SMEs main bank choice and organizational structure Evidence

SME’s main bank choice and organizational structure: Evidence from France Hiba EL HAJJ CHEHADE and Ludovic VIGNERON GERME – ESA UNIVERSITY OF LILLE 2 France H. EL HAJJ CHEHADE and L. VIGNERON NTU International Conference on Finance- December 2006 NTU International conference on finance

Question n Do SMEs choose their main bank according to its organizationnal structure? H. EL HAJJ CHEHADE and L. VIGNERON 2006 NTU International conference on finance 2

Related literature n Distinction between relationship and transactional lending depends on the type of information produced and used in each one: ¨ SOFT info VS HARD info. (Stein (2002), Berger, Klapper et Udell (2001), etc. ) n This is related to the bank’s organizational structure: ¨ hierarchical VS decentralized (Stein, 2002) H. EL HAJJ CHEHADE and L. VIGNERON 2006 NTU International conference on finance 3

Related literature (continued) n Empirical results ¨ Positive relationship between bank’s size and its customer’s size (Berger et Udell (1996), Strahan et Weston (1996), etc. ). ¨ Hierarchical banks are less likely to provide relationship financing and thus to finance opaque firms. (Berger, Klapper et Udell (2001)). ¨ SMEs dealing with large banks are more likely to be rationed (Berger, Miller, Petersen, Rajan and Stein (2002)) H. EL HAJJ CHEHADE and L. VIGNERON 2006 NTU International conference on finance 4

Hypotheses n 2 hypotheses to be tested ¨ H 1 : The probability that a firm chooses a decentralized main bank increases with its informational opacity. ¨ H 2 : The opaque firms that have hierarchical main banks are more likely to be credit constrained than those who have decentralized banks. H. EL HAJJ CHEHADE and L. VIGNERON 2006 NTU International conference on finance 5

DATA n From DIANE, a sample of 6 258 SME working with 182 different banks. n The sample SMEs are less than 5 years old. n This data is completed by banks characteristics extracted from Bankscope H. EL HAJJ CHEHADE and L. VIGNERON 2006 NTU International conference on finance 6

Dependent Variables n The decentralization of the main bank is estimated by 3 different measures: ¨ Bank’s size measured by the logarithm of total assets. ¨ A dummy equal to 1 if the bank has a decentralized structure. ¨ The Ratio personnel expenses over total assets. H. EL HAJJ CHEHADE and L. VIGNERON 2006 NTU International conference on finance 7

Summary of our empirical model The choice Informational between a opacity decentralized and a centralized bank n. Bank’s size n. Dummy «decentralized » n. Ratio: personnel expenses over total assets H. EL HAJJ CHEHADE and L. VIGNERON n. Firm’s size SA « Share company» (accounts certification) n Rationing Other explicative factors or control variables Taxes over total debt n 2006 NTU International conference on finance n Financial risk Specialization of banks n n Market 8

Estimations n Model 1: OLS n Model 2: Logit n Model 3: OLS H. EL HAJJ CHEHADE and L. VIGNERON 2006 NTU International conference on finance 9

Main Results n Regressions on the full sample Bank’s size Dummy Decent Personnel expenses / TA Informational tranparency -Firm’s size -SA > 0 and significant < 0 and significant 0, 463*** 0, 129* -0, 000127*** Rationing > 0 and significant < 0 and significant 0, 647*** -0, 583*** Financial risk (Score) Bank’s sector specialisation H. EL HAJJ CHEHADE and L. VIGNERON -0, 644*** -0, 320*** No significant effect >0 and significant 0, 007*** < 0 and significant >0 and significant for: Manufacturing, and Trade Services and Trade 2006 NTU International conference on finance < 0 and significant -0, 000163*** No significant effect >0 and significant for: Manufacturing and Services 10

Main Results (continued) n Regressions run on the sub-samples : SA and non SA ¨ Rationing n n ¨ n Non significant for transparent firms (SA) Significant coefficient for the opaque firms (non SA) For the rest of variables, the results are the same as previous ones. Regressions without « firm’s size » ¨ For Market dummy: the coefficient becomes significant. H. EL HAJJ CHEHADE and L. VIGNERON 2006 NTU International conference on finance 11

Conclusion n Purpose of the paper: to bring explanatory elements of the choice made by a company of its main bank. n The results of the different regressions support the two predictable hypotheses. They shows that: Opaque firms choose main banks, which have an organizational structure adapted to their needs: decentralized banks that are able to provide them a relationship financing. ¨ Opaque firms that choose to deal with hierarchical banks face rationing problems. ¨ n Importance of our research problem in the context of the current financial restructurings: What are the consequences of such restructurings on the financing of small and medium-sized firms. ¨ Is there really a decrease in the number of the decentralized banks and thus a decline in the financing of small and medium-sized firms. ¨ H. EL HAJJ CHEHADE and L. VIGNERON 2006 NTU International conference on finance 12

Dependent variable : Taxes over total debt Intercept 0, 864*** (58, 664) Bank debt/ total debt -0, 329*** (-28, 937) Ln (total firm’s assets) -0, 0628*** (-37, 334) Manufacturing -0, 0671*** (-7, 433) Services 0, 07877*** (9, 172) Trading -0, 116*** (-13, 134) Construction -0, 0840*** (-8, 248) Fisher 727, 213*** Adjusted R² 0, 411 Condition number 15, 635 H. EL HAJJ CHEHADE and L. VIGNERON 2006 NTU International conference on finance 13