Leverage Dynamics over the Business Cycle Michael Halling†
Jin Yu‡
∗
Josef Zechner§
February, 2011
Abstract There remains broad disagreement about what the important drivers of capital structure dynamics are. This paper sheds new light on this question by studying the business cycle dynamics of leverage ratios, using a comprehensive sample of firms from 19 countries. Our main finding is that book leverage evolves pro-cyclically but market leverage counter-cyclically. This empirical pattern is difficult to reconcile with any single theoretical capital structure model. We also find that leverage dynamics are dramatically different for financially constrained and unconstrained firms and that the origin of law, as a proxy for the development of public markets, matters. Finally, we find that leverage adjusts more slowly to target levels during recessions and that this effect is amplified for firms from civil law countries.
JEL Classifications: G32, G15. Keywords: Empirical corporate finance, capital structure dynamics, business cycle variation.
∗ Acknowledgment: Preliminary and incomplete. We welcome comments, including references to related papers we inadvertently overlooked. We thank Johann Reindl for excellent research assistance. All errors are our responsibility. † University of Utah (email:
[email protected]) ‡ University of New South Wales (email:
[email protected]) § Vienna University of Economics and Business (CEPR and ECGI) (email:
[email protected])
Leverage Dynamics over the Business Cycle Abstract There remains broad disagreement about what the important drivers of capital structure dynamics are. This paper sheds new light on this question by studying the business cycle dynamics of leverage ratios, using a comprehensive sample of firms from 19 countries. Our main finding is that book leverage evolves pro-cyclically but market leverage counter-cyclically. This empirical pattern is difficult to reconcile with any single theoretical capital structure model. We also find that leverage dynamics are dramatically different for financially constrained and unconstrained firms and that the origin of law, as a proxy for the development of public markets, matters. Finally, we find that leverage adjusts more slowly to target levels during recessions and that this effect is amplified for firms from civil law countries.
JEL Classifications: G32, G15. Keywords: Empirical corporate finance, capital structure dynamics, business cycle variation.
1
1
Introduction
There is an ongoing intense debate on what the main determinants of capital structure dynamics are. In particular, there is little agreement in the academic literature on the following questions. Are capital structure dynamics driven by changing firm characteristics, and if yes, which ones? Are changing capital market conditions, or changing investor sentiment important? Or are capital structure dynamics largely driven by random events? Empirical examinations of firms’ dynamic capital structure behavior face significant challenges. For example, existing tests have recently been criticized for ignoring the effects of transactions costs, for selection and survivorship biases or mechanical mean reversion (see, for example, Strebulaev (2007), Leary and Roberts (2005), Chang and Dasgupta (2009), or Shyam-Sunder and Myers (1999) and Chen and Zhao (2007)). One possible remedy for these challenges is to exploit natural experiments (see, e.g. Welch (2011)). Such natural experiments occur in periods in which certain candidate explanatory variables change significantly due to some exogenous event. One can then analyze how such events affect corporate capital structure choices. In this paper we analyze capital structure dynamics by focussing on natural experiments provided by economic recessions. During recessions most of the main theoretical determinants of firms’ financial structure usually experience significant changes. For example, during recessions corporate cash flows drop for many firms, equity capital of financial intermediaries is reduced, equity valuation levels and the term-structure of interest rates usually change. We first provide an overview of the main theories of dynamic capital structure choice and their implied empirical hypotheses. In this discussion we focus on the influence of business cycle variation on firms’ target capital structures and choices of debt maturities (long-term vs. short-term). Because of transactions costs or other market frictions, firms will not always be right at their target leverage ratios but might only partially adjust towards their targets. This partial adjustment or, more specifically, the speed of adjustment represents another dimension of our analysis. In our empirical study, we use stock prices and annual firm-level accounting data, combined with business cycle data for 19 countries. From a statistical point of view, the design of the empirical analysis is driven by the goal to include a sufficiently large number of recession year observations. This is achieved by obtaining data for several countries, instead of focus-
2
ing on a single country only.1 Thus, we also very carefully assign reported balance sheet information to recessions taking into consideration each firm’s fiscal year. In addition to analyzing a large number of recessions, our data panel also allows us to identify country characteristics that influence differences in capital structure dynamics. Our empirical results generate several novel and interesting insights. First, we find that market target leverage ratios are generally countercyclical. This contradicts traditional tradeoff models, which predict constant target ratios. It also is at odds with models that focus on time-varying collateral values, capital shocks to financial intermediaries or temporary freezing of debt markets, which predict pro-cyclical market leverage ratios. In contrast, the counter-cyclicality is more in line with dynamic agency models and possibly some market timing models. We also find evidence that firms adjust upwards (downwards) their book debt levels during expansions (recessions). This pro-cyclicality of book values is more in line with tradeoff models but clearly contradicts pure market timing based theories, since they would imply that firms should only issue equity and also should buy back debt during expansions. Finally, we find many highly significant effects of the business cycle on firms’ dynamic financing behavior for different leverage definitions and different subsets of firms. For example, we find that leverage dynamics differ across countries by their legal origin and the development of their capital markets. Thus, a pure irrelevance view of financing behavior, as proposed by some academics, seems to be clearly rejected by our data. As far as speed of adjustment to target leverage is concerned, we find a strong impact of business cycle variation that is consistent with theory: speed of adjustment is, in general, significantly slower during recessions than during expansions. We document, however, that this decrease in speed of adjustment during recessions is much less pronounced in countries with well developed public markets. This result is consistent with an interpretation that these markets are more robust during recessions and freeze to a lesser extent, thus enabling firms to consistently adjust their leverage to appropriate target levels. There have only been few empirical studies on capital structure and macroeconomic determinants. Korajczyk and Levy (2003) find evidence that book and market target leverage are counter-cyclical for relatively un1 If our study focused on the US, we would only end up with 5 recessions after 1975. Three of these recessions are less than 12 months long: 1/1980 to 7/1980, 7/90 to 3/91 and 3/2001 to 11/2001. Thus, the statistical power to discriminate between expansions and recessions using yearly balance sheet information would be very low.
3
constrained firms, but pro-cyclical for relatively constrained firms.2 Our conclusion about target leverage dynamics of unconstrained firms is quite different: while we also find that market leverage is counter-cyclical, we find strong evidence that book leverage is actually pro-cyclical. This difference is most probably explained by differences in empirical design: most importantly, we employ a more direct definition of business cycle variation. Our results are consistent with Korajczyk and Levy (2003) in the sense that we also find that constrained and unconstrained firms are differently affected by business cycle variation. We also extend their study by including short-term leverage, by analyzing the business cycle dynamics of speed of adjustment estimates and by evaluating the influence of capital market characteristics on these dynamics. Few empirical papers have looked at the relationship between speed of adjustment to target leverage ratios and macroeconomic conditions. Consistent with our results, Cook and Tang (2009) find, using only US data, that the speed of adjustments towards target capital structures is faster in booms than in recessions. Drobetz and Wanzenried (2006) find a similar result for 90 Swiss firms. The rest of the paper is organized as follows: Section 2 discusses three strands of theory that have predictions for leverage dynamics and summarizes our hypotheses; Section 3 summarizes the data and our empirical design; Section 4 reports results regarding target leverage ratios and Section 5 presents the results regarding speed of adjustment estimates. Section 6 concludes.
2
Literature and hypotheses
The literature on capital structure dynamics can be classified into three broad strands: issuer-driven tradeoff models, capital market-driven models of time-varying supply of financial intermediation and models driven by behavioral biases of investors and/or managers. In this section we summarize these strands of literature and explain which empirical hypotheses regarding firms’ capital structure dynamics they generate. In our discussion we define leverage as pro-cyclical (counter-cyclical) if it increases (decreases) during economic expansions (contractions). 2 They proxy for business cycle variation using 2-year corporate profit growth, 2-year equity market return and commercial paper spread.
4
2.1
Issuer-driven capital structure dynamics
The first strand of literature is by now well developed and focuses on capital structure dynamics determined by time-varying firm characteristics. The first advances were made by Merton (1974) and Black and Cox (1976) and extended by Kane, Marcus, and McDonald (1984) and Kane, Marcus, and McDonald (1985), and Leland (1994) to consider the tradeoff between a tax benefit of debt and costly bankruptcy in continuous-time. These models constitute an important first step towards accounting for dynamic aspects of capital structure choice by introducing a stochastic asset value process, but they do not allow firms to adjust their leverage over time. The next generation of dynamic capital structure models explicitly allows firms to increase or decrease leverage over time. The basic framework developed in Fischer, Heinkel, and Zechner (1989a) and Fischer, Heinkel, and Zechner (1989b) has been extended in several directions.3 Although firms do not respond instantaneously to changes in their profitability, due to transactions costs, these model predict a constant target capital structure, irrespective of whether the majority of firms experiences increasing or decreasing profitability, i.e. irrespective of the business cycle. Note that, strictly interpreted, the models only make predictions about leverage dynamics expressed in market values, since they do not separately model the evolution of book values of debt, equity, and assets. Obviously asset market values generally fluctuate more than their book values. Thus, if one assumes that firms’ asset book values move less than their market values over the business cycle, then the above models implicitly predict increasing (decreasing) book target leverage ratios during expansions (contractions). The models also have implications for their propensity to move to the target leverage ratio when their profitability either increases or decreases. While equityholders have an incentive to call existing debt to subsequently issue a higher amount in response to increased profitability, they are generally reluctant to call debt to reduce debt levels after decreasing profitability. This result is driven by the fact that calling debt after adverse changes in a firm’s profitability will generally lead to wealth transfers from equity to debt holders. Many dynamic tradeoff models therefore imply that equityholders do not find it optimal to decrease leverage in response to decreasing 3 For example, by modeling firm dynamics via stochastic cash flows rather than stochastic asset values (Goldstein, Ju, and Leland (2001), Dangl and Zechner (2004), and Strebulaev (2007)), and by allowing for investment (Mello and Parsons (1992), Mauer and Triantis (1994), Morellec (2001), Moyen (2004), Childs, Mauer, and Ott (2005),Hennessy and Whited (2005), Hennessy and Whited (2007), Moyen (2007), Titman and Tsyplakov (2007), and Gamba and Triantis (2008)).
5
profitability (see, for example, Dangl and Zechner (2004) and Dangl and Zechner (2007)). Thus, dynamic tradeoff models predict asymmetric adjustments towards the target leverage ratio: Firms respond to increased profitability by increasing leverage but are reluctant to adjust in response to decreasing profitability.4 Firms’ reluctance to reduce debt in response to lower profitability can be mitigated in two ways. First, instead of considering only perpetual debt, as most of the models referred to above do, one can consider finite debt maturities. As shown by Dangl and Zechner (2007), firms may then decide not to roll over expiring short term debt in bad times, since this could be too expensive. Thus, short-term debt induces firms to effectively reduce leverage in bad times by not fully rolling over short term debt. Second, instead of calling debt at the call price, firms may be able to renegotiate debt after a drop in profitability. In the presence of bankruptcy costs, it may be rational for bondholders to accept equityholders’ renegotiation offers, even if this implies partial debt forgiveness. This has been demonstrated, for example, by Mella-Barral and Perraudin (1997), Mella-Barral (1999), and Hege and Mella-Barral (2005). Such renegotiation offers may not be feasible if the debt is public and held by dispersed investors (for a discussion of related issues, see, for example, Gertner and Scharfstein (1991)). Thus, the countercyclicality of leverage during downturns can be mitigated by shorter debt maturities and if large blocks of debt are non-public. Summarizing our discussion so far, the following hypotheses emerge from dynamic tradeoff models: Firms market leverage target ratios are constant, whereas their book leverage ratios are pro-cyclical. Also, the speed of adjustment towards the target ratio is asymmetric: it is higher during expansions and lower during contractions. The latter effect should be mitigated if corporate debt is short-term and if large blocks are non-public. More recently, dynamic capital structure models have been extended to allow for the business cycle to affect either the level or the drift of the state variable. For example, the level of corporate cash flows and/or its drift may decrease (increase) during recessions (expansions). These effects on the model parameters influence firms’ target leverage ratios. During economic expansions, for example, the expected costs of bankruptcy are lower for a given debt level and the expected tax benefit may be higher. Thus, as shown by Bhamra, Kuehn, and Strebulaev (2009), target leverage ratios during expansions are higher for plausible model parameterizations. 4
The relationship between firm profitability/cash flow and leverage dynamics in the presence of transactions costs is illustrated for simulated economies in Strebulaev (2007).
6
Opposite results are found by Hackbarth, Miao, and Morellec (2006). They argue that the present value of future cash-flows is pro-cyclical and that this effect dominates the pro-cyclical choice of debt, leading overall to countercyclical leverage ratios. Thus, according to this literature, the predicted dynamics of market leverage ratios are ambiguous — they can be pro-cyclical or counter-cyclical. Gamba and Triantis (2008) develop a dynamic model that incorporates both, pecking order concepts and the trade-off between tax benefit against financial distress.5 Since it is more costly for a firm to adjust its debt than its cash holdings, their model predicts that a matured firm, which has already reached its steady state, maintains a constant debt level and adjusts its cash holdings, i.e. net debt level, in response to the time-varying profitability. Therefore, leverage ratios are counter-cyclical but net leverage ratios are more pro-cyclical. As Gamba and Triantis (2008) only discuss the levels of debt and cash, we interpret their predictions to apply to both, book and market, leverage ratios. Yu (2010) presents another dynamic model that incorporates aspects of pecking order and tradeoff theory. He allows firms to simultaneously make dynamic leverage and investment decisions over the business cycle. Again, the business cycle affects the model parameters. In the absence of investment, this model also implies pro-cyclical leverage, as in Bhamra, Kuehn, and Strebulaev (2009). However, the explicit consideration of investment opportunities changes this result, since investments are funded according to a version of the pecking order. I.e. investment is first funded via internally generated cash, then via debt and only in exceptions via equity. In his model, both the trade-off between tax benefit against bankruptcy costs and the pecking order of financing affect a firm’s capital structure and investment policies. Financially constrained firms have to follow the pecking order when they are at external financing margin but unconstrained firms have the privilege to trade-off tax benefit against bankruptcy costs. This generates counter-cyclicality of book and market leverage for firms that are financially constrained. These are likely to be small, growth, and/or highly levered firms. The models discussed so far focus largely on the tradeoff between tax benefits of debt versus bankruptcy costs. There are other papers which derive implications for leverage dynamics driven by time-varying collateral values, by time-varying agency costs or by dynamic versions of the pecking 5
Though in their model only risk-less debt contracts are studied, a firm’s borrowing capacity is constrained by its assets fire sale discount in financial distress.
7
order theory. Several models explore the effect of borrowers’ deteriorating balance sheets on their ability to obtain bank loans. In these papers firms’ access to outside debt is limited by the amount of collateral they have. If a recession leads to a drop in the value of the assets which can be pledged as collateral, this also limits debt financing. Examples for models that explore this are Bernanke and Gertler (1989), Calomiris and Hubbard (1990), Gertler (1992), Greenwald and Stiglitz (1992), Kiyotaki and Moore (1997), and Shleifer and Vishny (1992). These models predict pro-cyclical financial structures, since firms’ access to external debt markets is limited during contractions due to decreased collateral values. Levy and Hennessy (2007) model time varying agency costs and their effects on leverage dynamics. In this model an entrepreneur must hold a minimum share of the firm’s equity and keep the firm’s leverage below some critical upper bound to commit not to divert earnings and/or assets. In this setup firms are more highly equity financed during expansions and more levered during contractions. The model predicts that these counter-cyclical swings would be sharper for firms with good governance and for firms which are less financially constrained.
2.2
Capital market-driven capital structure dynamics
Another, somewhat less developed strand of literature focuses on capital market driven determinants of firms’ financial structures. There are at least two potential channels, through which such “supplyside” effects can arise. First, raising external capital requires the services of intermediaries, either by directly relying on funding via bank loans, private debt placements, private equity, etc. or by relying on intermediaries as underwriters in the primary market for corporate securities or as market makers in the secondary market. Second, liquidity in the secondary markets for corporate securities may change over the business cycle and thus have an effect on firms capital structure choice. If the costs of intermediation and market illiquidity vary with the business cycle, then firms’ leverage dynamics should reflect such variations. The first supply side channel mentioned above has been explored by analyzing shocks to intermediaries’ capital. If intermediaries’ lending capacities are limited by the amount of capital, for example via regulation, then an adverse shock may have a direct effect on the amount of credit banks can supply to businesses. Holmstrom and Tirole (1997) develop a model, where intermediaries are exposed to capital shocks. They show that poorly capital8
ized firms that have limited access to the public debt and equity markets are affected most by adverse shocks to intermediaries’ capital. Other analyses of this potential effect on capital structure are presented by Bernanke and Blinder (1992), Romer, Romer, Goldfeld, and Friedman (1990), or Kashyap, Stein, and Wilcox (1993).6 To the extent that economic expansions are correlated with higher amounts of capital for financial intermediaries, this implies pro-cyclical leverage dynamics. A young but growing literature explores time varying secondary market liquidity. Several recent papers provide models of secondary markets for corporate debt which are characterized by search costs. Examples are Ericsson and Renault (2006) and Duffie, Garleanu, and Pedersen (2007). In these models, search costs may increase during recessions. If the expected costs of trading in the secondary market is discounted in the issue price, then these models are consistent with lower leverage during contractions, i.e. procyclicality. Note that this only applies to firms borrowing via publicly held bonds, not bank debt. Hennessy and Zechner (2011) show that secondary debt markets become particularly fragile during contractions. In such situations multiple equilibria arise where secondary debt markets may freeze. Hennessy and Zechner (2011) show that this leads to financing cycles, where firms use less debt during contractions.7 Again this predicts pro-cyclical leverage for firms which rely on public debt.
2.3
Market-timing driven capital structure dynamics
Deviations from fundamental valuations may also influence capital structure choices. For example, Graham and Harvey (2001) find in a survey that the majority of CFOs state that the amount by which their stock is over- or undervalued plays an important role when deciding whether to issue equity or not. The effect of investors with limited rationality on financial markets has been analyzed theoretically, for example, by Fischer and Merton (1984), De Long, Shleifer, Summers, and Waldmann (1990), Morck, Shleifer, and Vishny (1990), and Blanchard, Rhee, and Summers (1993), and Stein (1996). According to this literature, firms can actively exploit misvaluations by timing their equity and debt issues. Specifically, firms time IPOs and seasoned stock offerings to take advantage of high market valuation levels and/or a run-up in their stock price (for some empirical evidence on such tim6
For an interesting empirical study of this channel, see Leary (2009). Bhamra, Kuehn, and Strebulaev (2009) also model exogenously varying debt market illiquidity during contractions. This leads to pro-cyclical leverage dynamics. 7
9
ing, see, for example, Pagano, Panetta, and Zingales (1998)). Furthermore, according to this literature, corporate debt is issued when equity valuation levels are low and/or interest rates are low (see, for example, Baker and Wurgler (2002) for empirical evidence on the market timing view of capital structure dynamics). According to this literature, firms should exhibit strongly counter-cyclical leverage ratios, both in terms of market values and in terms of book values. According to the market timing hypothesis firms should issue equity at the peak of an expansion, when equity valuation levels reach their highest levels, thereby reducing their leverage. At the trough of a contraction, equity valuations and interest rates are usually low, and firms would therefore prefer to issue debt, thus increasing their leverage further.8
2.4
Hypotheses
We now summarize the main hypotheses generated by the three strands of literature discussed above. The discussion of the previous chapter illustrates that there is considerable heterogeneity among the predictions of the mentioned models. In general, we distinguish hypotheses regarding business cycle dynamics of target leverage ratios and business cycle differences of speed of adjustment estimates towards these target ratios. An important complication arises because many theory models don’t explicitly distinguish between book and market leverage ratios or focus on either one and are silent on what should happen to the other one. According to our interpretation, in many cases models predict the same dynamics for book and market leverage ratios.9 Standard tradeoff theory is an exception, as it predicts constant market leverage ratios that imply pro-cyclical book leverage ratios. Wherever possible we will try to carefully disentangle our predictions and results according to this dimension. As far as the dynamics of target leverage are concerned, we evaluate the following specific hypotheses: 8 There is potentially another, behaviorally based channel of demand driven capital structure dynamics. This occurs when managers go through sentiment waves. I.e. sometimes they are overconfident, and issue debt. At other times they issue equity. One could even think of sun spot equilibria, where managers want to behave in a way that is similar to their peers. I.e. if some firms issue debt, other firms follow to imitate them. 9 In the case of pro-cyclical dynamics this argument is easy to see, as market leverage can only be pro-cyclical if book leverage is even more pro-cyclical. In the case of counter-cyclical market leverage, the implied prediction about book leverage is ambiguous. Nevertheless, as long as the theory does not explicitly suggest procyclical book leverage, we also assume the prediction of counter-cyclical book leverage.
10
H-TL 1a: Book and Market leverage ratios are pro-cyclical over the business cycle according to (i) dynamic tradeoff models without10 and, partly, with explicit business cycle state variables11 , (ii) models of time-varying collateral values and (iii) capital market driven capital structure models. H-TL 1b: In contrast, market and book leverage ratios are counter-cyclical over the business cycle in (i) (some) dynamic tradeoff models with explicit business cycle state variables, (ii) hybrid dynamic capital structure models that include pecking order and tradeoff concepts, (iii) in dynamic agency cost models, and (iv) in market timing driven capital structure models. These general hypotheses can be refined along the following dimensions. H-TL 2: Net book and market leverage ratios are pro-cyclical according to hybrid dynamic capital structure models that include pecking order and tradeoff concepts. H-TL 3a: Book and market leverage ratios are more pro-cyclical for financially constrained firms than for unconstrained firms according to (i) models of time-varying collateral values and (ii) capital constraints in the banking sector. H-TL 3b: Book and market leverage ratios (especially short-term) are more counter-cyclical for constrained firms according to (i) dynamic tradeoff models with BC variables, investment and pecking order components and (ii) time-varying agency costs. H-TL 4a: Long-term book and market debt ratios should be more pro-cyclical (less counter-cyclical) than short-term leverage ratios if information asymmetries increase during recessions. H-TL 4b: Long-term book and market debt ratios should be less pro-cyclical (more counter-cyclical) than short-term leverage ratios if transactions costs of rolling over short-term debt are higher during recessions, and 10 To be precise, standard dynamic tradeoff models predict constant market leverage ratios. We included the predictions from this model class in this hypothesis in order to keep the number of hypotheses manageable. 11 This class of dynamic capital structure models has ambiguous predictions: Bhamra, Kuehn, and Strebulaev (2009), for example, support pro-cyclicality, while Hackbarth, Miao, and Morellec (2006) suggest opposite dynamics. The latter paper, thus, belongs to Hypothesis H-TL 1b.
11
debt maturity is largely a tradeoff between some benefits of short-term debt, such as a commitment to debt reductions or the signaling effect. H-TL 5a: Firms in common law (civil law) countries exhibit more procyclical (counter-cyclical) leverage dynamics for book and market values in dynamic tradeoff models without investment. H-TL 5b: Firms in countries with higher proportions of public debt exhibit more counter-cyclical book and market leverage dynamics according to the debt renegotiation literature. In addition to target leverage ratios we also analyze the speed of adjustment (SOA) of leverage ratios towards these target leverage ratios. Here are the hypotheses regarding the speed of adjustment. H-SOA 1: The speed of adjustment towards the target ratio is higher during expansions and lower during contractions according to dynamic tradeoff models without explicit business cycle state variables. H-SOA 2: The difference between the speed of adjustment towards the target leverage ratio in expansions and in contractions is lower for firms with large proportions of public debt according to the debt renegotiation literature.
3
Data and Empirical Design
3.1
Data and Sample
Our source of business cycle data is Economic Cycle Research Institute (ECRI)’s international cycle dates. We use the business cycle chronologies file, which includes countries from America, Europe, Asia Pacific, Africa, and Middle East regions. In order to have information on both business cycle dates and firm-level variables, we end up with 19 countries, ranging from developing to developed economies and from common-law to civillaw countries. Specifically, these countries are: Australia, Austria, Brazil, Canada, China, France, Germany, India, Italy, Japan, Korea, Mexico, New Zealand, Spain, Sweden, Switzerland, Taiwan, UK, and USA. We use Worldscope to obtain annual firm-level accounting data. Our sample period is from 1983 to 200912 . In our subsequent analysis, we remove 12
In our sample some countries have shorter period of data available than others. We don’t have firms from all countries for all years between 1983 and 2009. However, our first observations are in 1983 and last observations are in 2009.
12
China from our database because there are no recessions during the sample period. Variable definitions are given in Appendix A and summarized in Table 1. Financial firms are usually regulated and hence their leverage choices ought to be quite different from other industrial firms. For this reason and following the literature, we remove all financial firms, i.e. all firms with WSIC between 4300 and 4400 are deleted from our sample. We also drop firm-year observations such that either of the following conditions are met: (i) zero total assets value, (ii) zero market capitalization, (iii) total debt greater than total asset, (iv) market asset less than cash, (v) total asset less than cash, and (vi) negative cash. Following Korajczyk and Levy (2003), we assume that a firm is financially constrained if the following conditions are fulfilled: (1) the firm does not have a net purchase of debt or equity and does not pay dividends within the event window, and (2) the firms market to book ratio, defined as the sum of the market capitalization of equity and the book value of debt, divided by the book value of assets, at the fiscal year end is greater than one. In our empirical analysis, we consider each of eight leverage ratios (four based on book values and four based on market values) as a dependent variable. Book total leverage ratio (bl) is the total debt to total assets ratio. Book long-term leverage ratio (bl lt) is the long-term debt to total assets ratio. Book short term leverage ratio (bl st) is the short-term debt to total assets ratio. Book net leverage ratio (bl net) is the total debt less cash 13 to total assets less cash ratio. Market value based leverage ratios (Market total leverage ratio (ml), Market long term leverage ratio (ml lt), Market short term leverage ratio (ml st) and Market net leverage ratio (ml net)) are calculated in a similar fashion by only replacing total assets in the above definitions by market value. A recession dummy that is 1 during recessions and 0 otherwise (Rec) is our key independent variable. More precisely, this dummy is equal to 1 if a firm’s entire fiscal year overlaps with a recession.14 We also control for other variables, which have been widely used in the literature, including the logarithm of Net Sales (sales), market to book ratio (market to book), EBITDA to total assets ratio (profitability), PPE to total assets ratio (tangibility), industry median leverage ratio (industry median), dividend payer dummy 13
The cash variable includes cash and short-term investments. This definition is a relatively conservative way of identifying recessions. There are, however, two advantages: (i) the definition is most precise in aligning yearly firm data with recession information, and (ii) the definition requires that recessions last for at least 12 months and, thus, filters out “less severe”recessions. 14
13
(dividend payer), which equals 1 if a firm pays out dividend and 0 otherwise. Last but not least, lagged leverage ratios are used to capture the persistence in leverage dynamics. We further drop observations with (i) negative net sales, (ii) book net leverage ratio of less than -1, and (iii) market net leverage ratio of less than -1. We do allow firms, at some point in time, to be cash savers, i.e. carrying a negative net leverage ratio, rather than borrowers. However, we remove firm-year observations with net leverage ratios less than -1 because such firms hold a tremendous amount of cash relative to their other type of assets and hence are unlikely to be normal industrial firms. Table 2 summarizes the descriptive statistics of our key variables of interest. Means and standard deviations of variables for the entire sample are given in Panel A. Panel B shows that leverage ratios, in particular market leverage ratios, tend to be higher in recessions than in expansions. Panel C presents the summary statistics of our variables when the sample is split by financial constraints. Financially unconstrained firms are more levered than constrained firms. Further, constrained firms tend to have higher market to book ratios as well as lower profitability and tangibility than unconstrained companies. As shown in Panel D, firms from developing countries, on average, have higher leverage than firms from developed economies. The intuition is that developing countries tend to have less developed equity capital markets and poorer corporate governance leading to higher cost of equity capital. Panel E lists the summary statistics of our variables when the sample is split by origin of law. In general, leverage ratios of civil-lawcountry firms are higher than leverage ratios of common-law-country firms. Again, this is likely due to the fact that common-law countries, on average, enjoy good corporate governance, which tilts firms’ external finance choice towards equity finance.
3.2
Empirical Specification
In our empirical analysis we follow Flannery and Rangan (2006) and estimate dynamic partial adjustment capital structure models (DPACS-Model) including firm fixed effects.15 These models contemporaneously estimate time-varying target leverage ratios and estimates of the speed of adjustment with which actual leverage ratios move towards target leverage ratios. In this research we, in particular, examine the effect of business cycles on a firm’s target leverage ratio as well as its speed of adjustment towards the 15
See Chang and Dasgupta (2009) and Iliev and Welch (2010) for critical discussions of these models.
14
target. Specifically, our dynamic partial adjustment capital structure model (DPACS-model) is given by, exp rec lrj,i,t+1 = ρrec lrj,i,t Ij,t+1 +ρexp lrj,i,t Ij,t+1 +αRecj,t+1 +βXj,i,t +δj,i +ej,i,t+1 , (1) rec = 1 (I exp = where lr ∈ {bl, bl lt, bl st, bl net, ml, ml lt, ml st, ml net}, Ij,t+1 j,t+1 1) during recessions (expansions), countries are indexed by subscript j, firms by i, and periods by t, and δj,i is a firm fixed effect. The speed of adjustment estimate (SOA-estimate) is defined as λrec = 1 − ρrec (λexp = 1 − ρexp ) during recessions (expansions). Given the above specification, λrec (λexp ) measures the fraction of the difference between a firm’s actual and its target leverage ratios, both of which are time-varying, that has been closed during recessions (expansions). As far as time-varying target leverage is concerned, recall that Rec is the recession dummy. Xj,i,t is a vector of lagged firm-level characteristics, i.e.
X=
sales market to book profitability tangibility industry median dividend payer
.
(2)
The coefficient α and the coefficient vector β capture the influence of the business cycle variable and the above explanatory variables on target leverage.16
4
Empirical Results: Target Leverage Ratios
In this section, we present our empirical results concentrating on insights with regard to the relationship between business cycles and firms’ target leverage ratios.
4.1
Total Debt
Several of the hypotheses discussed in subsection 2.4 predict differences between leverage ratios during recessions and during expansions. Table 3 16
The estimates of α and β are also affected by λ (see equation (4) in Flannery and Rangan (2006)). In the results section, we report unadjusted coefficient estimates. Where necessary, we show and discuss the adjustment in the text (i.e., when we compare estimates of α and β across subsamples).
15
shows the results from dynamic partial adjustment capital structure models (DPACS-Models) for book and market leverage ratios. In this subsection, we focus on the coefficients in front of the recession dummy.17 In DPACSModels, these coefficients can be interpreted as measures of the impact of recessions on target leverage ratios. Our main hypotheses, H-TL 1a and H-TL 1b, suggest that from a theoretical point of view both pro-cyclical (negative coefficient) and countercyclical (positive coefficient) patterns of book and market leverage can be justified. Standard trade-off models without investment variables even predict no effect of business cycle variation on market leverage ratios. The column labeled “All”in Table 3 Panel A summarizes our results for book leverage ratios for the entire sample of firms. In this case, we find a statistically significant, negative coefficient. This implies that overall book leverage seems to behave in a pro-cyclical way — it shrinks during recessions and increases during expansions. For market leverage (Table 3 Panel B), in contrast, we observe strong counter-cyclicality. This effect seems to be driven by the decrease in market valuation (the denominator in the ratios) during recessions. Overall, our results — pro-cyclicality of book leverage and countercyclicality of market leverage — are only partly consistent with dynamic capital structure theories. Pro-cyclical book leverage is broadly speaking consistent with dynamic tradeoff models, with models of time-varying collateral values and capital market driven capital structure models. But counter-cyclical market leverage is inconsistent with these kind of models. Some dynamic tradeoff models that explicitly model business cycle state variables and time-varying agency models predict counter-cyclical market leverage ratios. But, then these versions of dynamics tradeoff models are inconsistent with pro-cyclical book leverage. Other models that are consistent with counter-cyclical market leverage dynamics are hybrid dynamic capital structure models that include pecking order and tradeoff concepts, dynamic agency costs models and market timing driven capital structure models. But, again, these models usually also predict counter-cyclical book leverage. Thus, to this end, we are unable to identify a specific dynamic capital structure theory that is entirely consistent with our empirical results. Another set of hypotheses addresses the influence of financial constraints 17
The results with respect to our standard controls and their influence on leverage ratios are broadly consistent with the existing literature. These coefficients and their variation across samples and leverage definitions are, however, not the main interest of our paper. Thus, for reasons of brevity, we don’t discuss them in detail in the current version of the paper.
16
on the dynamics of capital structure. As in the case of total debt, theory supports both pro-cyclical (H-TL 3a, models of time-varying collateral values and capital constraints in the banking sector) and counter-cyclical (H-TL 3b, dynamic tradeoff models with BC variables, investment and pecking order components and models with time-varying agency costs) patterns. The columns labeled “Unconstrained”and “Constrained”in Table 3 show the appropriate results. Following theoretical predictions and common intuition, our results suggest that business cycle variation affects the capital structure of unconstrained firms differently than the one of financially constrained firms. While we find that book leverage is significantly pro-cyclical for unconstrained firms, we find a positive (counter-cyclicality) but insignificant coefficient for the subset of financially constrained firms. Ignoring for a moment that the coefficient is insignificant, our empirical results support hypothesis H-TL 3b. The more pronounced (but statistically insignificant) counter-cyclicality of financially constrained firms in the case of book leverage implies that these firms seem to be less responsive to issue more debt during expansions than financially unconstrained firms. This finding has several potential explanations. First, financially constrained firms are likely to be more risky, and thus have wider recapitalization bands (i.e., recapitalize less frequently) than less risky unconstrained firms, according to classic dynamic tradeoff theories (e.g. Fischer, Heinkel, and Zechner (1989a)). Second, constrained firms may be forced to issue debt with shorter maturities, which must be rolled over in shorter periods (this aspect will be discussed in more detail later on). However, a high short-term debt ratio creates a particularly significant funding risk if access to debt markets is difficult during recessions. Third, financially constrained firms may experience much sharper freezes of their primary debt markets during recessions than financially unconstrained firms. That is, during recessions debt markets may be characterized by a “flight to quality”, which affects constrained firms’ access to debt markets more severely than unconstrained firms. Thus, constrained firms are reluctant to increase their leverage during expansions. In the case of market leverage, we find a related result: unconstrained firms show a strong counter-cyclical pattern but, in the case of financially constrained firms, we find no effect. Thus, again the patterns are quite different but this specific result is somewhat surprising, as we would have expected to find rather more than less counter-cyclicality in the case of financially constrained firms. Note, however, that the sample size in case of financially constrained firms is much smaller than in the case of unconstrained firms. Thus, the results with respect to financially constrained firms 17
must be interpreted cautiously.18 Our last set of hypotheses with respect to target leverage ratios addresses the influence of a country’s legal origin and its degree of development on the leverage dynamics across the business cycle. The idea is that in common-law countries, owners of publicly traded stocks or bonds are better protected, and thus a larger fraction of firms’ funding comes directly from the capital market rather than from banks. By contrast, in civil law countries and developing countries19 a smaller fraction of corporate funding is obtained via issues of public equity and public corporate debt, and more funding occurs via bank loans. While standard dynamic tradeoff models suggest that firms in common law (civil law) countries exhibit more pro-cyclical (countercyclical) leverage dynamics (H-TL 5a), models in the debt renegotiation literature argue in the opposite direction (H-TL 5b). Columns 4 to 7 of Table 3 summarize the results. The only sub-sample for which we find statistical significance in the case of book leverage is the sample of firms from developed economies. As before we find pro-cyclical leverage dynamics. If we split the sample by legal origin, we unfortunately lose significance. Nevertheless, point estimates of the coefficients point into opposite directions: they imply pro-cyclical patterns for firms from common law countries and counter-cyclical patterns for firms from civil law countries. This last result, although statistically insignificant, would be consistent with dynamic tradeoff theory (Hypothesis H-TL 5a). In the case of market leverage we find consistently significant and positive coefficients implying counter-cyclical dynamics across all four sub-samples. Comparing the relative magnitudes of the coefficient estimates suggests a stronger influence of recessions on firms from common law countries (and firms from developing countries). This result, in contrast to the one for book leverage, seems to be more in support of the debt renegotiation literature.
4.2
Net Leverage
In this section, we interpret cash and marketable securities as negative debt and, accordingly, subtract them from total debt. Thus, the variable of interest in this section is net leverage. In contrast to book leverage, we find a strongly significant and positive coefficient for the recession dummy in the 18
We are in the process of looking into other definitions of financial constraints that would provide is with a larger sample of constrained firms. 19 Note that our current sample of countries shows a very homogenous distribution among common law and civil law countries but covers comparatively few observations from developing countries.
18
case of net book leverage (Table 4 Panel A). For net market leverage (Table 4 Panel B) we find the same pattern, namely strong counter-cyclicality. Thus, in contrast to our hypothesis, H-TL 2, net leverage seems to be countercyclical, in general. Considering the results for total book debt, the effect on net book debt must be driven by the level of cash reserves that seem to be strongly reduced during recessions (pushing net book leverage upward). Thus, firms build up financial slack during expansionary times. Then, during recessions, they use slack to partly fund capital expenditures and/or cash distributions to shareholders. One would expect that this effect is particularly pronounced for financially constrained firms. Columns 2 and 3 of Table 4 compare the results for financially constrained and unconstrained firms. In the case of book values, the results confirm our above conjecture: net book debt ratios are more strongly affected by the business cycle (i.e., more counter-cyclical) for financially constrained than unconstrained firms. The coefficient of interest in Table 4 Panel A is nearly four times as large for the sub-sample of financially constrained firms.20 In the case of market net leverage, we don’t find this result: estimated coefficients are very similar.21 Although we don’t have specific hypotheses for the relationship between business cycle variation, net leverage ratios and legal origin/economic development of a firm’s home country, we included the empirical results in Table 4. The results don’t show a lot of variation across samples: we find consistently significant counter-cyclical dynamics. The only exception are firms from developing countries and book net leverage where the estimated coefficient is negative but insignificant.
4.3
Short-Term Debt vs. Long-Term Debt
In this subsection we wish to gain deeper insights into the dynamics of firms’ financial maturity structures over the business cycle. As discussed in subsection 2.4, informational asymmetries may require firms to choose short maturities. If a recession makes the presence of informational asymmetries more likely, then we should expect debt maturity to become shorter (Hypothesis H-TL 4a). By contrast, if transactions costs of rolling over short-term debt are higher during recessions, and debt maturity is largely 20
For this cross-sample comparison we have to adjust the reported coefficients by dividing them by 1-β (β is an average of the coefficients estimated for recessions and expansions, βrec and βexp ). In this case the adjustment has basically no effect on the comparison. 21 And after controlling for the partial adjustment effect, the counter-cyclicality of unconstrained firms becomes even stronger compared to financially constrained firms.
19
a tradeoff between some benefits of short-term debt, such as a commitment to debt reductions as shown by Dangl and Zechner (2007) or the signaling effect, then we would expect debt maturity to increase in recessions (Hypothesis H-TL 5a). Thus, according to the latter theories, long-term debt ratios should be more pro-cyclical (or less counter-cyclical) than short-term leverage ratios. The opposite prediction is generated by purely informationally driven models of maturity or for the Dangl and Zechner (2007) model with constant transactions costs. Looking at book values (see Table 5 Panel A and Table 6 P anel A), we find no effect of business cycle variation on long-term leverage but significantly pro-cyclical dynamics for short-term leverage ratios. This evidence seems to be consistent with hypothesis H-TL 4b indicating that debt maturity choice reflects a tradeoff between benefits of short-term debt versus the transactions costs of rolling it over. In the case of market ratios, this pattern disappears and both, long-term and short-term, debt ratios behave in a counter-cyclical way. In the previous section we were discussing empirical results regarding the business cycle dynamics of financially constrained firms. One of the previous hypotheses (H-TL 3b) suggests that short-term leverage ratios are more counter-cyclical for financially constrained firms. We find empirical evidence consistent with this specific hypothesis. Table 6 shows that short-term book leverage is pro-cyclical for unconstrained firms and counter-cyclical for constrained firms. The latter effect is not statistically significant, though. In the case of short-term market ratios, we find counter-cyclical patterns for unconstrained firms and no effect for financially constrained firms. To summarize, short-term leverage dynamics are in general very different for financially unconstrained and constrained firms.22 Putting these results together, it seems that constrained firms rely more on issuing short-term debt during recessions to finance their investments while unconstrained firms reduce their short-term debt. The latter effect is likely due to unconstrained firms having higher internally generated cash flows. Thus, they either do not need significant amounts of external funding during recessions or they can also rely on external equity. Another explanation could be that the spread between cost of long term debt and cost of short term debt is larger (smaller) in recessions (expansions) due to timevarying agency costs of borrowing. This effect is reinforced for constrained 22 In the case of long-term leverage we observe this result only for market ratios. In terms of book ratios, long-term capital ratios are unaffected by the business cycle for both, financially constrained and unconstrained, firms.
20
firms. Therefore, during economic downturns, constrained firms substitute short term debt for long term debt and unconstrained firms do the opposite. Finally, we split our sample into firms from developed countries and developing countries on the one hand and common law countries and civil law countries on the other hand. For long-term debt ratios these splits have no effect: we find no business cycle dynamics for book values and countercyclicality for market values across all samples. In the case of short-term leverage, the picture is similar for market values. In the case of book values, we document counter-cyclicality for firms from common law countries and pro-cyclicality for firms from civil law countries. The latter effect is not statistically significant, though. One interpretation of these findings is that the banking sector in common-law countries is less affected by recessions than in civil-law countries. Thus, in common-law countries firms can partly replace expiring long-term public debt or bank-debt with short-term bank debt. In civil-law countries, banks are unwilling or unable to provide more short-term funding during recessions.
5
Empirical Results: Speed of Adjustment
In the presence of transactions costs, firms are usually not at their optimal target leverage ratios at any given point in time. In several dynamic capital structure models capital structure adjustments are lumpy in the sense that firms do not adjust until a boundary is reached, at which point a full adjustment towards the target capital structure occurs (see, e.g. Fischer, Heinkel, and Zechner (1989a). In other dynamic models, partial adjustments also occur, for example when firms choose the financing of new investments such that they move towards their target capital structure (e.g. DeAngelo, DeAngelo, and Whited (2010)). Flannery and Rangan (2006) show that even if firms’ capital structure adjustments are lumpy, dynamic partial adjustment models can capture actual firm behavior quite well. A high speed of adjustment can be interpreted that firms do not allow their actual leverage ratios to wander far from its target before they make adjustments. Thus, an interesting question is whether the business cycle affects the speed of adjustments towards leverage targets. If transactions costs associated with capital structure adjustments are higher during recessions, we should therefore expect the empirical estimates of the speed of adjustment to be lower (see hypothesis H-SOA 1). We will also explore whether the relationship between the business cycle and the speed of adjustment depends on whether firms are financially constrained,
21
whether they are located in developed or in developing markets, whether the results differ between common law and civil law countries (this comparison is directly related to hypothesis H-SOA 2) and whether the results differ between long-term debt and short-term debt ratios. Empirically, we focus our attention on the coefficient of the lagged leverage ratios, which we estimate separately for recessions and expansions. Subtracting each of these coefficients from 1 yields the appropriate speed of adjustment estimates (SOA-estimates) during recessions and expansions, respectively. Economically, these SOA estimates can be translated into halflives of the influence of a shock.23 In the literature there is some controversy about US-based SOA-estimates (see Iliev and Welch (2010) for a summary): Flannery and Rangan (2006) report 34%, Lemmon, Roberts, and Zender (2008) find 25%, Huang and Ritter (2009) document 23%, Fama and French (2002) estimate SOAs within the range of 7 to 18%, and Welch (2004), finally, argues that there is no adjustment. Our SOA-estimates vary considerably across leverage ratio definitions and firm samples but, overall, tend towards the upper boundary of the values reported in the above list. Every single estimate is positive, below 1 and statistically significantly different from zero. Our most important result is that, across all specifications, all measures of leverage and all samples we find very strong evidence that the speed of adjustment estimates are lower during recessions (i.e., coefficients in the regression are higher). These differences are significant with very few exceptions. There is not a single case in which we observe the opposite effect with statistical significance (there are only two cases in which the SOA is faster during recessions). Thus, these results strongly support our main hypothesis (H-SOA 1). Of course, there is quite a bit of dispersion in the differences between SOA-estimates in recessions and expansions across samples and leverage variables. In the next few paragraphs we will discuss some of this variation in more detail. One source of variation is based on the theory of debt renegotiations. According to this theory the difference between the speed of adjustment towards the target leverage ratio in expansions and in contractions should be lower for firms with large proportions of public debt (see hypothesis H-SOA 2). As before, we use the legal origin of a firm as a proxy for the proportion of public debt in its capital structure. The assumption is that firms from common law countries have more public debt on average than firms from civil law countries. 23
A SOA-estimate of x% corresponds to a half-life of log(0.5)/log(1 − x).
22
We find strong evidence for this hypothesis across all leverage definitions. Out of the 8 leverage definitions that we analyze, the difference between the speed of adjustment towards the target leverage ratio in expansions and in contractions is considerably lower for firms from common law countries than for firms from civil law countries in 6 of the 8 cases.24 For example, in the case of total market leverage the speed of adjustment is about 8% slower (this means that capital structure adjustment takes about half a year longer) during recession for firms from civil law countries while it is only about 1% and statistically even insignificantly slower for firms from common law countries. These pronounced differences between firms from common law and civil law countries also support the interpretation that, capital markets in commonlaw countries are more robust during recessions and freeze to a lesser extent, thus enabling firms to consistently adjust their leverage to appropriate target levels. In contrast, in civil law countries, increased transactions costs, or market freezes during recessions seem to significantly slow down the firms’ adjustments to their target ratios. Such an interpretation could also potentially explain differences in SOAestimates across the business cycle between firms from developed and developing markets. We would expect that speed of adjustment estimates decrease more during recessions for firms from developing countries. This expectation is surprisingly not confirmed in the data: differences of SOAestimates across the business cycle are often very similar for firms from developing and developed countries. The only exception are SOA-estimates for long-term leverage: in this case our results suggest that firms from developing countries adjust equally fast during recessions and expansions (the point estimates of the difference are even slightly negative).25 Finally, we look into differences of SOA-estimates across the business cycle for financially constrained and unconstrained firms. Our intuition would be that constrained firms are more affected by business cycle variation and, thus, that speed of adjustment decreases more during recessions for constrained than unconstrained firms. The empirical evidence in this respect is mixed. For book leverage ratios, with the exception of short-term book leverage, we observe comparable differences.26 In the case of market ratios, 24
The only cases, in which the differences are similar, are book leverage and book longterm debt. 25 This is a somewhat puzzling result that we will investigate in more detail in future revisions of the paper. 26 Consistent with intuition the levels of SOA-estimates are consistently lower for financially constrained firms than for unconstrained firms. This indicates that, in general,
23
our intuition is strongly confirmed, as business cycle related differences in SOA-estimates are much more pronounced for constrained firms than for unconstrained firms. The differences are particularly stark in the case of short-term leverage. In the case of short-term market leverage financially constrained firms’ speed of adjustment decreases by 32.6% (or the half-life of a shock increases by more than 1.1 years).
6
Conclusion
There is a large theoretical literature on dynamic capital structure choice. As we have discussed in Section 2.4, these theoretical models do not at all agree on the predicted dynamics of leverage over the business cycle. On the other hand, the existing empirical literature on this issue is very sparse and also not conclusive. This paper tries to bridge this gap and to provide a careful, empirical analysis of leverage ratio dynamics over the business cycle. In this analysis, we look at four different book and market leverage ratios (total debt, net debt, short-term debt and long-term debt) and split our sample of firms by various dimensions: financially constrained vs. unconstrained firms, firms from common law and civil law countries, and firms from developed and developing countries. Table 7 summarizes our empirical results across all these dimensions in a very aggregate and concise way. It also highlights our main conclusions. In the case of target leverage dynamics our main finding is that book leverage behaves in a pro-cyclical but market leverage in a counter-cyclical way. Especially, the counter-cyclicality of market leverage is very robust across leverage definitions and firm samples (i.e., the large majority of symbols in Table 7 Panel A, in rows starting with “ml”, is “+”). The procyclicality of total book leverage extends to short-term book leverage (except for firms from common law countries) but changes to counter-cyclicality in the case of net leverage and no-cyclicality in the case of long-term debt. These empirical patterns are difficult to reconcile with any single capital structure model: the counter-cyclicality of market leverage, for example, is consistent with time-varying agency models; the pro-cyclicality of book leverage is consistent with dynamic tradeoff models. Further, we find that leverage dynamics are different for financially constrained and unconstrained firms: book net leverage is roughly four times more counter-cyclical for financially constrained than unconstrained firms; short-term book leverage is counter-cyclical for constrained firms and proconstrained firms adjust more slowly to their target ratios than unconstrained firms.
24
cyclical for unconstrained firms. We also find that leverage dynamics vary by origin of law: short-term book leverage, for example, is counter-cyclical for firms from common law countries and pro-cyclical for firms from civil law countries. Summarizing our SOA-results, we document two main findings. First, SOA-coefficients are lower (i.e., observed leverage adjusts more slowly to target levels) during recessions (i.e., the large majority of symbols in Table 7 Panel B is “+”). Thus, firms manage their financial structures more actively during expansions. Second, there are pronounced differences between firms from common law countries and from civil law countries. In the former case, the business cycle has a much smaller impact on leverage speeds of adjustment. This could potentially reflect that well developed markets for publicly traded equity and corporate debt are better able to provide external funding during economic downturns.
25
A
Variables Definitions
Long Term Debt (lt debt) refers to all interest bearing financial obligations, excluding amounts due within one year. Short Term Debt (st debt) is the portion of debt payable with one year including current portion of long term debt and sinking fund requirements of preferred stock or debentures. Total Debt (tt debt) is the sum of its long term debt and short term debt. Net Sales are gross sales and other operating revenue less discounts, returns and allowances. Cash & Short Term Investments (cash) represents the sum of cash and short term investments. Market Capitalization (mc) equals to the product of Market Price and Common Shares Outstanding. Assets’ market value (ma) is the market capitalization of the firm plus its total debt. Total Assets (ta) are the sum of total current assets, long term receivables, investment in unconsolidated subsidiaries, other investments, net property plant and equipment and other assets. EBITDA is the earnings of a company before interest expense, income taxes and depreciation. PPE is gross property, plant and equipment less accumulated reserves for depreciation, depletion and amortization. Dividend per Share is the total dividends per share declared during the calendar year for US corporations and fiscal year for non-US firms. Industry Group (WSIC) is a four digit numeric code assigned to the company to represent its industry group. We use the first two digits to classify firms to different industry groups. Industry Median Leverage Ratio (industry median) is the median leverage ratio of an industry to which firms belong. Recession dummy (Rec) equals 1 if a firm’s entire fiscal year overlaps with a recession and 0 otherwise. Countries in our sample are categorized to common-law or civil-law countries based on Treisman (2000) and grouped to developed (advanced) or developing economies according to IMF (2010). We summarize these variables in Table 1 on page 33. Common-law dummy (Common Law) equals 1 if a country is classified as a common-law country and 0 otherwise. Developed dummy (Developed) is 1 if a country is a developed one and 0 otherwise.
26
References Baker, Malcolm, and Jeffrey Wurgler, 2002, Market timing and capital structure, The Journal of Finance 57, 1–32. Bernanke, Ben, and Mark Gertler, 1989, Agency costs, net worth, and business fluctuations, The American Economic Review 79, 14–31. Bernanke, Ben S., and Alan S. Blinder, 1992, The federal funds rate and the channels of monetary transmission, The American Economic Review 82, 901–921. Bhamra, H. S., L.-A. Kuehn, and I. A. Strebulaev, 2009, The aggregate dynamics of capital structure and macroeconomic risk, forthcoming in the Review of Financial Studies. Black, Fischer, and John C. Cox, 1976, Valuing corporate securities: Some effects of bond indenture provisions, The Journal of Finance 31, 351–367. Blanchard, Olivier, Changyong Rhee, and Lawrence Summers, 1993, The stock market, profit, and investment, The Quarterly Journal of Economics 108, 115–136. Calomiris, Charles W., and R. Glenn Hubbard, 1990, Firm heterogeneity, internal finance, and ‘credit rationing’, The Economic Journal 100, 90– 104. Chang, Xin, and Sudipto Dasgupta, 2009, Target behavior and financing: How conclusive is the evidence, The Journal of Finance 64, 1767–1796. Chen, Long, and Xinlei Zhao, 2007, Mechanical mean reversion of leverage ratios, Economics Letters 95, 223–229. Childs, Paul D, David C Mauer, and Steven H Ott, 2005, Interactions of corporate financing and investment decisions: The effect of agency conflicts, Journal of Financial Economics 76, 667–690. Cook, D. O., and T. Tang, 2009, Macroeconomic conditions and capital structure adjustment speed, forthcoming in the Journal of Corporate Finance. Dangl, Thomas, and Josef Zechner, 2004, Credit risk and dynamic capital structure choice, Journal of Financial Intermediation 2, 183–204.
27
, 2007, Debt maturity and the dynamics of leverage, Working paper. De Long, J Bradford, Andrei Shleifer, Lawrence H. Summers, and Robert J. Waldmann, 1990, Noise trader risk in financial markets, The Journal of Political Economy 98, 703–738. DeAngelo, Harry, Linda DeAngelo, and Toni M Whited, 2010, Capital structure dynamics and transitory debt, Journal of Financial Economics 99, 235–261. Drobetz, W., and G. Wanzenried, 2006, What determines the speed of adjustment to the target capital structure?, Applied Financial Economics 16, 941–961. Duffie, Darrell, Nicolae Garleanu, and Lasse Heje Pedersen, 2007, Valuation in over-the-counter markets, Review of Financial Studies 20, 1865–1900. Ericsson, Jan, and Olivier Renault, 2006, Liquidity and credit risk, The Journal of Finance 61, 2219–2250. Fama, E. F., and K. R. French, 2002, Testing trade-off and pecking order predictions about dividends and debt, The Review of Financial Studies 15, 1–33. Fischer, Edwin O., Robert Heinkel, and Josef Zechner, 1989a, Dynamic capital structure choice: Theory and tests, The Journal of Finance 44, 19–40. , 1989b, Dynamic recapitalization policies and the role of call premia and issue discounts, Journal of Financial and Quantitative Analysis 24, 427–446. Fischer, Stanley, and Robert Merton, 1984, Macroeconomics and finance: The role of the stock market, Carnegie-Rochester Conference Series on Public Policy 21, 57–108. Flannery, M. J., and K. P. Rangan, 2006, Partial adjustment toward target capital structures, Journal of Financial Economics 79, 469–506. Gamba, Andrea, and Alexander J Triantis, 2008, The value of financial flexibility, The Journal of Finance 63, 2263–2296. Gertler, Mark, 1992, Financial capacity and output fluctuations in an economy with multi-period financial relationships, The Review of Economics Studies 59, 455–472. 28
Gertner, Robert, and David Scharfstein, 1991, A theory of workouts and the effects of reorganization law, The Journal of Finance 46, 1189–1222. Goldstein, Robert, Nengjiu Ju, and Hayne Leland, 2001, An ebit-based model of dynamic captial structure, The Journal of Business 74, 483– 512. Graham, John R., and Campbell R. Harvey, 2001, The theory and practice of corporate finance: evidence from the field, Journal of Financial Economics 60, 187–243. Greenwald, Bruce C., and Joseph E. Stiglitz, 1992, Financial market imperfections and business cycles, The Quarterly Journal of Economics 108, 77–114. Hackbarth, D., J. Miao, and E. Morellec, 2006, Capital structure, credit risk, and macroeconomic conditions, Journal of Financial Economics 82, 519–550. Hege, Ulrich, and Pierre Mella-Barral, 2005, Repeated dilution of diffusely held debt, The Journal of Business 78, 737–786. Hennessy, Christopher A, and Toni M Whited, 2005, Debt dynamics, The Journal of Finance 60, 1129–1165. , 2007, How costly is external finance?, The Journal of Finance 62, 1705–1745. Hennessy, Christopher A., and Josef Zechner, 2011, A theory of debt market illiquidity and leverage cyclicality, forthcoming in the Review of Financial Studies. Holmstrom, Bengt, and Jean Tirole, 1997, Financial intermediation, loanable funds, and the real sector, The Quarterly Journal of Economics 112, 663–691. Huang, R., and J. R. Ritter, 2009, Testing theories of capital structure and estimating the speed of adjustment, Journal of Financial and Quantitative Analysis 44, 237–271. Iliev, Peter, and Ivo Welch, 2010, Reconciling Estimates of the Speed of Adjustment of Leverage RatiosWorking Paper. IMF, 2010, World Economic Outlook Database,International Monetary Fund http://www.imf.org/external/pubs/ft/weo/2010/02/weodata/groups.htm. 29
Kane, Alex, Alan J. Marcus, and Robert L. McDonald, 1984, How big is the tax advantage of debt?, The Journal of Finance 39, 841–853. , 1985, Debt policy and the rate of return premium to leverage, Journal of Financial and Quantitative Analysis 20, 479–499. Kashyap, Anil K., Jeremy C. Stein, and David W. Wilcox, 1993, Monetary policy and credit conditions: Evidence from the composition of external finance, The American Economic Review 83, 78–98. Kiyotaki, Nobuhiro, and John Moore, 1997, Credit cycles, The Journal of Political Economy 105, 211–248. Korajczyk, R. A., and A. Levy, 2003, Capital structure choice: Macroeconomic conditions and financial constraints, Journal of Financial Economics 68, 75–109. Leary, Mark T, 2009, Bank loan supply, lender choice, and corporate capital structure, The Journal of Finance 64, 1143–1185. Leary, Mark T., and Michael R. Roberts, 2005, Do firms rebalance their capital structures?, The Journal of Finance 60, 2575–2619. Leland, Hayne E, 1994, Corporate debt value, bond covenants, and optimal capital structure, The Journal of Finance 49, 1213–1252. Lemmon, M. L., M. R. Roberts, and J. F. Zender, 2008, Back to the beginning: Persistence and the cross-section of corporate capital structure, The Journal of Finance 63, 1575–1608. Levy, Amnon, and Christopher A. Hennessy, 2007, Why does capital structure choice vary with macroeconomic conditions, Journal of Monetary Economics 54, 1545–1564. Mauer, David C, and Alexander J Triantis, 1994, Interactions of corporate financing and investment decisions: A dynamic framework, The Journal of Finance 49, 1253–1277. Mella-Barral, Pierre, 1999, The dynamics of default and debt reorganization, Review of Financial Studies 12, 535–578. , and William Perraudin, 1997, Strategic debt service, The Journal of Finance 52, 531–556.
30
Mello, Antonio S., and John E. Parsons, 1992, Measuing the agency cost of debt, The Journal of Finance 47, 1887–1904. Merton, Robert C., 1974, On the pricing of corporate debt: the risk structure of interest rates, The Journal of Finance 29, 449–470. Morck, Randall, Andrei Shleifer, and Robert W. Vishny, 1990, Do managerial objectives drive bad acquisitions?, The Journal of Finance 45, 31–48. Morellec, Erwan, 2001, Asset liquidity, capital structure, and secured debt, Journal of Financial Economics 61, 173–206. Moyen, Nathalie, 2004, Investment-cash flow sensitivities: Constrained versus unconstrained firms, The Journal of Finance 69, 2061–2092. , 2007, How big is the debt overhang problem?, Journal of Economic Dynamics and Control 31, 433–472. Pagano, Marco, Fabio Panetta, and Luigi Zingales, 1998, Why do companies go public? an empirical analysis, The Journal of Finance 53, 27–64. Romer, Christina D., David H. Romer, Stephen M. Goldfeld, and Benjamin M. Friedman, 1990, New evidence on the monetary transmission mechanism, Brookings Papers on Economic Activity 1990, 149–213. Shleifer, Andrei, and Robert W. Vishny, 1992, Liquidation values and debt capacity: A market equilibrium approach, The Journal of Finance 47, 1343–1366. Shyam-Sunder, Lakshmi, and Stewart C. Myers, 1999, Testing static tradeoff against pecking order models of capital structure, Journal of Financial Economics 51, 219–244. Stein, Jeremy C., 1996, Rational capital budgeting in an irrational world, The Journal of Business 69, 429–455. Strebulaev, Ilya A., 2007, Do tests of capital structure theory mean what they say?, The Journal of Finance 62, 1747–1787. Titman, Sheridan, and Sergey Tsyplakov, 2007, A dynamic model of optimal capital structure, Review of Finance 11, 401–451. Treisman, Daniel, 2000, The causes of corruption: a cross-national study, Journal of Public Economics 76, 399–457. 31
Welch, I., 2004, Capital structure and stock returns, Journal of Political Economy 112, 106–131. Welch, Ivo, 2011, A critique of recent quantitative and deep-structure modeling in capital structure research and beyond, Working Paper. Yu, Jin, 2010, Captial structure, corporate investment, and macroeconomic conditions, Working Paper.
32
Variable Long Term Debt Short Term Debt
Net Sales Cash & Short Term Investments Market Capitalization Total Assets
EBITDA 33
PPE Dividends per Share
Industry Group Recession
Financial Constraints
Common-law Developed
Table 1: Variables Definitions Definition All interest bearing financial obligations, excluding amounts due within one year. The portion of debt payable with one year including current portion of long term debt and sinking fund requirements of preferred stock or debentures. Gross sales and other operating revenue less discounts, returns and allowances. The sum of cash and short term investments. Market Price ∗ Common Shares Outstanding The sum of total current assets, long term receivables, investment in unconsolidated subsidiaries, other investments, net property plant and equipment and other assets. The earnings of a company before interest expense, income taxes and depreciation. Gross Property, Plant and Equipment less accumulated reserves for depreciation, depletion and amortization. The total dividends per share declared during the calendar year for US corporations and fiscal year for Non-US corporations. A four digit numeric code assigned to the company to represent its industry group. Recession dummy is set to 1 if for a given firm-fiscal-year observation all the past 12 fiscal months are in a recession and 0 otherwise. (1) the firm does not have a net purchase of debt or equity and does not pay dividends within the event window, and (2) the firm’s marekt to book ratio, defined as the sum of the market capitalization of equity and the book value of debt, divided by the book value of assets, at the fiscal year end should be greater than one. See Treisman (2000). See IMF (2010).
Source(Worldscope Code) Worldscope(WC03251) Worldscope(WC03051)
Worldscope(WC01001) Worldscope(WC02001) Worldscope(WC08001) Worldscope(WC02999)
Worldscope(WC18198) Worldscope(WC02501) Worldscope(WC05101)
Worldscope(WC06011) Economic Cycel Research Institute
Worldscope
Treisman (2000) IMF (2010)
Table 2: Summary Statistics Panel A: Entire Sample
Total
Variable book leverage ratio (total debt) book leverage ratio (long term debt) book leverage ratio (short term debt) book leverage ratio (net debt) market leverage ratio (total debt) market leverage ratio (long term debt) market leverage ratio (short term debt) market leverage ratio (net debt) sales market to book profitability tangibility dividend payer
Source: ALL.dta
34
N 250476 250476 250476 250476 250476 250476 250476 250476 250476 250476 250476 250476 250476
mean 0.26 0.15 0.10 0.13 0.30 0.17 0.13 0.18 13.60 2.19 0.01 0.33 0.59
sd 0.19 0.16 0.12 0.31 0.25 0.19 0.17 0.34 3.32 182.50 10.37 0.23 0.49
Panel B: Recession vs. Expansion recession 0
1
Variable book leverage ratio (total debt) book leverage ratio (long term debt) book leverage ratio (short term debt) book leverage ratio (net debt) market leverage ratio (total debt) market leverage ratio (long term debt) market leverage ratio (short term debt) market leverage ratio (net debt) sales market to book profitability tangibility dividend payer book leverage ratio (total debt) book leverage ratio (long term debt) book leverage ratio (short term debt) book leverage ratio (net debt) market leverage ratio (total debt) market leverage ratio (long term debt) market leverage ratio (short term debt) market leverage ratio (net debt) sales market to book profitability tangibility dividend payer
Source: ALL.dta
35
N 227134 227134 227134 227134 227134 227134 227134 227134 227134 227134 227134 227134 227134 23342 23342 23342 23342 23342 23342 23342 23342 23342 23342 23342 23342 23342
mean 0.25 0.16 0.10 0.13 0.29 0.17 0.12 0.18 13.44 2.24 0.01 0.33 0.58 0.27 0.14 0.14 0.14 0.38 0.18 0.20 0.22 15.15 1.75 0.02 0.34 0.68
sd 0.19 0.16 0.12 0.31 0.25 0.19 0.16 0.33 3.26 190.99 10.87 0.23 0.49 0.20 0.14 0.14 0.31 0.28 0.18 0.20 0.41 3.42 49.41 2.14 0.21 0.47
Panel C: Unconstrained vs. Constrained constrained 0
1
Variable book leverage ratio (total debt) book leverage ratio (long term debt) book leverage ratio (short term debt) book leverage ratio (net debt) market leverage ratio (total debt) market leverage ratio (long term debt) market leverage ratio (short term debt) market leverage ratio (net debt) sales market to book profitability tangibility dividend payer book leverage ratio (total debt) book leverage ratio (long term debt) book leverage ratio (short term debt) book leverage ratio (net debt) market leverage ratio (total debt) market leverage ratio (long term debt) market leverage ratio (short term debt) market leverage ratio (net debt) sales market to book profitability tangibility dividend payer
Source: ALL.dta
36
N 241373 241373 241373 241373 241373 241373 241373 241373 241373 241373 241373 241373 241373 9103 9103 9103 9103 9103 9103 9103 9103 9103 9103 9103 9103 9103
mean 0.26 0.16 0.10 0.14 0.31 0.18 0.13 0.19 13.69 2.16 0.02 0.34 0.62 0.15 0.11 0.04 -0.12 0.10 0.07 0.03 -0.02 11.16 3.08 -0.12 0.26 0.00
sd 0.19 0.16 0.12 0.30 0.25 0.19 0.17 0.34 3.30 185.88 10.55 0.23 0.49 0.20 0.17 0.10 0.39 0.16 0.13 0.07 0.22 2.66 13.92 2.59 0.23 0.00
Panel D: Developed vs. Developing developed 0
1
Variable book leverage ratio (total debt) book leverage ratio (long term debt) book leverage ratio (short term debt) book leverage ratio (net debt) market leverage ratio (total debt) market leverage ratio (long term debt) market leverage ratio (short term debt) market leverage ratio (net debt) sales market to book profitability tangibility dividend payer book leverage ratio (total debt) book leverage ratio (long term debt) book leverage ratio (short term debt) book leverage ratio (net debt) market leverage ratio (total debt) market leverage ratio (long term debt) market leverage ratio (short term debt) market leverage ratio (net debt) sales market to book profitability tangibility dividend payer
Source: ALL.dta
37
N 15597 15597 15597 15597 15597 15597 15597 15597 15597 15597 15597 15597 15597 234879 234879 234879 234879 234879 234879 234879 234879 234879 234879 234879 234879 234879
mean 0.30 0.18 0.12 0.24 0.40 0.23 0.17 0.33 14.34 1.51 0.14 0.41 0.65 0.25 0.15 0.10 0.12 0.29 0.17 0.12 0.17 13.55 2.24 0.00 0.33 0.59
sd 0.20 0.16 0.11 0.25 0.29 0.21 0.18 0.33 2.20 27.39 0.62 0.21 0.48 0.19 0.16 0.12 0.31 0.25 0.18 0.16 0.33 3.37 188.32 10.71 0.23 0.49
Panel E: Common law vs. Civil law common law 0
1
Variable book leverage ratio (total debt) book leverage ratio (long term debt) book leverage ratio (short term debt) book leverage ratio (net debt) market leverage ratio (total debt) market leverage ratio (long term debt) market leverage ratio (short term debt) market leverage ratio (net debt) sales market to book profitability tangibility dividend payer book leverage ratio (total debt) book leverage ratio (long term debt) book leverage ratio (short term debt) book leverage ratio (net debt) market leverage ratio (total debt) market leverage ratio (long term debt) market leverage ratio (short term debt) market leverage ratio (net debt) sales market to book profitability tangibility dividend payer
Source: ALL.dta
38
N 109285 109285 109285 109285 109285 109285 109285 109285 109285 109285 109285 109285 109285 141191 141191 141191 141191 141191 141191 141191 141191 141191 141191 141191 141191 141191
mean 0.26 0.12 0.14 0.12 0.34 0.16 0.18 0.19 15.60 1.10 0.08 0.31 0.72 0.25 0.18 0.07 0.13 0.27 0.19 0.08 0.18 12.06 3.04 -0.04 0.35 0.49
sd 0.18 0.12 0.13 0.29 0.26 0.16 0.18 0.36 3.11 10.43 2.01 0.19 0.45 0.20 0.18 0.11 0.32 0.25 0.20 0.14 0.31 2.55 242.89 13.70 0.25 0.50
Table 3: Total Leverage Dynamics - All Countries Panel A. The dependent variable is the total debt to total (book) asset ratio. Independent variables are self-explanatory by their names and lagged one period (year). Standard errors are shown in parentheses. Coefficients that are significantly different from zero at the 1%, 5%, and 10% levels are marked with ∗ ∗ ∗, ∗∗, and ∗, respectively.
Leverage Regression Models
bl(exp) bl(rec) sales market to book 39
profitability tangibility industry median dividend payer recession dummy cons R2 N
All 0.589*** (0.002) 0.654*** (0.004) 0.004*** (0.000) 0.000** (0.000) -0.000** (0.000) 0.015*** (0.002) 0.090*** (0.010) 0.004*** (0.001) -0.002* (0.001) 0.024*** (0.005) 0.356 211776
Unconstrained 0.572*** (0.002) 0.641*** (0.004) 0.004*** (0.000) 0.000*** (0.000) -0.000 (0.000) 0.012*** (0.003) 0.092*** (0.010) 0.002*** (0.001) -0.004*** (0.001) 0.040*** (0.005) 0.341 202673
Constrained 0.909*** (0.007) 0.960*** (0.022) 0.004*** (0.001) -0.000** (0.000) -0.002 (0.001) 0.004 (0.008) 0.095** (0.037) 0.017*** (0.006) 0.002 (0.005) -0.067*** (0.013) 0.798 9103
Developed 0.590*** (0.002) 0.655*** (0.004) 0.005*** (0.000) 0.000** (0.000) -0.000 (0.000) 0.013*** (0.003) 0.088*** (0.010) 0.004*** (0.001) -0.002* (0.001) 0.018*** (0.005) 0.357 199345
Developing 0.574*** (0.008) 0.613*** (0.032) 0.002*** (0.001) 0.000 (0.000) -0.001 (0.001) 0.044*** (0.010) 0.129*** (0.040) -0.001 (0.003) -0.006 (0.010) 0.049*** (0.015) 0.349 12431
Common Law 0.553*** (0.003) 0.610*** (0.009) 0.005*** (0.000) 0.000* (0.000) -0.000 (0.000) 0.021*** (0.003) 0.071*** (0.014) 0.003** (0.001) -0.001 (0.003) 0.032*** (0.007) 0.311 117424
Civil Law 0.652*** (0.003) 0.707*** (0.004) 0.003*** (0.000) 0.000 (0.000) -0.001 (0.001) 0.004 (0.004) 0.108*** (0.013) 0.007*** (0.001) 0.000 (0.001) 0.014* (0.007) 0.440 94352
Test of the Asymmetry in Speed of Adjustment (SOA) between Recessions and Expansions bl(rec) − bl(exp)
0.065*** (0.004)
0.069*** (0.004)
0.051** (0.022)
0.066*** (0.004)
0.040 (0.032)
0.057*** (0.009)
0.055*** (0.004)
Panel B. The dependent variable is the total debt to total (market) asset ratio. Independent variables are self-explanatory by their names and lagged one period (year). Standard errors are shown in parentheses. Coefficients that are significantly different from zero at the 1%, 5%, and 10% levels are marked with ∗ ∗ ∗, ∗∗, and ∗, respectively.
Leverage Regression Models
ml(exp) ml(rec) sales market to book profitability 40 tangibility industry median dividend payer recession dummy cons R-squared N
All 0.597*** (0.002) 0.667*** (0.004) 0.014*** (0.000) 0.000*** (0.000) 0.000 (0.000) 0.019*** (0.003) 0.018*** (0.005) 0.008*** (0.001) 0.029*** (0.002) -0.077*** (0.006) 0.391 211776
Unconstrained 0.591*** (0.002) 0.661*** (0.004) 0.014*** (0.000) 0.000*** (0.000) 0.001** (0.000) 0.017*** (0.003) 0.022*** (0.005) 0.005*** (0.001) 0.028*** (0.002) -0.073*** (0.006) 0.387 202673
Constrained 0.598*** (0.008) 0.781*** (0.029) 0.005*** (0.001) 0.000 (0.000) -0.001 (0.001) -0.001 (0.009) -0.057*** (0.017) 0.010 (0.006) 0.000 (0.005) -0.022** (0.011) 0.596 9103
Developed 0.601*** (0.002) 0.672*** (0.004) 0.016*** (0.000) -0.000 (0.000) -0.000 (0.000) 0.016*** (0.003) -0.014*** (0.005) 0.008*** (0.001) 0.028*** (0.002) -0.100*** (0.007) 0.396 199345
Developing 0.489*** (0.009) 0.552*** (0.029) 0.004*** (0.001) 0.000*** (0.000) -0.003 (0.002) 0.087*** (0.016) 0.497*** (0.022) -0.002 (0.005) 0.048*** (0.016) -0.021 (0.019) 0.363 12431
Common Law 0.582*** (0.003) 0.590*** (0.009) 0.016*** (0.001) -0.000 (0.000) -0.001 (0.000) 0.023*** (0.004) 0.036*** (0.007) 0.011*** (0.002) 0.039*** (0.003) -0.096*** (0.007) 0.347 117424
Civil Law 0.618*** (0.003) 0.696*** (0.005) 0.010*** (0.001) 0.000*** (0.000) -0.005** (0.002) 0.007 (0.005) -0.009 (0.007) 0.006*** (0.001) 0.028*** (0.002) -0.034*** (0.010) 0.450 94352
Test of the Asymmetry in Speed of Adjustment (SOA) between Recessions and Expansions ml(rec) − ml(exp)
0.070*** (0.004)
0.071*** (0.004)
0.183*** (0.029)
0.071*** (0.004)
0.063** (0.029)
0.009 (0.009)
0.078*** (0.004)
Table 4: Net Leverage Dynamics - All Countries Panel A. The dependent variable is the net debt to total (book) asset ratio. Independent variables are self-explanatory by their names and lagged one period (year). Standard errors are shown in parentheses. Coefficients that are significantly different from zero at the 1%, 5%, and 10% levels are marked with ∗ ∗ ∗, ∗∗, and ∗, respectively.
Leverage Regression Models
bl-net(exp) bl-net(rec) sales market to book 41
profitability tangibility industry median dividend payer recession dummy cons R2 N
All 0.517*** (0.002) 0.566*** (0.005) 0.006*** (0.000) 0.000 (0.000) -0.001 (0.000) 0.020*** (0.004) 0.135*** (0.012) 0.001 (0.001) 0.012*** (0.001) -0.031*** (0.007) 0.294 211776
Unconstrained 0.507*** (0.002) 0.565*** (0.004) 0.005*** (0.000) 0.000*** (0.000) -0.000 (0.000) 0.019*** (0.004) 0.127*** (0.012) -0.000 (0.001) 0.010*** (0.001) -0.008 (0.007) 0.288 202673
Constrained 0.495*** (0.014) 0.555*** (0.051) 0.006* (0.004) 0.001** (0.001) -0.003 (0.005) -0.028 (0.039) 0.273** (0.111) 0.016 (0.027) 0.039** (0.019) -0.169*** (0.049) 0.252 9103
Developed 0.516*** (0.002) 0.565*** (0.005) 0.007*** (0.001) 0.000 (0.000) -0.001 (0.000) 0.017*** (0.004) 0.133*** (0.012) 0.002 (0.001) 0.012*** (0.001) -0.049*** (0.008) 0.293 199345
Developing 0.529*** (0.009) 0.593*** (0.036) 0.001 (0.001) 0.000 (0.000) -0.002 (0.002) 0.054*** (0.013) 0.238*** (0.041) -0.006 (0.004) -0.009 (0.010) 0.049*** (0.016) 0.311 12431
Common Law 0.481*** (0.003) 0.493*** (0.009) 0.008*** (0.001) 0.000 (0.000) -0.001 (0.000) 0.014** (0.006) 0.108*** (0.017) 0.003 (0.002) 0.013*** (0.003) -0.031*** (0.010) 0.250 117424
Civil Law 0.588*** (0.003) 0.638*** (0.005) 0.003*** (0.001) 0.000 (0.000) -0.003 (0.002) 0.027*** (0.006) 0.163*** (0.015) 0.005*** (0.001) 0.011*** (0.001) -0.023** (0.011) 0.384 94352
Test of the Asymmetry in Speed of Adjustment (SOA) between Recessions and Expansions bl-net(rec) − bl-net(exp)
0.049*** (0.004)
0.058*** (0.004)
0.060 (0.050)
0.050*** (0.004)
0.063* (0.036)
0.012 (0.009)
0.050*** (0.004)
Panel B. The dependent variable is the net debt to total (market) asset ratio. Independent variables are self-explanatory by their names and lagged one period (year). Standard errors are shown in parentheses. Coefficients that are significantly different from zero at the 1%, 5%, and 10% levels are marked with ∗ ∗ ∗, ∗∗, and ∗, respectively.
Leverage Regression Models
ml-net(exp) ml-net(rec) sales market to book profitability 42 tangibility industry median dividend payer recession dummy cons R2 N
All 0.550*** (0.002) 0.668*** (0.004) 0.011*** (0.001) 0.000*** (0.000) 0.000 (0.000) 0.037*** (0.005) 0.092*** (0.007) 0.008*** (0.001) 0.014*** (0.002) -0.095*** (0.008) 0.330 211776
Unconstrained 0.552*** (0.002) 0.668*** (0.004) 0.012*** (0.001) 0.000*** (0.000) 0.001 (0.001) 0.036*** (0.005) 0.089*** (0.007) 0.006*** (0.001) 0.014*** (0.002) -0.094*** (0.009) 0.330 202673
Constrained 0.380*** (0.011) 0.506*** (0.041) 0.001 (0.002) -0.000 (0.000) -0.000 (0.003) 0.040* (0.021) 0.143*** (0.044) 0.021 (0.015) 0.012 (0.010) -0.051** (0.025) 0.239 9103
Developed 0.552*** (0.002) 0.670*** (0.004) 0.013*** (0.001) -0.000* (0.000) -0.001* (0.001) 0.033*** (0.005) 0.068*** (0.007) 0.008*** (0.001) 0.014*** (0.002) -0.115*** (0.009) 0.331 199345
Developing 0.479*** (0.010) 0.583*** (0.029) 0.004*** (0.001) 0.000*** (0.000) -0.004 (0.003) 0.115*** (0.019) 0.491*** (0.028) -0.004 (0.006) 0.031** (0.015) -0.016 (0.022) 0.321 12431
Common Law 0.532*** (0.003) 0.585*** (0.009) 0.014*** (0.001) -0.000 (0.000) -0.001 (0.001) 0.030*** (0.006) 0.085*** (0.010) 0.015*** (0.002) 0.022*** (0.003) -0.102*** (0.010) 0.288 117424
Civil Law 0.571*** (0.003) 0.698*** (0.005) 0.007*** (0.001) 0.000*** (0.000) -0.009*** (0.003) 0.051*** (0.008) 0.097*** (0.010) 0.005** (0.002) 0.011*** (0.002) -0.067*** (0.015) 0.378 94352
Test of the Asymmetry in Speed of Adjustment (SOA) between Recessions and Expansions ml-net(rec) − ml-net(exp)
0.118*** (0.004)
0.117*** (0.004)
0.126*** (0.042)
0.118*** (0.004)
0.104*** (0.029)
0.053*** (0.009)
0.126*** (0.004)
Table 5: Long Term Leverage Dynamics - All Countries Panel A. The dependent variable is the long term debt to total (book) asset ratio. Independent variables are self-explanatory by their names and lagged one period (year). Standard errors are shown in parentheses. Coefficients that are significantly different from zero at the 1%, 5%, and 10% levels are marked with ∗ ∗ ∗, ∗∗, and ∗, respectively.
Leverage Regression Models
bl-lt(exp) bl-lt(rec) sales market to book 43
profitability tangibility industry median dividend payer recession dummy cons R2 N
All 0.520*** (0.002) 0.552*** (0.005) 0.004*** (0.000) -0.000 (0.000) -0.000** (0.000) 0.020*** (0.002) 0.102*** (0.011) -0.000 (0.001) 0.001 (0.001) 0.001 (0.005) 0.272 211776
Unconstrained 0.507*** (0.002) 0.542*** (0.005) 0.004*** (0.000) -0.000 (0.000) -0.000 (0.000) 0.017*** (0.002) 0.105*** (0.011) -0.001 (0.001) 0.000 (0.001) 0.010** (0.005) 0.260 202673
Constrained 0.748*** (0.010) 0.775*** (0.031) 0.003** (0.001) -0.000** (0.000) -0.001 (0.002) 0.026** (0.011) 0.018 (0.062) 0.022*** (0.008) 0.000 (0.006) -0.025 (0.017) 0.583 9103
Developed 0.522*** (0.002) 0.555*** (0.005) 0.004*** (0.000) -0.000** (0.000) -0.001*** (0.000) 0.017*** (0.002) 0.101*** (0.011) 0.000 (0.001) 0.001 (0.001) -0.003 (0.005) 0.274 199345
Developing 0.469*** (0.009) 0.434*** (0.043) 0.004*** (0.001) -0.000 (0.000) -0.001 (0.001) 0.052*** (0.009) 0.188*** (0.050) -0.006** (0.003) 0.003 (0.008) 0.005 (0.013) 0.245 12431
Common Law 0.511*** (0.003) 0.541*** (0.009) 0.004*** (0.000) -0.000** (0.000) -0.001** (0.000) 0.020*** (0.003) 0.030** (0.015) -0.002* (0.001) 0.001 (0.002) 0.032*** (0.006) 0.260 117424
Civil Law 0.536*** (0.003) 0.565*** (0.005) 0.004*** (0.000) 0.000 (0.000) -0.004*** (0.001) 0.022*** (0.003) 0.253*** (0.015) 0.002*** (0.001) 0.001 (0.001) -0.042*** (0.007) 0.302 94352
Test of the Asymmetry in Speed of Adjustment (SOA) between Recessions and Expansions bl-lt(rec) − bl-lt(exp)
0.033*** (0.005)
0.035*** (0.005)
0.027 (0.031)
0.033*** (0.005)
-0.036 (0.042)
0.030*** (0.009)
0.029*** (0.005)
Panel B. The dependent variable is the long term debt to total (market) asset ratio. Independent variables are self-explanatory by their names and lagged one period (year). Standard errors are shown in parentheses. Coefficients that are significantly different from zero at the 1%, 5%, and 10% levels are marked with ∗ ∗ ∗, ∗∗, and ∗, respectively.
Leverage Regression Models
ml-lt(exp) ml-lt(rec) sales market to book profitability 44 tangibility industry median dividend payer recession dummy cons R2 N
All 0.508*** (0.002) 0.569*** (0.005) 0.011*** (0.000) 0.000 (0.000) -0.000* (0.000) 0.029*** (0.003) 0.100*** (0.008) 0.000 (0.001) 0.014*** (0.001) -0.086*** (0.005) 0.278 211776
Unconstrained 0.504*** (0.002) 0.563*** (0.005) 0.011*** (0.000) -0.000 (0.000) -0.001** (0.000) 0.028*** (0.003) 0.104*** (0.008) -0.001 (0.001) 0.014*** (0.001) -0.085*** (0.006) 0.274 202673
Constrained 0.543*** (0.009) 0.717*** (0.032) 0.004*** (0.001) 0.000 (0.000) -0.000 (0.001) 0.009 (0.009) -0.072** (0.033) 0.014** (0.006) -0.003 (0.005) -0.014 (0.012) 0.496 9103
Developed 0.512*** (0.002) 0.572*** (0.005) 0.012*** (0.000) -0.000** (0.000) -0.001*** (0.000) 0.026*** (0.003) 0.078*** (0.008) 0.001 (0.001) 0.014*** (0.001) -0.102*** (0.006) 0.282 199345
Developing 0.422*** (0.010) 0.419*** (0.037) 0.008*** (0.001) 0.000 (0.000) -0.003* (0.002) 0.080*** (0.013) 0.698*** (0.042) -0.008** (0.004) 0.045*** (0.012) -0.089*** (0.017) 0.250 12431
Common Law 0.501*** (0.003) 0.541*** (0.009) 0.013*** (0.001) -0.000* (0.000) -0.001*** (0.000) 0.023*** (0.004) 0.108*** (0.010) -0.000 (0.001) 0.021*** (0.003) -0.084*** (0.007) 0.263 117424
Civil Law 0.518*** (0.003) 0.586*** (0.005) 0.008*** (0.000) 0.000** (0.000) -0.006*** (0.002) 0.039*** (0.004) 0.102*** (0.012) 0.001 (0.001) 0.012*** (0.001) -0.072*** (0.008) 0.308 94352
Test of the Asymmetry in Speed of Adjustment (SOA) between Recessions and Expansions ml-lt(rec) − ml-lt(exp)
0.060*** (0.005)
0.059*** (0.005)
0.174** (0.032)
0.060*** (0.005)
-0.004 (0.037)
0.039*** (0.009)
0.068*** (0.005)
Table 6: Short Term Leverage Dynamics - All Countries Panel A. The dependent variable is the short term debt to total (book) asset ratio. Independent variables are self-explanatory by their names and lagged one period (year). Standard errors are shown in parentheses. Coefficients that are significantly different from zero at the 1%, 5%, and 10% levels are marked with ∗ ∗ ∗, ∗∗, and ∗, respectively.
Leverage Regression Models
bl-st(exp) bl-st(rec) sales market to book 45
profitability tangibility industry median dividend payer recession dummy cons R2 N
All 0.381*** (0.002) 0.486*** (0.005) 0.002*** (0.000) 0.000** (0.000) -0.000** (0.000) 0.009*** (0.002) 0.106*** (0.014) -0.006*** (0.001) -0.002** (0.001) 0.034*** (0.004) 0.149 211776
Unconstrained 0.373*** (0.002) 0.480*** (0.005) 0.001*** (0.000) 0.000*** (0.000) -0.000* (0.000) 0.006*** (0.002) 0.105*** (0.015) -0.007*** (0.001) -0.002*** (0.001) 0.041*** (0.004) 0.146 202673
Constrained 0.521*** (0.015) 0.714*** (0.066) 0.002* (0.001) 0.000 (0.000) -0.001 (0.001) 0.013 (0.009) 0.154* (0.080) -0.003 (0.006) 0.003 (0.005) -0.011 (0.011) 0.221 9103
Developed 0.382*** (0.002) 0.487*** (0.005) 0.002*** (0.000) 0.000** (0.000) -0.000 (0.000) 0.009*** (0.002) 0.101*** (0.015) -0.006*** (0.001) -0.002** (0.001) 0.030*** (0.004) 0.150 199345
Developing 0.361*** (0.010) 0.462*** (0.043) 0.000 (0.001) 0.000 (0.000) -0.001 (0.001) 0.007 (0.008) 0.200*** (0.064) -0.006*** (0.002) -0.004 (0.007) 0.058*** (0.010) 0.136 12431
Common Law 0.297*** (0.003) 0.345*** (0.012) 0.001*** (0.000) 0.000** (0.000) -0.000 (0.000) 0.013*** (0.003) 0.151*** (0.019) -0.003*** (0.001) 0.004** (0.002) 0.025*** (0.004) 0.081 117424
Civil Law 0.482*** (0.003) 0.573*** (0.005) 0.002*** (0.000) 0.000 (0.000) -0.006*** (0.001) 0.002 (0.003) 0.049** (0.021) -0.006*** (0.001) -0.001 (0.001) 0.045*** (0.006) 0.254 94352
Test of the Asymmetry in Speed of Adjustment (SOA) between Recessions and Expansions bl-st(rec) − bl-st(exp)
0.106*** (0.005)
0.107*** (0.005)
0.192*** (0.066)
0.105*** (0.005)
0.101** (0.044)
0.048** (0.012)
0.090*** (0.005)
Panel B. The dependent variable is the short term debt to total (market) asset ratio. Independent variables are self-explanatory by their names and lagged one period (year). Standard errors are shown in parentheses. Coefficients that are significantly different from zero at the 1%, 5%, and 10% levels are marked with ∗ ∗ ∗, ∗∗, and ∗, respectively.
Leverage Regression Models
ml-st(exp) ml-st(rec) sales market to book profitability 46 tangibility industry median dividend payer recession dummy cons R2 N
All 0.430*** (0.002) 0.551*** (0.004) 0.006*** (0.000) 0.000** (0.000) 0.000 (0.000) 0.013*** (0.003) 0.141*** (0.009) -0.007*** (0.001) 0.011*** (0.001) -0.018*** (0.005) 0.210 211776
Unconstrained 0.426*** (0.002) 0.547*** (0.005) 0.006*** (0.000) 0.000*** (0.000) 0.000 (0.000) 0.012*** (0.003) 0.143*** (0.010) -0.009*** (0.001) 0.011*** (0.001) -0.016*** (0.005) 0.208 202673
Constrained 0.352*** (0.012) 0.678*** (0.070) 0.002** (0.001) -0.000 (0.000) -0.001 (0.001) 0.006 (0.006) 0.058* (0.033) -0.001 (0.005) 0.002 (0.003) -0.006 (0.008) 0.187 9103
Developed 0.432*** (0.002) 0.554*** (0.004) 0.008*** (0.000) -0.000*** (0.000) -0.001*** (0.000) 0.012*** (0.003) 0.109*** (0.010) -0.007*** (0.001) 0.011*** (0.001) -0.038*** (0.005) 0.212 199345
Developing 0.385*** (0.010) 0.496*** (0.034) -0.001 (0.001) 0.000*** (0.000) -0.000 (0.002) 0.026** (0.012) 0.532*** (0.039) -0.009** (0.004) 0.017* (0.010) 0.065*** (0.014) 0.195 12431
Common Law 0.362*** (0.003) 0.409*** (0.012) 0.007*** (0.000) -0.000** (0.000) -0.001** (0.000) 0.015*** (0.003) 0.184*** (0.014) -0.003** (0.001) 0.013*** (0.002) -0.038*** (0.005) 0.121 117424
Civil Law 0.492*** (0.003) 0.601*** (0.005) 0.005*** (0.001) 0.000*** (0.000) -0.008*** (0.002) 0.004 (0.005) 0.072*** (0.013) -0.008*** (0.001) 0.014*** (0.001) 0.010 (0.009) 0.303 94352
Test of the Asymmetry in Speed of Adjustment (SOA) between Recessions and Expansions ml-st(rec) − ml-st(exp)
0.121*** (0.004)
0.121*** (0.004)
0.326*** (0.070)
0.122*** (0.004)
0.111*** (0.034)
0.047*** (0.012)
0.109*** (0.005)
Table 7: Summary of Empirical Findings Panel A. Cyclicality of Target Leverage Ratios. In the table a “+” implies a positive coefficient of REC dummy and hence a counter-cyclical pattern; a “− ”implies a pro-cyclical pattern; and a “0 ” implies a non-cyclical pattern. ± is defined based on a 10% significance level.
Leverage Regression Models
47
bl ml bl-net ml-net bl-lt ml-lt bl-st ml-st
All − + + + 0 + − +
Unconstrained − + + + 0 + − +
Constrained 0 0 + 0 0 0 0 0
Developed − + + + 0 + − +
Developing 0 + 0 + 0 + 0 +
Common Law 0 + + + 0 + + +
Civil Law 0 + + + 0 + 0 +
Panel B. Asymmetry in Speed of Adjustment (SOA) between Recessions and Expansions. In the table a “+” implies a positive difference; a “− ”implies a negative difference; and a “0 ” implies no difference. ± is defined based on a 10% significance level.
Leverage Regression Models
bl ml bl-net ml-net bl-lt ml-lt bl-st ml-st
All + + + + + + + +
Unconstrained + + + + + + + +
Constrained + + 0 + 0 + + +
Developed + + + + + + + +
Developing 0 + 0 + 0 0 + +
Common Law + 0 + + + + + +
Civil Law + + + + + + + +
