1

Capital Structure Decisions and Family Ownership: Zero-Leverage Firms,

Firm Size Effects and the Global Financial Crisis

Joaquim J.S. Ramalho

Department of Economics and CEFAGE, University of Évora

Email: jsr@uevora.pt Address: Largo dos Colegiais 2, 7004-516 Évora

Telephone: + 351 266 740 800

Rui M.S. Rita School of Business Science, Setúbal Polytechnic Institute

Email: rui.rita@esce.ips.pt Address: Campus do IPS – Estefanilha

2910 503 Setúbal Telephone: + 351 265 709 300

Jacinto Vidigal da Silva Department of Management and CEFAGE,

University of Évora Email: jsilva@uevora.pt

Address: Largo dos Colegiais 2, 7004-516 Évora Telephone: + 351 266 740 800

Abstract

In this paper we examine the influence of family ownership, firm size and the current financial

crisis on two distinct financial leverage decisions: the probability of using debt; and, conditional

on its use, the proportion of debt issued. Overall, we find a significant positive effect of family

ownership on both decisions; a significant positive effect of firm size on the probability of

family firms using debt; a significant increment on the probability of using debt after 2008 for

all types of firms; and a significant reduction (increment) on the relative amount of debt used

by small (large) levered firms after 2008.

Keywords: family firms, financial crisis, firm size, capital structure, zero leverage.

2

1. Introduction

The study of capital structure exhibited a notable development from 1958 with the

pioneering contribution of Modigliani and Miller. From that time, and for many decades,

research was centered on the study of large quoted companies, mainly in the USA. However,

the economic literature has registered the great importance of both small and medium-sized

enterprises (SMEs) and family-owned firms for the economy. Regarding the latter type of firms,

it is estimated that about 70-80% of European companies (European Commission, 2008) and

90% of USA businesses (Shanker and Astrachan, 1996) are family-owned. Moreover, in

Europe, family firms represent around half of all employment in the private sector and

contribute to gross national product between a minimum of 20% and a maximum of 70%,

depending on the definition of a family firm used; see European Commission (2008).

The main aim of this paper is to analyze the impact of family ownership on firms’ financing

decisions. So far, the evidence collected seems to be contradictory. For example, while some

researchers found that family firms tend to use more debt than their non-family counterparts

(e.g., King and Santor, 2008; Schmid, 2013), others provided evidence on the opposite

relationship (e.g., Agrawal and Nagarajan, 1990; McConaughy, Matthews and Fialko, 2001).

These conflicting findings may be explained, for example, by the use of different definitions of

family firms, by the different economic and institutional context of the firms analysed in each

study (e.g., different countries, different time frames) or by the inclusion or not of the so-called

zero-leverage firms in the analysis (see, e.g., Ramalho and Silva, 2009; Strebulaev and Yang,

2013), but may also be due to methodological problems. For instance, in some cases it appears

that the effects of family ownership and firm size on capital structure decisions may have not

been properly disentangled, given that many empirical papers focussed on a particular group of

size-based firms. Similarly to most of the previous studies, in this paper we use a single

definition of family firm and analyse a single country, but, in contrast to them, we are able, by

using a suitable dataset and appropriate econometric methodology, to undertake an empirical

analysis that isolates the effect of family ownership from the economic context (namely, the

effects of the recent global financial crisis), the zero-leverage phenomenon and firm size issues.

The great heterogeneity of firms with a family influence creates difficulties in finding a

consensual definition of a family firm; for a summary of various definitions of family firms

used in the literature, see Chrisman, Chua and Sharma (2005), Chua, Chrisman and Sharma

(1999), Sharma (2004) and Westhead and Cowling (1998). Recently, González, Alexander and

Trujillo (2013) identified three perspectives to analyze the influence of the family factor on

3

business: management, ownership and control. Throughout this paper we use a definition that

integrates these three perspectives, considering a firm to be family-owned when one individual

or a family possesses more than 50% of its capital and some of its members are directly involved

in management. We focus on the Portuguese case, where family firms account for 70% to 80%

of the business sector, around two thirds of GNP and 50% of employment (European

Commission, 2008).

The first major difference between this paper and previous empirical analyses of family

firms’ financing decisions concerns the dataset used, which comprises firms of all sizes. The

sample is partitioned into micro/small and medium/large firms, which allows us to investigate

whether the effect of family ownership on firms’ capital structure differs across these size-based

groups of firms. As far as we know, this issue has not been directly investigated by any other

paper, since most previous studies focused on particular groups of firms, either small (e.g.,

Coleman and Carsky, 1999; López-Gracia and Sánchez-Andújur, 2007) or large firms (e.g.,

Agrawal and Nagarajan, 1990; Anderson, Mansi and Reeb, 2003; Brailsford, Oliver and Pua,

2002; Mishra and McConaughy, 1999). A related problem is that in papers focussing on small,

family firms, it has been relatively common to explain the financial behavior of small (family)

businesses using arguments that apply directly to family (small) companies but not to any kind

of small (family) firm. However, family-owned firms have characteristics (e.g., overlapping of

roles between being a member of the family and also being a manager of the firm) and aims

(e.g., transferring ownership to the following generations) that differentiate them from firms

controlled by another type of shareholder, irrespective of their dimension.

A second difference is that, because our sample is relative to the period 2006-2012, we are

also able to examine how the current global financial crisis affected the capital structure and

the financing decisions of family and non-family firms of different sizes. To the best of our

knowledge, although there are a few studies analyzing the effects of financial crises on capital

structure decisions, including the current crisis (e.g., Campello, Giambona, Graham and

Harvey, 2011; Campello, Graham and Harvey, 2010; Cowling, Liu and Ledger, 2012; Dang,

Kim and Shin, 2014; McGuinness and Hogan, 2016; Vermoesen, Deloof and Laveren, 2013),

no one has so far analyzed the impact of this recent crisis on the financing decisions of family

firms. Given that Portugal is one of the countries in the European Union more affected by the

current global financial crisis, our data is ideal for such analysis.

Another aspect that is extensively investigated in the paper but so far has not merited any

attention in the financial literature on family-owned firms’ financing decisions concerns the

4

zero-leverage behavior that is common to many small and large firms. Some authors (e.g.,

Ramalho and Silva, 2009; Strebulaev and Yang, 2013) have recently argued that zero-leverage

behavior is a persistent phenomenon and that the factors that determine whether a firm uses

debt at all may be different from those that determine how much debt is used by firms that do

use debt. This paper investigates whether zero-leverage is also a common behavior for family-

owned firms and estimates separately the influence of family ownership on both the

participation (to use debt or not) and amount (conditional on using debt, how much to use)

decisions. In this analysis, we use the most recent econometric techniques available to deal with

binary (participation decision) and fractional (amount decision) panel data, which contrasts

markedly with the methodology employed in previous studies, which often limited themselves

to univariate analysis and, in all cases where multivariate analysis was performed based on

panel data, failed to use econometric methods appropriate for dealing with two basic

characteristics of leverage ratios: by definition, they are bounded by 0 and 1; and, as mentioned,

many firms have null leverage ratios.

The remainder of this paper is organized as follows. Section 2 applies the most common

capital structure theories to the specific context of family-owned firms and formulates the

empirical hypotheses about their financing behavior that are tested in the paper. Section 3

describes the sample and performs a preliminary analysis of the impact of family ownership

and other factors on financial leverage. Section 4 discusses the econometric methodology used

and presents the main empirical results obtained. Section 5 contains some concluding remarks.

2. Capital Structure Decisions and Family Ownership

This section starts with an overview of classical capital structure theories, namely the

pecking-order, agency costs and trade-off theories, and how they may be adapted to the

particular case of family businesses. Then, theories on the effects of firm size and financial

crises and the behavior of zero-debt firms are also briefly reviewed. Finally, the main

hypotheses to be tested in the empirical part of the paper are formulated.

2.1. Classical Capital Structure Theories

The pecking order theory (Myers, 1984; Myers and Majluf, 1984) is one of the most popular

capital structure theories. According to this theory, firms tend to adopt a perfect hierarchical

order of financing: first, they use internal resources and then, if external financing is required,

they prefer debt to outside equity. In general, this behavior is explained in terms of information

asymmetries between firms’ managers and potential outside financiers, which limit firms’

5

access to external finance. In the context of family businesses, the same reasoning obviously

applies and, in fact, given the informational opacity that characterizes most family firms, the

asymmetric information problem tends to be even more serious. In addition, specific demand

side issues related to the characteristics of family businesses may be put forward to reinforce

the importance of the pecking-order theory in this particular framework. This is because one of

the distinctive features of family firms is that often managers are simultaneously owners, which

implies that they have a great ability to modify a firm’s asset portfolio and to use benefits and

channel funds to themselves and the family. Therefore, to preserve this situation, the financing

decisions of owner-managers are frequently driven by the desire to minimize interference in

their businesses and avoid the discipline inherent to the use of external funds (Mishra and

McConaughy, 1999). Moreover, see Brenes, Madrigal and Requena (2011), succession is one

of the main concerns in family firms. Thus, family firms tend to be more conservative, seeking

less external finance, even if that means the loss of opportunities to growth, to prevent dilution

of family control and avoid jeopardizing future generations (Blanco-Mazagatos, Quevedo-

Puente and Castrillo, 2007; López-Gracia and Sánchez-Andújar, 2007). Hence, retained

earnings and personal savings lie in the first place of their preference of financing sources and,

if internal funds are not enough, they will prefer debt to outside equity, because debt means a

lower level of intrusion and, thus, a lower risk of losing control and decision-making power,

exactly as predicted by the pecking-order theory.

The agency costs theory (Jensen and Meckling, 1976) states that the optimal capital

structure of each firm depends on the value of debt that mitigates the conflicts between

stockholders and managers, on the one hand, and stockholders and debtholders, on the other. In

the specific context of family firms, the former type of agency cost is expected to be minimal,

since concentrated ownership and owner-management naturally aligns the owner-managers’

interests about growth opportunities and risk. Therefore, the incentives for issuing debt as a

means of reducing the free cash at managers’ disposal are much less important for family firms.

In contrast, that same close alignment of management’s and shareholders’ interests in family

firms, and the consequent added flexibility of changing the asset base and greater opportunity

to consume perquisites, exacerbates the debtholder-shareholder conflict within the firm, due to

issues like parental altruism and self-control problems (Schulze, Lubatkin, Dino and Buchholtz,

2001), implying higher monitoring costs. Hence, more stringent lending conditions, such as a

higher interest rate and more collateral requirements, may be imposed by lenders on family

firms. Clearly, combining both types of agency costs, family firms are expected to use less

external finance than their non-family counterparts. However, this negative effect of family

6

control on the use of debt may be attenuated by other characteristics of family firms and family

shareholders (e.g., undiversified portfolios, concern about firm and family reputation, longer

investment horizons, desire to pass the firm onto their descendants), which, if recognized by

the lender, may reduce monitoring costs and contribute to more favourable lending conditions.

Indeed, many authors claim that debt holders view family ownership as an organizational

structure that better protects their interests by ensuring continuity and stability; see inter alia

Bopaiah (1998), who provides evidence that family ownership is associated with greater

availability of credit, and James (1999) and Anderson, Mansi and Reeb (2003), who found a

lower cost of debt financing for family firms.

Regarding the trade-off theory, its central aspect is that firms set a target level for their debt-

equity ratio that balances the tax advantages of additional debt (Modigliani and Miller, 1958)

against the costs of possible financial distress and bankruptcy originated by an excessive

amount of debt (Kraus and Litzenberger, 1973). Several papers suggest that family firms use

resources more efficiently (e.g., Anderson and Reeb, 2003a; Chu, 2009; Lee, 2006; Maury,

2006), which may include the development of strategies based on debt usage aimed at reducing

the tax burden. However, note that in family firms similar gains may be obtained simply by

transferring cash income from the firm to the family, thereby reducing the net profits of the

business (Haynes, Walker, Rowe and Hong, 1999). On the other hand, the costs of insolvency

tend to be higher for family firms because of the greater involvement of family owners in their

businesses (Blanco-Mazagatos, Quevedo-Puente and Castrillo, 2007). Indeed, loss of self-

esteem, self-employment and personal assets are particularly relevant issues for family firms.

Overall, the trade-off theory suggests that ceteris paribus family firms may have a different

optimal capital structure, but it is not clear whether the specificities of family firms will lead

them to use more or less external finance.

2.2. Firm Size, Zero-Leverage Firms and Financial Crises

All classical capital structure theories predict a positive relationship between debt usage and

firm size. For instance, according to the trade-off theory, because larger firms tend to be more

diversified, their probability of bankruptcy is relatively smaller (Warner, 1977); and according

to the pecking-order theory, larger firms find it easier to raise debt as in their cases informational

asymmetries are less severe (Myers, 1984). Therefore, all empirical capital structure studies

based on regression analyses include as control variable a proxy for firm size, namely a

quantitative variable such as log(sales) or log(assets). When the study covers firms of very

7

different sizes, it is also common to include in the regression model dummy variables

representing different size-based group of firms. Indeed, as argued by Scherr and Hulburt

(2001), smaller firms are not simply larger firms scaled down: they differ, among other things,

in taxability, flexibility, economies of scale and financial market access. Given the impossibility

of obtaining separate proxies for all these differences, the use of size-based dummy variables

in addition to the quantitative variable measuring firm size is expected to provide a better

approximation of the predicted positive effect of firm size on leverage.

Recently, it has been argued that the positive effect of firm size on leverage becomes

negative when unlevered firms are excluded from the analysis. In particular, Faulkender and

Petersen (2006), Kurshev and Strebulaev (2007), Ramalho and Silva (2009) and Strebulaev and

Yang (2013) found that while larger firms are more likely to have some debt, conditional on

debt issuance, larger firms are typically less leveraged. Moreover, Strebulaev and Yang (2013)

provided evidence that zero-leverage behavior is an important and persistent phenomenon.

Overall, these findings suggest that firm size may affect in inverse ways the participation and

amount debt decisions: on the one hand, it influences positively the probability of a firm having

debt; on the other hand, it affects negatively the (relative) amount of debt issued by levered

firms. Clearly, neither the classical capital structure theories can explain this double effect of

firm size on leverage, nor standard regression models can capture it. In theoretical terms, a

possible explanation based on the costs of external financing was put forward by Kurshev and

Strebulaev (2007). According to them: (i) because they are more affected in relative terms by

the existence of fixed costs in debt issuance, smaller firms opt more frequently for no leverage

and have longer intervals between re-financings, which explains the positive effect of firm size

on the participation debt decision; and (ii) smaller firms choose higher leverage at the moment

of refinancing to compensate for less frequent rebalancing, which explains the negative effect

of firm size on the amount debt decision. Regarding empirical work, the distinct effects that

firm size has on the two decisions is accommodated by the two-part model proposed by

Ramalho and Silva (2009), which uses a binary regression model to explain the probability of

a firm of raising debt and a fractional regression model to explain the relative amount of debt

issued by levered firms and includes in both regressions the quantitative and dummy size

variables mentioned above.

There is a vast literature on the consequences of financial crises over the economy. In

general, any financial crisis seems to reduce firms’ self-financing capacity and its access to

bank credit and new equity (Bernanke, 1983; Kaminsky and Reinhart, 1999; Lemmon and

Roberts, 2010; Chava and Purnanandam, 2011), leading often to debt crises (Borensztein and

8

Panizza, 2009; Levy-Yeyate and Panniza, 2011; Furcery and Zdzienicka, 2012). One of the

most severe economic crises on record is still affecting the advanced world. In Europe, the

financial crisis originated by the 2008 U.S. subprime mortgage transformed into sovereign debt

crises in 2010 in several countries, including Portugal, which had to apply for a bail-out

program in 2011. At the moment, there are already a few studies on the consequences of this

crisis on firms’ leverage decisions. For example, Dang, Kim and Shin (2014) presents evidence

that this crisis had a greater impact on the financing policies of firms with more debt, less size

and greater information asymmetry, while Vermoesen, Deloof and Laveren (2013) found that,

in Belgium, the reduction in the credit supply originated by the global financial crisis had a

greater impact on small firms' investment due to their lower financing capacity. No study of the

effects of the current financial crisis on the capital structure of either family firms or Portuguese

firms seems to be available.

2.3. Empirical Hypotheses

Given that the literature on family firms and capital structure present various arguments

justifying possible differences in the financing decision process of family-owned and non-

family-owned firms, it is not surprising that several empirical studies have focused on the

investigation of whether family firms use a lower or a higher level of debt. So far, the evidence

provided in those papers is mixed, as the following examples illustrate: (i) Agrawal and

Nagarajan (1990) found that all-equity listed USA firms exhibit more extensive family

involvement than levered firms and McConaughy, Matthews and Fialko (2001) observed more

conservative levels of debt in large family firms in the USA; (ii) Coleman and Carsky (1999)

found virtually no differences between family-owned and non-family-owned small US

businesses in the usage of various credit products and Anderson and Reeb (2003b) report that

S&P 500 family firms use similar levels of debt to non-family firms; and (iii) King and Santor

(2008) show that Canadian publicly listed family firms display higher financial leverage than

their non-family counterparts, while Schmid (2013) finds a similar result for several countries

in East Asia and Western Europe.

In this paper we test several hypotheses on the influence of family ownership on capital

structure decisions. The first two hypotheses are very general. First of all, we are interested in

investigating whether there is a relationship between family ownership and debt usage,

irrespective of its sign. Indeed, as discussed in the previous section, from a theoretical point of

view, a positive, a negative or no relationship are all plausible situations and there is no special

9

reason to think a priori that in Portugal some of the arguments put forward in the previous

sections make more sense than the others. Thus, our first hypothesis may be formulated as

follows:

Hypothesis 1: “Family ownership is a relevant factor in determining firms’ financing

decisions.”

Our second general hypothesis regards the stability of the relationship between family

ownership across the three factors discussed in Section 2.2. To the best of our knowledge, most

previous empirical studies on family firms’ capital structure focused on particular size-based

groups of firms and economic contexts and treated levered and unlevered firms in a similar

way, so they were unable to separate in a suitable manner the effects of all these factors on

capital structure choices. For instance, when the analysis is restricted to a sample of SMEs, it

is often argued that ‘family firms have limited sources of external financial capital because (…)

their size normally does not justify bond issues’ (Sirmon and Hitt, 2003). However, this is a

direct consequence of the dimension of the firm and not of its ownership, and so cannot be

generalized to any type of family firm.1 In this paper we analyze the influence of family

ownership conditional on, among other factors, firm size, zero-leverage firms and financial

crises and test the following hypothesis:

Hypothesis 2: “The influence of family ownership on firms’ financing decisions is stable

across small/large firms, levered/unlevered firms and before/after 2008.”

To further examine the issue of zero-leverage firms, we consider also a specific hypothesis

on this subject, investigating the influence of family ownership on both the participation and

amount debt decisions. The ‘two-part theory’ described by Kurshev and Strebulaev (2007) rely

exclusively on firm size: smaller firms are more affected by the existence of fixed costs in debt

issuance. Because this is true irrespective of a firm being family-owned or not, we conjecture

that that theory applies to all types of firms. Therefore, this paper tests the following hypothesis:

Hypothesis 3: “Firm size affects positively the probability of family firms using debt and

negatively the proportion of debt issued by levered family firms.”

10

Regarding the financial crisis, we conjecture that it may have had a double effect on

firms’ capital structure decisions. On the one hand, given the associated economic crisis, the

internal resources generated by firms are expected to have diminished, which may have forced

some of them to resort for the first time to debt. On the other hand, the reduction in the credit

supply originated by the global financial crisis is expected to have reduced the amount of debt

held by firms. We conjecture that this second effect was particularly important for smaller firms

and also for the more information-opaque family firms; see inter alia Beck, Demigurc-Kunt

and Maksimov (2008), Beck, Demirguc-Kunt, Laeven and Maksimov (2006), Berger and Udell

(1998) and Hyytinen and Vaananen (2006). Thus, we formulate the following hypothesis:

Hypothesis 4: “The financial crisis affected positively the probability of firms using debt,

but decreased the proportion of debt used by small, family levered firms.”

3. Data

This section describes the sample used in this study and performs a preliminary analysis of

the effect of family ownership on capital structure decisions. Zero-leverage firms, firm size

effects and the influence of the 2008 financial crisis are also discussed.

3.1. Sample

The dataset used in this paper was taken from the SABI database2, from which some

information was extracted about balance sheets, income statements, concentration of capital

and other characteristics of Portuguese non-financial firms for the period 2006 to 2012. Firms

with non-positive equity, sales or assets were discarded from the analysis as well as firms that

were created in 2004 or later. By applying these selection criteria, we aimed to obtain a sample

that included only effectively operational firms that were not in the final stage of their life-cycle

or at too early a stage. Only firms without missing data for the whole 7-year period were

considered. The final sample contains 9 220 firms and 64 540 observations, from which 27 660

are relative to the period pre-crisis (2006-2008) and 36 880 to the years 2009-2012.

There is no consensus in the empirical literature about how a family firm should be defined,

but, typically, criteria related to firm management, ownership and/or control have been used;

see González, Alexander and Trujillo (2013). The information available on SABI allows us to

consider these three perspectives and, following López-Gracia and Sánchez-Andújar (2007), to

classify as a family-firm those firms where an individual or a family is in possession of more

11

than 50% of the capital and is directly involved in firm management. As a result of this

definition, our sample comprises 4 752 family firms and 4 468 non-family firms. In the family

firms, family members possess on average 88% of the capital and the average number of

owners, managers and managers/owners is, respectively, 3.5, 3.7 and 1.4.

This paper also aims to clarify the effect of the size factor on the financing decision.

Therefore, the sample was divided into two groups: micro/small (small) and medium/large

(large) firms. The process for classification into the two groups followed the criteria of the

European Commission Recommendation of 6 May, number 2003/361/CE, which was

independently applied in each year. Therefore, a firm was classified as small in a given year if

it reported a number of employees under 50 and annual turnover or total assets no greater than

10 million euros. Otherwise, it was classified as a large firm. In each year, our sample comprises

between 7 396 and 7 550 small firms and between 1 670 and 1 824 large firms. Overall, 52 177

observations are relative to small firms and 12 363 to large firms.

As a summary measure of firm’s financing decisions, the ratio of long-term debt (debt with

a maturity of more than one year) to long-term capital assets (defined as the sum of long-term

debt and equity) is used. This measure of leverage was chosen because we are interested in

active capital structure choices of firms, while a non-trivial portion of short-term liabilities may

simply reflect day-to-day business arrangements rather than financial considerations; see Rajan

and Zingales (1995) for an extensive discussion on alternative measures of financial leverage.

As the sample contains mainly unquoted firms, the ratio was calculated based on book values.

Table 1 contains the breakdown of our sample by ownership and other factors. Clearly, on

average, family firms tend to be smaller in size. Nevertheless, around 32.1% of large firms are

family-owned and about 43.9% of small firms are classified as non-family firms, so this sample

is sufficiently diversified to allow us to analyze the separate effects of family ownership and

firm size on capital structure choices. Note also the large proportion of unlevered firms, 34.2%,

which are present in similar proportions in both family and non-family firms.

Table 1 about here

3.2. The impact of family ownership and other factors on financial leverage: a preliminary analysis

Table 2 presents some sample statistics for the selected measure of leverage by ownership

type and firm size for different groups of firms. Considering both levered and unlevered firms,

family firms seem to have similar mean leverage ratios to their non-family counterparts. For

12

both groups of firms, it appears there is a positive relationship between firm size and leverage:

on the one hand, large firms have on average greater mean leverage ratios; on the other hand, a

much higher proportion of small firms are unlevered. This positive relationship seems to hold

after the beginning of the financial crisis. Interestingly, large family firms seem to be clearly

more prone to use debt than their non-family counterparts, while in the case of small firms the

influence of family ownership is less relevant.

Table 2 about here

When the analysis is restricted to levered firms, the relationships suggested by the figures

reported in Table 2 are substantially different. Indeed, conditional on the use of debt, the

greatest leverage ratios are displayed by small firms. These conflicting results for both family

and non-family firms justify our option for studying separately the determinants of the decisions

on using debt or not and how much debt to use. Moreover, by excluding unlevered firms, the

average book leverage ratio increases from 23.6% (family firms) and 23.7% (non-family firms)

to 36.6% and 35.5%, respectively, which suggests that the stylized fact that on average firms

have low leverage ratios relative to what could be expected from various models of capital

structure (Strebulaev and Yang, 2013) may not apply to levered firms and may be better

explained by investigating, as we do in the paper, why some firms tend not to have debt at all.

The 2008 financial crisis does not seem to have changed substantially the mean leverage of

both family and non-family firms. Nevertheless, overall mean leverage ratios increased a little

after the beginning of the crisis, while the corresponding figures for levered firms decreased

between 0.013 (non-family firms) and 0.027 (family firms). More importantly, it seems that all

types of firms (family/non-family, small/large) became clearly more prone to issue debt after

2008.

Overall, this preliminary analysis gives full support to Hypothesis 3 and some support to

Hypothesis 1 and Hypothesis 4. On the other hand, unlike stated in Hypothesis 2, the influence

of the family ownership factor on leverage decisions does not seem to be uniform across

different groups of firms.

4. Econometric analysis

The relationships suggested by Table 2 may be due, at least partially, to other factors

mentioned in the literature as being determinants of capital structure. This section uses

13

regression techniques to investigate whether, once those factors are controlled for, those

relationships remain significant for explaining both the probability of a firm using debt and,

conditional on the former decision, the amount of debt to use.

4.1. Explanatory variables

Characterizing the influence of family ownership on capital structure decisions, isolating it

from the effects of firm size and the 2008 financial crisis, is the main aim of this paper.

Therefore, in the regression analysis that follows we introduce a full set of interaction terms

that allows a clear separation of the effects of the three factors. Each interaction term is defined

as the product of three dummy variables: Family, which equals one if the firm is classified as a

family firm, and is zero otherwise; Small, which equals one if the firm is classified as small,

and is zero otherwise; and DuringCrisis, which equals one if the observation is relative to the

period 2009-2012, and is zero otherwise. In the tables that will be presented later with

estimation results, in order to facilitate their reading, these interaction variables will be denoted

using the terms Family / Non-Family, Small / Large and DuringCrisis / BeforeCrisis, with the

obvious meanings.

On the other hand, over the years, empirical studies on capital structure produced a long list

of factors that are also expected to influence financial leverage decisions. Hence, some of the

most common of those factors are used in this paper as control variables: Size, measured by the

natural logarithm of assets;3 Profitability, the ratio between earnings before interest and taxes

and total assets; Tangibility, the proportion of fixed assets in total assets; Growth, the yearly

percentage change in total sales; Age, the number of years since the foundation of the firm;

Liquidity, the sum of cash and marketable securities, divided by total assets; and eight industry

dummies: Manufacturing Non Equipment; Manufacturing Equipment; Firm Services;

Agriculture and Mining; Construction, Sales; Transportation; and Accommodation. Some of

these variables are expected to have a positive impact on leverage ratios (Profitability, in the

case of the trade-off and agency costs theories; Growth, in the pecking-order theory; Age, in the

agency costs theory; and Tangibility and Size, in all cases), while others are expected to have a

negative effect (Growth, in the trade-off and agency costs theories; and Profitability, Age and

Liquidity, in the pecking-order theory). See inter alia Ramalho and Silva (2009) for a detailed

explanation of all these possible effects.

Table 3 reports some descriptive statistics for the continuous explanatory variables,

showing that non-family firms have on average greater size, profitability, growth and maturity,

14

while family-owned firms present a higher level of liquidity and asset tangibility, which may

be a consequence of the mitigated stockholder-manager agency problems that arise in this

context, as discussed in Section 2. Moreover, firms with zero-leverage ratios have a smaller

proportion of tangible assets, a lower growth rate and are smaller and older than levered firms.

In contrast, they show greater average profitability and liquidity. There are also relevant

differences between small and large firms (e.g., the latter are older, less liquid and more

profitable) and over time periods (e.g., after 2008, the profitability of small firms and the growth

rate of sales for all firms decreased significantly). Most of the mentioned differences are

statistically relevant, which emphasizes the importance of using regression analysis to study the

effect of family ownership on capital structure choices.

Table 3 about here

4.2. Econometric Models

In this paper we are interested in explaining separately both the probability of a firm using

debt and the (relative) amount of debt used by levered firms. Therefore, we use two different

types of econometric models: for the former decision, binary choice models; for the latter

decision, fractional regression models. In both cases, because we are mainly interested in the

effects of (mostly time-invariant) dummy variables, we consider only pooled and random-

effects models.

Let y be the ratio of long-term debt to long-term capital assets, 0 ≤ y < 1, and z be a binary

indicator that takes the values of unity and zero for levered and unlevered firms, respectively.

Then:

1for0 10for 0

Pooled binary choice models for the probability of observing a levered firm are given by:

1| , ,

where is the vector of explanatory variables observed for firm in year t, is the vector of

the parameters of interest and ∙ is a distribution function. The random-effects panel data

models allow for firm-specific effects, assuming that they are normally distributed:

1| , , ~ 0, ,

where is the variance of the firm-specific effects . Both pooled and random-effects binary

regression models are estimated by maximum likelihood.

15

Fractional regression models for responses bounded in the unit interval were first proposed

by Papke and Wooldridge (1996) and further developed by Ramalho, Ramalho and Murteira

(2011) and, for the case of panel data, by Ramalho, Ramalho and Coelho (2016). The pooled

version is given by

| ,

where ∙ is some nonlinear function satisfying 0 ∙ 1 and, as such, may have the same

specification as in binary regression models. However, because in the fractional case there is

only a conditional mean assumption, the model is estimated by quasi-maximum likelihood. On

the other hand, a random-effects model may be written as

| ,

but, unlike the binary case, there are no simple methods to estimate directly this model.

However, because we use fractional regression models to explain the mean debt issued by

levered firms, estimation is performed based only on the sub-sample of firms that use debt.

Thus, no boundary values are observed for and, see Ramalho, Ramalho and Coelho (2016),

the following linearized random-effects model may be estimated:

| ,

where ∙ ∙ .

For both binary and fractional regression models, two alternative specifications were

considered for : a logit specification, where and, hence,

; and a cloglog specification, where 1 and, hence,

1 . The suitability of each specification is assessed using a RESET test,

which, in addition to be powerful against incorrect specification of and , is able to

detect a wide range of misspecifications, including omission of relevant (correlated or not with

the included regressors) covariates. Thus, the RESET test provides also an indirect way of

testing whether the unobserved firm-specific effects are correlated with the explanatory

variables (fixed effects) or not (random effects) and is used instead of the more traditional

Hausman test, which is not applicable in the binary panel data framework. See Ramalho and

Ramalho (2012) for an analysis of the RESET test in the context of binary response models.

4.3. Empirical Results

Tables 4 and 5 report the results obtained for four alternative binary and fractional

regression models, respectively. In each case, all models make identical predictions for the sign

16

of the relationships between the explanatory variables and the dependent variable. However,

there are slight divergences between the pooled and the random effects models in terms of the

statistical significance of some of those relationships. Given that the RESET test rejects the null

hypothesis of the correct specification of some of the pooled models, from now on we focus

only on the results produced by the random effects models.

Table 4 about here

Table 5 about here

Based on the sign and significance of the coefficients associated to the control variables,

our results suggest that the pecking-order theory provides a reasonable explanation of the

capital structure decisions of Portuguese firms. Indeed, the effects on leverage of the control

variables Size (+), Tangibility (+), Age (-) and Liquidity (-) are consistent, in both binary and

fractional models, with the pecking-order theory: as informational asymmetries between

managers and outside investors are typically less severe for larger firms, these firms generally

have better access to the debt market; firms with a larger proportion of tangible assets may find

it easier to issue debt, since it is easier for the lender to establish the value of this kind of assets

in informational opaque firms; and, assuming that firms prefer internal sources of finance, as

the pecking-order theory does, both older firms (which they tend to accumulate retained

earnings) and more liquid firms are expected to require less debt. Moreover, the findings

relative to the two remaining control variables do not contradict the pecking-order theory,

although they do not (entirely) corroborate it either: the positive effect found for Growth in only

significant for levered firms, while Profitability is never a relevant variable.

Most of the interaction terms are statistically significant, which suggests that, in fact, the

classical pecking-order theory does not suffice to explain firms’ financial behavior. Family

ownership, other size effects and the financial crisis may be also important determinants of their

financing policies. Because the interpretation of the estimated coefficients for the interaction

variables is not direct in Tables 4 and 5, Table 6 presents for the logit models the results in a

simpler manner, allowing immediate comparisons across different groups of firms.4

Table 6 about here

The first panel of Table 6 shows clearly that family firms tend to have significantly more

debt than their non-family counterparts, which supports Hypothesis 1. This relationship is valid

17

both for the probability of using debt and the amount of debt used by levered firms, for the

whole period in analysis and for both large and small firms, although in the latter case the effect

is not statistically significant. Therefore, Hypothesis 2 is, at least partially, supported: the effect

of family ownership holds across a range of different groups of firms and scenarios.

The results in the second panel of Table 6 reveal that the effects of firm size on leverage

have not been uniform across groups and over time. Moreover, in most cases the sign of the

effect is the opposite of the one found for the quantitative variable Size included in the

regression, which suggests that, as commented on before, smaller firms are not simply larger

firms scaled down and further investigation, based on new variables characterizing different

aspects related to firm size, is needed. Nevertheless, there is one thing that seems to be clear:

large family firms are more prone to use debt than their smaller counterparts. Hence, there is

only limited evidence on Hypothesis 3.

The 2008 financial crisis led to a clear increment on the probability of firms’ using debt,

see the third panel of Table 6. All types of firms are now more prone to issue debt, probably

because less internal resources were generated as a direct consequence of the economic crisis

after 2008. On the other hand, while the proportion of debt hold by small levered firms

diminished, it increased for large levered firms. This suggests that, on the one hand, smaller

firms were more affected by the reduction in credit supply after 2008 (although more small

firms issue debt, they do it in smaller amounts), particularly in the case of family firms; and

that, on the other hand, many large firms may have been forced to resort to debt due to

increasing difficulties in raising external equity. Overall, Hypothesis 4 is validated.

5. Conclusions

In this paper, we analyzed the effect of family ownership, firm size and the 2008 financial

crisis on the capital structure decisions of levered and unlevered Portuguese firms. The results

reveal that family ownership affects positively both the probability of using debt and the

conditional amount of debt issued. Our results also show the importance of studying

simultaneously the influence of family ownership and firm size on financial leverage and that

failing to do so may bias the conclusions and be at the origins of some contradictory results

found in earlier studies. For example, while the average leverage ratio of family firms (23.6%)

is lower than that of non-family firms (23.7%) and the proportion of family firms that use long-

term debt (59.5%) is also lower than in the non-family case (62.8%), see Table 2, in a regression

analysis controlling for several factors including size category we conclude that family

18

ownership actually influences positively both the probability of using long-term debt and the

conditional amount of debt issued. This suggests that in previous studies not controlling

appropriately for firm size, opposite effects for the family ownership may have been found

simply because family firms are, on average, smaller than non-family firms. Indeed, our

investigation also revealed that larger family firms are more prone to use debt than small family

firms.

The positive effect found for the family ownership factor provides support to those theories

that claim that family firms have the capacity of having higher leverage due to the higher

continuity and stability of their organizational structure, which is implied by factors like

concern with family reputation and longer investment horizons and may lead to more favorable

lending conditions; see inter alia Anderson, Mansi and Reeb (2003), Bopaiah (1998) and James

(1999). The more efficient management of resources that characterizes many family firms (see,

e.g., Andersen and Reeb, 2003a; Maury, 2006; and Chu, 2009) is another factor that may

explain the positive impact of family ownership on financial leverage. Nevertheless, even after

controlling for family ownership, classical determinants of capital structure are still relevant

and we found that their effects conform to some extent to the pecking-order theory.

Regarding the 2008 financial crisis, it appears that it led to an increasing proportion of firms

issuing debt. In terms of debt amount, we found distinct behaviours for small and large firms:

while the average leverage ratios of the former group decreased, those of the latter increased.

Both supply- and demand-side effects of the 2008 financial crisis seem to be the cause of these

changes. On the one hand, the reductions in the credit supply originated by the crisis led to a

reduction of average leverage ratios, particularly of smaller firms. On the other hand, the

reduction in firms’ earnings originated by the economic crisis and the increased difficulty in

raising external equity implied that firms increased their demand for debt.

1 Conversely, some previous studies focusing on generic SMEs may have used abusively theoretical arguments that are specific for family firms. For instance, Ramalho and Silva (2009) explain some of the results found for micro and small firms based on the assumption that most of those firms are family businesses. 2 SABI – Analysis System of Iberian Balance Sheets – is the largest database of financial information about Portuguese firms and is produced by Bureau Van Dijk and managed by Informa, S.A. and BvD. 3 Note that size is included in two different ways in the analysis, both as a quantitative variable (assets) and as a nominal variable (size-based group of firms), since it is assumed that the effects of firm size may vary depending on whether the firm is in fact small or large sized. 4 Considering cloglog instead of logit models does not change any of the conclusions of the paper. To understand how the values in Table 6 were calculated, consider the first figure of this table (0.100), which measures the effect of family ownership for small firms in the period before the financial crisis. The value reported is simply the difference between the coefficients associated to the variables Family * Small * BeforeCrisis (0.254) and Non-Family * Small * BeforeCrisis (0.154).

19

References

Agrawal, A. and Nagarajan, N.J. (1990), “Corporate capital structure, agency costs, and

ownership control: the case of all-equity firms”, Journal of Finance, 45(4), 1325-1331.

Anderson, R.C., Mansi, S.A. and Reeb, D.M. (2003), “Founding family ownership and the

agency cost of debt”, Journal of Financial Economics, 68(2), 263-285.

Anderson, R.C. and Reeb, D.M. (2003a), “Founding-family ownership and firm performance.

Evidence from the S&P 500”, Journal of Finance, 58(3), 1301-1328.

Anderson, R.C. and Reeb, D.M. (2003b), “Founding-family ownership, corporate

diversification, and firm leverage”, Journal of Law and Economics, 46(2), 653-684.

Beck, T., Demirgüç-Kunt, A. and Maksimovic V. (2008), “Financing patterns around the world:

Are small firms different?”, Journal of Financial Economics, 89, 467-487.

Beck, T., Demirgüç-Kunt, A., Laeven, L. and Maksimovic V. (2006), “The determinants of

financing obstacles?”, Journal of International Money and Finance, 25, 932-952.

Berger, A.N. and Udell, G.F. (1998), “The economics of small business finance: the roles of

private equity and debt markets in the financial growth cycle”, Journal of Banking &

Finance, 22(6-8), 613-673.

Bernanke, B.S. (1983), “Nonmonetary effects of the financial crisis in propagation of the Great

Depression”, American Economic Review, 73, 257-276.

Blanco-Mazagatos, V., Quevedo-Puente, E. and Castrillo, L.A. (2007), “The trade-off between

financial resources and agency costs in the family business: an exploratory study”, Family

Business Review, 20(3), 199-213.

Bopaiah (1998), “Availability of credit to family business”, Small Business Economics, 11, 75-

86.

Borensztein, E., and Panizza U. (2009), “The costs of sovereign default”, IMF Staff Papers,

56(4), 683-741.

Brailsford, T.J., Oliver, B.R. and Pua, S.L.H. (2002), “On the relation between ownership

structure and capital structure”, Accounting and Finance, 42, 1-26.

Brenes, E. B., Madrigal, K. and Requena, B. (2011), "Corporate governance and family

business performance”, Journal of Business Research, 64, 280-285.

Campello, M, Graham, J.R. and Harvey, C.R. (2010), “The real effects of financial constraints:

evidence from a financial crisis”. Journal of Financial Economics, 97(3), 470-487.

20

Campello, M., Giambona, E., Graham, J., and Harvey, C. (2011), “Liquidity management and

corporate investment during a financial crisis”, Review of Financial Studies, 24, 1944–

1979.

Chava, S. And Purnanandam, A. (2011), “The effect of banking crisis on bank-dependent

borrowers”, Journal of Financial Economics, 99, 116-135.

Chrisman, J.J., Chua, J.H. and Sharma, P. (2005), “Trends and directions in the development

of a strategic management theory of the family”, Entrepreneurship Theory and Practice,

29 (5), 555-576.

Chu, W. (2009), “The influence of family ownership on SME performance: evidence from

public firms in Taiwan”, Small Business Economics, 33(3), 353-373.

Chua, J. H., Chrisman, J.J. and Sharma, P. (1999), “Defining the family business by behavior”,

Entrepreneurship Theory and Practice, 23(4), 19-39.

Coleman, S. and Carsky, M. (1999), “Sources of capital for small family-owned businesses:

evidence from the national survey of small business finances”, Family Business Review,

12(1), 73-86.

Cowling, M., Liu, W. and Ledger, A. (2012), “Small business financing in the UK before and

during the current financial crisis”, International Small Business Journal, 30(7), 778–

800.

Dang, V.A, Kim, M. and Shin, Y. (2014), “Asymmetric adjustment toward optimal capital

structure: Evidence from a crisis”, International Review of Financial Analysis, 33, 226-

242.

European Commission (2008), Overview of Family Business Relevant Issues, Final Report.

Available at:

http://ec.europa.eu/DocsRoom/documents/10389/attachments/1/translations/en/renditio

ns/pdf.

Faulkender, M. and Petersen, M.A., (2006), “Does the source of capital affect capital

structure?”, Review of Financial Studies, 2006, 19(1), 45-79.

Furceri, D., and Zdzienicka, A. (2012), “How costly are debt crises?”, Journal of International

Money and Finance, 31(4), 726–42.

González, M.G., Alexander, P.C. and Trujillo, M.A. (2013), "Family firms and debt: risk

aversion versus risk of losing control", Journal of Business Research, 66, 2308-2320.

Haynes, G.W., Walker, R., Rowe, B.R. and Hong, G.-S. (1999), “The intermingling of business

and family finances in family-owned businesses”, Family Business Review, 12(3), 225-

239.

21

Hyytinen, A. and Väänänen, L. (2006), “Where do financial constraints originate from? An

empirical analysis of adverse selection and moral hazard in capital markets”, Small

Business Economics, 27, 323-348.

James, H.S. (1999), “Owner as manager, extended horizons and the family firm”, International

Journal of the Economics of Business, 6(1), 41-55.

Jensen, M.C. and Meckling, W.H. (1976), “Theory of the firm: managerial behaviour, agency

costs, and ownership structure”, Journal of Financial Economics, 3(4), 305-360.

Kaminsky, G.L. and Reinhart, C.M. (1999), “The twin crisis: the causes of banking and balance

of payments problems”, American Economic Review, 89, 473-500.

King, M.R. and Santor, E. (2008), “Family values: ownership structure, performance and

capital structure of Canadian firms”, Journal of Banking and Finance, 32, 2423-2432.

Kraus, A., and Litzenberger, R. (1973), “A state-preference model of optimal leverage”,

Journal of Finance, 28, 911-920.

Kurshev, A. and Strebulaev, I.A. (2007), “Firm size and capital structure”, mimeo.

Lee, J. (2006), “Family firm performance: further evidence”, Family Business Review, 19(2),

103-114.

Lemmon, M. And Roberts, M.R. (2010), “The response of corporate financing and investment

to change in the supply of credit”, Journal of Financial and Quantitative Analysis, 45,

555-587.

Levy-Yeyati, E. L. and Panizza U. (2011), “The elusive costs of sovereign defaults,” Journal

of Development Economics, 94(1), 95–105.

López-Gracia, J. and Sánchez-Andújar, S. (2007), “Financial structure of the family business:

evidence from a group of small Spanish firms”, Family Business Review, 20(4), 269-287.

Maury, B. (2006), “Family ownership and firm performance: empirical evidence from Western

European corporations”, Journal of Corporate Finance, 12, 321-341.

McConaughy, D.L., Matthews, C.H. and Fialko, A.S. (2001), “Founding family controlled

firms: performance, risk, and value”, Journal of Small Business Management, 39(1), 31-

49.

McGuinness, G. and Hogan, T. (2016), “Bank credit and trade credit: evidence from SMEs over

the financial crisis”, International Small Business Journal, 34(4), 412-445.

Mishra, C.S. and McConaughy, D.L. (1999), “Founding family control and capital structure:

the risk of loss of control and the aversion to debt”, Entrepreneurship Theory and

Practice, 23, 53-64.

22

Modigliani, F. and Miller, M.H. (1958), “The cost of capital, corporation finance and the theory

of investment”, American Economic Review, 48(3), 261-297.

Myers, S. C. (1984), “The capital structure puzzle”, Journal of Finance, 39(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,

13(2), 187-221.

Papke, L.E. and Wooldridge, J.M. (1996), “Econometric methods for fractional response

variables with an application to 401(k) plan participation rates”, Journal of Applied

Econometrics, 11(6), 619-632.

Rajan, R. and Zingales, L. (1995), “What do we know about capital structure? Some evidence

from international data”, Journal of Finance, 50, 1421-1460.

Ramalho, E.A. and Ramalho, J.J.S. (2012), “Alternative versions of the RESET test for binary

response index models: a comparative study”, Oxford Bulletin of Economics and

Statistics, 74(1), 107-130.

Ramalho, E.A., Ramalho, J.J.S. and Coelho, L. (2016), “Exponential regression of fractional-

response fixed-effects models with an application to firm capital structure”, Journal of

Econometric Methods, forthcoming.

Ramalho, E.A., Ramalho, J.J.S. and Murteira, J. (2011), “Alternative estimating and testing

empirical strategies for fractional regression models”, Journal of Economic Surveys,

25(1), 19-68.

Ramalho, J.J.S. and Silva, J.V. (2009), “A two-part fractional regression model for the financial

leverage decisions of micro, small, medium and large firms”, Quantitative Finance, 9(5),

621-636.

Scherr, F.C. and Hulburt, H.M. (2001), “The debt maturity structure of small firms”, Financial

Management, 30(1), 85-111.

Schmid, T. (2013), “Control considerations, creditor monitoring, and the capital structure of

family firms”, Journal of Banking and Finance, 37(2), 257-272.

Schulze, W.S., Lubatkin, M.H., Dino, R.N. and Buchholtz, A.K. (2001), “Agency relationships

in family firms: theory and evidence”, Organization Science, 12(2), 99-116.

Shanker, M.C. and Astrachan, J.H. (1996), “Myths and realities: family businesses’

contribution to the U.S. economy”, Family Business Review, 9(2), 103-124.

Sharma, P. (2004), “An overview of the field of family business studies: current status and

directions for the future”, Family Business Review, 17(1), 1-36.

23

Sirmon, D.G. and Hitt, M.A. (2003), “Managing resources: linking unique resources,

management, and wealth creation in family firms”, Entrepreneurship Theory and

Practice, 27(4), 339-358.

Strebulaev, I.A. and Yang, B. (2013), “The mystery of zero-leverage firms”, Journal of

Financial Economics, 109(1), 1-23.

Vermoesen, V., Deloof, M. And Laveren, E. (2013), “Long-term debt maturity and financing

constraints of SMEs during the Global Financial Crisis”, Small Business Economics, 41,

433-448.

Warner, J.B. (1977), “Bankruptcy costs: some evidence”, Journal of Finance, 32(2), 337-347.

Westhead, P. and Cowling, P. (1998), “Family firm research: the need for a methodological

rethink”, Entrepreneurship Theory and Practice, 23(1), 31-56.

24

Table 1 – Sample Family firms Non-family firms All firms # % # % # % By leverage: Zero-leverage firms 11 744 35.3 10 351 33.1 22 095 34.2 Levered firms 21 520 64.7 20 925 66.9 42 445 65.8 By Size: Small firms 29 290 88.1 22 887 73.2 52 177 80.8 Large firms 3 974 11.9 8 389 26.8 12 363 19.2 By Period: 2006-2008 14 256 42.9 13 404 42.9 27 660 42.9 2009-2012 19 008 57.1 17 872 57.1 36 880 57.1 Total 33 264 100.0 31 276 100.0 64 540 100.0

25

Table 2 – Leverage Small firms Large firms All firms Family Non-family Family Non-family Family Non-family

Mean leverage ratios Both levered and zero-leverage firms

2006-2008 0.224 0.220 0.252 0.247 0.227 0.228 2009-2012 0.239 0.240 0.274 0.257 0.243 0.245 All years 0.233 0.232 0.265 0.253 0.236 0.237

Only levered firms 2006-2008 0.392 0.374 0.325 0.339 0.382 0.363 2009-2012 0.359 0.354 0.329 0.340 0.355 0.350 All years 0.372 0.361 0.328 0.339 0.366 0.355

Percentage of zero-leverage firms 2006-2008 42.8 41.0 22.7 27.0 40.5 37.2 2009-2012 33.5 32.0 16.7 24.4 31.4 30.0 All years 37.5 35.6 19.2 25.5 35.3 33.1

26

Table 3 – Average values for the non-dummy explanatory variables

2006-2008 2009-2012 All

Small Large Small Large

Family

Non-family

Family Non-

family Family

Non-family

Family Non-

family

Family Non-

family All firms

Size 13.861 14.145 15.492 16.215 13.976 14.264 15.643 16.323 14.124 14.767 Profitability 0.044 0.043 0.046 0.051 -0.053 -0.006 0.037 0.050 -0.005 0.024 Tangibility 0.273 0.264 0.292 0.291 0.257 0.248 0.278 0.262 0.266 0.260 Growth 0.135 0.232 0.107 1.640 0.039 0.056 0.018 0.020 0.077 0.287 Age 19.269 20.596 25.447 27.312 22.853 24.419 28.077 29.859 21.989 24.385 Liquidity 0.105 0.096 0.072 0.069 0.112 0.102 0.089 0.081 0.106 0.093

Zero-leverage firms Size 13.610 13.893 15.098 16.118 13.740 14.002 15.231 16.286 13.773 14.417 Profitability 0.045 0.045 0.052 0.052 0.005 0.002 0.044 0.049 0.026 0.029 Tangibility 0.249 0.224 0.235 0.188 0.221 0.203 0.223 0.177 0.234 0.207 Growth 0.178 0.287 0.129 0.087 -0.005 0.061 0.009 -0.001 0.084 0.144 Age 19.538 21.286 26.260 28.827 24.145 25.460 28.612 32.542 22.246 24.954 Liquidity 0.135 0.131 0.117 0.112 0.158 0.149 0.146 0.135 0.146 0.137

Levered firms Size 14.049 14.320 15.608 16.251 14.094 14.387 15.725 16.335 14.315 14.940 Profitability 0.043 0.042 0.044 0.051 -0.082 -0.010 0.036 0.051 -0.022 0.022 Tangibility 0.292 0.292 0.309 0.328 0.276 0.270 0.289 0.290 0.284 0.287 Growth 0.102 0.193 0.100 2.213 0.061 0.054 0.019 0.027 0.073 0.357 Age 19.067 20.118 25.209 26.753 22.203 23.928 27.969 29.991 21.849 24.104 Liquidity 0.083 0.072 0.059 0.053 0.089 0.079 0.078 0.063 0.084 0.071

27

Table 4 – Binary regression models for the probability of using debt

Variable Pooled models Random effects models

Logit Cloglog Logit Cloglog Size 0.373*** 0.217*** 0.860*** 0.536***

(0.018) (0.011) (0.032) (0.021) Profitability -0.005 -0.001 -0.007 -0.005

(0.013) (0.002) (0.009) (0.005) Tangibility 1.700*** 0.943*** 2.676*** 1.638***

(0.097) (0.053) (0.166) (0.103) Growth 0.000 0.000 0.000 0.000

(0.000) (0.000) (0.000) (0.000) Age -0.010*** -0.006*** -0.017*** -0.010***

(0.001) (0.001) (0.002) (0.002) Liquidity -2.620*** -1.866*** -2.569*** -1.734***

(0.127) (0.087) (0.189) (0.125) Non-Family * Large * DuringCrisis 0.208*** 0.110*** 0.316*** 0.212***

(0.052) (0.030) (0.107) (0.063) Non-Family * Small * BeforeCrisis 0.131* 0.075* 0.154 0.092

(0.072) (0.043) (0.122) (0.077) Non-Family * Small * DuringCrisis 0.598*** 0.353*** 0.990*** 0.604***

(0.071) (0.042) (0.121) (0.076) Family * Large * BeforeCrisis 0.425*** 0.270*** 0.632*** 0.397***

(0.103) (0.056) (0.174) (0.109) Family * Large * DuringCrisis 0.884*** 0.492*** 1.326*** 0.782*** (0.108) (0.055) (0.179) (0.108) Family * Small * BeforeCrisis 0.121* 0.063 0.254** 0.163** (0.072) (0.044) (0.126) (0.081) Family * Small * DuringCrisis 0.600*** 0.357*** 1.074*** 0.658*** (0.071) (0.042) (0.125) (0.080) Constant -5.332*** -3.424*** -12.480*** -8.295***

(0.325) (0.195) (0.588) (0.380) RESET test 34.40*** 0.390 0.17 1.42 Number of observations 64 540 64 540 64 540 64 540

Notes: ***, ** and * denote regression coefficients and test statistics which are significant at 1%, 5% and 10%; all models include industry dummies.

28

Table 5 – Fractional regression models for the probability of using debt

Variable Pooled models Random effects models

Logit Cloglog Logit Cloglog Size 0.003 0.004 0.088*** 0.067*** (0.011) (0.009) (0.016) (0.013) Profitability -0.002*** -0.001*** 0.000 0.000 (0.001) (0.000) (0.000) (0.000) Tangibility 0.930*** 0.732*** 1.288*** 1.069*** (0.051) (0.040) (0.076) (0.062) Growth 0.001*** 0.000*** 0.000*** 0.000*** (0.000) (0.000) (0.000) (0.000) Age -0.008*** -0.006*** -0.017*** -0.014*** (0.001) (0.001) (0.001) (0.001) Liquidity -1.778*** -1.464*** -1.154*** -1.020*** (0.101) (0.095) (0.103) (0.093) Non-Family * Large * DuringCrisis 0.077*** 0.058*** 0.090** 0.080** (0.027) (0.022) (0.038) (0.034) Non-Family * Small * BeforeCrisis 0.176*** 0.137*** 0.188*** 0.150*** (0.039) (0.031) (0.051) (0.044) Non-Family * Small * DuringCrisis 0.149*** 0.116*** 0.182*** 0.150*** (0.037) (0.030) (0.049) (0.043) Family * Large * BeforeCrisis -0.027 -0.022 0.169** 0.152*** (0.049) (0.040) (0.067) (0.058) Family * Large * DuringCrisis 0.060 0.047 0.300*** 0.273*** (0.045) (0.036) (0.062) (0.054) Family * Small * BeforeCrisis 0.265*** 0.208*** 0.347*** 0.285*** (0.040) (0.032) (0.054) (0.047) Family * Small * DuringCrisis 0.170*** 0.134*** 0.271*** 0.233*** (0.039) (0.031) (0.053) (0.046) Constant -0.690*** -0.913*** -2.590*** -2.529*** (0.196) (0.159) (0.280) (0.239) RESET test 3.183* 0.007 2.00 0.00

Number of observations 42 445 42 445 42 445 42 445 Notes: ***, ** and * denote regression coefficients and test statistics which are significant at 1%, 5% and 10%; all models include industry dummies.

29

Table 6 – Random effects logit regression models Binary Fractional

Family ownership effects (Family firms – Non-family firms) 2006-2008 2009-2012 2006-2008 2009-2012 Small firms 0.100 0.084 0.159*** 0.088*** (0.078) (0.076) (0.037) (0.034) Large firms 0.632*** 1.010*** 0.169** 0.210***

(0.174) (0.179) (0.067) (0.061) Size effects (Large firms – Small firms)

2006-2008 2009-2012 2006-2008 2009-2012 Family firms 0.378*** 0.252* -0.178*** 0.030 (0.146) (0.153) (0.054) (0.047) Non-family firms -0.154 -0.673*** -0.188*** -0.092* (0.122) (0.121) (0.051) (0.048)

Crisis effects (During crisis – Before crisis) Small Large Small Large Family firms 0.820*** 0.694*** -0.076*** 0.132*** (0.051) (0.164) (0.021) (0.045) Non-family firms 0.836*** 0.316*** -0.005 0.090**

(0.060) (0.107) (0.024) (0.038) Notes: ***, ** and * denote regression coefficients which are significant at 1%, 5% and 10%.