Real Effects of Bank Governance Bank Ownership and
Real Effects of Bank Governance: Bank Ownership and Firm Level Innovation Rainer Haselmann Katharina Marsch Beatrice Weder di Mauro 15 th Dubrovnik Economic Conference June 24 - June 27, 2009
Motivation n n High government involvement in banking sector since financial crisis Financial intermediaries select entrepreneurs – choice affects rate of technological progress (King and Levine 1993 QJE; Levine and Zervos AER 1998) Banking development stimulates the introduction of innovations (Benfratello et al. JFE 2008) Are public or private financial intermediaries better in selecting innovative projects? 2
Question Theory is ambivalent about the effect of public bank ownership on technological progress: n n n Public banks might foster innovation because of market failures (e. g. asymmetric information/moral hazard/positive externalities) Government bankers’ incentives can result in a misallocation of financial resources (e. g. politicians follow personal goals; government banks want to secure employment – La Porta et al. JF 2002; Sapienza JFE 2004) 3
Contribution Germany is used as laboratory n n n Industrialized country (e. g. Khwaja and Mian QJE 2005: Pakistan) German financial sector is bank-based Large public banking sector Innovative economy (innovative SMEs) Unique dataset Methodology n n n Model relationship bank selection Determining local bank supply of sample firms 4
Question Why do firms not simply switch their main lender if a certain ownership type is beneficial for their innovation preferences? n n Asymmetric information and moral hazard are large in the process of innovation financing (Carpenter and Petersen EJ 2002) Main lender (relationship bank) collects information on borrower to moderate asymmetric information and moral hazard problem (Diamond REStud 1984) Hold-up problem is especially important for information opaque projects such as innovation financing 5
Findings n n A firm’s probability to innovate is affected by the ownership of its main lender A firm’s probability to innovate is about 13 percent higher if the main lender is a private compared to a government banker (after controlling for firm specific characteristics and selectivity bias) The ownership of the main lender affects the probability to innovate to a larger extend for smaller firms Innovators with a private main lender (as compared to a government main lender) produce more innovations 6
Agenda n Motivation n Data and Descriptives n Methodology n Empirical Results n Conclusion 7
Data and Descriptives Financials n n n Bureau van Dyck‘ Amadeus dataset for German manufacturing firms 9, 310 firms (32, 839 firm-year observations) for 1993 -2006 Innovation ability n n n Patent filings from European patent office (EPO) Citations to measure relative importance of patent filing Lending relationship n n Credit registry from the Deutsche Bundesbank (Mio-Evidenz) Every lending relationship exceeding 1. 5 M Euros in a quarter Remaining sample ~ 6, 500 firms Supply of local bank branches n n Address of all bank branches (Banken-Verlag Medien Gmb. H) Geocoding of addresses Great-Circle-Distance of 3 km (~28 km 2) and 10 km (~ 314 km 2) 8
Data and Descriptives All firms Private Public Difference (p-value) Observations 12, 343 (firm-year) 7, 444 4, 899 2, 545 Innovative (mean) 0. 342 0. 384 0. 278 0. 106 (0. 000) Employees (mean) 1, 687 2, 010 987 1, 023 (0. 000) 9
Data and Descriptives 10
Agenda n Motivation n Data and Descriptives n Methodology n Empirical Results n Conclusion 11
Selectivity bias n n Firms may choose a certain type of bank depending on their innovation ability Idea: Instrument for firms’ main lender selection by determining supply of local bank branches Assumption: Geographic distance is an important determinant for the choice of main lender (Degryse and Ongena JF 2005, Peterson and Rajan JF 2002) Private banks do not have branches in all regions – regional principle for public banks 12
13
Methodology n Firm i has a choice to innovate or use an existing technology. Innovation decision of firm i: (outcome equation) n n innovation decision of firm i (1/0) ownership of main lender (1 if government bank is main lender/ 0 if private bank is main lender) vector of controls (firm and industry characteristics) coefficient of interest 14
Methodology n To control for selectivity bias introduce bivariate probit model (Heckman 1978). A firm’s main lender selection can be modeled as follows: (selection equation) n n n is vector of instruments Two binary decisions (4 states of the world) Full maximum likelihood bivariate probit estimation 15
Agenda n Motivation n Data and Descriptives n Methodology n Empirical Results n Conclusion 16
Results - Selection n Two conditions need to be met for our instrument to be valid: 1. ) Instrument has to be important determinant of firm‘s choice of a main lender 2. ) Instrument must not be a determinant of firm‘s decision to innovate n Bank and firm location should not be endogenously determined: n n Regional principle: Rural areas tend to be overbanked by public banks Moving for manufacturing firms is costly especially for small firms and those with a high proportion of fixed assets (high tangibility) 17
Results - Innovation Bivariate probit estimates: 18
Results - Innovators and # of patents 19
Results - Robustness tests n Use a 10 km radius of distance around each firm n Use alternative definitions of relationship lender n Alternative estimation method (2 SCML) n Use sample with firm with high tangibility ratio 20
Agenda n Motivation n Data and Descriptives n Methodology n Empirical Results n Conclusion 21
Conclusion n n Providing external finance is key mechanism through which banks affect economic growth Probability of a firm to innovate is about 13 percent higher if the main lender is a private compared to a government bank Public bankers’ incentives are manifold which is adverse impact on selecting innovative projects Government ownership of banks might comes at the cost of lower innovation in the long run 22
Appendix Dr. Rainer Haselmann Johannes Gutenberg University Mainz – June 17 th, 2009 23
Results - Selection equation for different samples sizes: Dr. Rainer Haselmann Johannes Gutenberg University Mainz – June 17 th, 2009 24
Results - Innovation Bivariate probit estimates – high tangible assets: Dr. Rainer Haselmann Johannes Gutenberg University Mainz – June 17 th, 2009 25
Data and Descriptives Dr. Rainer Haselmann Johannes Gutenberg University Mainz – June 17 th, 2009 26
Results Robustness Locate banks in a 10 km radius around each firm: Dr. Rainer Haselmann Johannes Gutenberg University Mainz – June 17 th, 2009 27
Results Robustness Using alternative definitions of relationship lender: Dr. Rainer Haselmann Johannes Gutenberg University Mainz – June 17 th, 2009 28
Dr. Rainer Haselmann Johannes Gutenberg University Mainz – June 17 th, 2009 29
Results Innovation and Firm Size 2 SCML estimates: Dr. Rainer Haselmann Johannes Gutenberg University Mainz – June 17 th, 2009 30
Related work n n n Herrera and Minetti JFE (2007) find that relationship finance (duration of credit relationship) promotes innovation finance Benfratello, Schiantarelli, and Sembenelli JFE (2008) show that local banking development matters for the probability of corporate innovations Atanassov, Nanda, and Seru (2005) show that large firms actually prefer market based finance over relationship lending Dr. Rainer Haselmann Johannes Gutenberg University Mainz – June 17 th, 2009 31
- Slides: 31