Are the determinants of capital structure country or firm specific?

Abstract

In this paper we investigate the capital structure determinants of Greek, French, Italian, and Portuguese small and medium-sized enterprises (SMEs). We compare the capital structures of SMEs across countries and differences in country characteristics, asset structure, size, profitability, risk, and growth and how these may impact capital structure choices. The results show that SMEs in these countries determine their capital structure in similar ways. We attribute these similarities to the country institutional and financial characteristics and the commonality of their civil law systems. However, structural differences arise due to firm specific effects. We find that size is positively related to leverage while the relationship between leverage and asset structure, profitability and risk is negative. Growth is not a statistically significant determinant of leverage for any of the four countries. Our main conclusion is that firm-specific rather than country facts explain differences in capital structure choices of SMEs.

1 Introduction

The increasing interest in studies focusing on the small and medium sized enterprises (SMEs) sector is to a large extent driven by the recognition that SMEs are important engines of economic growth. While much country-level and microeconomic research has been devoted in assessing the importance of SMEs in the economic development and industrialization process (Snodgrass and Biggs 1996; Beck et al. 2005), a cross-country comparative analysis of similarities or differences in the capital structure determinants of SMEs is a topic that has not yet been adequately explored. In this study we address the following questions: Are there any substantial differences in the capital structure choices of SMEs across European countries? If differences do exist, are they due to country- or firm-specific factors? We provide answers to these questions by applying panel data methods to a sample of firms from four EU countries—France, Greece, Italy, and Portugal—for the period 1998–2002.

So far, the literature has focused either on the capital structure determinants of the SMEs of a single country (see for example Van der Wijst and Thurik 1993; Sogorb-Mira 2005; Bartholdy and Mateus 2005) or on cross-country comparisons of capital structure determinants for listed companies (see for example Rajan and Zingales 1995 for the G7 countries; Booth et al. 2001 for 10 developing countries; Ozkan 2001; Bancel and Mitoo 2004; Gaud et al. 2004 for European firms).

An important question in capital structure theory relates to the extent that firms’ financing decisions are driven by their own characteristics rather than being the result of the institutional environment in which they operate (see Rajan and Zingales 1995; Hall et al. 2004). The nature of the firms’ asset structure, the riskiness of the different types of debt, increasing financial distress costs, and the agency cost of outside debt associated with high debt levels have all been suggested as creating firm-specific capital structures (see Watson and Wilson 2002). Given the economic significance of capital structure decisions, an understanding of the relative importance of firm- versus country-specific effects in determining those decisions is an extremely valuable area of research. In addition, the increasing importance of SMEs in economic activity explains why financial policy and the capital structure of the SMEs is a major area of policy concern (Jordan et al. 1998).

Furthermore, it is of interest to examine whether empirical evidence from capital structure choices of large listed companies can also be applied to small firms. Over the last 30 years, the small business research literature has grown significantly resulting in a recognition and acceptance of the idea that small business is specific (Torres and Julien 2005). This specificity concept is dual. First, it refers to differences with large firms; for example, the capital structure decisions of SMEs are constrained by access to financial and credit markets. Second, it refers to the large heterogeneity of the small business field itself. Hence it is important not only to investigate SMEs separately from large companies, but also to test whether this inherent heterogeneity accounts for firm-specific differences within the broad SMEs field. In carrying out the empirical analysis of SMEs capital structure choices we focus on some of the most important regressors that have been used for large firms (Rajan and Zingales 1995; Booth et al. 2001; Ozkan 2001; Gaud et al. 2004).

Comparisons among countries in relation to SMEs’ capital structure determinants are rare in international research. In our knowledge, the only studies that tackle this issue are those of Hall et al. (2004) who investigate the capital structure of SMEs in eight European countries, and Daskalakis and Psillaki (2008) who compare Greek and French SMEs. Hall et al. (2004) report cross-country differences in SME capital structure arguing that these differences are likely to be due to firm- rather than country-specific effects. Daskalakis and Psillaki (2008) provide quantitative evidence establishing that indeed firm-specific factors are responsible for the differences in the capital structure determinants of SMEs for France and Greece.

The aim of this paper is to extend the previous research on country- versus firm-specific differences in capital structure choices of SMEs. In particular, we broaden the Daskalakis and Psillaki (2008) study by examining whether their findings on the relative importance of country- versus firm-specific determinants of capital structure also apply (1) to a larger sample of institutionally similar countries and (2) to a larger set of conditioning variables. This extension is important for the following two reasons. First, by choosing countries with similar institutional characteristics, we test whether the SMEs’ specificity paradigm actually holds. In addressing this question we focus on civil law countries offering low investor and creditor protection. Second, unlike Daskalakis and Psillaki (2008) and Hall et al. (2004), we explicitly account for the role of business risk on contracting costs and thereby on the SMEs’ debt to equity choices. Risk is particularly important within the SMEs context, because it is directly associated with the SMEs’ death rate, which is much higher when compared to large business.

An important question in the SMEs’ literature is the definition of what is really considered to be an SME as these companies are often very heterogeneous.¹ Regarding the SMEs’ definition, we adopt the European Commission SME definition of 1996 ² in which SMEs are defined as companies that employ less than 250 employees, have either an annual turnover not exceeding 40 millions euros, and/or an annual balance-sheet total not exceeding 27 millions euros, and are independent and privately held. SMEs represent 99.8% of the European Union (EU) companies and 66% of the total European employment (European Commission Report 2002).

The structure of the paper is as follows. Section 2 provides a brief background of the institutional and financial features of the four countries. Section 3 presents the different capital structure theories and provides the link from theory to the empirical hypotheses used in this study. Section 4 describes the definition of variables, the data used, and the econometric model employed. Section 5 discusses the empirical results and presents an inter-country comparison. Section 6 concludes the paper.

2 Institutional and financial characteristics

In this section we provide background information on each country’s financial and institutional features and link these characteristics to the finance choices that their SMEs face. The importance of the institutional setting (bankruptcy law, fiscal treatment, ownership concentration, and accounting standards) in identifying the fundamental determinants of capital structure is emphasized by Rajan and Zingales (1995), LaPorta et al. (1998), Beck et al. (2004a), and Beattie et al. (2006).

Beck and Demirguc-Kunt (2006) argue that financial and institutional weaknesses may affect firm size and particularly SMEs from growing to their optimal size.³ Most of the SMEs are microfirms, as shown in Table 1. Microfirms (1–9 employees) comprise more than 93% of all SMEs in the sample of the four countries. On average these firms employ no more than three persons.

Table 1 SMEs in EU-19

Bancel and Mittoo (2004) and LaPorta et al. (1998) classify Greece, Portugal, and Italy among the so-called French Law countries regarding their legal system (civil-law). According to these authors the legal system plays a key role in the availability of external financing in a country. Bancel and Mittoo (2004) argue that the common law system provides better quality of investor and creditor protection than civil-law systems, and among the civil-law systems the French system provides the least protection. Legal structures with little investor and creditor protection exacerbate information asymmetries and contracting costs. These costs tend to be particularly acute in the case of SMEs. As indicated in the European Commission (2003) report on bankruptcy the four countries that we study have in place similar rescue procedures⁴ for potentially viable businesses. However, Portugal and Greece have stricter regimes regarding restrictions, disqualifications, or prohibitions on those who are subject to bankruptcy proceedings.

Beck et al. (2004a) have compiled financial and institutional indicators for several countries. Table 2 summarizes these indicators⁵ for our sample of countries.

Table 2 Financial and institutional indicators

Private Credit and Market Capitalization are two indicators of financial sector development. Table 2 shows that France has a more developed financial sector than the other three countries, while Italy’s financial depth is similar to Portugal’s.

Table 2 also presents information on different features of the institutional system. Judicial Efficiency, Contract Enforcement, and Legal Formalism are summary measures of the efficiency of the legal system in relation to the business environment. France is shown to have in place a more efficient legal system than the other three countries and be less “tolerable” in the process of dispute resolution, in the sense that procedures are quicker and “legal windows” are fewer. Portugal and Italy seem to be more tolerable in case of dispute resolution.

The Corruption and Property Rights indicators provide additional information on the likely influence of the institutional environment on the firm’s financial choices. Strong property rights protection, the absence of corruption, and the efficient enforcement of laws facilitate external funding of firms. According to Beck et al. (2004b) better protection of property rights increases external financing of small firms significantly more than it does for large firms. Property Rights captures the degree of legal protection of private property and the probability that the government will expropriate private property.

Financing tools such as leasing and factoring can be useful in facilitating greater access to finance in the absence of well-developed institutions. In France, Spain and Portugal, leasing is used more often than overdrafts (see Observatory of European SMEs 2003). Italy and France use trade credit more frequently than other European countries. Bank loans seem to be important for Greece and France, where Italy uses more often overdrafts. Factoring seems to be especially important in France.⁶ Furthermore, the ratio of financial debt to total (balance sheet) debt (all debt bearing financial charges to the balance sheets; i.e. trade credit is not included) is rather small in France and Italy, suggesting again that trade credit is used more frequently in these two countries.

We conclude that the four countries seem to present similar financial and institutional characteristics. They all have bank-rather than capital markets-oriented financial systems regarding the provision of external finance. As indicated above bank finance comprises the dominant source of external finance for SMEs. The similarities of the legal and financial systems allow us to hypothesize that in case we find any differences in the determinants of capital structure, these differences are more likely to occur due to firm-specific rather than country-specific factors.

3 Capital structure theory and empirical hypotheses

Leverage determination in the SME theoretical framework does not substantially differ from capital structure theory of the large listed firms, though there are some specific characteristics that should be noted. Capital structure theory dates back to the seminal work of Modigliani and Miller (1958) and their famous “capital structure irrelevance” proposition in relation to the value of firms operating in perfect markets. Subsequent literature placed much emphasis on relaxing the assumptions made by Modigliani and Miller, in particular considering agency costs (Jensen and Meckling 1976; Myers 1977; Harris and Raviv 1990), signaling (Ross 1977), asymmetric information (Myers and Majluf 1984; Myers 1984), product/input market interactions (Brander and Lewis 1986; Titman 1984), corporate control considerations (Harris and Raviv 1988), and taxes (Bradley et al. 1984).

The current state in studies of capital structure comprises a wide variety of theoretical approaches but no theory is universally accepted and practically applied (Myers 2001; Harris and Raviv 1991). According to Myers (2001, p. 81) “There is no universal theory of the debt-equity choice, and no reason to expect one. There are several useful conditional theories however.” The major reason why financing matters includes taxes, differences in information, and agency costs. The different theories of optimal capital structure depend on which economic aspect and firm characteristic we focus on.⁷ For example, the free cash flow theory emphasizes agency costs, the tradeoff theory puts emphasis on taxes, and the pecking order theory focuses on differences in information. The different theories based on taxes, information, and agency costs are relevant for financing choices. However as Myers (2001, p. 99) stressed “Yet none of the theory gives a general explanation of financing strategy.” According to the same author (p. 91) “a fact or a statistical finding is often consistent with two or more competing capital structure theories.” Our main concern is to study SMEs’ capital structure determinants that would cover most of the approaches developed in the capital structure theoretical field. Thus, in the following subsections we analyze how different approaches help us develop hypotheses regarding the main capital structure determinants.

3.1 Asset structure and debt

The agency costs theory is premised on the idea that the interests of the firm’s financial managers and its shareholders are not perfectly aligned. This brings us to Jensen’s (1986, p. 323) free cash flow theory where “The problem is how to motivate managers to disgorge the cash rather than investing it below the cost of capital or wasting it on organizational inefficiencies.” In other words complete contracts cannot be written. Jensen and Meckling (1976) argued for the inevitability of agency costs in corporate finance. Agency costs can also exist from conflicts between debt and equity investors. These conflicts arise when there is a risk of default. The risk of default may create what Myers (1977) defined as “underinvestment” or “debt overhang” problem. Agency costs must be carefully considered within the SME context. There are actually no (or very few) agency costs of equity, because managers are, most likely, also the owners of an SME.⁸ However, agency conflicts between shareholders and lenders may be particularly severe (see Van der Wijst 1989; Ang 1992). The existence of asymmetric information and agency costs may induce lenders to require guarantees materialized in collateral (Myers 1977; Scott 1977; Harris and Raviv 1990). The type of assets that a firm possesses can be considered as an ambiguous factor for the determination of the debt-equity ratio. In fact, as Acs and Isberg (1996) pointed out, capital structure choice is conditioned by the firm’s asset specificity and by implication large and small firms respond differently. This particular factor is closely related to the notion of financial distress costs. Specifically, the costs of financial distress depend on the types of assets that a firm employs. For example, if a firm retains large investments in land, equipment and other tangible assets, it will have smaller costs of financial distress than a firm that relies on intangible assets. Scott (1976) argues that a firm determining the optimal capital structure will issue as much secured debt as possible, because the agency costs of secured debt are lower than those of unsecured debt. Securable assets are considered as being tangible assets such as plant and machinery. Furthermore, Van der Wijst and Thurik (1993) suggest that fixed assets are generally considered to offer more security than current assets. Thus, firms with more tangible assets should issue more debt. This leads to our first testable hypothesis:

H1:

Asset structure will be positively related to debt

However, large holdings of tangible assets may imply that a firm has already found a stable source of return which provides more internally generated funds and discourage it from turning to external financing. Thus, a negative relationship between leverage and asset structure indicates that firms that employ lots of tangible assets seem to rely more on internal funds generated from these assets. Hall et al. (2004) and Sogorb-Mira (2005) find a negative relationship between the short-term debt and the asset structure and a positive relationship between long-term debt and asset structure. In our study, we do not differentiate between long-term and short-term debt. Thus a negative relationship between asset structure and leverage would imply that firms use more short-term debt in their capital structures than long-term debt. Following this line of reasoning we could also hypothesize that:

H1bis:

Asset structure will be negatively related to debt

3.2 Firm size and debt

The tradeoff theory suggests that the optimal capital structure for any particular firm will reflect the balance (at the margin) between the tax shield benefit of debt and the increasing agency and financial distress costs (bankruptcy or reorganization costs) associated with high debt levels (Jensen and Meckling 1976; Myers 1977; Harris and Raviv 1990). The tradeoff theory cannot account for the correlation between high profitability and low debt ratios. When firms are profitable, they should prefer debt to benefit from the tax shield. In addition, if past profitability is a good proxy for future profitability, profitable firms will be less subject to rationing and other supply constraints. However, note that there are some specific capital structure considerations that cannot fit into the SMEs context. Pettit and Singer (1985) have pointed out that tax considerations are of little attention for SMEs because these firms are less likely to generate high profits and therefore are less likely to use debt for tax shields. Myers (1984, p. 588) after a review of the available empirical work concluded that: “there was no study clearly demonstrating that a firm’s tax status has a predictable, material effect on its debt policy. I think that the wait for such a study will be protracted.”

From a financial distress perspective, as larger firms are more diversified they are expected to go bankrupt less often than smaller ones (Warner 1977; Ang et al. 1982; Pettit and Singer 1985), so size must be positively related to leverage. As stated in Sect. 2 the optimal size of firms may be influenced by the quality of a country’s financial and legal system. Thus, the relationship between firm size and institutional development is likely to depend on the relative importance of these two effects. In addition, larger firms may be able to incur lower transaction costs associated with debt. Finally, information costs are lower for larger firms because of better quality (accuracy and transparency) of financial information. In fact, according to the Observatory of European SMEs (2003/2) inadequate company information is often mentioned as one of the main factors hampering bank finance to SMEs.⁹ Therefore, our second hypothesis can be formulated as:

H2:

Firm size will be positively related to debt

However, Rajan and Zingales (1995, p. 1451) have pointed out that the effect of size on equilibrium leverage is more ambiguous: size may also be a proxy for the information outside investors have, which should increase their preference for equity relative to debt. Following this line of reasoning the foregoing hypothesis could be established as:

H2bis:

Firm size will be negatively related to debt

3.3 Growth and debt

Myers (1977) argues that firms with growth potential will tend to have lower leverage. According to the same author growth opportunities can produce moral hazard effects and can push firms to take more risk. This may explain why firms with important growth opportunities will be considered as risky and face difficulties in raising debt capital on favorable terms. Based on Myers (1977) we propose that:

H3:

Growth will be negatively related to debt

On the other hand growth will push firms into seeking external financing, as firms with high growth opportunities are more likely to exhaust internal funds and require additional capital (Michaelas et al. 1999). Growth is likely to put a strain on retained earnings and push the firm into borrowing. Furthermore, the cash flows of firms whose values are most likely to remain stable in the future are predictable and their capital requirements can be financed by debt more easily than those of firms with growth potential (Psillaki and Mondello 1996). This is in line with Acs and Isberg (1996) who found that innovation for small firms is associated with higher levels of debt. Beck and Demirguc-Kunt (2006) analyze the relationship between the economic and financial environment and the SMEs’ access to finance and note that there is substantial evidence that small firms face larger growth constraints and have less access to formal sources of external finance, implying that leverage will be positively related to size. Consequently:

H3bis:

Growth will be positively related to debt

3.4 Profitability and debt

The pecking order theory, developed by Myers and Majluf (1984) and Myers (1984) is a consequence of information asymmetries existing between insiders of the firm and outsiders (i.e. the capital market). Managers adapt their financing policy to minimize the associated costs. More precisely they will prefer internal financing to external financing and risky debt to equity. In such context information asymmetries are relevant only for external financing. According to that theory more profitable firms have more internal financing available. In fact according to the pecking order theory where internal cash flows (retained earnings) are the preferred form of financing new investments, we should expect a negative relationship between leverage and profitability (Harris and Raviv 1991; Rajan and Zingales 1995; Booth et al. 2001). Based on this we propose the following hypothesis:

H4:

Profitability will be negatively related to debt

In a trade-off theory framework an opposite conclusion is expected. Tax shield benefits of debt will induce profitable firms to use more debt (Jensen and Meckling 1976; Myers 1977; Harris and Raviv 1990). Based on the latter we formulate the following:

H4bis:

Profitability will be positively related to debt

The pecking order theory is especially appropriate for small and medium sized firms. These firms do not typically aim at a target debt ratio; instead, their financing decisions follow a hierarchy, with a preference for internal over external finance, and for debt over equity. Thus, according to the pecking order theory, many SMEs would tend to borrow more and more in case their investment needs are typically well in excess of internally generated cash flows. Changes in debt ratios are therefore driven by their need to obtain external funds rather than an attempt to achieve an optimal capital structure. For example, Ang (1991), Holmes and Kent (1991), and Watson and Wilson (2002), emphasized that the pecking order theory can be easily applied to SMEs. SMEs are often opaque and have important adverse selection problems that are explained by credit rationing and therefore bear high information costs (Psillaki 1995). These costs can be considered nil for internal funds but are very high when issuing new capital, whereas debt lies in an intermediate position. Moreover, SMEs are often managed by very few managers whose main objective is to minimize the intrusion in their business and avoid the discipline inherent in financing options other than internal funds. That is why internal funds will lie in the first place of their preference of financing. In case internal funds are not enough, SMEs will prefer debt to new equity mainly because debt means lower levels of intrusion and, most importantly, lower risk of losing control and decision-making power than new equity.

3.5 Risk and debt

Leverage increases the volatility of the net profit. More risky firms can lower the volatility of the net profit by reducing the level of debt. A negative relation between risk and leverage is expected from a pecking order theory perspective: firms with high volatility on earnings try to accumulate cash to avoid under investment issues in the future. From an agency costs or asymmetric information theory perspective we expect a positive relationship (Harris and Raviv 1990; Ross 1977). On the basis of the previous discussion we expect that:

H5:

Risk will be negatively related to debt

In conclusion we have seen from the discussion in Sects. 2 and 3 that financial intermediary development and the efficiency of the legal system affect financial choices of firms and are also likely to affect, asset structure, size, growth, profitability, and firm’s probability of default.

4 Data and the empirical model

4.1 Data and definition of variables

The dependent variable that we use is the debt to assets ratio (DR _i,t ). The debt ratio is defined as the ratio of total liabilities divided by the total assets of the firm (e.g. Rajan and Zingales 1995). Total liabilities include leasing, accounts payable and accounts receivable, namely the trade credit which is an important means of finance for SMEs. For this reason we consider this broader definition. Total liabilities contain both long-term and short-term liabilities.

Following the empirical hypotheses formulated in Sect. 3 the first determinant of capital structure we consider is the asset structure of the firm (AS _i,t ). We measure the asset structure as the ratio of tangible assets divided by the total assets of the firm (e.g. Titman and Wessels 1988; Rajan and Zingales 1995; Frank and Goyal 2003).

The next explanatory variable we use is the size of the firm (SIZE _i,t ). Size is computed as the logarithm of sales (e.g. Titman and Wessels 1988; Rajan and Zingales 1995; Ozkan 2001).

The third determinant of capital structure we use refers to firm’s growth (GROWTH _i,t ), calculated as the annual change on earnings.

Another variable that we consider is profitability (PROFIT _i,t ). We measure profitability by pre-interest and pre-tax operating surplus divided by total assets (e.g. Fama and French 2002; Titman and Wessels 1988).

Finally, we consider the effect of risk (RISK _i,t ) on the firm’s capital structure. Several authors have included a measure of risk as an explanatory variable of the leverage level (Titman and Wessels 1988; MacKie-Mason 1990; Booth et al. 2001). We measure risk as the squared deviation of each year’s earnings before taxes from the period average (Castanias 1983; MacKie-Mason 1990).

We use panel data of SMEs in France, Greece, Italy, and Portugal over the period 1997–2002. As per standard practice, we exclude firms from financial and insurance companies.¹⁰ All data were extracted from the Amadeus database. Amadeus is a database provided by Bureau Van Dijk and harmonizes accounting data from different countries which permits cross-country analysis. The Amadeus database is based on public information.

All firms included in our sample fulfill the criteria of an SME as described in the introductory section. There are 320 Italian SMEs resulting in 1,600 observations; 52 Portuguese firms resulting in 260 observations; 1,252 Greek firms, resulting in 6,260 observations; and 2006 French SMEs resulting in a 10,030 observations balanced panel.¹¹

4.2 The model

We use a balanced panel model. There are several advantages¹² associated with the use of panel models in comparison to the cross-sectional models employed in most previous capital structure studies. First the use of panel data reduces collinearity among the explanatory variables thus improving the precision of econometric estimates. Second, panel data models can take into account a greater degree of the heterogeneity that characterizes firms. This is particularly important in our context as we are specifically interested in controlling for the presence of firm-specific effects in making comparison between the capital structure determinants across countries. Third, panel models also allow for the presence of dynamic effects. Our panel is considered as “typical” in the sense that there are large numbers of cross-sectional units and only a few periods (5 years). The hypothesis that is tested independently for each sample is that the capital structure of the firm expressed by the ratio of total liabilities to total assets depends upon its asset structure, size, growth rate, profitability, and its risk.

$$ {\text{DR}}_{i,t} = \beta _0 + \beta _1 {\text{AS}}_{i,t} + \beta _2 {\text{SIZE}}_{i,t} + \beta _3 {\text{GROWTH}}_{i,t} + \beta _4 {\text{PROFIT}}_{i,t} + \beta _5 {\text{RISK}}_{i,t} + \varepsilon _{i,t} $$

(1)

where: DR _i,t , the debt to assets ratio of firm i at time t; AS _i,t , the asset structure of firm i at time t; SIZE _i,t , the size of firm i at time t; GROWTH _i,t , the percentage change in earnings of firm i between time t and t − 1; PROFIT _i,t , the profitability of firm i at time t; RISK _i,t , the risk of firm i at time t; ɛ _i,t , the error term.

5 Empirical results and cross-national comparisons

In this section we try to get answers to the questions of Sect. 1: Are there any substantial differences in the capital structure choices of SMEs across European countries? If differences do exist, are they due to country- or firm-specific factors? As we mentioned in the theoretical part of this paper (Sect. 3) if there were no country-specific influences on capital structure we would expect the relationships between leverage and asset structure, size, growth, profitability, and finally risk to be uniform across countries. Table 3 presents descriptive statistics of the capital structure and its determinants for France, Greece, Italy, and Portugal.

Table 3 Descriptive statistics

From Table 3 we can see that Italian SMEs seem to maintain the highest leverage in their capital structure, whereas the French SMEs have the lowest debt to assets ratio. This can be explained by the fact that our dependent variable is inclusive of trade credit and, as we mentioned in Sect. 2, Italian firms use this indirect source of financing to the highest extent among all European countries. Another possible explanation may be the fact that Italy seems to be more tolerable in case of dispute resolution. Greek firms are the largest in this sample, whereas the French, Italian, and Portuguese are more or less of similar size. Greek SMEs have the lowest ratio of tangible assets to total assets, while French SMEs seem to maintain more tangible assets in their asset structure. Greek firms seem to be more profitable, whereas Italian SMEs present the highest growth rate. Note that the growth and the profitability rate of the Portuguese SMEs were negative during the period we examine. Greek firms exhibit higher earnings volatility than firms operating in the other countries of our sample. Standard deviations seem to be similar for all variables across countries, except for growth in France, which is much higher than in other countries.

The econometric analysis results are presented in Table 4. The estimation method that we use is Period SUR (Seemingly Unrelated Regression)-pooled EGLS (Estimated Generalized Least Squares). This method¹³ corrects for both arbitrary period serial correlation and period heteroskedasticity between the residuals for a given cross-section. Standard errors and covariances are calculated with (Panel Corrected Standard Error) cross-section weights (PCSE) to obtain robust estimate of the cross-section residual (contemporaneous) covariance matrix. We find that the results are robust to an instrumental variables estimator so there is no obvious endogeneity problem.

Table 4 Empirical results

Asset structure is significant and negatively correlated with leverage except for Portugal. Thus, hypothesis (H1bis) is verified and firms that maintain a large proportion of tangible assets in their total assets tend to use less debt than those which do not. A possible explanation is that firms with lots of tangible assets may have already found a stable source of return which provides them with more internally generated funds and discourage them from turning to external financing.

We find a positive relationship between size and leverage, though significant only for France, Greece, and Portugal verifying hypothesis (H2) for these countries. Our findings are in line with those of Titman and Wessels (1988), Ozkan (2001), Gaud et al. (2005), and Sogorb-Mira (2005) who report results which are consistent with the notion that larger firms have higher debt ratios. As discussed in Sect. 3 this relationship reflects the effects of both asymmetric information and distress costs in the determination of capital structure. From the discussion in Sect. 2 it appears that different perceptions on information asymmetry may be country-specific and related to the structure and size of the financial markets. Larger firms use more debt as they are less informationally opaque and less risky.

The effect of earnings growth is not significant for all countries. Hence we find no evidence that growth is an important determinant of capital structure in our sample.

Profitability is negatively related to leverage for France, Greece, and Italy and is not statistically significant for Portugal. This result supports hypothesis (H4) and this negative relationship leads us to the conclusion that firms that generate relatively high internal funds, generally tend to avoid gearing. This finding is consistent with the pecking order theory which argues that firms prefer internal financing from external. Our findings are in line with Harris and Raviv (1991), Rajan and Zingales (1995), Booth et al. (2001), Gaud et al. (2005), and Ozkan (2001) for large companies and Van der Wijst and Thurik (1993), Hall et al. (2004), and Sogorb-Mira 2005 for SMEs.

Finally, risk is negatively related to leverage and significant only for France and Portugal, supporting hypothesis (H5) for these countries. The negative relationship between risk and leverage is very much as expected, as the riskier the firm, the less debt burden is expected to be carried so as to avoid the possibility of not being able to meet its financial obligations. This result corroborates the evidence reported by Marsh (1982) and Ozkan (2001). Costs of monitoring devices and covenants’ enforcement are likely to be comparatively high for small and young firms, because of the lack of formal financial control and the firms’ flexibility to change their assets (Van der Wijst and Thurik 1993).

The main conclusion of the preceding analysis is that there seem to be similarities in the determinants of capital structure across our sample of countries as the signs of the coefficients are the same for the statistically significant variables. These results are consistent with the discussion in Sect. 2 regarding the similarities of the financial and legal features of these countries. However, even if the direction of the relationship is the same, structural differences may still arise due to possible differences in the magnitude of the coefficients in the country regressions. To test this we apply an F-test by pooling the data for all four countries together to estimate a panel model which restricts the coefficients of the capital structure determinants to be the same for all countries. We then form the F-statistic appending the information obtained from the (unrestricted) individual country models. The F-statistic for this test is given by:

$$ F = \frac{{({\text{RSS}}_{\text{W}} - {\text{RSS}}_{{\text{GR}}} - {\text{RSS}}_{{\text{FR}}} - {\text{RSS}}_{{\text{IT}}} - {\text{RSS}}_{{\text{PO}}} )/3k}} {{({\text{RSS}}_{{\text{GR}}} + {\text{RSS}}_{{\text{FR}}} + {\text{RSS}}_{{\text{IT}}} + {\text{RSS}}_{{\text{PO}}} )/(n - 4k)}} $$

where: RSS_W, residual sum of squares for the restricted model, containing all SMEs; RSS_FR, residual sum of squares for the model containing the French SMEs; RSS_GR, residual sum of squares for the model containing the Greek SMEs; RSS_IT, residual sum of squares for the model containing the Italian SMEs; RSS_PO, residual sum of squares for the model containing the Portuguese SMEs; n, number of observations; k, number of variables.

The value of the F-statistic¹⁴ is equal to 47.84 which is highly significant at conventional levels. We therefore conclude that there are indeed statistical differences in the structure of the relationship between the dependent variable and the regressors across the four countries. It appears that these differences may be due to the differential effects of size and growth on the debt ratio but they may also reflect other country- or firm-specific effects. To investigate this question further we turn next to estimate the panel models by controlling for the presence of fixed (firm-specific) effects in the capital structure relationship.¹⁵

The results of panel regressions for the individual countries and the combined country (restricted model) results are presented in Table B in the Appendix. The new F-statistic is equal to 0.15 which is clearly not significant at conventional levels leading us to the conclusion that firm heterogeneity does affect capital structure determination. This finding actually provides empirical support for the conjecture of Hall et al. (2004) that differences in capital structure across countries are due to firm- rather than country-specific effects and is also consistent with the arguments put forth by Myers (1984). It also corroborates with the results of Balakrishnan and Fox (1993) who found that firm-specific effects contribute most of the variance in leverage.

6 Conclusion

In this paper we analyze the determinants of capital structure for France, Greece, Italy, and Portugal using panel data methods. We find similarities in the determinants of capital structure across our sample countries as the signs of the coefficients are the same for the statistically significant variables. Specifically, as expected size is positively related to leverage. Asset structure is negatively correlated with leverage. Thus, firms that maintain a large proportion of tangible assets in their total assets tend to use less debt than those which do not. Profitability is also negatively related to leverage which is consistent with the pecking order theory that argues that firms prefer internal financing from external. We also find that leverage and risk are negatively related, as expected, which means that the riskier the firm, the less debt burden is expected to carry, so as to avoid the possibility of not being able to meet its financial obligations. Finally, our results show that the growth variable is not statistically significant for any of the four countries in our sample.

Our first conclusion is that there seem to be similarities in the determinants of capital structure across our sample countries. We attribute these similarities in the determinants of capital structure to the commonality of the institutional and legal characteristics of the four countries. However, even if the direction of the relationship is the same, structural differences may still arise due to possible differences in the magnitude of the coefficients in the country regressions. Test results suggest that there are indeed statistical differences in the structure of the relationship between the dependent variable and the regressors.

Our second and main conclusion is that firm rather than country factors explain differences in the intensity of capital structure choices. This finding is obtained by estimating a fixed effects model which controls for firm-specific effects. The results show that firm heterogeneity does account for emerging differences in capital structure determination. Thus we provide evidence indicating that in the case of France, Greece, Portugal, and Italy firm-specific effects, rather than country-specific effects, are responsible for the main differences in the determinants of capital structure thereby supporting the SMEs’ firm-specificity paradigm. This is a rather important result since a lot of previous work (e.g. Rajan and Zingales 1995; Beattie et. al. 2006) has focused on country rather than firm-specific factors in explaining differences in capital structure. Although we have been able to extend the empirical validity of firm relative to country effects on leverage to a larger sample of countries, we maintain that our results and conclusions are limited to the specific sample used to conduct our research and to the specific capital structure determinants used. In future research it will be of interest to examine the robustness of the firm- versus country-specific effects in a sample of common law countries. It will also be of interest to explore further this issue by comparing the capital structure determinants for large and small firms across different countries.

Appendix

In Table A we present the regression results from pooling the observations of all four countries to get the residual sum of squares for the calculation of the F-test. Table B shows the results of the fixed-effects regressions.

Table A France, Greece, Italy, and Portugal

Table B Fixed effects regressions