The valuation effects of long-term

changes in capital structure

Ce´cile Carpentier

Faculty of Administrative Science, School of Accountancy, Laval University,

Que´bec, Canada

Abstract

Purpose – This paper aims to test the irrelevance proposition whereby changes in capital structure

do not affect firm value.

Design/methodology/approach – The long-run effect of changes in capital structure on firm value

is examined, using a sample of 243 French firms over the period 1987-1996.

Findings – The null hypothesis cannot be rejected. No evidence is found to support a significant

relationship between the changes in debt ratios and the changes in value. To assess the strength of this

finding, control for reversion towards the target debt level induced by the static trade-off theory is

introduced. Similar results were obtained.

Originality/value – This paper is one of the first to analyze the long-term relationship between

financial structure changes and value, and to propose a direct test for the irrelevance proposition.

Keywords Financing, Capital structure, Debt capital, Targets, France

Paper type Research paper

The irrelevance proposition introduced by Modigliani and Miller (1958) states that in a

perfect market, capital structure has no impact on firm value. Although market

imperfections such as taxes, agency costs and asymmetric information have frequently

been cited in defense of the relevance of the capital structure, 40 years later we have

still not truly ascertained or clarified the relationship between capital structure and

firm value (Myers, 1993, 2001).

As Giner and Reverte (2001) noted, little empirical research has analyzed the

usefulness of leverage in equity valuation. However, numerous papers have explored the

short-term effects of various financing announcements. Notably, the survey conducted by

Harris and Raviv (1991) shows that stock market reaction to the announcement of

leverage-increasing (leverage-reducing) transactions tends to be positive (negative).

Research in this area predominantly adopts the event study methodology, and considers a

few transaction days surrounding the announcement date. More recently, the long-term

impact of new issues has been analyzed. Researchers generally agree that initial public

equity offerings and seasoned equity offerings are followed by long-run abnormal

underperformance. Long-run abnormal negative performance is also associated with

convertible debt offerings and public bond offerings (see Hovakimian, 2004, for a survey).

Nonetheless, the announcement of a debt or equity issue or an open repurchase

program can hardly be considered a permanent change in capital structure. Indeed,

according to Fama and French (2004), most firms issue, repurchase or do both, every

year. Several studies have shown that many firms use a “target debt ratio”. Dissanaike

et al. (2001) report that approximately 60 percent of the firms in their sample follow a

debt target; Frank and Goyal (2003) find strong evidence of mean reversion in leverage

in their sample, and the target leverage story is also confirmed by Hovakimian et al.

The current issue and full text archive of this journal is available at

www.emeraldinsight.com/1743-9132.htm

IJMF

2,1

4

International Journal of Managerial

Finance

Vol. 2 No. 1, 2006

pp. 4-18

q Emerald Group Publishing Limited

1743-9132

DOI 10.1108/17439130610646144

(2001). Fama and French (2002) put forth statistically reliable evidence of the mean

reversion of leverage, albeit at a low rate. Antoniou et al. (2002) conclude that firms in

France, Germany and the UK adjust their debt ratio to attain their target capital

structure. The abnormal returns following the announcement of debt or equity issues is

thus more easily associated with an information effect than with the impact of a change

in the financial structure. If the target leverage story adequately accounts for the firm’s

financial decisions, any change induced by a new issue would be offset by a

subsequent issue of the complementary financing instrument, a pattern affirmed by

Mayer and Sussman (2005). Thus, the long-term abnormal return on debt, IPO and

seasoned equity issue are also inconclusive with respect to the financial

structure-valuation relationship. The studies mentioned above rely on stock market

returns to assess the effect on financing decisions, and most researchers restrict their

investigation to a very short-term window. Neither event studies nor long-term

analysis of stock return after new issues yield compelling conclusions regarding the

relationship between financial structure and value. These studies generally do not

account for confounding events (simultaneous or subsequent public security offerings),

or for internal or private financing opportunities. The analysis of the valuation effect of

capital structure or changes therein thus necessitates an alternate approach.

Accordingly, Giner and Reverte (2001) propose a more direct test to analyze the

irrelevance proposition. By applying the Ohlson valuation model, they show that

deviation from the optimal debt level is perceived negatively by investors. Using panel

data, Giner and Reverte analyze the influence of debt level on share prices in relation to

several economic arguments. In this paper, I adopt the complementary perspective of

changes in both prices and debt level. I analyze the effect of capital structure changes

on firms’ relative stock prices, using economic arguments such as the target debt ratio.

Changes in the capital structure are considered in a long-term perspective. The

hypothesis that capital structure changes do not affect firm value is tested. Based on a

sample of 243 French firms during the 1987-1996 period, this hypothesis cannot be

rejected because both the decrease and the increase in leverage can be associated with

positive or negative changes in value. To assess the strength of this finding, the

reversion process of the debt ratio towards its target level is also taken into account

(according to the static trade-off theory). There is no evidence of a significant

relationship between the variation in debt and the variation in value.

To my knowledge, this paper is one of the first to analyze the long-term relationship

between financial structure changes and value, and to propose a direct test for the

irrelevance proposition. The remainder of the paper is organized as follows: the first

section is devoted to a brief survey of the two competitive theories of capital structure.

Part two presents the hypothesis and methodology. Part three describes the sample,

while Part four discusses the results. Part five contains the conclusion.

1. Literature review

The static trade-off theory maintains that firms select an optimal capital structure by

trading off the advantages of debt financing against its cost. The optimum debt level

maximizes firm value and should become a target debt level. According to

Shyam-Sunder and Myers (1999):

. . .this static trade-off theory quickly translates into empirical hypotheses. For example, it

predicts reversion of the actual debt ratio towards a target or optimum (. . .).

Changes in

capital structure

5

The target-adjustment pattern is not a specific feature of US firms. Using UK data,

Dissanaike et al. (2001) show that the firms adjust towards a target ratio. Hovakimian

et al. (2001) confirm that:

. . .when firms adjust their capital structures, they tend to move towards a target ratio that is

consistent with theories based on trade-off between the costs and benefits of debt.

They conclude that the static trade-off theory has a much greater time series and

cross-sectional explanatory power than the alternative theory of capital structure (the

pecking order theory), which states that firms rank financing sources by their

sensitivity to asymmetric information. Accordingly, firms prefer internal financing,

followed by debt and then by external financing by equity. Shyam-Sunder and Myers

(1999, p. 221) assert that in this model:

. . . changes in debt ratios are driven by the need for external funds, not by an attempt to

attain an optimal capital structure.

Remolona (1990) confirms that German, Japanese, British and American firms act in

accordance with the pecking order theory. Allen (1993) finds mixed evidence in

Australia in support of the pecking order theory, but acknowledge that further work is

required in this area.

The opposition between these two models of financing is not clear-cut, in spite of the

extensive empirical research that attempts to determine which of these two models

better corresponds with reality. For example, Shyam-Sunder and Myers (1999) argue

that the pecking order theory is an excellent first order descriptor of corporate finance

behavior. However, Frank and Goyal (2003) contradict these results, and found no

out-of-sample validity for Shyam-Sunder and Myers’ propositions. Hovakimian et al.

(2004) suggest that these two conceptual frameworks should be integrated rather than

opposed in order to understand the firm’s financial decisions. As summarized by Fama

and French (2002, p. 20):

. . . in the trade-off and pecking order models, the exogenous driving variables for the level of

leverage are the profitability of assets in place, investment opportunities, non-debt tax shields

and volatility.

These variables are used to estimate target ratios, which are the central point of the

static trade-off theory. Based on evidence from the field, Graham and Harvey (2001)

find some support for both the pecking order and the static trade-off theory. In a

similar paper, Bancel and Mittoo (2004) observe that two main considerations drive the

behavior of European firm managers facing policy decisions: impact on the financial

statements and financial flexibility. Financial flexibility is achieved by selecting the

timing of issues. This observation is confirmed by Baker and Wurgler (2002), who

conclude that capital structure is the cumulative outcome of past attempts to time the

equity markets. Welch (2004) denies managers’ willingness to apply an active financial

policy. In his “inertia” story, the capital structure is determined primarily by external

stock market influences, not by internal corporate optimizing decisions. Assuming that

the timing or inertia proposition adequately describes financial strategies (or the lack

thereof), there should be no relationship between changes in debt ratios and changes in

value.

In this paper, firms simultaneously consider target ratios and factors linked to the

pecking order theory, as evidenced by field studies. More specifically, neither of these

IJMF

2,1

6

theories is excluded: firms establish a target ratio following the static trade-off theory,

but can temporarily diverge from the optimal level, either for historical reasons or for

considerations linked to the pecking order theory. Similarly, the reversion to the target

process may be affected by considerations linked to the pecking order theory.

2. Hypothesis and methodology

I test the irrelevance proposition whereby the capital structure has no impact on firm

value. A direct consequence of this proposition is that a change in the financial

structure should be neutral in terms of value, except for a short-term informational

effect. Therefore, according to the irrelevance proposition, within a long-term

perspective a change in capital structure has no significant effect on the value of the

firm. Consequently, the null hypothesis may be stated as follows:

H0. Capital structure changes do not affect firm value.

This proposition is assessed over a nine-year period, using bivariate tests and

multivariate regressions. However, since the sample comprises a mix of firms going

towards (away from) the target debt level, the analysis of the entire sample may be

biased, which would subsequently indicate that debt and value are unrelated. The

static trade-off theory posits that firm value increases (decreases) as the financial

structure moves closer to (away from) the target. Note that some authors consider that

the target can change (Hovakimian et al., 2001; Frank and Goyal, 2003). In this pattern,

an observed change in leverage can be either a target move or an adjustment of the

debt level to, or from, the unchanged target. In this paper, I do not try to disentangle

these various possibilities; the target ratio is assumed to remain unchanged throughout

the study period[1].

Given that the reversion process requires a measure for the target, Hovakimian et al.

(2001) and Fama and French (2002) use a two-step model in which the target debt ratio

is first estimated by regressing observed debt ratios on the variables used in the

classical cross-sectional studies of financial structure. Several of these variables are

generally associated with the pecking order theory. Although I applied a similar

approach to the data set, the coefficients of determination obtained were generally very

low, estimated coefficients were non-significant, and resulted in a spurious

relationship. Consequently, following Hovakimian (2004), the industry mean

leverage is used as a proxy for the target debt ratio[2].

3. Data and sample description

3.1 Data

Although most of the studies relative to capital structure focused on the American

market, out of sample analyses could be particularly informative when legal and

institutional contexts differ. Clayton et al. (1999) find that on a market value basis,

French companies tend to use a higher proportion of total debt and a higher proportion

of institutional debt (non-spontaneous funds) than US companies. La Porta et al. (1997)

underscore that the French legal structure is differentiated by less defence of investors’

rights compared with other legal systems. They show that countries with poorer

investor protections have smaller and narrower capital markets. In such markets,

equity offerings can be more limited and changes in capital structure should be more

complicated than in the USA. For this reason, changes in the capital structure of

Changes in

capital structure

7

French firms can potentially have a more significant effect on value. Financing choices

of French firms have barely been examined, and the few such investigations analyzed

cross-sectional data exclusively (Rajan and Zingales, 1995).

To allow for a long-term test of the irrelevance proposition, a series of significant

changes in the capital structure is required, and periods of large stock market

fluctuations observed in many countries during the 1997-2001 period must be avoided.

The French firms met both conditions. Indeed, one can observe both increases and

decreases in leverage during this period, but no overall tendency can be detected. This

discrepancy of financial strategies warrants an empirical investigation, and is also

conducive to a study of the long-term relationship between leverage and value. The

sample of French firms seemed to satisfy the second condition as well, since 1987-1996

is a moderate-growth-in-returns period. The CAC 40 capital index advanced from 1,000

to 2,316 between 1987 and 1996 (annual rate of return of 9.7 percent), and then climbed

to 5,958 in 1999 (annual rate of return of 37.03 percent)[3].

The accounting and stock market data originate from the Corporate Information on

the World’s Leading Companies, provided by Worldscope/Disclosure. The sample

initially consisted of all firms recorded in the database for the 1987-1996 period. From

an initial 660 firms reporting consolidated financial statements, the sample is restricted

to firms with available market-to-book ratios for the whole period. The final sample

thus comprised 243 observations.

3.2 Measurement of variables

Following Rajan and Zingales (1995), debt is measured using book value[4], and the

capital structure is proxied by the total debt ratio. This total debt ratio is computed by

adding the long-term debt to the short-term debt and the current portion of the

long-term debt[5]; this amount is then divided by total assets. Financial debt, rather

than total liabilities, is used in this study owing to the preponderance of accounts

payable in the French firms’ financial statements, representing a portion of their

working capital (Rajan and Zingales). The estimated total debt ratio is very close to

that reported by Rajan and Zingales for 1991 (24.76 percent versus 25 percent).

Table I shows a mean total debt ratio of 23.45 percent. The mean equity ratio[6] is

33.29 percent[7]. Table II shows the industrial distribution of the sample and illustrates

the differences between the mean debt levels across the various industries. These mean

debt ratios at the industry level will be used as a proxy for the target debt ratio in the

following sections.

At times, the mean variation masks large variations at the firm level. Table III

shows that 113 firms are increasing their total debt ratios, whereas 130 are reducing

the ratio.

The variation in leverage for each firm (DTDi) was calculated using the growth rate

of leverage between the beginning (t ¼ 1) and the end of the period (t ¼ 9):

DTDi ¼

TDi9

TDi1 _ _2 1 ً1ق

Changes in value (DMBi) are estimated in the same manner as for debt:

DMBi ¼

MBi9

MBi1 _ _2 1 ً2ق

IJMF

2,1

8

Although Tobin’s Q is often used as a proxy for firm value, this variable cannot be

calculated for the French firms because it is not mandatory to report replacement cost

in France. Consequently, the market-to-book ratio (MB) defined as the ratio of market

capitalization to common shareholders’ equity is used. This variable is an indicator of

the goodwill that is created by the firm and recognized by the market. For example,

Fama and French (2002) use V/A to proxy firm value, where V and A are the market

and the book value of the total assets of the firm respectively. Since the debt is

estimated according to the book value, this ratio roughly equals the market-to-book

ratio. Harvey et al. (2004) use a similar proxy.

Industrial distribution SIC Total debt Total debt

Industry From To Number Mean Median

Primary and building 1 19.99 16 13.93 12.93

Consumer goods manufacturing 20 31.99 79 26.52 26.57

Industrial manufacturing 32 35.99 31 21.53 20.78

Technology 36 39.99 41 21.96 20.79

Transportation 40 47.99 11 34.26 31.53

Utilities 48 49.99 7 19.04 16.89

Trade 50 59.99 36 22.65 20.93

Services 70 87.99 22 22.25 18.94

Note: The industry classification is based on SIC codes

Table II.

Descriptive statistics

Capital structure ratios

Total debt ratio Long-term debt ratio Equity ratio

Year Number of firms Mean Mean Mean

% % %

1 243 21.50 13.68 32.47

2 243 23.14 13.63 32.20

3 243 23.78 13.79 32.58

4 243 24.33 13.96 32.93

5 242 24.76 14.65 32.83

6 243 24.62 14.39 33.66

7 240 23.49 14.20 34.16

8 242 22.65 12.94 34.21

9 243 22.75 13.15 34.54

Mean 23.45 13.82 33.29

Note: The capital structure ratios are calculated for 243 non-financial French firms for which

accounting and market data are available for the period of 1987 to 1996. Data come from the

Corporate Information on the World’s Leading Companies database, provided by

Worldscope/Disclosure. The total debt ratio is the sum of the long-term debt and the short-term

debt and the current portion of the long-term debt to the book value of assets. The long-term debt

ratio is the long-term debt to the book value of assets. The equity ratio is the common shareholders’

equity to the book value of assets

Table I.

Descriptive statistics

Changes in

capital structure

9

4. Empirical results

4.1 Bivariate tests of the irrelevance proposition

In order to show the impact of the firm’s debt choice on its market valuation, the

total sample is divided into four sub-samples, on the basis of debt and value

changes. Table III summarizes the distribution of the observations in the four

groups, based on changes in leverage and in market-to-book ratio. In each of the

two sub-samples, namely the increasing and the decreasing leverage sub-groups,

observations are also split in two other sub-samples distributed very closely to their

expected frequency under the null hypothesis of independence between debt and

value changes. The null hypothesis of a random distribution of data among the four

groups cannot be rejected.

This bivariate non-parametric test reveals that it is impossible to reject the

hypothesis of independence between changes in debt level and changes in value.

However, this method does not consider the magnitude of the changes.

Increase in the

market-to-book ratio

Decrease in the

market-to-book ratio Total

Increase in total debt:

Observed frequency 41 72 113

Theoretical frequency 44 69

Chi square contribution 0.17 0.11

Decrease in total debt:

Observed frequency 53 77 130

Theoretical frequency 50 80

Chi square contribution 0.15 0.09

Total absolute value 94 149 243

Total relative value (%) 39 61 100

Chi square 0.52

Notes: For each firm, the variation in leverage is measured by the total debt ratio at the end of the

period (t = 9) divided by the total debt ratio at the beginning (t = 1) minus one. For each firm, the

variation in value is measured by the market-to-book ratio at the end of the period (t = 9) divided by

the market-to-book ratio at the beginning (t = 1) minus one. Chi square test of the null hypothesis: the

changes in debt are independent from the changes in value. Chi square is equal to:

x 2 ¼X

2

i¼1X

2

j¼1

ًFOij 2 FTijق2

FTij

where FO and FT stand for the observed frequency (the actual number of firms in each part of the

Table: 41, 72, 53, 77) and the theoretical frequency (expected number of firms if the null hypothesis is

true), i and j stand for an increase (decrease) in debt and an increase (decrease) in the market-to-book

ratio. If the total debt variation distribution is independent from the market-to-book ratio variation

distribution, the expected theoretical frequency is 44 firms (94 *113/243), should both debt and value

increase. The Chi square contribution is the sum of the differences between observed frequencies and

theoretical frequencies. The null hypothesis is rejected at the 5 percent error level if the Chi square is

greater than 3.84. The calculated Chi square is 0.52

Table III.

Distribution of the 243

firms according to

changes in total debt and

value between 1987 and

1996

IJMF

2,1

10

4.2 Multivariate test of the irrelevance proposition

To assess the reliability of the hypothesis that capital structure changes are not related

to firm value, I use a multivariate parametric model. I control for profitability, growth,

size and asset tangibility, which are known to influence the value of the firm. Table IV

defines the variables, and Table V presents descriptive statistics and Table VI presents

the correlation coefficients.

A multivariate test of the irrelevance proposition can be performed based on an

ordinary least squares regression of the determinants of value, including the change in

leverage. The model is:

Abbreviation Definition

DTDi Change in leverage of company i

Growth rate of total debtb measured by (total debti9/total debti1) 2 1

DMBi Change in value of company i

Growth rate of market-to-book measured by (market-to-booki9/market-to-booki1) 21

PROFi Profitability (nine-year mean value) of company i

S net incomei/S shareholders’ equityi (nine years)

GROWTHi Growth rate of total assets between year 1 and 9 of company i

(total assetsi9/total assetsi1) 2 1

SIZEi Size (nine-year mean value)

Log (1/9 S net sales or revenuesi (nine years))

INTi Intangible ratio (nine-year mean value) of company i

1/9 S intangible ratioi (nine years)

With intangible ratioi ¼ intangible assetsi/total assetsi

DUMi Dummy variable

DUMi ¼ 1 if DGAPi ,0: Debt level of company i reverts towards the target

DUMi ¼ 0 if DGAPi .0 or if the absolute value of DGAP is more than 2:c Debt level

moves away from the target

With:

DGAPi ¼ (GAPi9/GAPi1) 2 1

GAPi9 ¼ debt ratio of the company i at the end of the period (year 9) – the industry

mean over the period

GAPi1 ¼ debt ratio at the beginning of the period (year 1) – the industry mean over

the period

Industry mean over the period ¼ mean of the total debt ratio for all the firms in the

gross sampled over the nine years

Notes: aThe ratios are estimated on an annual basis and the mean is then calculated, aside from

profitability. Net income and shareholders’ equity are added first, to avoid the influence of extreme

data on the mean ratio. Net income excludes special items. If these special items are taken into account,

the mean profitability increases but similar results are obtained; bThe total debt ratio is computed by

adding the long-term debt to the short-term debt and the current portion of the long-term debt. This

amount is then divided by total assets; cIf the change in GAP is negative, the debt ratio moves towards

its target debt level, except if the absolute variation in GAP is larger than two. In such case, the debt

ratio moves away from the target debt level. A negative relationship with value is consequently

expected, which is also true when the change in GAP is positive; dThe gross sample includes all the

French firms that are not in the financial, insurance or real estate sectors, and that have at least three

years of accounting data available (382 firms)

Table IV.

Definition of the variables

using accounts of the

Corporate Information on

the World’s Leading

Companies database,

compiled by

Worldscope/Disclosurea

Changes in

capital structure

11

DMBi ¼ b1 b2PROFi b3GROWTHi b4DTDi b5SIZEi b6INTi 1i ً3ق

When the change in debt affects the value of the firm, the coefficient of this variable

should be significant. The model is applied to three groups of firms: the whole sample,

the sample restricted to firms with positive changes in debt and the sample restricted

to firms with negative changes in their debt ratios. Normally, it is easier for a firm to

increase its indebtedness rather than increasing shareholders’ equity; this situation

could affect the relationship between the change in debt and the change in value.

Table VII shows that determination coefficients range from 6.12 percent in the case

of the sample restricted to positive changes in debt to 7.66 percent in the case of the

whole sample. The only significant coefficients are profitability and size, which are

positively related to changes in value in each sample. For the whole sample, and for the

nine-year period studied, when the mean profitability increases by 1 percent, the value

increases by 0.66 percent. There is no evidence of a significant relationship between the

change in debt and the change in value. In light of these multivariate results, the

irrelevance proposition cannot be rejected.

Variable Number Mean Median Standard deviation Minimum Maximum

DTD 242 0.9210 20.0480 5.4994 20.9943 68.5263

DMB 242 0.0087 20.2063 0.7789 23.5071 3.6071

PROF 242 0.0557 0.0950 0.2839 22.8031 0.6680

GROWTH 237 0.2124 0.0964 0.8226 20.0983 12.2819

SIZE 242 14.8081 14.7643 1.9016 10.2578 19.0631

INT 242 0.0776 0.0509 0.0834 0 0.4408

DUM 242 0.4793 0 0.5006 0 1

Note: One extreme observation (DTD of 662.33) comes from a non-operating firm in the beginning of

the period. This firm (Lagardere SCA) is subsequently excluded from the analyses

Table V.

Descriptive statistics for

the sample of 243 French

firms (Variables are

defined in Table IV)

Pearson correlation coefficients

DTD DMB PROF GROWTH SIZE INT DUM

DTD 1 20.0526 20.0174 0.0645 20.1287 0.0158 0.1561

(0.42) (0.79) (0.32) (0.05) (0.81) (0.02)

DMB 1 0.2326 20.0572 0.1652 20.0269 20.0694

(,0.01) (0.38) (0.01) (0.68) (0.28)

PROF 1 0.0308 20.0022 0.0674 0.0120

(0.64) (0.97) (0.30) (0.85)

GROWTH 1 0.0541 0.2248 0.0455

(0.41) (,0.01) (0.49)

SIZE 1 0.3162 0.0659

(,0.01) (0.31)

INT 1 0.1269

(0.05)

DUM 1

Notes: Numbers in parentheses are the probabilities that the correlation coefficients are different from 0.

One extreme observation (DTD of 662.33) comes from a non-operating firm in the beginning of the

period. This firm (Lagardere SCA) is subsequently excluded from the analyses

Table VI.

Descriptive statistics for

the sample of 243 French

firms (Variables are

defined in Table IV)

IJMF

2,1

12

To assess the strength of the lack of a significant relationship between debt and

value, I control for the static trade-off framework, including a dummy variable that

reflects the reversion towards the target, as defined in Table IV. The model is as

follows:

DMBi ¼ b1 b2PROFi b3GROWTHi b4DTDi b5SIZEi b6INTi

b7DUMi 1i ً4ق

I expect a positive relationship between the dummy variable and the change in value,

because the static trade-off theory posits that firm value should increase (decrease) as

the financial structure moves closer to (away from) the target. The model is tested on

three groups of firms: the whole sample, the sample restricted to firms reverting

towards the target debt level and the sample restricted to firms moving away from

the target. Table VIII shows the results. There is no other significant coefficient. The

null hypothesis that there is a lack of relationship between the change in debt and the

change in value cannot be rejected.

4.3 Robustness check

To assess the strength of these results, a multivariate non-parametric model is applied.

The logit estimation method links changes in values of the independent variables to

increasing or decreasing probability of occurrence of the event being modeled by the

dependent variable. Using a dichotomous variable as a dependent variable makes it

possible to bypass potential error of measures of this variable. With the logit, the

Whole sample

Positive variations in

the debt ratio

Negative variations in

the debt ratio

Parameters Coefficient Coefficient Coefficient

Intercept 21.1006 20.6646 21.4819

(22.80) * * * (21.30) (22.33) * * *

PROF 0.6638 0.7729 0.6337

(3.97) * * * (2.96) * * * (2.79) * * *

GROWTH 20.0530 20.0428 20.0496

(20.87) (20.72) (20.16)

DTD 20.0025 20.0045 0.1674

(20.28) (20.51) (0.59)

SIZE 0.0763 0.0490 0.1044

(2.82) * * * (1.37) (2.49) * * *

INT 20.7393 21.0726 20.4315

(21.17) (21.37) (20.39)

Number 237a 110 b 127 c

Adjusted R square 0.0766 0.0612 0.0698

Notes: Numbers in parentheses are the Student’s t and are significant for an accepted error risk of 10

percent *, 5 percent * *, and 1 percent * * *. a5 observations are lost due to the lack of dependent or

explanatory variables; b2 observations are lost due to the lack of dependent or explanatory variables;

c3 observations are lost due to the lack of dependent or explanatory variables. The model is:

DMBi ¼ b1 b2PROFi b3GROWTHi b4DTDi b5SIZEi b6INTi 1i

Table VII.

Linear regression results

for the model explaining

the change in value. The

sample comprises 242

non-financial French

firms between 1987 and

1996. The dependent

variable is the change in

value (DMBi) as defined

in Table IV

Changes in

capital structure

13

results are less sensitive to the distribution of value changes, since I assign a value of 1

to a positive change in the market-to-book ratio for any given change, and 0 otherwise.

Tables IX and X show the results. The Chi square is never significant, except for the

sample restricted to positive variations in the debt ratio. Profitability and size are

positively related to changes in value in each sample, but the coefficient is either not

significant or weakly significant, and the coefficient associated with the change in debt

is not significant[8].

In light of these multivariate results, the null hypothesis cannot be rejected. Hence, it

seems impossible to find a strong statistical link between changes in debt and the

variation in the market-to-book ratio, regardless of whether I control for reversion

towards the target.

5. Conclusion

This empirical investigation demonstrates that the irrelevance proposition cannot be

rejected. Bivariate and multivariate tests do not show a significant relationship

between changes in value and changes in leverage. With all things being equal,

changes in capital structure apparently do not explain changes in the value of the

sample of French firms. The lack of relationship is observed even when the direction of

change in financial structure is accounted for. The firm’s movement to or away from its

sectoral target level does not alter this conclusion.

Whole sample Reverting towards target Moving away from target

Parameters Coefficient Coefficient Coefficient

Intercept 21.0760 21.4631 20.6893

(22.74) * * * (22.87) * * * (21.10)

PROF 0.6657 0.6232 0.7021

(3.99) * * * (3.20) * * * (2.23) * *

GROWTH 20.0527 20.0458 20.2180

(20.87) (20.78) (20.80)

DTD 20.0006 0.0008 20.0244

(20.06) (0.10) (20.39)

SIZE 0.0780 0.0953 0.0530

(2.88) * * * (2.74) * * * (1.23)

INT 20.6485 20.6279 20.4875

(21.02) (20.73) (20.50)

DUM 20.1262

(21.27)

Number 237a 112b 125c

Adjusted R square 0.0790 0.1054 0.0322

Notes: Numbers in parentheses are the Student’s t and are significant for an accepted error risk of 10

percent *, 5 percent * *, and 1 percent * * *. a5 observations are lost due to the lack of dependent or

explanatory variables; b4 observations are lost due to the lack of dependent or explanatory variables;

c1 observation is lost due to the lack of dependent or explanatory variables.

For the whole sample, the model is:

DMBi ¼ b1 b2PROFi b3GROWTHi b4DTDi b5SIZEi b6INTi b7DUMi 1i

For each sub-sample, the model is:

DMBi ¼ b1 b2PROFi b3GROWTHi b4DTDi b5SIZEi b6INTi 1i

Table VIII.

Linear regression results

for the model explaining

the change in value. The

sample includes 242

non-financial French

firms between 1987 and

1996. The dependent

variable is the change in

value (DMBi) as defined

in Table IV

IJMF

2,1

14

The lack of a relationship between debt and value affirms the propositions of

Modigliani and Miller (1958) and Miller (1977). This result is also consistent with the

timing capital structure story (Baker and Wurgler, 2002), the inertia proposition

(Welch, 2004) and the evidence from the field. However, my empirical results contradict

the observations of Giner and Reverte (2001, p. 310), who conclude that the debt level of

a firm is taken into account by investors in order to determine stock prices. Several

observations can be put forth to explain this contradiction. The first explanation is

linked to the differing samples, periods under study and even countries under

examination. Second, the empirical methodology is quite different from that of the

above authors. Changes are analyzed over a long period, whereas Giner and Reverte

use a cross-sectional model. The valuation models are different, but one can observe

that dividing both sides of the simple form of the Ohlson valuation model used by

Giner and Reverte by the book value of equity yield the relation between the

price-to-book ratio and the return on equity, which is the basis of my analysis.

Furthermore, the contextual variables analyzed in this study are not identical to those

of Giner and Reverte, even though the return on equity (and for a robustness check, the

return on assets) is included in the model. The more plausible explanation is that while

a cross-sectional relationship between value and debt may exist, the numerous factors

that affect firm value in the long run overshadow the debt-value relationship. Clearly,

the relationship between leverage and value, in both the short run and the long run,

warrants more extensive analysis.

Whole sample Positive variations of debt ratio Negative variations of debt ratio

Parameters Coefficient estimate Coefficient estimate Coefficient estimate

Intercept 2.3444 2.8950 1.9919

(4.42) * * (3.16) * (1.43)

PROF 20.5940 23.6463 20.0553

(0.99) (2.02) (0.01)

GROWTH 0.1731 1.1163 20.6480

(0.26) (1.01) (0.63)

DTD 0.0196 0.0192 20.0439

(0.33) (0.27) (,0.01)

SIZE 2 0.1460 2 0.1926 2 0.1132

(3.62) * (2.87) * (1.07)

INT 3.2767 4.7429 2.5548

(2.98) * (2.86) * (0.75)

Number 237a 110b 127c

Chi square 8.1222 11.6263 * * 1.8390

Notes: The changes in leverage (DTDi) and the explanatory variables are defined in Table III.

Numbers in parentheses are the Wald coefficient of Chi square, which is significant for an accepted

risk level of * 10 percent, * * 5 percent, and * * * 1 percent. a5 observations are lost due to the lack of

dependent or explanatory variables; b2 observations are lost due to the lack of dependent or

explanatory variables; c3 observations are lost due to the lack of dependent or explanatory variables.

The estimated model is:

Li ¼ lnً Pi

12Piق ¼ b1 b2PROFi b3GROWTHi b4DTDi b5SIZEi b6INTi 1i

Pi is the probability of observing a variation in the market-to-book ratio of the firm (DMBi):

Pi = P (yi = 1), with yi = 1 if DMBi .0

Table IX.

Logit regression results

for the model explaining

the change in value.

A positive estimated

parameter is associated

with a more likely

decrease in value (group 0)

Changes in

capital structure

15

Notes

1. There is no significant variation in mean debt ratio for the entire sample, which comprises a

mix of increased, decreased and unchanged leverage ratios. Similarly, the mean industrial

ratio used as an estimator of the target ratio seems to be statistically unchanged between the

beginning and the end of the period studied.

2. Hovakimian uses medians. Reported results were obtained with the mean ratio as a proxy.

The use of medians rather than means does not change the results.

3. Moreover, a more recent period has not been used for the following reasons: The French

stock market exhibits strong variation in value during the 1997-1999 period (157.29

percent), and a 48.58 percent decline during the 2000-2002 period. The condition of stability

in value required for the model is clearly not met; The variations in value were stronger for

technology and communications companies, which represent 17 percent of the sample. The

sector effect thus induced can hardly be incorporated in the empirical pattern; The 2000-2001

period is known as a hot issue market period in France (Derrien, 2005). During these years,

numerous firms issued common shares, propelled by very high stock market valuations.

Concomitantly, high market values induced changes in capital structure. This phenomenon

can create spurious correlation in the model; During the 1997-2001 period, the target ratio

estimated as the mean of the sample can hardly be considered constant. However, this is one

of the hypotheses of the model; and updating the sample period would require accounting

Whole sample Reverting towards target Moving away from target

Parameters Coefficient estimate Coefficient estimate Coefficient estimate

Intercept 2.3056 3.7301 1.2807

(4.25) * * (5.19) * * (0.61)

PROF 20.6178 20.0909 24.1499

(1.01) (0.02) (2.10)

GROWTH 0.1876 0.1097 0.6702

(0.27) (0.14) (0.66)

DTD 0.0147 0.0096 20.1102

(0.20) (0.08) (0.26)

SIZE 20.1496 20.2352 20.0616

(3.78) * (4.45) * * (0.30)

INT 3.0741 4.1058 2.0588

(2.61) (2.00) (0.61)

DUM 0.2389

(0.73)

Number 237a 112b 125c

Chi square 8.8551 6.2715 6.0107

Notes: The changes in leverage (DTDi) and the explanatory variables are defined in Table III.

Numbers in parentheses are the Wald coefficient of Chi square, which is at a significant level for an

accepted risk level of * 10 percent, * * 5 percent, and * * * 1 percent. a5 observations are lost due to the

lack of dependent or explanatory variables; b4 observations are lost due to the lack of dependent or

explanatory variables; c1 observation is lost due to the lack of dependent or explanatory variables.

The estimated model for the whole sample is:

Li ¼ lnً Pi

12Piق ¼ b1 b2PROFi b3GROWTHi b4DTDi b5SIZEi b6INTi b7DUMi 1i

For each sub-sample, the model is:

Li ¼ lnً Pi

12Piق ¼ b1 b2PROFi b3GROWTHi b4DTDi b5SIZEi b6INTi 1i

Pi is the probability of observing a variation in the market-to-book ratio of the firm (DMBi):

Pi = P (yi = 1), with yi = 1 if DMBi .0

Table X.

Logit regression results

for the model explaining

the change in value.

A positive estimated

parameter is associated

with a more likely

decrease in value (group 0)

IJMF

2,1

16

and market data for a period of 15 years (1987-2001), but mergers, delistings and other

events during the tumultuous 1997-2001 period result in the disappearance of more than half

of the sample’s firms.

4. It is highly problematic to use market value to measure debt in France due to the small bond

market, which is restricted to the largest firms.

5. Long-term debt represents all interest-bearing financial obligations, excluding amounts due

within one year. Short-term debt and the current portion of long-term debt represent the

portion of debt payable within one year, including current portion of long-term debt, notes

payables arising from short-term borrowings, and various forms of financial debt.

6. The equity ratio is common shareholders’ equity divided by total assets. Common equity

represents common shareholders’ investment in a company, and includes common stock

value, retained earnings and capital surplus.

7. Firms with accounting data only for the eight years are included, when the fiscal year end

was modified. The number of firms consequently varies during the period.

8. To assess the robustness of the results using panel data but keeping a long-term approach, I

also split the sample into two five-year sub-periods (year 1-5 and year 5-9) and I replicated

the analysis on a total sample of 486 observations. Similar results were obtained. In order to

control for the possible endogeneity problem arising from the fact that the accounting rate of

return and the growth of total assets can be related simultaneously to the change in value

and the change in leverage, I supplemented the analysis with a two-stage least square

simultaneous equations model whereby the first equation explains the changes in value and

the second equation the changes in debt by the classical determinants of leverage (growth,

profitability and size). Similar results were obtained.

References

Allen, D.E. (1993), “The pecking order hypothesis: Australian evidence”, Applied Financial

Economics, Vol. 3 No. 2, pp. 101-12.

Antoniou, A., Guney, Y. and Paudyal, K.N. (2002), “Determinants of corporate capital structure:

evidence from European countries”, paper presented at the EFMA 2002 London Meetings,

available at: http://ssrn.com/abstract=302833

Baker, M. and Wurgler, J. (2002), “Market timing and capital structure”, Journal of Finance,

Vol. 57 No. 1, pp. 1-32.

Bancel, F. and Mittoo, U.R. (2004), “Cross-country determinants of capital structure choice:

a survey of European firms”, Financial Management, Vol. 33 No. 4, pp. 103-32.

Clayton, R., Hofler, R.A. and McClure, K.G. (1999), “International capital structure differences

among the G7 nations: a current empirical view”, European Journal of Finance, Vol. 5 No. 2,

pp. 141-64.

Derrien, F. (2005), “IPO pricing in hot market conditions: who leaves money on the table?”,

Journal of Finance, Vol. 60 No. 1, pp. 487-521.

Dissanaike, G., Lambrecht, B. and Saragga Seabra, S.S. (2001), “Differentiating debt target from

non-target firms: an empirical study on corporate capital structure”, JIMS working paper,

University of Cambridge, Cambridge, available at: http://ssrn.com/abstract=293103

Fama, E.F. and French, K.R. (2002), “Testing trade-off and pecking order predictions about

dividends and debt”, Review of Financial Studies, Vol. 15 No. 1, pp. 1-33.

Fama, E.F. and French, K.R. (2004), “Financing decisions: who issues stock?”, CRSP working

paper, No. 549, available at: http://ssrn.com/abstract=429640

Changes in

capital structure

17

Frank, M.Z. and Goyal, V.K. (2003), “Testing the pecking order theory of capital structure”,

Journal of Financial Economics, Vol. 67 No. 2, pp. 217-48.

Giner, B. and Reverte, C. (2001), “Valuation implications of capital structure: a contextual

approach”, European Accounting Review, Vol. 10 No. 2, pp. 291-314.

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 Nos 2-3, pp. 187-243.

Harris, M. and Raviv, A. (1991), “The theory of capital structure”, Journal of Finance, Vol. 46

No. 1, pp. 297-355.

Harvey, C.R., Lins, K.V. and Roper, A.H. (2004), “The effect of capital structure when expected

agency costs are extreme”, Journal of Financial Economics, Vol. 74 No. 1, pp. 3-30.

Hovakimian, A. (2004), “The role of target leverage in security issues and repurchases”, Journal

of Business, Vol. 77 No. 4, pp. 1041-71.

Hovakimian, A., Opler, T. and Titman, S. (2001), “The debt-equity choice”, Journal of Financial

& Quantitative Analysis, Vol. 36 No. 1, pp. 1-24.

La Porta, R., Lopez-De-Silanes, F., Shleifer, A. and Vishny, R.W. (1997), “Legal determinants of

external finance”, Journal of Finance, Vol. 52 No. 3, pp. 1131-50.

Mayer, C. and Sussman, O. (2005), “A new test of capital structure”, paper presented at the AFA

2005 Philadelphia Meetings, available at: http://ssrn.com/abstract=643388

Miller, M.H. (1977), “Debt and taxes”, Journal of Finance, Vol. 32 No. 2, pp. 261-75.

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, pp. 261-97.

Myers, S.C. (1993), “Still searching for optimal structure”, Journal of Applied Corporate Finance,

Vol. 6 No. 1, pp. 4-14.

Myers, S.C. (2001), “Capital structure”, Journal of Economic Perspectives, Vol. 15 No. 2, pp. 81-102.

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, pp. 1421-60.

Remolona, E.M. (1990), “Understanding international differences in leverage trends”, Quarterly

Review – Federal Reserve Bank of New York, Vol. 15 No. 1, pp. 31-43.

Shyam-Sunder, L. and Myers, S.C. (1999), “Testing static tradeoff against pecking order models

of capital structure”, Journal of Financial Economics, Vol. 51 No. 2, pp. 219-44.

Welch, I. (2004), “Capital structure and stock returns”, Journal of Political Economy, Vol. 112 No. 1,

pp. 106-31.

Corresponding author

Ce´cile Carpentier can be contacted at: cecile.carpentier@fsa.ulaval.ca

IJMF

2,1

18

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com

Or visit our web site for further details: www.emeraldinsight.com/reprints