Discussion of Relationship and Transaction Lending in a Crisis Philipp Schnabl NYU Stern, CEPR, and NBER USC Conference December 14, 2013
Summary 1 Research Question How does relationship lending vary over the business cycle? 2 Methodology 3 Results Develop model of relationship lending after an aggregate shock Empirically analyze relationship lending using Italian lending data from the 2008 financial crisis Model predicts that relationship lenders charge higher spreads during normal times and lower ones during bad times Find empirical support using within-firm and within-bank estimator
Overview Contribution 1 First to explore the role of relationship lending during crises 2 Provides a testable model of multiple lending relationships 3 Uses within-firm and within-bank estimator in a novel setting Discussion 1 Provide intuition on the role of multiple lending relationships 2 Discuss identification strategy using within-estimator
Time Line: Transaction Bank Only t=0 t=1 t=2 Bank observes firmspecific risk p and offer competitive interest rate r 0 (p) Aggregate state S is realized Firm payoff is realized Firm payoff is realized and cash flows are distributed Success Failure Firm pays r 0 (p) Liquidation Roll-over Cannot observe type (H,L) Offers roll-over at interest rate r 1 (p,s)
Transaction Banks Only Bank offers a roll-over at rate r 1 (p, S) iff: r 0 (p) = E[r 1 (p, S)] V H υ s 1 Firm cannot promise more than pledgable income V H υ s 2 Perfect competition equalizes time-0 and time-1 interest rate 3 Time-0 lender cannot renegotiate (needs to be paid off in full) Creates a debt overhang problem at roll-over 1 If time-0 interest rate r 0 (p) is too high, there is no roll-over in the bad state 2 Leads to inefficient liquidation
Transaction Bank Only Interest rate r(p) at time 0 High risk firms (default in bad aggregate state) p Low risk firms (always repay) Firm-specific risk p
Multiple Lending Relationships Assumptions: 1 Assume T-banks have lower funding cost than R-banks firms prefer borrowing from T-Banks 2 R-banks can prevent debt overhang firms borrow to provide R-banks an incentive for monitoring at time 1 Equilibrium: 1 Firms borrow just enough from R-banks to prevent debt overhang 2 All roll-overs are done by R-banks; charge lower interest rate than T-bank during crises 3 If R-bank refuses roll-over, firm must be L-type and is liquidated No debt overhang with both T- and R-banks
Comment: Debt overhang in T-Lending Debt overhang is usually associated with arm s-length financing and dispersed holdings Not entirely clear how to think about this in traditional banking Probably most applicable to loan products that are eventually sold or securitized (e.g., syndicated loans) More discussion on role of debt overhang for empirical tests
Identification strategy Natural Experiment from Italy Use Lehman default as exogenous shock to Italian Banks Exploit credit registry data that covers all firm-bank relationships Identify T-lending based on informational distance (bank and firm headquartered in the same region) Naive estimator: FirmOutcome i,t = α i + delta t + βtbank i,t Crisis t + ɛ Identification concern: Bank-firm matching proxies for firm quality
Identification strategy Solution: Estimate effect of T-banking at loan-level (not firm-level) Use within-firm and within-bank estimator: FirmOutcome i,b,t = α it + γ bt + βtbank i,b,t Crisis t + ɛ 1 α it controls for all firm-level shocks (e.g., change in investment opportunities) 2 γ bt controls for all bank-level shocks (e.g., exposure to int l funding markets) Implemented using separate estimation by year
T-banks charge lower spreads during bad times (Table 3) Table 3 T-banking and R-banking in good times and bad times Variables Interest rate good time (2007:q2) (I) Interest rate bad time (2010:q1) (II) Log Loans good time (2007:q2) (III) Log Loans bad time (2010:q1) (IV) T-Bank -0.0805*** 0.1227*** -0.2753*** -0.3129*** (0.0174) (0.0210) (0.0123) (0.0110) Bank fixed effects yes yes yes yes Firm fixed effects yes yes yes yes Number of obs. 184,859 184,859 184,859 184,859 Adjusted R-squared 0.529 0.585 0.426 0.473 Notes: The models in column (I) and (III) are estimated in 2007:q2; those in columns (II) and (IV) in 2010:q1. The dummy T-Bank takes the value of 1 if the loan is granted by a transactional bank. The coefficients represent the difference relative to relationship banking (R-banks). Parameter estimates are reported with robust standard errors in brackets (cluster at individual firm level). The symbols *, **, and during *** represent regular significance times levels of 10%, 5%, and 1% respectively. Coefficients for fixed effects are not reported. T-banks charge lower spreads during normal times and higher ones T-banks loan size is smaller during regular times
Comment: Extensive Margin of Lending Theory predicts that T-banks stop lending But analysis appears to be conditional on the loan relationship being in the data pre and post-lehman Can induce large sample bias during a crisis in which many banks stop lending Analyze extensive margin of lending
Comment: Estimate in one regression Two time periods are measured separately But explanatory variable (T-Bank) should be measured prior to the shock Does not measure change in lending for the same firm Estimate two time periods in one regression
Conclusion Novel model on relationship lending 1 First paper to analyze relationship lending after a crisis 2 Incorporates multiple lending relationships Main comments 1 Discussion of debt overhang assumption for empirical test 2 Also analyze extensive margin and estimate in one regression