Firm Industry Affiliation and Multiple Bank Relationships

Following Mikkelson and Partch (1986) and James (1987) a large literature has studied cumulative abnormal returns on borrower stocks around bank loan announcements for example (see also e.g., Lee and Sharpe (2009)).

Though most common, relationship multiplicity remains puzzling because a sole lender should be able to provide the most efficient monitoring of borrowers. The involvement of multiple banks in lending to a single borrower will lead to a duplication of their monitoring efforts (Diamond (1984); Ramakrishnan and Thakor (1984); Boyd and Prescott (1986)) and may weaken the incentives of each individual lender to monitor due to the opportunities for free-riding (Carletti (2004); Carletti et al. (2007)). An exclusive bank relationship may further provide greater (intertemporal) flexibility in contract setting (Boot and Thakor (1994); Petersen and Rajan (1995); Dewatripont and Maskin (1995)), potentially helping borrowers in obtaining access to financing and overcoming adverse economic conditions, or allow for easier product bundling (Calomiris and Pornrojnangkool (2009)) and economies of scope.

I.e., Bosnia and Herzegovina, Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Macedonia (Fyrom), Republic Of Moldova, Montenegro, Poland, Romania, Russian Federation, Serbia, Slovakia, Slovenia, Ukraine. Most of these countries were repeatedly covered in other studies on firm access to bank financing and that rely on a combination of other datasets including Amadeus (e.g., Giannetti and Ongena (2009)), Kompass (e.g., Giannetti and Ongena (2012)) and BEEPS (e.g., Brown et al. (2011a, b); Popov and Udell (2012); Ongena et al. (2013); Ongena and Popov (2012)). These countries also have consistent reporting of all relevant information, including bank relationships.

Exceptions are Degryse and Ongena (2007) who study the impact of local banking market competition on the industry specialization of individual bank branches in Belgium, and recently Paravisini et al. (2014) who study matched credit-export data from Peru and show that firms that expand exports to a destination market tend to expand borrowing disproportionately more from banks specialized in that destination market.

We find it rather unlikely that the choice of the bank relationship arrangement determines the choice of the number of industries the firm operates in. We think the latter choice is likely to be even more long term and strategic than the choice of banks the firm utilizes and fail to see any reasonable instrument on this account. We acknowledge, however, that in general access to adequate financial resources may help the firm to expand into other industries (e.g., Villalonga (2004)).

In Detragiache et al. (2000) and Guiso and Minetti (2010) firm and country characteristics determine a two-staged corporate decision to have multiple versus single bank relationships and once in the multiple-bank region the optimal number of bank relationships. Using their theoretical model as an alternative identification strategy, and methodologically following Cerqueiro (2009), Popov and Ongena (2011) and Brown et al. (2011a), who extend Heckman (1979), we also estimated three-stage selection models that accounted at once for the sequence of corporate choices pertaining to: (1) having a bank or not (reported), (2) the multiplicity or singularity of the bank relationship arrangement, and (3) the number of bank relationships. Though we focus on a single year in these exercises the estimates are similar with respect to the main results reported here.

As in the tables we will star the estimated coefficients as follows: *** significant at 1 percent, ** significant at 5 %, * significant at 10 %.