Heterogeneity in the Speed of Capital Structure Adjustment across Countries and over the Business Cycle

• Published date: 01 November 2015
• DOI: http://doi.org/10.1111/eufm.12048
• Date: 01 November 2015

Heterogeneity in the Speed of Capital

Structure Adjustment across Countries

and over the Business Cycle

Wolfgang Drobetz

Hamburg Business School, University of Hamburg, Von‐Melle‐Park 5, 20146 Hamburg, Germany

E-mail: wolfgang.drobetz@wiso.uni-hamburg.de

Dirk C. Schilling

HDI‐Gerling Industrial Insurance Company, 10 Fenchurch Street, London EC3M 3BE, UK

E-mail: dirk.c.schilling@gmail.com

Henning Schröder

Hamburg Business School, University of Hamburg, Von‐Melle‐Park 5, 20146 Hamburg, Germany

E-mail: henning.schroeder@wiso.uni-hamburg.de

Abstract

This study analyses the heterogeneity in the speed of capital structure adjustment.

Using a doubly‐censored Tobit estimator that accounts for mechanical mean

reversion in leverage ratios, the speed of adjustment is 25% per year in a large

international sample, supporting the economic relevance of the trade‐off theory.

Differences in the adjustment speed across financial systems are attributable to

differences in the costs of adjustment. Macroeconomic and micro‐level supply‐side

constraints also affect the dynamics of leverage. Firms adjust more slowly during

recessions, and the business cycle effect on adjustment speed is most pronounced

for financially constrained firms in market‐based countries.

Keywords: capital structure, speed of ad justment, institutional arrangem ents,

business cycle, dynamic panel methods

JEL classification: G30, G32

1. Introduction

If firms pursue target leverage ratios, how quickly do they adjust back to their target

capital structure subsequent to leverage shocks? How do institutional differences across

We thank two anonymous referees, Manuel Rocha Armada, Gonul Colak, Ettore Croci,

John Doukas (the editor), Yilmaz Guney, Ottorino Morresi, and Tatjana Puhan as well as

participants of the 2012 Financial Management Association (FMA) European Conference in

Istanbul, the 2013 Midwest Finance Association (MFA) meeting in Chicago, and the 2014

European Financial Management (EFMA) Annual Meeting in Rome for helpful comments.

Correspondence: Henning Schröder.

European Financial Management, Vol. 21, No. 5, 2015, 936–973

doi: 10.1111/eufm.12048

© 2014 John Wiley & Sons Ltd

countries and financial systems affect the speed of adjustment? What impact do

macroeconomic and micro‐level supply‐side constraints have on the speed of adjustment?

And how does the interaction between institutional characteristics, macroeconomic

conditions, and financial constraints affect the dynamics of leverage?

There is still no consensus in the literature on the question how fast firms are able to

adjust back to their target capital structure subsequent to leverage shocks. Huang and

Ritter (2009, p. 239) call this question ‘perhaps the most important issue in capital

structure research.’An estimate of the speed of adjustment can help to sort out major

theories that explain the dynamics of capital structure. A positive speed of adjustment may

be interpreted as evidence for the existence of a target leverage ratio, or more generally, a

dynamic trade‐off model of capital structure. For example, Fischer et al.’s (1989) dynamic

trade‐off model shows that even small adjustment costs can lead to large swings in firms’

capital structures. While any variant of the dynamic trade‐off model with low or moderate

adjustment costs implies a positive adjustment speed, the pecking order theory predicts no

measurable adjustment (Fama and French, 2002). Instead, there is no target leverage ratio,

and leverage changes according to the financing deficit (Myers and Majluf, 1984).

Market‐timing theories even support a negative speed of adjustment. If firms respond to

increasing stock prices by issuing equity, the measured adjustment speed will be lower

than zero (Baker and Wurgler, 2002; Dittmar and Thakor, 2007).

Time‐varying estimates for the speed of capital structure adjustment may contribute to

our understanding about leverage dynamics in an even broader context. In an influential

study, Lemmon et al. (2008) report that corporate leverage is stable and driven mostly by

time‐invariant determinants. Their results are consistent with the unexpectedly low

estimates for (average) adjustment speed in most empirical studies (see Section 2 for a

review). In contrast, DeAngelo and Roll (2013) find that many firms have high and low

leverage at different times and that capital structure stability is the exception. Significant

variation in the speed of adjustment estimates and in the coefficients measuring the impact

of firm‐level characteristics on target leverage over the business cycle would lend support

to these latter findings for instability of corporate leverage.

Adjustment speed depends on (i) the costs of deviating from the target capital structure

and (ii) the costs of adjusting back to the target. Financial managers must assess the trade‐

off between the costs of being off the target leverage ratio and the costs of adjustment. On

the one hand, the financial status of a firm, such as the degree of target deviation or the

magnitude of the financing deficit, has an impact on the speed of adjustment (Faulkender

et al., 2012). On the other hand, both the costs of deviating from the target leverage ratio

and the costs of adjustment are affected by a firm’s institutional, legal, and financial

environment (Antoniou et al., 2008; Öztekin and Flannery, 2012). Finally, macroeco-

nomic conditions may have an impact on firms’time‐varying abilities to readjust

subsequent to a leverage shock, as recession periods are often accompanied by a shortage

of capital supply (Cook and Tang, 2010; Halling et al., 2012). Moreover, during

recessions most of the traditional capital structure variables experience significant shocks,

thus changes in the macroeconomic state will affect leverage dynamics. Overall, firm‐

level, country‐level, and macroeconomic factors are likely to be responsible for the

observed heterogeneity in the speed of adjustment.

In our study, we use a comprehensive sample of firms from the G7 countries and

explore the heterogeneity in the speed of adjustment in two ways. First, we make cross‐

country comparisons to determine whether there are differences between bank‐based and

market‐based financial systems with respect to adjustment speed. For example, firms in

© 2014 John Wiley & Sons Ltd

Heterogeneity in the Speed of Capital Structure Adjustment 937

countries with a bank‐oriented financial system tend to suffer from less liquid capital

markets, making it more expensive for them to issue new or to retire outstanding securities

and to rebalance after a leverage shock. Second, we contribute to the literature by

comparing the speed of adjustment in different macroeconomic states. On the one hand,

adverse selection costs vary over the business cycle, implying that issuing (or retiring)

securities becomes more expensive and that external relative to internal financing costs

increase during economic downturns. On the other hand, the business cycle may influence

the aggregate supply of capital, thereby affecting financing choices at the economy level.

To the extent that firms depend on external financing to adjust their capital structure, these

effects will increase firms’rebalancing costs and slow down their speed of adjustment

particularly during financial crisis episodes.

In addition to measuring heterogeneity in the speed of adjustment, we also provide a

methodological contribution. By imposing a large set of different dynamic panel

estimators on a regime‐switching partial adjustment model for international data, our

results can be interpreted as an out‐of‐sample test given that these estimators have been

tested mostly on US data. To mitigate the biases inherent in virtually all estimators for the

speed of adjustment (Chang and Dasgupta, 2009; Iliev and Welch, 2010), we employ a

new estimator for dynamic panel models introduced by Elsas and Florsysiak (2014) and

compare it to the more standard estimators used in other recent studies. This fractional

dependent variables estimator (DPF‐estimator) exhibits the smallest bias in their US

sample and delivers adjustment speed estimates that support the trade‐off theory of capital

structure. Comparing the estimates from all commonly used dynamic panel estimators, we

find that that the mean speed of adjustment is closest in magnitude to the DPF‐estimator.

The empirical results enhance our understanding of firms’capital structure dynamics.

Based on our full sample of firms from the G7 countries and estimating the regime‐

switching partial adjustment model using the DPF‐estimator, the estimated speed of

adjustment in our regime‐switching partial adjustment model is approximately 25% per

year. The implied half‐life of the average shock of roughly 2.5 years supports the

economic relevance of the trade‐off theory. As expected, the speed of adjustment is

significantly faster in market‐based countries than in bank‐based countries. The

differences in adjustment speed across countries are attributable to differences in both

the costs of deviation from target and the costs of adjustment back to the target, although

the latter effect seems to be much stronger. Furthermore, the macroeconomic environment

has an impact on the speed of adjustment. Firms adjust more slowly during bad

macroeconomic states, and the business cycle effect is most pronounced for financially

constrained firms in countries with a market‐based financial system.

The remainder of our study is as follows: Section 2 provides a brief literature overview.

Section 2 discusses the econometric problems involved in estimating adjustment speeds

in the framework of dynamic panel models. Section 4 describes the data. Section 5

compares the speed of adjustment across countries and explores the influence of different

institutional arrangements. Section 6 analyses the heterogeneity in the speed of

adjustment over the business cycle. Finally, Section 7 concludes and provides an outlook

for future research.

2. Review of the Literature

Modigliani and Miller (1958) conclude that a company’s capital structure is irrelevant for

its valuation. Their original framework is very restrictive and implies no adjustment to any

© 2014 John Wiley & Sons Ltd

938 Wolfgang Drobetz, Dirk C. Schilling and Henning Schröder