleverage firms tend to outperform. Contradicting the results in Strebulaev and Yang (2012), they suggest that zero-leverage firms are more conservative, self-disciplined, and prudent in making decisions.

7 6 While most of the related studies focus on debt conservatism in general, we explicitly analyze debt-free firms and provide several explanations for the increasing prevalence of zero-leverage, such as IPO and industry effects as well as increasing asset risk. Moreover, we are the first to show that zero-leverage is an international phenomenon, albeit country-specific factors exert strong influences. Our study further contributes to the literature by analyzing both supplyand demand-side explanations for the zero-leverage puzzle. While there have been attempts to differentiate between constrained and unconstrained zero-leverage firms (Byoun et al., 2011; Devos et al., 2012), our framework is novel in that it entails a detailed comparison between the two groups of zero-leverage firms based on alternative measures of financial constraints. Finally, we examine the investment and cash holding decisions of previously unlevered firms. Most important, the financing decisions of firms that maintain zero-leverage only for a short period of time are attributable to their preference for financial flexibility; these firms strategically use their debt capacity and start to issue debt when good investment opportunities arise. The remainder of our study is structured as follows: Section 2 provides descriptive statistics and presents stylized facts about the zero-leverage puzzle across countries. Section 3 analyzes firm- and country-level determinates for the decision to maintain zero-leverage. Section 4 and Section 5 provide an analysis of the capital supply-side and demand-side explanations for zero-leverage, respectively. Finally, Section 6 concludes and provides an outlook for further research. 2. Data Description and Stylized Facts 2.1 Definition of Variables and Descriptive Statistics In order to analyze firms that maintain a zero-leverage policy, we collect annual balance sheet and market data of listed firms from the United States, Canada, the United Kingdom, Germany, and Japan from the Compustat Global database over the time period from 1989 to Our sample consists of active and inactive publicly traded industrial firms and avoids a survivorship bias. Given the specific nature of their businesses, financial and utility firms (with SIC and ) are omitted from the sample. Firms that do not have an industrial sector- or a country code in the database as well as firms with non-consolidated balance sheets are excluded. Within this basic specification, our sample consists of 13,897 consolidated firms (8,692 active and 5,205 inactive) with 176,907 firm-year observations for the whole sample.

8 7 Tables A1 and A2 provide an overview of all variables used in our empirical analyses together with a detailed description of their construction principles. We exclude firm-year observations with missing information on total assets and market value. Furthermore, following Frank and Goyal (2003), we recode deferred taxes, purchase of treasury shares, and preferred stock to zero if firm-year observations are missing. All constructed variables are winsorized at the 1% and the 99% tails in order to reduce outlier observations. After these additional data cleaning steps, our final panel includes 13,309 industrial firms with a total of 151,638 firm-year observations. Furthermore, we follow Devos et al. (2012) and Strebulaev and Yang (2012) and define book leverage as the ratio of the sum of short- and long-term liabilities to total assets. Zero-leverage observations are firm-year observations without leverage. The number of firms included in our sample differs across countries. For example, we observe a strong increase in the number of German firms over the time. The Compustat Global database includes only 100 German firms in the year 1989, whereas by the year 2010 the number of firms increases to 461. The major reason for this sharp increase in the number of listed firms and coverage is the strong European IPO activity during our sample period (Giudici and Roosenboom, 2004; Bessler and Seim, 2012; Vismara et al., 2012). The number of Japanese firms also increases strongly, from 944 firms in 1989 to 2,535 firms in In contrast, the number of sample firms in the Anglo-Saxon countries, i.e., the United States, Canada, and the United Kingdom, rises at a lower rate, from 2,669 in 1989 to 3,450 in Table I provides descriptive statistics for all variables. Most important, the median book leverage ratio (the ratio of interest bearing debt to total assets) is 19.1% for the full sample. Japan boasts the highest median book leverage ratio (23.2%), followed by Canada (19.7%) and the United States (18.6%). Germany (15.8%) and the United Kingdom (14.3%) exhibit the lowest book leverage ratios. These differences in cross-country leverage are similar to those reported by Rajan and Zingales (1995), Gungoraydinoglu and Öztekin (2011), and Fan et al. (2012). [Insert Table I here] 2.2 Stylized Facts about the International Zero-Leverage Phenomenon As shown in Figure 1 and Table II, the large number of zero-leverage firms and the sharp increase in the percentage of firms without debt is an international phenomenon. On average, 12.25% of all firm-year observations are classified as zero-leverage during our sample period.

9 8 Nevertheless, cross-country analyses reveal substantial differences. While about 14% of all firm-year observations in the Anglo-Saxon countries are zero-leverage, on average, this fraction is 8.62% and 5.63% in Germany and Japan, respectively. Most important, Figure 1 and Table II document a strong increase in the percentage of zero-leverage firms over time. Using all observations in the sample, only 5.14% of the firms are classified as zero-leverage in 1989, but this value increases to 14.97% by The increase is most pronounced in the Anglo- Saxon countries. For example, the percentage of zero-leverage firms in the United States and the United Kingdom increases to above 20% by the end of our sample period. These crosscountry differences are explored in Section 3.3 in more detail. [Insert Table II and Figure 1 here] In order to examine whether size is an important determinant for the decision to follow a zeroleverage policy, we sort all firm-year observations into size quartiles. Figure 2 depicts the percentage of zero-leverage firms in each size group over time. As expected, small firms are most likely to renounce the use of debt. At the end of the sample period, more than 27% of all firms in the smallest size quartile (Q1), which Fama and French (2008) refer to as micro caps, pursue a zero-leverage policy. This finding is consistent with capital structure theories that incorporate motives related to agency costs and/or asymmetric information (see Section 3.2). Nevertheless, the zero-leverage phenomenon is not confined to the smallest firms, and there is a pronounced upward trend in other size quartiles as well. While the fraction of zero-leverage firms in the largest size quartile (Q4) increases slightly but never rises above 5%, it goes up to more than 10% (Q3) and 15% (Q2) in the two middle size quartiles. [Insert Figure 2 here] DeAngelo and Roll (2012) argue that the increased incidence of low leverage firms in recent years is due to a surge in listing by young growth firms that have little or no debt. In fact, our sample period is characterized by varying activity on national IPO markets, and a main reason for the sharp increase in the number of listed firms and coverage is the strong IPO activity. Fama and French (2001) suggest that the change in the characteristics of newly listed firms is attributable to a decline in the cost of equity, which allowed firms with remote cash flow expectations to raise public equity. Therefore, we analyze if a new listing vintage effect is able to explain the large increase in the number of firms which tap the public equity market and re-

10 9 nounce the use of debt. 4 Four listing groups are defined according to a firm s IPO date. The first group comprises all firms listed before 1989; the second group includes all firms listed between 1989 and 1993; the third group between 1994 and 2003; and the fourth group between 2004 and These four groups roughly represent the aggregate IPO cycles during our sample period. Figure 3 depicts the evolution of the percentage of zero-leverage firms in the four different listing groups for the full sample and the individual countries. In fact, the zero-leverage phenomenon is more pronounced in the more recent vintage periods. While the fraction of zero-leverage firms in the pre-1989 listing group exhibits little variation over time, each vintage group starts with a higher percentage of zero-leverage firms. Moreover, the fractions of zero-leverage firms in the more recent listing groups are strongly increasing over time. Although the percentage of zero-leverage firms decreases in the most recent years in all vintage groups, the ordering of the different vintage groups remains unchanged. 5 An exception to this is the United States, where the fraction of zero-leverage firms in the last vintage group is lower than that in the preceding one. A potential explanation is the sharp decline in IPO activity in the United States following the technology bubble burst in 2000 (Ritter et al., 2012). An alternative explanation could be an increasing debt capacity during the up-markets just before the outbreak of the financial crisis in In contrast, Canada and Germany experienced a second IPO wave between 2004 and 2007, potentially explaining the higher fraction of zeroleverage firms in the last vintage group. Overall, these results indicate that the upward trend in the percentage of zero-leverage firms is partly driven by firms in more recently listed groups, thus by new IPO firms entering the sample. [Insert Figure 3 here] In a related analysis, we investigate whether our findings for vintage effects are directly related to firm age. Firm age is measured as the difference between the actual calendar year and a firm s IPO date. The IPO date is obtained from merging the available IPO dates in the Compustat Global and the Thomson One databases. We classify a firm as an IPO firm if it was 4 5 Related to this conjecture, Byoun et al. (2011) document that several years prior to becoming debt-free zeroleverage firms rely mainly on external equity in order to reduce debt when the stock market valuation is high. Custódio et al. (2011) use a vintage approach to analyze the declining debt-maturity of US firms. In results not reported, we use different (fixed-length) listing periods with similar outcomes. In a robustness check for the United Kingdom, we exclude all IPOs from our sample that are listed on the Alternative Investment Market or AIM (Espenlaub et al., 2009). Again, our results remain qualitatively unchanged.

11 10 listed over the preceding three years (and as established otherwise). 6 Figure 4 depicts the evolution of the fraction of zero-leverage firms in three different age groups for the full sample and the individual countries: (i) the oldest firms listed before 1989; (ii) IPO firms (not older than three years and listed 1989 or later); (iii) non-ipo firms (older than three years and listed 1989 or later). Young IPO firms exhibit the highest fraction of zero-leverage firms, and it is increasing sharply over time from roughly 6% in 1989 to almost 25% in Nevertheless, the percentage of zero-leverage firms is increasing in the subsample of non-ipo firms as well. Taken together, the results on firm size and age indicate that a large number of zero-leverage firms are small and young. According to Hadlock and Pierce (2010), size and age are particularly useful predictors for the level of a firm s financial constraints. Therefore, in Section 4 we use their constraint measure (SA-index) in order to analyze capital supply-side effects and differentiate between financially constrained and unconstrained zero-leverage firms. [Insert Figure 4 here] Our stylized facts so far suggest that changes in the sample composition explain a large part of the increase in the percentage of zero-leverage firms. However, as we do not observe that newly listed firms in each vintage year start debt-free and then initiate using debt as they mature, the increase in the percentage of zero-leverage firms cannot be fully captured by listing vintage and age effects. Therefore, we analyze if changes in the overall industry composition also contribute to explaining the increasing percentage of zero-leverage firms in the more recent listing groups. If riskier industries have become relatively larger because of newly-listed IPO firms, an increase in newly-listed technology firms in the more recent years may explain the vintage effect and contribute to the increase in the percentage of zero-leverage firms. In a first step, we assign firms to the 10 Fama and French (1997) industries based on their four-digit SIC codes. 7 Figure 5 shows the evolution of the percentage of zero-leverage firms in main industrial sectors (excluding financial and utility firms) for both the full sample and the individual countries. Zero-leverage firms are not limited to certain industries. Nevertheless, a common observation across all countries is that zero-leverage firms are concentrated in the 6 7 As a robustness check, we use a five year period in the secondary market to classify IPO firms, and the results remain qualitatively unchanged. See If SIC codes are not available in Compustat, we use the GICS (Global Industry Classifications Standard) codes to assign a firm to an industry sector.

12 11 healthcare and the information technology sectors. In the full sample, the percentage of zeroleverage firms in these two sectors increases to approximately 32% and 37%, respectively, by the end of the sample period. This high concentration of zero-leverage firms is consistent with Wessel and Titman s (1988) conjecture that firms with unique or specialized products (having large research and development expenditures and high selling expenses) are likely to impose higher costs on their customers, workers, and suppliers in the event of liquidation, and thus choose extremely conservative debt ratios. 8 The pattern also follows along Shleifer and Vishny s (1992) conjecture that non-redeployability of assets (and their lower ex post liquidation values) creates ex ante incentives to reduce leverage in order to mitigate the possibility of forced sales at prices below value in best use. Moreover, there are several country-specific industry effects. For example, the percentage of zero-leverage firms in the telecommunication sector is high in Germany, which is attributable to privatization efforts in this sector and firms going public during the late 1990s. In both the United Kingdom and Canada, the zero-leverage phenomenon is pronounced in the energy and materials sectors. This increase in the percentage of zero-leverage firms is accompanied by significant IPO activity in these two sectors in both countries. For example, in our sample the materials sector accounts for 34% of all Canadian IPOs in 2003, 48% in 2004, and 43% in 2005 (in terms of the number of issues), and, as such, this heavy IPO activity in the materials sector accounts for the increase in the percentage of zero-leverage firms (to more than 30%) during these years. Similar patterns are observable for the materials and the energy sectors in the United Kingdom. All in all, the country-specific industry effects seem to be driven to a large extent by the IPO waves in the different countries. [Insert Figure 5 here] In a second step, we compare the actual percentage of zero-leverage firms with the percentage of zero-leverage firms when holding the industry weights constant over time. Using the 1989 market capitalization weights, Figure 6 plots the yearly percentage of zero-leverage firms against the value-weighted average across industries over time. The two lines for the full sample start to diverge in the early 1990s, when the actual percentage of zero-leverage firms increases more than that based on the constant 1989 weights. This difference increases to rough- 8 Froot s (1992) case study of Intel Corporation is an early example from the information technology sector. In the late 1980s and early 1990s, Intel chose a policy of zero net-debt given its highly competitive and technologically risky environment. Similarly, firms in the pharmaceutical sector are potentially threatened by excessive damages claims, and thus they follow extremely conservative leverage policies.

13 12 ly five percentage points by Nevertheless, there is a strong increase in the percentage of zero-leverage firms in both groups. If industry composition effects were able to fully capture the zero-leverage phenomenon, the line comprising the 1989 market capitalization weights should not exhibit a strong upward trend. As a result, the international zero-leverage phenomenon cannot be purely driven by IPO firms shifting to industries where extreme debt conservatism is more commonly adopted. [Insert Figure 6 here] With increasing business risk, the prevalence of extreme debt conservatism is more likely to increase. In addition, if there has been a trend for increasing business risk over time, this effect contributes to explaining the zero-leverage phenomenon. A zero-leverage firm s stock return volatility reflects its business risk; thus stock return volatility equals asset volatility. For all other (debt-financed) firms, we follow Frank and Goyal (2009) and compute their asset volatility by unleveraging the annual volatility of stock returns (see Table A1). Annual standard deviations are derived from monthly stock returns (matched from Thomson Datastream). Figure 7 depicts the evolution of asset volatility over the four vintage periods (as defined in Section 2.2) for both zero-leverage and debt-financed firms. There are two observations here: First, asset risk is substantially higher in the zero-leverage group in comparison with the group of debt-financed firms. And second, there is a general trend of increasing asset return volatility over time (Campbell, et al., 2010; Wei and Zhang, 2006); thus the vintage effect seems strongly related to this change in asset risk. 9 Therefore, in addition to IPO and industry effects, higher asset return risk is another driver of the increasing propensity to maintain a zero-leverage policy over time and in the different vintage periods. Firms with higher business risks are more likely to engage in risk management. In addition to their risk management choices on the asset side of the balance sheet, their capital structure constitutes another layer of corporate risk management (Stulz, 1996). By reducing the amount of debt in its capital structure (or even completely deleveraging), a firm reduces shareholders total risk exposure because equity represents a residual claim and offers an all-purpose risk cushion against losses. Equity provides protection against risks that are difficult to anticipate or measure (Meulbroek, 2002). In this sense, our findings with regards to asset risk are con- 9 We use all months in a vintage group to compute asset return volatility. The effect is even more pronounced if we use only the first three years of return data for each vintage group.

14 13 sistent with our earlier observation that zero-leverage firms tend to be concentrated in the more opaque information technology and healthcare sectors. [Insert Figure 7 here] Taken together, these stylized facts offer explanations (albeit incomplete and interrelated ones) for the strong increase of zero-leverage firms over time. The percentage of zero-leverage firms changes with firm size, age, and industry. The vintage effect, changes in industry composition and higher business risk further contribute to explaining the surprising increase in the percentage of zero-leverage firms across all sample countries. However, the fraction of zero-leverage firms is increasing in the older vintage groups and in the sample based on the (constant) 1989 industry composition as well, and thus there remain unexplained parts of the puzzle. 3. Firm- and Country-level Determinants of Zero-Leverage 3.1 A Propensity Model Using Standard Capital Structure Variables This section starts by quantifying the roles of changing firm characteristics and the potentially increasing propensity to follow a zero-leverage policy in explaining the zero-leverage puzzle. We adopt the approach used by Fama and French (2001), Bates et al. (2009), and Denis and Osobov (2008). In a first step, we run yearly logistic regressions using the full sample to estimate the probability of firms exhibiting zero-leverage during a base period. The dependent binary variable is set equal to 1 for a firm adopting a zero-leverage policy in year t (and 0 otherwise). Our explanatory variables are the four standard capital structure variables (Rajan and Zingales, 1995; Frank and Goyal, 2009): profitability, market-to-book ratio, size, and tangibility (see Table A1). In a second step, we calculate the probability for each firm to maintain a zero-leverage policy based on these characteristics in each year (starting in 1994) using the average annual coefficients estimates from the base period. The expected percentage of zero-leverage firms is obtained by averaging the individual probabilities across firms in each year and multiplying the result by 100. As the probabilities associated with firm characteristics are fixed at their base period values, variation in the expected percentage of zeroleverage firms after 1993 is attributable to the changing firm characteristics. Any difference between the expected percentage and the actual percentage of zero-leverage firms measures the firms propensity to maintain a zero-leverage policy.

15 14 Table III shows the results of our out-of-sample logistic regression. Controlling for the changes in firm characteristics, changes in the unexpected proportion of zero-leverage firms reflect changes in the propensity to follow extreme debt conservatism. At the beginning of the forecasting period, the difference between the actual and the expected percentage is small, indicating that the coefficients obtained from the base period are good predictors for the expected fraction of zero-leverage firms. However, the actual percentage of zero-leverage firms is higher than the expected percentage, and the difference increases over time. This observation suggests that there is an increasing propensity to follow a zero-leverage policy. [Insert Table III here] Moreover, the expected values barely change over time, indicating that the four standard capital structure variables would not allow for more zero-leverage observations. However, the actual percentage of zero-leverage firms is increasing sharply over time. Using the estimated coefficients from the base period regression on firm characteristics in any given sample year after the base period systematically underestimates the actual fraction of debt-free firms. These findings suggest considering additional variables to explain the zero-leverage phenomenon. 3.2 A More Comprehensive Analysis of Firm Fundamentals Table IV reports the evolution of the percentage of zero-leverage firms for a large set of firm characteristics over three-year subperiods. For each variable, we divide firms into three groups using the 30 th and 70 th percentiles as breakpoints. We also test whether there is a significant time trend in the different subperiods. The discussion is structured along major determinants of capital structure theories, such as taxes, agency problems and asymmetric information. [Insert Table IV here] Taxes The tax system is a primary factor for capital structure choices (de Jong et al., 2008; Fan et al., 2012). As tax deductions are to a large extent generated by interest payments, it is not surprising that many zero-leverage firms exhibit high tax payments, as shown in Panel A of Table IV. Moreover, even their non-debt tax shields are smaller compared to leveraged firms. This finding is hard to explain, because non-debt tax shields are the only possibility for zero-leverage firms to reduce their tax obligations (De Angelo and Masulis, 1980).

16 Signaling In the presence of information asymmetry, Ross (1977) argues that investors should choose larger levels of debt as a signal of higher quality and that profitability and leverage are positively related. An alternative way to incorporate signaling is through abnormal earnings. Barclay et al. (1995) show that firms with higher abnormal earnings carry more secured debt in order to control for the underinvestment problem, and thus zero-leverage firms should exhibit low abnormal earnings. 10 As shown in Panel A of Table IV, the mean percentage of zeroleverage firms is 11.0% in the group of low abnormal earning firms and 9.0% in the group of high abnormal earning firms. Moreover, the mean percentage of zero-leverage firms is 16.1% in the group of low profitability firms, and with 13.8% it is only slightly lower in the group of high profitability firms. This observation is inconsistent with the notion that leverage and profitability are positively related and that firms choose higher debt levels as a signal of high quality. Overall, the decision to follow a zero-leverage policy does not appear to be driven by signaling considerations Asymmetric Information Based on adverse selection costs, the pecking order theory suggests that firms prefer internal funds to external funds and debt to equity. In contrast, the trade-off theory predicts lower leverage under high asymmetric information; with pronounced information asymmetries, the expected costs of financial distress increase and firms thus choose lower leverage ratios. We use tangibility as a proxy variable for the degree of asymmetric information (Wessels and Titman, 1988). 11 Panel A of Table IV indicates that zero-leverage firms feature lower tangibility. Research and development (R&D) expenses serve as an alternative proxy variable for asymmetric information. We classify firms whose R&D-to-assets ratio is above the 70 th percentile (below the 30 th percentile) in a given sample year as high R&D firms (low R&D firms). While the percentage of zero-leverage firms is only 10.3% (5.6%) for the low (medium) R&D group, on average, the high R&D group consists of 16.7% zero-leverage firms Abnormal earnings are defined as the ratio of the difference between the income before extraordinary items in time t and t 1 to the firms market value of equity at time t 1. However, high tangibility does not necessarily imply that industry-specific assets are highly liquid and redeployable during turbulent times (Shleifer and Vishny, 1992; Pulvino, 1998).

17 16 Firms with a low or no credit rating at all are expected to suffer from a higher degree of information asymmetry. Following Faulkender and Petersen (2006), we use a debt capacity variable (see Section 4 for more details) to proxy for the degree of asymmetric information and expect firms with low debt capacity to be more likely to pursue a zero-leverage policy. Indeed, the average percentage of zero-leverage firms is almost six times larger for firms with a low debt capacity (17.5%) than for firms with a high debt capacity (3.6%). Overall, we observe cross-sectional variation in zero-leverage firms that is consistent with the asymmetric information hypothesis. The relationship between zero-leverage and the firm characteristics tangibility, R&D expenses as well as credit ratings supports findings in related studies, such as Dang (2011) and Byoun et al. (2011). Most important, as indicated by the magnitude of the estimated time trend coefficients, the increase in the fraction of zero-leverage firms is partly attributable to groups of firms with highly pronounced information asymmetry Agency Costs Lower leverage minimizes the agency costs of debt, such as underinvestment (Myers, 1977) and asset substitution (Leland and Toft, 1996). If zero-leverage firms attempt to mitigate underinvestment problems by following an extremely conservative debt policy, we expect them to exhibit high growth options, high payouts, and to rely heavily on external equity financing in order to retain their growth options. While the respective time trends in Panel A of Table IV are all estimated significantly, high market-to-book firms contain the highest fraction of zeroleverage firms (20.4%), on average. High asset growth firms have a higher percentage of debtfree firms (12.6%) than small asset growth firms (11.7%), albeit the difference is only marginal. Therefore, growth opportunities exert a positive (albeit weak) influence on the percentage of zero-leverage firm. In the absence of interest and amortization payments, dividend payouts are the only way to smooth the earnings of zero-leverage firms. Debt conservative firms that do not pay dividends may be prone to free cash-flow problems (Jensen, 1986). Alternatively, when coupled with high equity issuances, higher dividends may work as a means of establishing a reputation for moderation in expropriating wealth from shareholders (Byoun et al., 2011). Using a payout dummy variable that takes the value of 1 if a firm pays dividends in year t (and 0 otherwise), Table IV shows that the percentage of zero-leverage firms is higher among non-payers. In contrast, when only considering dividend-paying firms, both high and low payout firms exhibit

18 17 higher percentages of zero-leverage firms (14.9% and 17.7%) than firms in the medium range (6.6%). As a large proportion of zero-leverage firms do not pay dividends at all or pay low dividends, it seems that zero-leverage firms taken as a whole do not have a strong tendency to substitute dividend payments for leverage. 12 Finally, Panel A of Table IV confirms the hypotheses that zero-leverage firms rely heavily on external equity financing. On average, the percentage of zero-leverage firms is highest in the group of firms with the highest equity issuances (20.0%), and it strongly increases over time (from 10.8% to 25.9%) Multivariate Analysis So far, our analysis provides evidence that a vintage effect, changes in industry composition, and higher business risk contribute to explaining the increase in the fraction of zero-leverage firms. Supporting the evidence of DeAngelo and Roll (2012) that the existence of low leverage firms in recent years is due to a surge in listings by young growth firms, we document that the majority of the zero-leverage firms in our sample carry these attributes. As expected, our propensity model reveals that the standard capital structure variables are not able to capture this strong increase, and thus there seems to be an increasing propensity to adopt a zero-leverage policy. In order to examine if additional capital structure variables are able to explain the increasing propensity to follow a zero-leverage policy, we analyze all firm fundamentals in a multivariate setup and run logistic regressions. The dependent binary variable takes the value of 1 if firm i pursues a zero-leverage strategy in year t, and 0 otherwise. All explanatory variables are lagged by one year. Based on the stylized facts in Section 2.2, we control for industry effects by including 2-digit SIC code dummy variables (Faulkender and Petersen, 2006). Table V presents the results of the logistic regressions. The negative intercepts in all columns imply that zero-leverage firms are less common than debt firms. Column 1 shows the results when we only include the standard capital structure variables. Similar to the coefficients from the base period regression in the propensity model, size and tangibility decrease and the market-to-book ratio and profitability increase the probability to adopt a zero-leverage policy. In Column 2 we add all firm fundamentals from Table IV, including asset risk and dummy variables for the four vintage periods (see Section 2). Comparing the explanatory power of the lo- 12 This result contradicts the findings in Dang (2011) and Byoun et al. (2011) that zero-leverage firms exhibit higher payout ratios than their matching firms (either to address the agency costs of free-cash flow or to build reputation for addressing the agency problem of expropriation of outside shareholders; see section 4).

19 18 gistic regressions in Columns 1 and 2, the pseudo R-square increases from 15.1% to 25.7%, thus indicating that the additional firm fundamentals are predictors for the probability to maintain zero-leverage over and above the standard capital structure variables. The vintage dummy variables exhibit increasing coefficients, indicating that the probability of a firm to pursue a zero-leverage policy is higher for younger vintage periods (see Figure 3). As expected, the coefficients on asset risk and taxes are significantly positive. The coefficient on the payout dummy variable is significantly negative. Together with the positive coefficient on the payout ratio, this result indicates that there are two different kinds of zero-leverage firms: non-payout and high payout zero-leverage firms (Strebulaev and Yang, 2012). Conditional on being a payer firm, the positive coefficient indicates that a higher payout ratio increases the probability of maintaining a zero-leverage policy. Consistent with this positive payout coefficient, zero-leverage firms have significantly higher equity issuances than debt firms. In combination with their higher asset risk, this finding is consistent with the conjecture that the riskiest firms are constrained to use equity (Bolton and Freixas, 2000). [Insert Table V here] 3.3 Country-Level Regression Analysis The financial systems of our sample countries can be roughly divided into bank-oriented and capital-market-oriented systems. According to Allen and Gale (2001), the continental European countries and Japan are classified as bank-oriented financial systems, whereas the United States, Canada, and the United Kingdom are considered capital-market-oriented financial systems. Petersen and Rajan (2002) and Djankov et al. (2007) note that the most important aspect of corporate lending is information. If lenders know more about borrowers, their credit history, or other lenders to the firm, they are less concerned about the lemons problem and extend more credit. Relationship lending in bank-based countries implies that banks (as natural monitors) have privileged access to information (Leland and Pyle, 1977), and thus the asymmetric information problem will be less pronounced. Therefore, it is expected that leverage is lower and the percentage of zero-leverage firms is higher in countries with a market-oriented financial system in comparison to countries with a bank-based financial system. La Porta et al. (1998, 2002) argue that the common law system provides better external investor protection than the civil law system, resulting in better access to external finance and high-