«THE IMPACT OF FINANCIAL STRUCTURE ON FIRMS’ PROBABILITY OF BANKRUPTCY: A COMPARISON ACROSS WESTERN EUROPE CONVERGENCE REGIONS SUCCURRO, Marianna* ...»
THE IMPACT OF FINANCIAL STRUCTURE ON FIRMS’ PROBABILITY OF
BANKRUPTCY: A COMPARISON ACROSS WESTERN EUROPE
The aim of the paper is to investigate the impact of financial structure on firms’
probability of bankruptcy in Western Europe convergence regions. The empirical evidence shows that the financial structure is a key factor explaining firms’ bankruptcy, but while the debt, the cash flow and the profitability ratios are strongly significant in explaining firms’ failure, structure and operational ratios are not relevant factors of bankruptcy. Additional differences arise when we consider the countries separately: while debt and cash flow ratios are significant for bank based economies, they are not significant for market oriented countries.
Keywords: Financial Structure, Probability of Bankruptcy, Convergence Regions JEL classification: D92; E22; G33; L1.
1. Introduction A large amount of research has focused on the relationship between finance and bankruptcy but, to the best of the authors’ knowledge, there is no study that has verified if and to what extent the financial structure impacts the firms’ probability of bankruptcy in Western Europe convergence regions. This is the aim of our paper.
As extension of previous empirical research on both developed and developing countries, our analysis focuses on developing areas of developed economies. The regions included in the analysis share some common characteristics, but they differ along several dimensions including differences in market imperfections (Cleary, 2006), different economic and institutional framework, the origins of the legal systems (La Porta et al., 1998), industry concentration, and so on. World Bank rankings on the ease of doing business show great differences among the considered countries, ranging between rank 7 (United Kingdom) and rank 87 (Italy) (World Bank Doing Business Report, 2012).
Additional features differentiate convergence regions form the rest of the country. These differences would reinforce the generality of the results obtained and the conclusions reached.
While large amount of evidence exists on the relation between financial development, firms’ bankruptcy and growth both cross-countries and cross-industries (Levine, 2005;
Demirguc-Kunt and Maksimovic, 1998; Beck at al., 2005a; Beck at al., 2005b; Aghion et al., 2007; Jeon and Townsend, 2005), much less is known at the microeconomic level of the firm. In this context, the contribution of our research - which relies on accounting data collected from the Bureau van Dijk’s Amadeus database - is twofold. First, we highlight the role of the financial structure in explaining firms’ bankruptcy in Western Europe convergence regions. Second, following the most recent literature in this field, our * Marianna Succurro, firstname.lastname@example.org; Lidia Mannarino, email@example.com;
Department of Economics, Statistics and Financial Sciences, University of Calabria, Italy research is based on an econometric analysis which takes into account several financial ratios instead of the commonly used one-dimensional definition of the financial status.
Empirical studies on developed economies indentify several determinants of firms’ failure, like size, age, industrial characteristics and geographical location but financial variables are not always significant in explaining firms’ bankruptcy. We posit that, in Western Europe convergence regions, where the percentage of SMEs is relatively higher than in more developed regions, financial variables are equally if not more important than the other factors. Given the credit market imperfections, the access to financial market instruments is more limited for small and medium-sized enterprises than for large firms, which can benefit from reputation, privileged access to financial resources, economies of scale in their financing operations and access to stock markets. On one hand, these limitations for SMEs could seriously limit their expansion potential and, therefore, their future survival. On the other hand, firms operating in European convergence regions have also access to EU structural funds which would allow them to diversify their sources of financing and the associated risks. In this context, the final effect of diverse financial structures on firms’ bankruptcy probabilities is worth analyzing.
We can summarize our main empirical results as follows. First, the financial strength is a key factor explaining bankruptcy in Western Europe convergence regions. This result is in line with other studies on developing economies. With reference to additional control variables, size, age and industry are always significant in explaining firms’ probability of bankruptcy. This result, instead, is in line with other studies on developed economies.
Second, some differences arise from a deeper analysis of the financial ratios. While the debt ratio, the cash flow ratio and profitability are strongly significant, structure and operational ratios are not important factors explaining firms’ failure. Third, some differences arise when we consider the countries separately: while debt and cash flow ratios are significant for bank-based countries, they are not significant for United Kingdom, characterized by a developed financial market.
The paper is organized as follows. Section 1 presents a brief overview of the empirical literature on financial structure and bankruptcy. Section 2 illustrates our data and descriptive statistics. Section 3 discusses the empirical strategy and Section 4 focuses on empirical results. The last section concludes the work.
2. Literature review A large number of empirical studies have addressed the issue of financial status since it can significantly affect the firm’s investment and its ability to grow and stay in the market. Several studies stem from the finance and growth literature and are based on cross-country comparison that takes financial variables as given for all firms located in the same country and/or industry (Levine, 2005; Demirguc-Kunt and Maksimovic, 1998;
Beck at al., 2005a; Beck at al., 2005b; Aghion et al., 2007; Jeon and Townsend, 2005).
For what concerns microeconometric studies, financial status has been found to play an increasingly important role on various aspects of firms’ behavior1 such as their investment capacity, their employment and their R&D activities. A lot of evidence seems to exist also about the significant role played by financial constraints in conditioning
firms’ growth and survival (Zingales, 1998; Fotopoulos and Louri, 2000; Geroski and Gregg, 1997; Bunn and Redwood, 2003; Vartia, 2004; Nkurunziza, 2005). A first group of empirical studies relies on a one-dimensional definition of financial constraint, assuming that a single variable can effectively identify the existence of a constraint.
Specifically, several studies categorize firms according to an established characteristic (like dividend payout, size, age, location, group membership, debt rating) designed to measure the level of financial constraints faced by firms2 (Fazzari et al., 1988; Devereux and Schiantarelli, 1990; Gilchrist and Himmelberg, 1995; Kaplan and Zingales, 1997;
Kadapakkam et al., 1998; Greenaway et al., 2005; Cleary, 2006). On the base of the chosen segmenting variable, these researches analyze the impact of financial status on various aspects of firm’ behavior, often producing contradictory findings (Cleary, 2006, p.1561-1562). Departing from Altman (1968), a second group of empirical studies proposes a multivariate analysis for the financial status, often based on multiple discriminant analysis (MDA) or principal component analysis (PCA) which consider an entire profile of characteristics shared by a particular firm and transform them into a univariate statistic (Musso and Schiavo, 2007; Cleary 1999, 2006; Whited and Wu, 2006;
Ginoglou et al., 2002; Lamont et al., 2001). Firms are classified into groups on the base of this beginning-of-period synthetic index which, however, can result from very different financial structures.
An additional weakness of the earlier approaches lies in the choice of a single variable or a synthetic index to classify firms ex-ante (a priori classification). Our study, on the contrary, applies an ex-post classification by distinguishing failed firms (in bankruptcy or in liquidation) and not failed firms (active firms) at time t and including several variables
- each potentially important in affecting the probability of bankruptcy- that can give information on the firm’s financial structure in the previous years. More specifically, in line with Lamont et al. (2001), Cleary (1999, 2006), Whited and Wu (2006), Musso and Schiavo (2007) we use several financial ratios to identify the financial structure which, however, is not used to categorize firms ex-ante, but to estimate the probability of bankruptcy in an econometric analysis.
Moreover, while most microeconometric works use market data for listed enterprises or survey data where firms give self-assessment of their financial status (Winker, 1999;
Becchetti and Trovato, 2002; Holtz-Eakin et al., 1994), we use public balance sheet data for both listed and not-listed firms.
3. Descriptive analysis across Western Europe Convergence Regions This study uses two data sources. The first one is the EU regional policy online database which allows to indentify Western Europe Convergence Regions (Structural and Cohesion funding 2007-13). The selected convergence regions, illustrated by the red areas on Graph 1, include: Calabria, Campania, Puglia, Sicilia (Italy); Andalucia, Extremadura, Castilla-La Mancha, Galicia (Spain); Norte, Centro, Alentejo (Portugal);
Western Wales & The Valleys; South Western (UK); Mecklenburg-Vorpommern, Sachsen-Anhalt, Sachsen, Thüringen (Germany).
2 For a list of papers in chronological order and the segmenting variables used to distinguish among constrained and unconstrained firms see Musso and Schiavo (2007), Table 1 (p.15).
83 The second data source is the Amadeus database, published by Bureau van Dijk. It is a European financial database which includes more than 4 million firms’ accounting data in a standardized balance sheet format. The database includes both SME and large firms operating in all industries.
Our sample, which includes only manufacturing firms, is essentially made up of small and medium enterprises3, which constitute the 93% of the firms in Italy, the 97% in Spain, the 91% in Portugal, the 86% in Germany and the 72% in United Kingdom.
Graph 1A Western Europe Convergence Regions
Source: http://ec.europa.eu/regional_policy/atlas2007/index_en.htm The figure shows the selected convergence regions (RED AREAS): Calabria, Campania, Puglia, Sicilia (Italy); Andalucia, Extremadura, Castilla-La Mancha, Galicia (Spain); Norte, Centro, Alentejo (Portugal); Western Wales & The Valleys; South Western (UK); MecklenburgVorpommern, Sachsen-Anhalt, Sachsen, Thüringen (Germany).
Table 1 shows the percentage of survived (S) and failed (F) firms in each country by size and technological cluster. On the base of the Pavitt’s Taxonomy (Pavitt, 1984; Archibugi,
2001) and departing from NACE 2007 Classification, the main manufacturing sectors are grouped into four clusters with an increasing technological intensity (OECD, 2001;
OECD, 2003): Low Technology (LT); Medium-Low Technology (MLT); Medium-High Technology (MHT) and High Technology (HT).
By comparing the considered convergence regions, the percentage of failed firms is relatively higher in UK for SMEs and in Spain for large companies. In general, in Italy, 3 According to European Union Commission Recommendation 96/280/EC we classify the firms on the base of their relative size: micro and small firms (turnover10mln); medium firms (10mln eurosturnover50mln euros); large firms (turnover50 mln euros).
84 Succurro, M., Mannarino, L. Impact of Financial Structure of Firms on Bankruptcy In European Regions Portugal and UK small and medium enterprises go bankrupt more frequently than large companies. When we focus on the technological clusters, data show a higher percentage of failed firms in high-tech and medium-high-tech sectors in UK and in Italy than in the other countries.
Table 1A in the online Annex illustrates the composition by size – in each industrial cluster - of the two groups of active and failed firms. Data indicate a high firms’ size homogeneity in Italy, Spain and Portugal, where the presence of small firms is very high, independently from the technological intensity of the sector. In Germany and UK, on the contrary, large firms are relatively more numerous than in other countries.
4. Empirical strategy and Econometric specification Our study distinguishes survived firms and failed firms (in bankruptcy or in liquidation) at time t and includes several variables that can give information on their financial structure at time t-τ. Specifically, we consider all the active firms at time t-τ. The dependent variable takes value 1 if the firm, active at time t-τ, is in bankruptcy or in liquidation at time t, 0 otherwise. Failed firms are removed from the Amadeus database after 2 years, hence in our empirical analysis τ=2 years. The financial health is investigated at the start of the recent global financial crisis (2008), and firms’ probability of bankruptcy is evaluated during the next two years.
Following a consolidated methodology (Pederzoli and Torricelli, 2010; Zeitun et al., 2007; Ginoglou et al., 2002; Westgaard and Wijst, 2001, among others), we use a logit analysis to compute the probability of bankruptcy based on several financial ratios used to measure the financial structure of the firms. The aim of the research is to verify if and to what extent the selected financial ratios affect the probability to go bankrupt.
The logistic regression technique allows us to specify the probability of bankruptcy as a
function of a set of explanatory variables. In formal terms:
pi= Pr (Yi = 1) = F(xiβ) where pi is the probability that the dependent variable equals 1 (Y=1), F(.) is the logistic cumulative distribution function, xi is the set of explanatory variables thought to affect pi, and β are the regression coefficients.