Sie sind auf Seite 1von 55

Firms historical market valuation and capital structure

Evidence from firms on FTSE Eurofirst 300

Master thesis

Zigu Chen

Student Number: 0430935

Supervisor: Prof. Dr. Joost Driessen

September 2007

MSc Business Economics in Finance

Faculty of Economics and Business

University of Amsterdam
Abstract

The recent research shows that firms capital structure can be explained by the

historical market valuation. With the comparison test and the integrated test, the

market timing measure, which is effective in determining the dynamic change of the

capital structure by the previous empirical evidence, turns out to be insignificant.

Instead, the stock return measure, which is an alternative proxy of the historical

market valuation, proofs to have large influence on the changes in firms capital

structure both in the short term and the long term. The stock return measure, therefore,

is a better determinant for firms dynamic capital structure.

Keywords: market timing measure, stock return measure, annual effect, persistence
effect

2
Preface

How firms shape their capital structure has always been a hot topic in the field of

corporate finance. Over the decades, major and minor theories on capital structure are

proposed, trying to undo this mystery. The frequent and clustering corporate issuing

activities in recent years even complicate our understanding on firms choices

between debt and equity. The newly developed corporate theories on firms historical

market valuation attracts my attention to look at firms capital structure from a

historical perspective. The idea is quite new that it places firms motive on issuing

activities within a dynamic time-series framework. As indicated, the cumulative

market valuation could affect changes in firms capital structure in the following

period. By examining the efficiency of the two market valuation measures, I hope to

be able to present what kind of measure can be used as the estimation on the future

corporate pie.

Since the start of this thesis, I have received valuable advice, comments and

suggestions from a variety of people. I would like to take the opportunity to mention

some of them who played an important role. Without them this thesis could be

successfully completed.

First of all I would like to thank Joost Driessen and Tse-Chun Lin who are my

supervisors to guide the research. The internship at Movir/ING provides me a great

opportunity to have an additional insight into the business world, from which I would

be grateful to Peter de Bruijne. Finally, but not least important, I would like to thank

my mother and father, for their constant support and encouragement which I feel like

3
an enormous help.

Amsterdam

September 2007

Zigu Chen

4
Content

Abstract ..........................................................................................................................2
Preface............................................................................................................................3
Content...........................................................................................................................5
1. Introduction............................................................................................................6
2. Theories on capital structure..................................................................................9
2.1 Irrelevance theorem (Modigliani-Miller theorem)...........................................9
2.2 The trade-off theory .......................................................................................10
2.3 Pecking order theory ......................................................................................12
2.4 The managerial entrenchment theory.............................................................13
3. Current debate: theories on firms historical market valuation............................14
3.1 Market timing theory .....................................................................................14
3.1.1 Two versions: adverse selection versus mispricing ............................15
3.2 Historical stock return....................................................................................16
4. Variable construction ...........................................................................................17
4.1 Market timing measure ..................................................................................17
4.2 Stock return measure......................................................................................19
4.3 Profitability ....................................................................................................22
4.4 Asset tangibility .............................................................................................23
4.5 Firm size.........................................................................................................23
5. Data and summary statistics.................................................................................25
6. Comparison test ...................................................................................................28
6.1 Annual market timing effect. .........................................................................28
6.2 Annual stock return effect ..............................................................................31
7. Integrated test.......................................................................................................38
8. Persistence test .....................................................................................................42
9. Discussion ............................................................................................................46
9.1 The role of stock return..................................................................................47
9.2 The European evidence..................................................................................48
9.3 Impact by country-specific factors.................................................................49
9.4 Future improvement.......................................................................................49
10. Conclusion ...........................................................................................................50
References....................................................................................................................51
Appendix: Regression with country-specific factors...................................................54

5
1. Introduction

Traditional theories of corporate finance have been developed extensively to

explain the choices on capital structure. Until today, however, there has been no single

theory of capital structure to be able to completely explain the motives behind the

corporate issuing activities. At the start of the research on corporate finance, the

Modigliani and Miller theorem states that, in the efficient market with perfect capital

allocation, there is no opportunity cost in switching between equity and debt. The

pecking order theory argues that firms tend to trade off the costs and benefits by

taking additional debt. Myers and Majluf (1984) further proposed the pecking order

theory, which claims firms preferences as internal funds, then external debts and the

external equity financing as the last resort. Although distinct from each other, both

theories are regarded as the backbones for the substantial studies on capital structure,

trying to undo the mystery about the choice between debt and equity financing.

Recently, several studies have been trying to explain firms capital structure

using the historical market valuation. The most prominent one is the market timing

theory, which states that firms would issue shares when the market prices are high and

repurchase shares when the market prices are low. Therefore, managers would

actively reshape firms capital structure according to the historical market valuation.

On the other hand, the research on the historical stock returns argues that the

historical market valuation is a passive variable affecting firms choice between debt

and equity. Any change on firms capital structure is merely a mechanical response to

the historical market valuation. Despite both of them have been empirically proofed,

doubts arises about which theory plays a dominant role in explaining firms capital

6
structure from a historical perspective. This paper aims to compare the two theories

and find out whether firms historical market valuation could affect ones capital

structure in an active or a passive way.

In this paper, I focus on the use of measures to investigate the historical effect

on firms capital structure. While the active change between debt and equity is tested

by the market-to-book ratio, the passive response from market valuation is measured

by the historical stock return. Since there is no explicit literature to test the historical

effect on the European market, I choose firms on FTSE Eurofirst 300 as my data set

to make the comparison. First, the market-to-book ratio and the historical stock return

are used in the market timing test and the stock return test, respectively, to figure out

whether the historical market valuation could affect the annual changes on the capital

structures of European biggest firms. It shows that both of them could explain the

annual changes in firms capital structure, which is ambiguous to explain the

historical effect on firms capital structure. Whether the historical market valuation

could influence firms annual leverage change in an active or a passive way is not

clear. The doubts raised from the above empirical results lead me to integrate the two

measures into one test to find out the dominant variable between them. As the result

of the integrated test shows, the coefficients of the market-to-book ratio and the stock

return are -0.000117 and 0.9106, respectively. It is significant to indicate that the

stock return measure turns out to be the dominant determinant within the framework

of firms dynamic capital structure. The change in firms capital structure is mainly

due to a passive response to the cumulative stock returns. The high correlation

coefficient between the market timing measure and the stock return measure further

shows that the latter one is a better and more accurate determinant for firms leverage

changes. The final step consists of the use of the stock return measure to test the

7
persistent effect on the long-term changes in firms capital structures. Over the

ten-year horizon, the historical market valuation still imposes explanatory power to

firms capital structure changes.

This paper is organized as follows. The next section describes the core theories

on capital structure. Section 3 presents the current debate on how firms historical

valuation could affect the capital structure. Based on the two different views about the

historical market valuation, section 4 gives the descriptions for the variable

construction in the following empirical tests. It follows with the data and summary

statistics in Section 5. Section 6 starts the empirical part with the comparison tests

using two different measures, the market timing measure and the stock return measure.

Section 7 and section 8 conduct the integrated test and the persistence test,

respectively. Further discussion is presented in section 9 and section 10 concludes.

8
2. Theories on capital structure

Since Modigliani and Miller published their irrelevance theorem in 1958, the

theoretical and empirical researches have been developed to focus on the determinants

of firms financing decisions. It is important to analyze the firms capital structure that

it tells the managers how to maximize the firm value by using a combination of equity,

debt or the hybrid securities. Besides certain securities, firms value could be as well

affected by bankruptcy costs, asymmetric information and the agency costs. The more

variables involved to determine the choices on capital structure, the more complexity

it becomes for the research in corporate finance. Over the decades, major and minor

theories have been brought to challenge the basic assumption of the perfect capital

market on which the irrelevance theorem is based. The following sub-sections present

how theories on capital structure are developed.

2.1 Irrelevance theorem (Modigliani-Miller theorem)

Proposed by Modigliani-Miller in 1958, this theorem is regarded as the

cornerstone of the modern theory on capital structure. The irrelevance proposition

argues that, firms capital structure remains the same regardless of whether it finances

itself with debt or equity, assuming that there is a perfect market without transaction

costs, bankruptcy costs, and taxes. However, there is friction in reality. A firms value

could be differed by the capital structure it employs because of the inefficient or

segmented market in the real world. Investors may exploit the gains from switching

between equity and debt opportunistically. As the Modigliani-Miller theorem does not

hold in practice, the ongoing debates continue and try to figure out an appropriate

9
interpretation for firms capital structure in the real world.

2.2 The trade-off theory

The trade-off theory argues that firms trade off the costs and benefits by taking

additional debt. It makes a progress in relaxing the restrictive assumptions in the

Modigliani-Miller theorem and brings the debates much closer to the real world

characteristics. While adding various imperfections, including bankruptcy costs, taxes,

transaction costs and agency costs, it retains the assumption of a perfect market with

symmetric information. There are mainly two aspects. The first one is regarding to the

imperfections that could lead to an optimal trade-off. When there are higher taxes on

dividends, firms would finance itself with more debt (Modigliani and Miller (1963),

Miller and Scholes (1978)). When there are higher non-debt tax shields, firms would

take less debt instead (DeAngelo and Masulis (1980)). When higher costs of financial

distress are subjected, firms would change to more equity financing. On the other

hand, the second aspect sheds lights on the agency problem. Too much equity could

result in a substantial free cash flow and the interest conflicts between managers and

bondholders (Jensen (1986)). An example of the interest conflicts could be a situation

where managers are tempted to benefit the current shareholders at the expenses of

debtors by taking relatively riskier projects. This is the so-called risk shifting problem,

in which shareholders could benefit the upside potentials whereas leaving the

downside risk to debtors.

10
Figure 1 The trade-off between debt and equity

Market Value of
the Firm

Costofoffinancial
PV financial
distress

PV of tax shield

Firm value under fully


equity financing

Debt
Optimum D/E*

Explanation: as the leverage ratio increases, there is a trade-off between the tax shield and financial

distress costs, causing an optimum capital structure D/E*.

As indicated in Figure 1, the trade-off theory states that different sources of

firms finance should be balanced out to reach an optimal leverage ratio. As the

optimal ratio is a firm-specific variable, which largely depends on the firm

characteristics such as profitability, firm size and growth opportunities, the optimal

leverage ratio would adjust over time to any changes in firm characteristics. Since the

validity of the trade-off theory has been empirically tested (Rajan and Zingales (1995),

Titman and Wessels (1988)), it has been considered as the most promising theory on

capital structure nowadays.

11
2.3 Pecking order theory

The pecking order theory claims that there is no optimal capital structure,

assuming informational asymmetries between insiders and outsiders. Described by

Myers (1984) and Myers and Majluf (1984), a firms asset base can be decomposed

into assets in place and growth opportunities. Assets in place are firms physical assets

that are safe and can mostly be financed by debt. Growth opportunities are the future

projects and the overall risk profiles of the firms, which should be financed by equity.

Using external equity finance is more costly as insiders would have more information

about firms prospects than outsiders have. Because of higher risks associated with

growth opportunities rather than assets in place, a firm would finance a project using

internal funds first, followed by risky debts. Equity financing should only be issued as

the final resort. This is what pecking order theory predicts on firms financing puzzles.

The capital structure would therefore be a reflection on how the firm finances its

deficit over time.

Although widely accepted by economists in the 1990s, its fame faded because of

the loss of empirical validity (Fama and French (2002), Frank and Goyal (2003)).

Suggested by Lemmon and Zender (2004), the increase of relatively small firms

amongst the large public-listing corporations may be the cause of the weakening of

the pecking order theory. In order to prevent debt-overhand and underinvestment

problems, these small firms are more willing to use external finances, like equity, for

their investments.

12
2.4 The managerial entrenchment theory

The managerial entrenchment theory (Zwiebel (1996)) is a dynamic theory of

capital structure. It states that higher market valuations and profitable investment

opportunities could facilitate the equity finance. In the meanwhile, however, a large

proportion of equity financing could lead to an entrenched management. Managers

would be reluctant to rebalance the leverage ratio by raising more debt afterwards.

This shows a favor to the market timing theory1, where managers prefer to issue

equity when market valuations are high and do not rebalance subsequently. Managers

would have attempts to exploit existing investors by value-destroying decisions and

empire-building behavior. Since the management entrenchment theory has not been

studied heavily, its validity is still needed against empirical researches.

The theories discussed above have imposed their significant influence on the

determination of firms capital structure. Although their explanations differ from each

other, they all contribute to the solid development of the research in corporate finance.

Nowadays, the use of historical market valuation is trying to give a new explanation

to firms changes in their capital structures.

1
The market timing theory would be discussed in more details in Section 3.1.

13
3. Current debate: theories on firms historical market valuation

Having discussed the revolution of the theories on the capital structure, I am

now proceeding to the main focus of this paper, the impact of historical market

valuation on firms capital structure. There are mainly two different views on it. While

the market timing theory suggests that the historical market valuation could actively

affect the choices on debt and equity, the research on firms historical stock return

argues that any change in firms capital structure is merely a passive response to the

stock returns. In this section, the two different views are discussed as follows.

3.1 Market timing theory

Advocated by Baker and Wurgler (2002), the authors believe that a theory of

capital structure based on market timing could provide the most natural explanation

for the dynamic changes in firms capital structure. It argues that the current leverage

ratio is the cumulative outcome of managers past attempts to time the equity market.

After the issuance activities of debt or equity, firms would not actively rebalance their

leverage ratio to the target level. Any financing decision would have a persistent

impact on the capital structure instead. In other words, the change in firms capital

structure is induced by the issuance persistence because firms do not care about

adjusting their leverage ratio to a target level.

14
3.1.1 Two versions: adverse selection versus mispricing

There are two versions of equity market timing that would result in the similar

capital structure dynamics. The first one concerns about the rational investors and

managers, together with the adverse selection cost that varies across firms or across

time period. Evolved as a dynamic form of Myers and Majluf (1984), the adverse

selection version implies that the past variation in the market-to-book ratio would

have a long-lasting effect on firms choice between equity and debt. The assumption

under this dynamic version is that the costs of deviating from an optimal leverage

ratio are smaller than the resulting variation in issuance costs. The second version

involves the irrational investors with the perception of mispricing. Given that the

market-to-book ratio is inversely related to future returns on equity, managers would

issue equity when they perceive issuance costs are irrationally low and repurchase

equity when the costs are irrationally high. There is no optimal leverage ratio under

the second version of market timing theory, meaning that managers do not need to

reverse their decisions when firms appear to be correctly valued and the costs

associated with equity issuances appear to be normal. This could lead the temporary

fluctuations in market-to-book ratio to have persistent impacts on firms capital

structure.

Despite of the differences, the two versions of market timing theory can be well

explained by the single model presented in Baker and Wurgler (2002). The survey by

Graham and Harvey (2001) supports the primary assumption mentioned above, which

states that managers believe they are able to time the marketbut do not discriminate

immediately between the dynamic asymmetric information and the mispricing version

of market timing theory. This theory empirically proofs that the changes in firms

15
leverage ratio is the result of the active adjustment to the historical market valuation.

However, several researches argue that the change in firms leverage ratio is due to the

passive response to the historical stock returns. The specific reasons are presented in

the following sub-section.

3.2 Historical stock return

The paper by Welch (2004) suggests that firms market leverage moves closely

with the fluctuations in their stock prices. The cumulative stock return even shows

large and long lasting effect on firms capital structure. With the variable implied

debt ratio2, the empirical result posits a new view on the dynamic capital structure,

which claims that the change in firms capital structure is merely a mechanical

response to the historical stock return. The stock return is an endogenous factor that it

imposes both direct effect and the indirect effect through other proxies3 on firms

capital structure. Kayhan and Titman (2007) further confirmed this finding with a

similar variable k-year cumulative stock return. In the meanwhile, it shows that the

market timing measure initiated by Baker and Wurgler (2002) influences the capital

structure less. The magnitude of the market timing effect is relatively small compared

to the stock return effect.

As indicated above, the divergence of the debate mainly focuses on the impact

on the changes in firms capital structure. Although empirically proofed, there is still

doubt about which measure would be dominant in determining firms leverage

changes. To conduct the comparison tests, Section 4 firstly specifies the variables to

2
The explanation for the implied debt ratio is presented in more details in section 4.2.
3
The previously used proxies are asset tangibility, firm size, profitability, etc.

16
be used in the following empirical tests.

4. Variable construction

There are five variables to be used in the comparison test, the integrated test and

the persistence test in section 6, 7, 8, respectively. They are market timing measure,

stock return measure, profitability, asset tangibility and firm size. Each one is

discussed in more details in the sub-sections below.

4.1 Market timing measure

The concept of market timing suggests that the dynamic capital structure is due

to the cumulative attempts to time the equity market. An effective timing measure

should be related to the market mispricing in the short run and investment

opportunities in the long run. Several empirical evidences support the use of the

market-to-book ratio as a natural measure of the market valuation both in the short

and long run. As observed from the annul change in firms leverage ratio, the

once-lagged market-to-book ratio captures the idea that a firm raises external capital

when the market valuation is relatively high and vice versa. The actual leverage ratio

would be more likely to decrease after the raise of equity in the previous period

(Baker and Wurgler (2002), Kayhan and Titman (2007)). By investigating the hot- and

cold-market issuances, Alti (2006) further confirms that the impact of the

market-to-book ratio on leverage in the IPO year is more likely to be negative.

Although hot-market firms issue much more equity in the IPO year than cold-market

17
firms do, they do not differ in the way how they convert their new equity into illiquid

assets. The additional equity raised in the IPO year primarily affects the cash and

other illiquid assets on the balance sheet. This shows that market timers (e.g.

managers) would tap the equity market more than their desired capital needs

following the IPO. By using the U.S. firms existing between the year 1983 and 2002,

Hovakimian (2006) finds similarly annual market timing effect, indicating that

managers would take the chance on mispricing to raise equity financing in the short

run. To the contrary, the investment opportunities can be captured by the cumulative

change in the market-to-book ratio in the long term. The reason for the persistence

effect is that firms with higher market-to-book ratios are perceived to have higher

growth opportunities, and thus they are more likely to finance their current financial

deficit with additional equity. As a result, they reserve their borrowing capacity for the

future (Kayhan and Titman (2007)). However, not all the empirical results point to the

same direction. Several find that the persistent effect fades away after a short period

(Alti (2006), Hovakimian (2006)). The negative relationship between the

market-to-book ratio and the leverage ratio is just a transitory and small pattern on

firms capital structure. Past attempts to time the equity market are unlikely to be a

significant impact on the current capital structure.

In this paper, I would alternatively use the timing measure called external

finance weighted-average market-to-book ratio, which is developed by Baker and

Wurgler (2002). It is expressed as

M t 1
es + d s M
= t 1
B efwa ,t 1 s =0 e + d

B s
(1)
r r
r =0

where the summations are taken starting at the first observation year, and e and d

denote the net equity issues and net debt issues, respectively. The external financing

18
decision works better to captures the idea of the market timing, as it takes high values

for firms with external finance when the market-to-book ratio was high and vice versa.

When this weighting scheme is taken into account for capital structure choices, it

means that managers place greater importance on the weights for the years when

significant external financing decisions were being made, and consider them as

practical opportunities to change the leverage ratio in the future.

When estimating the annual changes in the leverage ratio, the external finance

weighted-average market-to-book ratio is equivalent to the one-year-lagged

market-to-book ratio. Since the total observation years in this estimation is one year,

the weight to the one-year-lagged market-to-book ratio is exactly equal to one.

Therefore, the annual timing measure in the estimation of the change in the capital

structure can be simplified as the one-year-lagged market-to-book ratio as follows.


M M
One-year = (2)
B efwa ,t 1 B t 1
When estimating the persistent changes in the leverage ratio, it is better to use this

weighted-average measure rather than the one-year-lagged market-to-book ratio

because the weighting variable well captures the cumulative historical effect of past,

within-firm variation in market-to-book ratio.

4.2 Stock return measure

Within the dynamic framework of the firms capital structure, the stock return is

observed to be explicitly correlated with the changes in firms leverage ratio. In the

survey of Graham and Harvey (2001), evidence suggests that the leverage ratio

decreases following an increase in stock prices because CFOs believe that they could

19
raise equity financing under more favorable terms. Hovakimian, Opler and Titman

(2001) use empirical results to show that firms tend to decrease their leverage ratio

through equity issuing when stock prices are increasing. By decomposing the capital

structure changes into effects caused by stock returns and effects caused by corporate

issuing activities, Welch (2004) shows that the stock return is the most influential

variable accounted for the long-term decision on corporate financing. Firms leverage

ratio is negatively related to the past stock returns. In the debt dynamics established

by Welch (2004), the stock price acts as a major force to let the leverage ratio

co-move with it. He further finds that the previously used proxies that seem to help

explain the debt dynamics is primarily due to the fact that they are closely correlated

with the omitted dynamics caused by stock price changes. Kayhan and Titman (2007)

further concludes that the cumulative stock returns act more strongly than the

traditional timing measures 4 under the market timing context. The stock return

therefore provides a better explanation for the changes in firms financing decisions.

To examine the direct effect of the stock return on the debt dynamics, I include

the variable implied debt ratio (IDR) as initiated by Welch (2004).


Dt
IDRt ,t + k = (3)
(Et (1 + xt ,t + k ) + Dt )
where D, E, and x are, respectively, the book debt, book equity and the cumulative

stock return. It is an implied debt ratio under the situation when the corporation

would issue neither debt nor equity in a certain time period. If the IDR can be an

influential factor to force the changes in the actual leverage ratio, we could state that

the dynamic changes of the capital structure is a mechanistic co-movement with the

4
The traditional timing measure here refers to the market-to-book ratio that is used to estimate the
how changes in the market valuation would affect the actual leverage ratio in the future period. The
related empirical evidence can be found in the papers by Baker and Wurgler (2002), Alti (2006), etc.

20
stock returns, instead of the active attempts to time the equity market suggested by the

market timing theory.

The use of the IDR is a reliable stock return measure under the debt dynamic

because it relates the total debt changes to the corporate issuing activities and the total

equity changes to the cumulative stock returns, respectively. The total amount of debt

changes is due to debt issuing, debt retirements and debt value changes. Therefore, the

cumulative debt changes can be expressed as

Dt + k = Dt + NDI t ,t + k (4)

where NDI is the net debt issuing activities from time t to t+k. Similarly, the total

amount of equity changes is mainly because of equity issuing, equity repurchases and

the equity value changes with stock returns. It can be described as

Et + k = Et (1 + xt ,t + k ) + NEI t ,t + k (5)

where NEI is the net equity issuing activities from time t to time t+k. By definition,

the actual market debt ratio evolves as

D D t +k ( )
D t + NDI t , t + k
A t +k
=
( ( ) )
(D t +k + E t +k ) D t + NDI t ,t +k + E t 1 + x t ,t +k + NEI t ,t +k = IDR t ,t +k
= (6)

Referred to the definition of the implied debt ratio IDR discussed above, the
D
actual market debt ratio would be equal to IDRt ,t + k if there is neither net
A t +k
debt issuing activities nor the net equity issuing activities. This first extreme case

would lead to the 100 % direct effect by the cumulative stock returns. The second

extreme case would be that there is totally no relationship between IDRt ,t + k and

D
. Stock returns cannot exert any influences on the evolution of the capital
A t +k

21
structure. Both extreme cases can be expressed as the following specification

estimates.

5 D D
= 0 + 1 + 2 IDRt ,t + k + t (7)
A t +k A t
When 1=0 and 2=1, the first extreme case applies. When 1=1 and 2=0, it indicates

the second extreme case. As Welch (2004) suggests, the variable IDR largely accounts

for the changes in the actual market leverage ratio. The intended inclusion of the

cumulative stock returns in the IDR is proofed to effectively estimate its explicit

correlation to the dynamic capital structure. As a result, the variable IDR could be

used as a safe measure to consider the direct stock returns impacts on the capital

structure dynamics.

4.3 Profitability

Since the findings in Rajan and Zingales (1995), profitability has been

considered as an explanatory variable to measure the changes in firms capital

structure. Defined as the sum of earnings before interest, taxes, depreciation and

amortization over the past periods, profitability is closely related to the availability of

internal funds. According to the pecking order theory, higher profitability may be

associated with less leverage since firms could use the least risky internal funds to

finance themselves (Marsh (1982)). It could affect the capital structure for tax reasons

as Myers (1984) noted. Auerbach (1979) indicated that more profitable firms would

be more likely to reduce their leverage ratios if there is a tax advantage associated

with the retaining equity. Alternatively, firms with lower profitability would be tended

5
According to the definition by Welch (2004), the variable (D/A)t+k and (D/A)t are initially called the
actual corporate debt ratio (ADR). It is the market leverage, defined as the book value of debt divided
by the sum of the book value of the debt and the market value of equity. Therefore, ADRt=Dt/(Et+Dt).

22
to indicate a debt issue. This payout effect is consistent with the attempts to minimize

transaction costs in their mixed financing and dividend policy decisions (Martin and

Scott (1974)). As indicated by the empirical results (Baker and Wurgler (2002)), it

shows that the persistent effect by profitability is long-lasting, which we cannot omit

in determining the dynamic capital structure.

4.4 Asset tangibility

Defined as the sum of plant, property and equipment, asset tangibility is used to

measure the collateral within a firm. Higher asset tangibility then could be associated

with higher leverage ratio, as the collateral within the firm could enhance its ability to

borrow, which in turn decreases the premium on default required by debtors. As

expected, the coefficient in the market timing test by Baker and Wurgler (2002)

appears to be positive. Besides the significant effect in the short run, asset tangibility

has a persistent effect on the capital structure in the long run (Baker and Wurgler

(2002), Hovakimian (2006)). Both the regressions on book leverage and market

leverage have over 95% confidence interval to explain the strong effect by asset

tangibility in the 10-year horizon. Therefore, the asset tangibility could not be omitted

in the OLS regression used for testing the historical market valuation effect on firms

capital structure.

4.5 Firm size

It is measured as the log of net sales. Size may be related to the chance

involving into a financial distress. Larger firms often have lower costs of debts and

may tend to increase their leverage (Dittmar and Thakor (2007)). By doing this, it is

23
less likely that the firm would involve in the financial disturbances and pay for the

high financial distress costs. The positive coefficient estimate for the firm size is

presented by Baker and Wurgler (2002) and Hovakimian (2006). Even in the longer

term, the value of t-statistics still shows a significant level for its impact on capital

structure (Baker and Wurgler (2002)). However, due to the fact that firm size is just

the physical characteristics of the firm, less attention has been placed to find out its

correlation with the capital structure in previous literatures.

So far, we have seen how theories on the capital structure are evolved. A better

understanding has been established by clarifying the economic relevance for the

variables chosen to conduct the following empirical tests. The following sections are

going to conduct the empirical research, aiming to find out which effect dominates

within the context of the dynamic capital structure, the market timing effect or the

historical stock return effect.

24
5. Data and summary statistics

My initial sample consists of all firms listed in the FTSE Eurofirst 300 Index

from the period 1991-2005. Data on firm characteristics and stock prices are obtained

from Thomson Datastream. I restrict the sample to exclude banks, firms in financial

services and insurance because their capital structures are likely to be significantly

different from firms in industry, service and natural sources. Firms with incomplete

data are dropped. My sample is further restricted to include those firms that have data

at least for three years in order to conduct our long-horizon persistence analysis.

Variable definitions are as follows. Book equity, E, is defined as book value of

the ordinary (common) equity in the company. Book debt, D, is defined as total assets

minus book equity plus convertible debt. My definition of book debt is different from

the definition in Baker and Wurgler (2002) and Kayhan and Titman (2007) for two

reasons. First, convertible debt should be classified as debt rather than as equity. In

most of the studies on capital structure, such as Rajan and Zingales (1995) and Fama

and French (2002), they include the convertible debt into the measures of book debt

and leverage. Second, the treatment of convertible debt in Baker and Wurgler (2002)

is not consistent. While they include the convertible debt into the book equity, they do

not include it into the measure of the market value of equity (Hovakimian (2006)).

This would artificially reduce firms market-to-book ratios with convertibles. Market

leverage, D/Am, is the book debt divided by the sum of the book value of debt and the

market value of equity, where the market value of equity is defined as the market

value of the ordinary equity (MV). Firm-year observations where the resulting book

25
leverage exceeds 100% are dropped. Market-to-book ratio, M/B, is defined as defined

as the market value of equity divided by the book value of equity.

Profitability is defined as EBITDA, which is earnings before interest, taxes,

depreciation and amortization. Asset tangibility, PPE, is measured by net plant,

property, and equipment. Firm size, is defined as the logarithm of net sales in million

euros. The variables EBITDA and PPE are normalized by the fiscal year-end total

assets.

Table 1 summarizes the leverage ratio and financing decisions for this sample.

All the variables are expressed in percentage terms except the market-to-book ratio

M/B and the firm size Log(sales). Over the 15 years, the book leverage declines

slightly, while the market leverage decreases more strongly. The trend of the

market-to-book ratio increases with age, which implied an increasing probability for

future investment using external finances. The other three variables, PPE/A,

EBITDA/A and Log (S) represent the firm-specific development patterns in the

consecutive 15 years. Asset tangibility does not change significantly, with no more

than 2% fluctuation around its mean in the long term. Profitability remains almost the

same, indicating that these European top firms maintain dominant positions and

receive stable revenues in the European market. The size trend is indeed an age effect,

as it shows a steady increase in my observation period. Since these European firms

have been developed into the biggest players in the European market, it is necessary

to maintain their dominant position by steadily increasing their market shares and firm

sizes. Therefore, their steady growth patterns in terms of firm size could reinforce

their market participation appeared on the European market.

26
Table 1 Summary Statistics of Capital Structure

Year N Market D/A (M/B) (PPE/A) (EBITDA/A) Log (S)

Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D.

1991 154 0.58 0.21 1.47 0.84 0.38 0.23 0.14 0.05 15.19 1.67

1992 159 0.59 0.21 1.48 0.81 0.39 0.22 0.13 0.06 15.36 1.38

1993 167 0.59 0.21 1.37 0.72 0.40 0.22 0.13 0.06 15.45 1.38

1994 172 0.53 0.20 1.73 0.77 0.41 0.23 0.14 0.06 15.54 1.34

1995 180 0.55 0.21 1.58 0.74 0.40 0.23 0.15 0.07 15.62 1.31

1996 185 0.52 0.21 1.79 0.96 0.40 0.23 0.14 0.06 15.69 1.29

1997 192 0.49 0.21 2.01 1.00 0.40 0.23 0.15 0.06 15.80 1.30

1998 196 0.43 0.20 2.76 1.66 0.40 0.22 0.15 0.06 15.84 1.29

1999 199 0.46 0.20 2.45 1.43 0.39 0.21 0.14 0.06 15.96 1.29

2000 205 0.46 0.21 2.45 1.67 0.37 0.21 0.15 0.08 16.06 1.35

2001 208 0.47 0.21 2.37 1.38 0.37 0.21 0.14 0.08 16.14 1.35

2002 211 0.47 0.21 2.39 1.51 0.37 0.21 0.13 0.09 16.14 1.32

2003 212 0.54 0.22 1.71 1.08 0.36 0.20 0.14 0.08 16.20 1.26

2004 213 0.48 0.20 2.14 1.28 0.35 0.20 0.16 0.09 16.26 1.25

2005 213 0.49 0.19 2.06 1.31 0.35 0.21 0.15 0.09 16.34 1.22

27
6. Comparison test

The debate in Section 3 shows the importance of the historical market valuation

on firms capital structure, despite the two distinct explanations. Previous literatures

mainly focus on the U.S. firms. Both the market timing measure and the historical

stock return measure turn out to be valid to indicate the impact of historical market

valuation on firms capital structure.

To see how historical market valuation could affect the capital structure on

European firms, I separately test the market timing effect and the stock return effect

on the annual change of the debt ratio.

6.1 Annual market timing effect.

As proposed by Baker and Wurgler (2002), a good measure on the annual

market timing effect is to look into how annual change of the actual leverage ratio is

affected by the timing variables. The following regression model gives us an insight

into it from the quantitative perspective.

D D M PPE EBITDA D
= c + 1 + 2 + 3 + 4 log(S)t1 + 5 + t (8)
A t A t 1 B t 1 A t 1 A t 1 A t1

In equation (8), the dependent variable is the annual change in leverage.

Depending on different purposes, it can be either the book leverage ratio (Baker and

Wurgler (2002), Alti (2006), Hovakimian (2006), Kayhan and Titman (2007)) or the

market leverage ratio (Baker and Wurgler (2002), Kayhan and Titman (2007)). The

28
first independent variable is the market-to-book ratio, which can be a measure for the

market valuation. Other independent variables consist of firm specific characteristics

such as the asset tangibility PPE/A, profitability EBITDA/A, and firm size log(S).

They are the key variables in numerous studies (Rajan and Zingales (1995),

Hovakimian (2006), Alti (2006)) as proxies for the choice on capital structure. The

last variable is the one-time-lagged leverage, which controls the boundary effect of

the leverage ratio6.

To be comparable with the stock return analysis in the Section 6.2, Table 2

reports the annual market timing effect on firms market leverage only. I run this

regression from the year 1991 to 2005 on the refined FTSE Eurofirst 300 sample. For

example, the leverage change between the year 1996 and 1997 is regressed on the

value of the independent variables in 1996. Excluding the incomplete firm data, the

regression sample consists of 2374 firm-year observations for the consecutive 15

years.

Table 2 Annual change in market leverage

Panel A: Regression result

Coefficient t-Statistic Prob.


C 0.0055 0.2132 0.8312
(M/B)t-1 -0.0110 -5.4175 0.0000
(PPE/A)t-1 0.0259 2.7802 0.0055
(EBITDA/A)t-1 -0.2068 -5.3520 0.0000
log(S)t-1 0.0033 2.2763 0.0229
(D/A)t-1 -0.0893 -6.4186 0.0000

6
As the range of the leverage ratio is between zero and one, the boundary effect of the leverage ratio
suggests that any attempt to go over the upper or the lower boundary is not allowed. If the leverage
ratio is close to one of the boundaries, the change in leverage can only be redirected to one direction.
As expected, the empirical result shows that the lagged leverage enters with a negative sign.

29
R-squared 0.0849

Panel B: Standard deviation of the right-hand side variables

Variable Standard deviation

(M/B)t-1 1.2330

(PPE/A)t-1 0.2103

(EBITDA/A)t-1 0.0680

log(S)t-1 1.3198

Inferred from the coefficient and the t-statistics in Table 2, the market-to-book

ratio shows a significant negative effect on the annual change in the market leverage.

A higher market-to-book ratio results in a lower market leverage. For example, a one

standard deviation increase in market-to-book ratio is associated with a 1.36 7

percentage-point decrease in market leverage. This finding is consistent with the idea

that firms tend to increase equity when their market valuations are high and

repurchase equity when the market valuations are low. By following the timing effect

in the financial market, firms can catch up the investment opportunity for an expected

higher return in the future, or reserve their borrowing capacity when the return on

investment is low. Compared to my result with the one presented in Baker and

Wurgler (2002), the coefficient of (M/B)t-1 in their finding is much higher, which is

-3.70 in the IPO year. The reason for the relatively low coefficient of (M/B)t-1 but

statistically significant in Table 2 could be that the U.S firms response to the historical

market valuation more efficient than the European firms when making decisions on

firms capital structure. The second largest effect is profitability (EBITDA/A)t-1. The

OLS result shows that firms profitability tends to decrease by 1.27% percentage
7
This is computed with the coefficient and the sample standard deviation of the independent variables.
-0.0136= -0.0110*1.2330, where 1.2330 is the standard deviation of the lagged market-to-book value
for t from 1991 to 2005.

30
points per standard deviation increase in the market leverage. As profitability is

associated with the availability of internal funds, a higher profitability may be

associated with less debt financing, which is consistent with the pecking order theory.

Both the asset tangibility PPE/A and the firm size Log (S) show modest effect on the

annual change in firms market leverage. It implies that tangible assets tend to

increase leverage (by 0.54 percentage points per standard deviation increases), and

size tends to increase leverage (by 0.44 percentage points per standard deviation

increases). Overall, the result above is similar to that in Baker and Wurgler (2002).

Importantly, the market-to-book ratio has a statistically negative effect on the market

leverage ratio annually. Baker and Wurgler (2002) interpret it as reflecting the

cumulative outcome of past attempts to time the equity market on the year-on-year

basis. In the short run, the European evidence above shows a firmly support to the

equity market timing theory.

6.2 Annual stock return effect

In section 6.1, it is shown that the market timing effect is significant in the short

term. As the one-year-lagged market-to-book ratio is accountable for the annual test, it

implies that the market valuation would actively influence the capital structure

through the perception by both rational and irrational investors. However, not all the

empirical evidence points to the same direction. The alternative measure of the

historical market valuation, the stock return, argues that the change in firms capital

structure is just a co-movement with the market valuation. Managers passively change

the mix of debt and equity according to the changes in firms historical stock returns.

My guess is that if the market timing theory is correct and reliable, the test with the

stock return should appear to be insignificant. To the contrary, if the stock return can

31
be an influential factor to force the changes in the actual leverage ratio, we could state

that the impact of the historical market valuation on the capital structure is passive

that the stock return forces mangers to change the mix of debt and equity. Therefore,

we can re-examine the validity of the market timing effect by invalidate the opposite

view.

As discussed in section 4.2, the stock price empirically shows an explicit link

with the capital structure. In practice, it is much more transparent to extract the signal

from the stock price to make decisions on the actual debt ratio, for any changes in

stock price are available immediately through the media. Based on the idea of Welch

(2004), the alternative market valuation measure is called the implied debt ratio (IDR),

which might be more direct to measure the dynamic relationship between the market

leverage ratio and the stock return. To be comparable with the market timing test in

section 6.1, I refine the test by Welch (2004) and express it as follows.

D D
= c + 1 X t ,t +1
A t +1 A t
PPE PPE
2 c + 2 c +1 X t ,t +1 +
A t A t
(9)
EBITDA EBITDA
+ 3c +
3c +1 X t ,t +1 + + t
A t A t
[ log(S ) +
4 c +1 log(S )t X t ,t +1 ]

4c t

where Xt,t+k IDRt,t+1 - (D/A) t. The variable 1 X t ,t +1 explains the direct co-movement

with the historical return. The variables within the square bracket [.] indicate the

indirect influence of the stock return through asset tangibility, profitability and firm

size, which can be referred from the incremental differences between the coefficients

n and n+1. Take profitability for example. Firms profitability was perceived to have

32
a negative relationship with leverage ratio in previous studies and my results in

section 6.1, respectively. If profitability does not have any incremental significance

when IDR is taken into account, it would only have indirect correlation with the

capital structures through its correlation with stock returns. Put differently, managers

would not actively decrease their leverage ratio when profitability is perceived to

increase. Instead, managers of higher profitable firms would have experienced a

period of higher stock prices, which in turn would decrease their debt ratios

accordingly. Table 3 shows the annual result in more details.

Table 3 Historical market valuation effect measured by implied debt ratio

Panel A: Regression Result

Variable t-Stat. Prob.

C 0.0804 4.0478 0.0001


X t,t+1 1.4128 4.6466 0.0000
(PPE/A)t 0.0199 2.5347 0.0113
(PPE/A)t * X t, t+1 0.0950 0.7459 0.4558
(EBITDA/A)t -0.1383 -5.5254 0.0000
(EBITDA/A)t * X t, t+1 -0.0150 -0.0407 0.9676
log(S)t 0.0057 4.8024 0.0000
log(S)t * X t,t+1 0.0317 1.7143 0.0866

R-square 0.4823

Panel B: Standard deviation of the right-hand side variables

variable Standard deviation

X t,t+1 0.0614

(PPE/A)t 0.2106

(EBITDA/A)t 0.0616

log(S)t 1.3302

33
First of all, the variable X t,t+k , which is the difference between the implied debt

ratio and the actual ratio, is largely significant as a market valuation measure. There is

a positive correlation between the annual change in the actual market leverage ratio

and the implied debt ratio. One standard deviation increase in Xt,t+1 would result in an

increase of 8.678 percentage points in the actual market leverage ratio. This effect is

much larger compared to the market-to-book ratio effect as a timing measure in

section 6.1. It shows a more direct link between the stock return and the capital

structure. Managers regard it as a signal to adjust their debt ratio to an optimal level.

This would in turn induce the manager to change the debt ratio mechanically.

Second, the use of firm characteristics as proxies in this model gives us more

insight into how they affect the choices between debt and equity. Not surprisingly, the

coefficients of single variables like asset tangibility, profitability and firm size all

show that they have large impact on the change in the debt ratio. The result in Table 3

is similar to what I have found in section 6.1, indicating that these proxies can be

consistently used as the determinants for the capital structure. The signs and

magnitudes are more or less the same as the traditional method for the market timing

tests. However, when the stock returns are properly controlled for, the annual

re-adjustment effect decreases. The incremental values between (PPE/A)t and

(PPE/A)t * X t, t+1 , (EBITDA/A)t and (EBITDA/A)t * X t, t+1, log(S)t and log(S)t * X

t,t+1 are relatively small across years, so that the causal-relationship among the capital

structure and asset tangibility, profitability and firm size, respectively, is just a passive

one. Therefore, the stock return acts as an endogenous factor to have effects on the

8
This is computed with the coefficient and the sample standard deviation of the independent variables.
0.0867 = 1.4128* 0.0614, where 0.0614 is the standard deviation of the implied debt ratio
X t,t+1.

34
capital structure. Asset tangibility, profitability and firm size influence the capital

structure through the indirect force by the stock return.

To be comparable with the annual market timing test in section 6.1, I

alternatively replace the market-to-book ratio (M/B)t-1 with X t, t+1 from equation (8) so

that the soundness of the measure X t,t+1 could be seen directly from the same

methodology. The model with the replacement of X t, t+1 can be expressed as follows.
D D PPE EBITDA D
= c + 1Xt,t +1 + 2 + 3 + 4 log(S )t 1 + 5 + t (10)
A t A t 1 A t 1 A t 1 A t 1

As indicated from Table 4, the change from (M/B)t-1 to X t, t+1 does not change the

significant effect by the cumulative stock return, which re-confirm the result from

section 6.2 that the implied debt ratio could be an effective measure of firms annual

leverage change.

Table 4 Annual market timing testreplace (M/B)t-1 with X t, t+1


t-Stat. Prob.

C 0.0684 3.7185 0.0002


X t,t+1 0.9422 39.4692 0.0000
(PPE/A)t 0.0174 2.4237 0.0155
(EBITDA/A)t -0.1417 -5.8043 0.0000
log(S)t 0.0049 4.4838 0.0000

R-square 0.4811

Seen from section 6.1 and 6.2, the results obscure my guess that one of the

theories should dominate the other, either the market timing theory or the historical

stock return effect, because each of them explains the economic relevance of firms

leverage changes in the opposite direction. In terms of coefficients, both tests

35
successfully indicate that the historical market valuation could largely explain the

annual change of firms capital structure, despite the use of different measures. There

is no reason to drop either the market-to-book ratio or the stock return as a

determinant for debt- or equity- issuing activities. However, the large differences in

R-square between the two tests raise the concern of excluding the least effective

measure of the historical market valuation in the empirical analysis. While the

R-square for the annual market timing test is 0.0850, the one for the annual stock

return test is 0.4823, which is much higher. The annual stock return test in section 6.2

tends to present a better evaluation of the relationship between firms historical

market valuation and their capital structure in terms of the value of R-square. Notice

that the difference between the two tests in terms of variables is the use of either

market-to-book ratio or the stock return, the following single-variable regressions

would shed lights on their individual effects on firms annual leverage changes.

D D M
Single-variable market-to-book ratio: = c + 1 + t (11)
A t A t 1 B t 1
D D
Single-variable stock return measure: = c + 1 X t,t +1 + t (12)
A t A t 1

Table 5 Single-variable regression: market-to-book ratio versus stock return

t-Stat. Prob.

market-to-book ratio
C 0.0371 9.8602 0.0000
(M/B)t-1 -0.0158 -9.9480 0.0000
R-square 0.0553
Stock return measure
C 0.0147 9.3569 0.0000
X t,t+1 0.9122 37.8530 0.0000

R-square 0.4585

36
There are two points to be mentioned from the results in Table 5. First, both the

market-to-book ratio and the stock return measure are effective determinants of firms

capital structure, which can be inferred from the coefficients. Compared to the result

in Table 2 and Table 3, the signs of (M/B)t-1 and X t,t+1 are the same while the

magnitudes are much larger. The single-magnitude effects posit stronger evidence to

show their significant influences on firms choice between debt and equity. Second,

the values of R-square for both measures decrease slightly in relative to those with

multi-variables in Table 2 and Table 3, which implies that the market-to-book ratio

and the stock return measure remain important variables in the annual market timing

test and the annual stock return test, respectively. Excluding either of them from the

separated tests in Section 6.1 and 6.2 could turn the empirical tests into insignificant

and inaccurate. The additional variables including asset tangibility, profitability and

firm size only impose moderate effects on firms annual leverage changes.

As a fact that the separation of the two tests does not show any explicit

domination effects between the two measures, the following section is to integrate the

two measures to find out the dominant measure for the capital structure from the

historical perspective. The integrated test in Section 7 could provide us an insight into

a better and more accurate estimation on firms leverage changes in terms of the

variable uses.

37
7. Integrated test

As long as the separated tests in section 6 could not provide us a convincing

result to show how market valuation affect firms capital structure, this section aims to

integrate the two measures, namely the market-to-book ratio and the implied debt

ratio to find out a clear explanation to the dynamic capital structure puzzle.

Based on the dynamic regression suggested in Baker and Wurgler (2002), the

annually integrated test is specified as follows.


D D M PPE EBITDA
= c + 1 + 2 X t ,t +1 + 3 + 4
A t +1 A t B t A t A t
(13)
D
+ 5 log(S )t + 6 + t
A t
Equation (10) is similar to the market timing test in section 6.1 except for an extra

variable X t ,t +1 . Any dominant effect between the two market valuation measures can

be shown directly from the integrated regression result in Table 6.

Table 6 Integrated TestAnnual result

Panel A: Regression result

t-Stat. Prob.

C 0.0839 4.0349 0.0001


(M/B)t -0.0001 -0.0688 0.9452
X t,t+1 0.9106 33.7614 0.0000
(PPE/A)t 0.0124 1.5934 0.1113
(EBITDA/A)t 0.0512 1.5725 0.1160
log(S)t -0.0033 -2.6402 0.0084
(D/A)t -0.05219 -4.7013 0.0000

R-square 0.4841

Panel B: Standard deviation of the right-hand side variables

38
variable Standard deviation

(M/B)t 1.2226

Xt,t+1 0.0609

(PPE/A)t 0.2173

(EBITDA/A)t 0.0603

log(S)t 1.3269

When both the market timing measure and the stock return measure are

incorporated into one test, X t,t+1 appears to be the most significant one accounting for

the historical market valuation. One standard deviation increase in Xt,t+1 would result

in an increase of 5.559 percentage points in the actual market leverage ratio. Similar

to the finding by Welch (2004), the coefficient for X t,t+1 is relatively higher than other

variables10. This indicates that the implied debt ratio is indeed a strong measure in

determining the relationship between firms historical market valuation and the

change in capital structure. The market-to-book ratio (M/B)t, on the other hand, takes

no effect on firms capital structure. The t-test of (M/B)t turns out to be insignificant

as a result. It is clear that the measure using firms stock return plays a dominant role

in explaining the historical effect on firms capital structure. Since the cumulative

stock return is the most direct and dynamic information of firms market value,

managers would regard it as a reliable determinant when making decisions on the

choice between debt and equity. As firms capital structure cannot be changed

frequently, the 1-year cumulative stock return is effective to inform managers any

necessary change in the market leverage. Over the annual horizon, the stock return is

9
This is computed with the coefficient and the sample standard deviation of the independent variables.
0.0555 = 0.9106* 0.0609, where 0.0609 is the standard deviation of the implied debt ratio X t,t+1.
10
In Welch (2004), the coefficient of X t,t+1 is 7.38 while those of other variables are between 0.06 to
2.54.

39
endogenous that any change in the market leverage is a mechanical response. Highly

levered firms would increase the stocks systematic risk, which implying a higher

return on stocks. Therefore, managers would regard the annual stock return as an

indicator to increase firms debt ratio. Seen from the coefficients for asset tangibility,

profitability and firm size, their effects on firms capital structure are slightly

decreased while the stock return shows the most significant influence. This further

confirm the regression result in Section 6.2 that asset tangibility, profitability and firm

size are passive determinants for firms capital structure. Stock return is a first-order

endogenous factor that it can affect the capital structure through the indirect impact on

asset tangibility, profitability and firm size.

In this integrated test, the market timing measure which appears to be significant

in previous studies turns out to be failure. Instead, the alternative measure of the

historical market valuation, the stock return, performs well to explain the annual

changes in market leverage. However, the answer is not clear when comparing with

the regression results in Section 6. Why does the market timing measure still show

significance in the separation test while there is no effect in the integrated test? The

doubt leads to the following correlation matrix, which could provide us insight into

the problem.

Table 7 Correlation Matrix

(M/B)t X t,t+1 PPE EBITDA Log(S)


(M/B)t 1 0.6973 -0.0943 0.3658 -0.0636
X t,t+1 0.6973 1 -0.0752 -0.1113 0.0843
PPE -0.0943 -0.0752 1 0.2245 -0.1742
EBITDA 0.3658 -0.1113 0.2245 1 -0.0109
Log(S) -0.0636 0.0843 -0.1742 -0.0109 1

40
Among all the variables, the correlation between (M/B)t and Xt,t+1 is relatively

high. Despite the differences in the variable construction, both measures aim to reveal

the market valuation in a quantitative way, which can be seen from Section 6. Due to

the high correlation, the integrated test in Table 6 properly controls the dominant and

dominated effects for both two measures. The measure of stock return appears to be

the dominant one, which correctly implies the impact of the historical market

valuation on firms capital choice. The significance of the market timing measure in

the separate test is just a coincident variable setting that the historical market-to-book

ratio appears to be an accurate proxy. In addition, the high correlation with the

measure by stock return suggests that most of the impact by market-to-book ratio in

the Table 2 is due to the implicit explanatory power by the stock return measure Xt,t+1.

41
8. Persistence test

As indicated in section 7, the stock return is a better measure to determine firms

capital structure from a historical perspective. To complete the research, this section

continues to conduct the persistence test using Xt,t+k. The test follows the same

methodology as that in section 6.2, but with a longer time horizon, 3-year, 5-year and

10-year, respectively. The specific test is expressed as follows.

D D
= c + 1 X t ,t + k
A t +k A t
PPE PPE EBITDA
2 c A + 2 c +1 A X t ,t + k + 3c A
(14)
t t t +
+
EBITDA t
+ 3c +1 X t ,t + k + 4 c log(S )t + 4 c +1 log(S )t X t ,t + k
A t

where k is the time horizon (k=3,5,10 years).

As seen from Table 8, the variable X t,t+k is largely significant as a measure of

the historical market valuation. Even in the longer time period, it shows a consistent

positive correlation with the change in the actual market leverage ratio. In the ten-year

persistence test, one standard deviation increase in X t,t+10 would result in an increase

of 17.75 percentage points in the actual market leverage ratio. This effect is much

larger compared to the annual test in section 6.2. It shows a more direct link between

the stock return and the capital structure in the long run. As X t,t+k assumes that the

accumulative stock return is an endogenous determinant on firms capital structure,

the result suggests that the market leverage responses to the stock return

mechanically.

42
Table 8 Persistence test

Panel A: Regression result

Three-year Five-year Ten-year

t-Stat. Prob. t-Stat. Prob. t-Stat. Prob.

C 0.2359 6.0488 0.0000 0.4390 7.0837 0.0000 0.5231 3.2763 0.0011


X t,t+k 1.9297 5.5278 0.0000 1.3560 4.3143 0.0000 1.4005 2.3111 0.0213

(PPE/A)t 0.0984 4.3389 0.0000 0.0445 2.9832 0.0029 0.1375 1.9964 0.0465
(PPE/A)t * X t, t+k 0.0688 0.5578 0.5771 0.0902 0.8989 0.3689 0.1711 0.6501 0.5160
(EBITDA/A)t -0.3304 -7.0085 0.0000 -0.4730 -5.6694 0.0000 -1.2359 -4.5814 0.0000
(EBITDA/A)t * X t, t+k -1.0735 -2.5364 0.0113 -0.8947 -2.2772 0.0230 -3.8272 -3.3096 0.0010

log(S)t 0.0162 6.9497 0.0000 0.0290 7.7671 0.0000 0.04059 4.1025 0.0000
log(S)t * X t,t+k 0.0433 2.2511 0.0245 0.0760 3.5360 0.0004 0.03939 1.0102 0.3130

R-square 0.4994 0.46440 0.3793

Panel B: Standard deviation of the right-hand side variables

Variable Standard deviation

Xt,t+3 0.1047

Xt,t+5 0.1317

Xt,t+10 0.1267

(PPE/A)t 0.2142

(EBITDA/A)t 0.0598

log(S)t 1.3341

Other variables including asset tangibility, profitability and firm size still affect

the capital structure strongly in the longer term. However, the effects by the cross

terms (PPE/A)t * X t, t+k, (EBITDA/A)t * X t, t+k and log(S)t * X t,t+k diversify as time

goes by. Only the variable (EBITDA/A)t * X t, t+k shows cross effect on firms capital

structure consistently. The incremental effect, which is the difference between

(EBITDA/A)t * X t, t+k and (EBITDA/A)t in the persistence test suggests that the stock

43
return has an indirect impact on firms profitability only partially in the long run. Part

of the change in the actual debt ratio can be accounted for the active response from

firms profitability changes and the remains comes from the passive response to the

change in the stock return. Furthermore, the negative coefficient of (EBITDA/A)t *

Xt,,t+k suggests non-adjustment. Firms with higher profitability tend to adjust less for

stock return-induced capital structure changes, which is consistent with the finding in

Welch (2004). Compared with the short-term result in section 6.2, the annual test

suggests no adjustment effect by firms profitability.

By using the stock return X t, t+k as a market valuation measure, we can find out

the implied changes in firms capital structure. The stock return performs as a passive

factor towards the dynamic change of firms capital structure. To clarify the

mechanical impact by stock returns, I conduct the following regression demonstrating

the single effect caused by the return-induced debt ratio. The cross-sectional

regression is

ADR t+k = c+1IDR t, t+k + 2ADR t+t (15)

where the reported coefficients and t-statistics are calculated from the time series of

cross-sectional regression (Fama-MacBeth regression). As seen from Table 9, the

coefficients of IDRt,t+k are statistically significant, whereas the ADRt adds no

economic or statistical power over the longer term. It means that the debt ratio drifts

largely with stock returns, but the old debt ratio ADRt has no tendency to affect the

current debt ratio after 5 years. Managers would believe in the cumulative stock return

to make decisions on firms capital structure rather than using one-year stock return,

which is remote and less correlated to the current market leverage. Overall, the

influence of stock returns through IDR is much more important that omitting the stock

return in the dynamic capital structure analysis would make it incomplete.

44
Table 9 Annual aggregate effects by return-induce debt ratio

Panel A: Annual test

t-Stat. Prob.

C -0.02960 -9.5814 0.0000

IDR t, t+k 0.4864 36.3941 0.0000

ADR t 0.5328 40.4851 0.0000

R-square 0.9529

Panel B: Long term test

Three-year Five-year Ten-year

t-Stat. Prob. t-Stat. Prob. t-Stat. Prob.

C 0.1243 18.4210 0.0000 0.1884 20.3624 0.0000 0.2990 14.8342 0.0000

IDR t, t+k 0.7408 30.4554 0.0000 0.7333 27.5152 0.0000 0.5947 12.1830 0.0000

ADR t 0.0863 3.4184 0.0006 0.0198 0.7133 0.4758 0.0165 0.3080 0.7582

R-square 0.7809 0.6758 0.5109

45
9. Discussion

So far, I have presented two market valuation measures to examine firms

dynamic capital structure. In the comparison tests presented in Section 6, both the

market timing measure and the stock return measure show statistical significance in

the tests of annual market timing effect and the annual stock return effect, respectively.

However, this would make the explanation of how firms historical market valuation

would affect ones capital structure ambiguous. On the one hand, the market timing

measure shows that the change in firms leverage ratio is an active response to the

market valuation. On the other hand, the stock return measure implies that any change

in firms capital structure is merely a mechanical adjustment to the historical stock

returns. Since the tests in Section 6.1 and 6.2 indicates the economic relevance in the

opposite directions and both shows significant result, an integrated test combining the

two measures continues to find out a better explanation about the historical market

valuation effect. The results in Section 7 tells that the market timing measure fails to

support the dynamic capital structure whereas the stock return measure succeeds in

explaining it. Due to the high correlation with the stock return measure, the market

timing measure is just a special variable setting11 that appears to show the large

impact on firms capital structure. Overall, the stock return measure can be used as a

sound measure explaining the relationship between firms historical market valuation

and their capital structure changes. Given the results above, this section presents a

number of additional specifications and a discussion of further improvement.

11
As indicated in Section 7, the market timing measure is highly correlated with the stock return
measure. In previous literature, the market timing measure successfully explaining the mystery of the
capital structure is due to the way how the variable is constructed. Since the stock return could be
alternatively used to gauge the market valuation and shows significance in my empirical tests, we can
safely use the stock return measure to conduct the analysis on the dynamic capital structure.

46
9.1 The role of stock return

The traditional market timing theory suggests that the capital structure is the

result of the cumulative attempts to time the equity market (Baker and Wurgler

(2002)). The market valuation plays a key role in motivating the dynamic changes of

firms capital structure. When the market valuation is high, managers would regard it

as an opportunity to raise equity to finance their investment projects. When the market

valuation is low, they would prefer to repurchase shares to reserve the investment

opportunities. The traditional market timing framework does a good job in revealing

the direct relationship between market-to-book ratio and firms capital structure.

However, when the stock return measure is incorporated into the empirical test, it fails

to deliver any significant evidence on how the market valuation could perform within

the context of the dynamic capital structure. Shown by the empirical evidence in

section 7 and 8, the use of the stock return measure can be an alternative market

valuation measure to explain the mystery. By reconstructing the multivariate

time-series model into a dynamic leverage regression, the return-induced debt ratio

succeeds to inform how the return-induced debt ratio affects firms actual leverage

ratio. Stock returns, or the historical market valuation, passively induce the changes in

the capital structure both in the short run and the long run. This could be called a

mechanical impact of the stock price on the dynamic framework of firms capital

structure. To the contrary, the significant impact by other proxies such as asset

tangibility and firm size is just a passive co-movement with the stock return. Put

differently, stock price indirectly influences the capital structure through the asset

tangibility channel and the firm size channel. Stock price here acts as an endogenous

factor to influence the capital structure. Interestingly, profitability plays a mixed role.

In the long run, part of its effect comes from the direct link with the capital structure,

47
through which it indicates the availability of firms internal funds. More profitable

firms would have more internal funds available for their investment so that they

would accordingly reduce their debt ratio. And part of the effect by profitability

comes from the co-movement with the stock return, which is the same as the asset

tangibility and the firm size. Moreover, profitability is the only variable showing

non-adjustment through the change in stock prices in the long run. Higher profitable

firms tend to adjust less to the target debt ratio, but let the leverage ratio flow with the

changes in the stock return. As discussed above, the stock return plays a dominant role

in determining the change in the capital structure. Almost all other proxies are just

passive factors within the context of the dynamic capital structure.

9.2 The European evidence

My findings are similar to what have been suggested by Welch (2004). The

historical market valuation effect is not only a short-term event, but also a long-term

impact on the changes in firms capital structure for both European and the U.S. firms.

One reason could be that my data sample consists of the 300 biggest firms in the

European market. They are all dominant players in different sectors, such as natural

resources, retails and autos. As most of them tend to develop into international

corporations rather than local-based ones, their capital structure would be more likely

to move towards the international type, which most big U.S. firms adopt. Therefore,

the patterns of the cumulative change in the capital structure would mostly tend to

move towards be the same type as the multinationals in the United States.

Furthermore, the historical market valuation tests in the short run and the long run

would be expected to produce significant results for both countries.

48
9.3 Impact by country-specific factors

As my sample is based on the index FTSE Eurotop 300, which consists of firms

in U.K., France, Germany, the Netherlands and other European countries, one could

argue that the country-specific factors could posit certain impacts on the dynamic

changes in firms capital structure. This could be referred to the differences in law

systems, the regulation environment and the progress in economic development. To

see whether this argument is correct or not, I conduct an additional regression test

with dummy variables related to the different countries. The result is presented in

more details in Appendix. Not surprisingly, the country-specific dummy variables do

not impose any significant impact on firms leverage changes. One reason could be

that the European countries are now integrating towards a single economic union,

within which the regulations, the currency and the economic development are moving

towards the same direction. The impact by country-specific factors could therefore be

minor on firms capital structure compared to the influence by firm-specific

characteristics.

9.4 Future improvement

Although we could see the implied changes of the stock return under the

framework of the dynamic capital structure, this needs further proof to show the

feasibility and the reliability in practice. A good theory is the one which could be both

empirically and theoretically supported. In my results, I propose the mixed role of

firms profitability. It actively affects the capital structure, whereas passively

co-moving with the stock return. But the magnitude and the sign of this movement is

still not clear. This would require further research on it.

49
10. Conclusion

To find out how the historical market valuation affects the dynamic change in

firms capital structure, two different measures, namely the market timing and the

historical stock return, are used to double-examine the influence. While the market

timing measure suggests an active adjustment to the market valuation, the stock return

measure argues that any change in firms capital structure would be a passive

response to the cumulative stock returns. With the comparison test, both measures

turn out to be statistically significant. This obscures my guess that one of them should

give insignificant result because the two measures indicate the economic relevance in

the opposite direction. When integrating the two measures into one test, only the stock

return measure turns out to be effective. By showing the correlation matrix, things

become clear that the market timing measure is highly correlated with the stock return

measure. The significant result from the single market timing test is due to the

implicit correlation with the stock return measure. As the stock return measure shows

the dominant effect in firms dynamic capital structure, we change to use the stock

return measure only to proceed with the persistence test, which indicates a

long-lasting effect on firms capital structure. The stock return measure, therefore, is a

better proxy in determining the dynamic capital structure. It could therefore provide a

more convincing explanation about the relationship between firms historical market

valuation and their leverage changes.

50
References

Alti, A. (2006). How persistent is the impact of market timing on capital structure?
Journal of Finance, Vol. 61, No. 4, 1681-1710.

Auerbach, A. J. (1979). Share valuation and corporate equity policy. Journal of Public
Economics, Vol. 11, No. 3, 291-305.

Baker, M. and Wurgler, J. (2002). Market timing and capital structure. Journal of
Finance, Vol. 57, No. 1, 1-32.

DeAngelo, H. and Masulis, R. (1980). Optimal capital structure under corporate and
personal taxation. Journal of Financial Economics, Vol. 8, No. 1, 3-29.

Dittmar, A. and Thakor, A. (2007). Why do firms issue equity? Journal of Finance,
Vol. 62, No. 1, 1-54.

Fama, E.F. and French, K.R. (2002). Testing trade-off and pecking order predictions
about dividends and debt. Review of Financial Studies, Volume 15, No. 1, 1-34.

Frank, M. Z. and Goyal, V. K. (2003). Testing the pecking order theory of capital
structure. Journal of Financial Eonomics, Vol. 67, No. 2, 217-248.

Graham, J. R. and Harvey, C. R. (2001). The theory and practice of corporate finance:
evidence from the field. Journal of Financial Economics, Vol. 60, No. 2-3.
187-243.

Hovakimian, A., Opler, T. and Titman, S. (2001). The debt-equity choice. Journal of
Financial and Quantitative Analysis, Vol. 36, No. 1, 1-24.

Hovakimian, A. (2006). Are observed capital structures determined by equity market


timing? Journal of Financial and Quantitative Analysis, Vol. 41, No. 1, 221-243.

Huang, R. and Ritter, J. R. (2005). Testing the market timing theory of capital
structure. Unpublished Working Paper.

51
Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and
takeovers. American Economic Review, Vol. 76, No. 2, 323-329.

Kayhan, A. and Titman, S. (2007). Firms histories and their capital structures.
Journal of Financial Economics, Vol. 83, No. 1, 1-32.

Lemmon, M. L., and Zender, J. F. (2004). Debt capacity and tests of capital structure
theories. Working Paper, University of Utah.

Marsh, P. (1982). The choice between equity and debt: an empirical study. Journal of
Finance, Vol. 37, No. 1, 121-144.

Martin, J. D. and Scott, D. F. (1974). A discriminant analysis of the corporate


debt-equity decision. Financial management, Vol. 3, No. 4, 71-79.

Miller, M. H. and Scholes, M. S. (1978). Dividends and taxes. Journal of Financial


Economics, Vol. 6, No. 4, 333-364.

Modigliani, F. and Miller, M. H. (1958). The cost of capital, corporation finance, and
the theory of investment. American Economic Review, Vol. 48, No. 3, 261-297.

Modigliani, F. and Miller, M. H. (1963). Corporate income taxes and the cost of
capital: a correction. American Economic Review, Vol. 53, No. 3, 433-443.

Myers, S. C. (1984). The capital structure puzzle. Journal of Finance, Vol. 39, No. 3,
575-592.

Myers, S. C. and Majluf, N. S. (1984). Corporate financing and investment decisions


when firms have information that investors do not have. Journal of Financial
Economics, Vol. 13, No. 2, 187-221.

Rajan, R.G. and Zingales, L. (1995). What do we know about capital structure? Some
evidence from international data. Journal of Finance, Vol. 50, No. 5, 1421-1460.

Titman, S. and Wessels, R. (1988). The determinants of capital structure choice.


Journal of Finance, Vol. 43, No. 1, 1-18.

Welch, I. (2004). Capital structure and stock returns. Journal of Political Economy,

52
Vol. 112, No. 1, 106-131.

Zwiebel, J. (1996). Dynamic capital structure under managerial entrenchment.


American Economic Review, Vol. 86, No. 5, 1197-1215.

53
Appendix: Regression with country-specific factors

In consideration of the influence by country-specific factors, the following

regression presents the empirical evidence on how the difference in countries could

affect the relationship between the historical market valuation and firms dynamic

capital structure. The country-specific factors are defined as dummy variables in the

following regression model, which is a revision of equation (9) in Section 6.2.

D D
= c + 1 X t ,t +1
A t +1 A t
PPE PPE EBITDA
2 c A + 2 c +1 A X t ,t +1 + 3c A

t t t +
+ (16)
EBITDA
+ 3c +1 X t ,t +1 + 4 c log(S )t + 4 c +1 log(S )t X t ,t +1
A t
5 AUS + 6 BEL + 7 DK + 8 FR + 9 FIN + 10 GRE +

11GRM + 12 IRE + 13 ITA + 14 NL + 15 NOR + 16 PORT + + t
SPN + SWE + SWZ + UK
17 18 19 20

Table 10 Annual leverage change with country dummies

Panel A basic variables from equation (9)

Variable t-Stat. Prob.

C 0.0301 1.1770 0.2394


X t,t+1 1.2373 3.8883 0.0001
(PPE/A)t 0.0228 2.9041 0.0037
(PPE/A)t * X t, t+1 0.0458 0.4101 0.6818
(EBITDA/A)t -0.0455 -1.8384 0.0662
(EBITDA/A)t * X t, t+1 -0.0368 -1.4314 0.1525
log(S)t 0.0386 4.6638 0.0000
log(S)t * X t,t+1 0.0229 2.7685 0.0057
Panel B country dummies
AUS 0.0004 0.0249 0.9801

54
BEL -0.0003 -0.0188 0.9850
DK -0.0074 -0.4405 0.6596
FR -0.0020 -0.1972 0.8437
FIN -0.0074 -0.4405 0.6596
GRE -0.0002 -0.0075 0.9940
GRM -0.0065 -0.6446 0.5193
IRE 0.0015 0.0828 0.9340
ITA 0.0019 0.1709 0.8644
NL -0.0067 -0.6011 0.5479
NOR 0.0013 0.0969 0.9228
PORT -0.0097 -0.5477 0.5840
SPN 0.0001 0.0124 0.9901
SWE 0.0075 0.6221 0.5339
SWZ 0.0005 0.0466 0.9628
UK -0.0042 -0.4363 0.6626
R-square 0.4759
The country dummies are defined as follows. AUS=Austria, BEL=Belgium, DK=Denmark,
FR=France, FIN=Finland, GRE=Greece, GRM=Germany, IRE=Ireland, ITA=Italy, NL=the
Netherlands, NOR=Norway, PORT=Portugal, SPN=Spain, SWE=Sweden, SWZ=Switzerland,
UK=United Kingdom.

Seen from the table above, the country-specific factors do not impose significant

influence on the firms leverage changes. The coefficients and the t-statistics are

rather small relative to those of the variables in Panel A. The major variables used to

gauge the market valuation effect still appear to be significant in terms of the

t-statistics and p-value. The additional sixteen country dummies do not change their

explanatory power about the relationship between firms historical market valuation

and their leverage changes. Furthermore, the value of R-square dose not change

largely compared to the result in Table 3. There is no incremental effect when the

additional 16 dummies are added into the model, which emphasizes that the

country-specific factors do not have influential effect on my empirical evidence.

55

Das könnte Ihnen auch gefallen