ساختار بازار اعتباری و غربالگری بانکی: آزمون غیر مستقیم بر روی داده های ایتالیایی

Based on a large panel of Italian SMEs, this paper focuses on the relationship between firms' default probability and the amount of bank debt they obtain, evaluating whether and to what extent this link is affected by the degree of competition characterizing the local credit market where firms operate. Using a dynamic panel estimator, we find that higher bank competition implies a stronger influence of firms' riskiness on bank financing to SMEs. We provide two plausible interpretations of this finding: one resorting to more accurate screening by more competitive banks; the other alleging lower market power of incumbent banks, which may restrict their willingness to finance riskier firms.

In the last decade an intense dispute in the economic literature has been centered on the question “is competition among banks good or bad?”. As Cetorelli (2001) claims, the need for such a debate would be unjustified if banks' role were simply to intermediate between supply and demand of credit. In that case, in fact, there would not be reasons to treat banks differently from other firms or to doubt that market power in credit markets is likely to lead to welfare losses, as asserted by the common wisdom. However banks perform other crucial functions in an economy, such as the screening of investment projects and, through this, the allocation of capital resources to the best social uses. Understanding how credit market structure may affect these additional banks' functions represents the meaning and focus of the banking competition debate. So far, studies on this topic have reached controversial results – on both theoretical and empirical grounds – thus calling for further research. This paper aspires to contribute to the empirical literature on banking competition by indirectly investigating the impact of credit market structure on banks' screening activity. More precisely, we focus on the relationship between default probability and bank debt at firm level and question if and to what extent this link is influenced by the degree of banking competition characterizing the local credit market in which firms operate. Our hypothesis may be stated as follows: if competition in credit markets affects banks' probability to screen – as shown, both theoretically and empirically, by several contributions in the literature on bank market structure, briefly reviewed in the next section – then it is reasonable to suppose that firms' riskiness (which is the core of bank screening) should have, ceteris paribus, a different effect both on the cost and quantity of credit to entrepreneurships, depending on the degree of bank competition. Since we lack information on loan interest rates at firm level, our hypothesis is investigated considering the relationship going from firms' default probability to their bank debt. We do not posit any a priori expectation on the sign of this relationship (thus leaving it to be an empirical finding) – since the theoretical implications of firms' characteristics, such as default risk, on lending volumes are not univocal (i.e. Stiglitz & Weiss, 1981versus de Meza & Webb, 1987; for a wide discussion on this debate see Cressy, 2002). To carry out the empirical analysis, we focus on Italian small and medium sized manufacturing firms (henceforth SMEs), which have little access to capital markets (either public equity or bond market) and are bound to ask credit from banks with branches in the same local market where they operate. Indeed, as Cesarini (2003) highlights, once internal funds are depleted, the banking channel is often the only way for Italian SMEs – usually facing high costs in employing arm's length finance (bond and Stock Exchange markets finance) – to gain access to external funds. Consistently with other contributions on the Italian banking system, we define 103 local markets corresponding to the existing administrative provinces. This disaggregation enables us to take advantage of an important feature of the Italian case. Indeed, Italian provinces are characterized by diversity in banking structures and this provides sufficient cross-sectional variability within a single institutional framework. Given this regulatory uniformity, there is no need to control for different regimes (Bonaccorsi di Patti & Dell'Ariccia, 2004). To measure banking competition in local credit markets, we employ the Herfindahl–Hirschman Index (HHI) on deposits which “represents a good proxy for competition in loan markets if the empirical investigation involves firms that largely borrow from local markets, that is if credit markets are local for the firms under consideration” (Petersen & Rajan, 1995 p. 418). As claimed above, this is the case for our sample units. The indicators of firms' default probability used in this paper have been computed by Moody's KMV on our sample data, via RiskCalc model. As argued by Moody's KMV, this model enables high precision and accuracy in evaluating private firm credit risk by using financial statements and, for listed firms, equity market-based information. The RiskCalc model is adopted by leading Italian banks as a benchmark for their internal credit risk estimates. In the empirical investigation we employ the cumulative EDF (Expected Default Frequency) measures – which are actual firms' default probabilities – within one, three and five years. The econometric analysis, implemented on a large set of micro-data running up to 2003 from 1995, is carried out by employing the dynamic panel estimator of Arellano & Bover, 1995 and Blundell & Bond, 1998, which allows to take into account the role of firm-specific effects (unobserved heterogeneity), as well as the “endogeneity” of a number of bank debt determinants. Our results seem to indicate that higher competition in local credit markets implies a stronger influence of firms' riskiness on the amount of bank debt granted to small and medium entrepreneurships (as well as on their bank debt growth). So that, since the relationship between firms' riskiness and bank debt is found negative, ceteris paribus, bank financing tends to be lower for riskier SMEs running in more competitive credit markets. In our view, a plausible interpretation of this evidence is that – as argued also by other contributions (i.e. Benfratello, Schiantarelli, & Sembenelli, 2006) – competitive pressures might force banks to perform more accurate screening, thus raising their efficiency in funds allocation. However, bank financing to riskier firms could be lower in more competitive credit markets for a reason unrelated to bank screening: higher bank competition may reduce the market power of incumbent banks, hence lowering their willingness to finance riskier firms – an explanation in line with the findings of Petersen, and Rajan (1995). As we argue below (see Section 5), both these interpretations may be considered likewise plausible, as well as not conflicting with each other, since the conceptual mechanisms they subtend may jointly represent the source of our evidence. The remainder of the paper is organized as follows. The next section presents a brief review of the literature on the economic effects of banking competition. Section 3 illustrates the econometric specification and the methodology adopted. Section 4 describes the data. Section 5 reports the results obtained and the robustness checks performed. Finally, Section 6 summarizes and concludes.

This paper has been concerned with the role of credit market structure on banks' screening activity, a topic disputed in the literature. We have dealt with this issue by empirically investigating whether and to what extent the link between a firm's default probability and the amount of bank debt it receives is affected by the degree of competition characterizing the local credit market where the firm operates. Our approach moves from the consideration that if competition in credit markets affects banks' probability to screen, as shown by the literature on bank market structure, then (ceteris paribus) the effect of firms' default probability on credit quantity to entrepreneurships should be different, depending on the degree of bank competition. The research has been conducted on a large panel of Italian SMEs for the period spanning the years from 1995 to 2003. Having focused on Italian SMEs is relevant since these firms have little access to capital markets and are bound to ask credit from banks with branches in the same local market where they operate. The methodology adopted to implement the analysis consisted in specifying a multiplicative interaction model, in which the impact of firms' riskiness profile on their usage of bank financing is made conditional on the level of local banking concentration. Moreover, both the role of firm-specific effects and the endogeneity of several determinants of bank debt have been taken into account. Our results suggest that in local credit markets where competition is more vigorous the influence of firms' riskiness on the amount of bank debt to small and medium entrepreneurships, as well as on firms' bank debt growth, is stronger. Since the relationship between firms' riskiness and bank debt is found negative, and ceteris paribus, bank financing appears to be lower for riskier SMEs running in more competitive credit markets. This evidence seems to be compatible with at least two explanations. On one hand, according to some contributions (i.e. Benfratello et al., 2006), competitive pressures might stimulate banks to perform more accurate borrowers' screening, so that credit market competition would be beneficial in raising banks' efficiency in funds allocation. On the other hand, regardless of screening activity (and in line with Petersen & Rajan, 1995 thesis), bank financing to riskier firms could be lower where credit competition is stronger because this latter might erode the market power of incumbent banks, thus reducing their incentives to finance higher riskier entrepreneurships. We have argued that both the above interpretations may be considered likewise plausible, and not necessarily conflicting with each other, as the conceptual mechanisms they subtend may jointly represent the source of our evidence. Nonetheless, to shed more light on the issue addressed in this paper, ongoing research is devoted to explore novel empirical strategies aimed at directly relating local credit market competition with both banks' incentives and ability to screen. From this research we also expect to draw some indications that may further corroborate the policy implication suggested by the results of the present paper, namely that the profound transformation process of the Italian banking industry – which started in the early 1990s and fostered competition in the sector (Angelini & Cetorelli, 2003) – might have improved funds allocation in the economy.