Banking environment and loan syndicate structure: a cross-country analysis

Abstract

What is the influence of the banking environment on bank syndicate structure? We provide empirical evidence on this issue by examining the influence of several banking environment characteristics, including the structure of the banking market, financial development, banking regulation and supervision and legal risk, on the structure of bank syndicates. The results of a cross-country analysis performed on a sample of 15 586 syndicated loan facilities for borrowers from 24 countries confirm that syndicate structure is influenced by banking environment in a way consistent with minimizing agency costs and efficient re-contracting objectives.

Notes

¹ A syndicated loan is a loan which is provided to the borrower by two or more banks that is governed by a single loan agreement.

² See Esty (2001) for a detailed description of the syndication process.

³ Lopez Iturriaga (2005) also finds that the legal system has an impact on bank debt.

⁴ Another candidate measure of syndicate structure was the concentration of retained shared of the loan by syndicate members, based on a Herfindahl–Hirschman index. However, information on lenders shares is very scarce in the dataset.

⁵ To avoid biased results, we do not distinguish the number of participants, as the same financial institution can have several roles in a syndicate, being simultaneously an arranger and a participant.

⁶ A more common proxy would be the cost to income ratio but such information is unavailable in our dataset.

⁷ The other variables considered for regulatory discipline and disclosure were the obligation to publicly disclose off-balance sheet items and the presence of a public or private credit registry, but we did not include them in our estimations as they account for more than 95% of the loans in the sample.

⁸ We match borrowers by their country, name and industry sector. This procedure reduces the size of the sample. Borrower's variables have a 1-year lag compared to the year of loan syndication.

⁹ Unfortunately, we do not have access to data for syndicated lending in the US and UK.

¹⁰ Using Poisson and negative binomial regressions lead to virtually the same results.

¹¹ For instance, the F-statistic increases from 43.82 (specification 2.1) to 97.51 (specification 2.3).

¹² Finally, using an alternative measure of syndicate structure – a Herfindahl–Hirschman index computed with the retained shared of the loan by syndicate member – gives results consistent with existing empirical evidence and with previous results obtained with the Number of Lenders and Number of Arrangers variables. Due to data availability, we were able to perform the regressions with loan and banking environment variables only on a reduced sample of 1023 loan facilities. Syndicates are more diffuse when bank concentration and bonds market are more important, while they are more concentrated when legal environment quality is higher.

¹³ We also performed several reduced form regressions omitting potentially endogenous variables such as Maturity, Guarantors or log(Loan Size), as a lack of data prevented us from using an instrumental variables approach. These regressions led to similar coefficients, and thus conclusions regarding banking environment variables.

¹⁴ Using other measures of borrower transparency, such as Moody's Senior Debt rating or a dummy variable equal to one if a Moody's or a S&P rating is available, led to similar results.

¹⁵ Overall, the results for the loan characteristics support prior findings by Lee and Mullineaux (2004), Bosch and Steffen (2007) and Sufi (2007).

¹⁶ Including alternative proxies for borrower risk, such as the Quick Ratio or the ratio of Net Income to Total Assets, also produces similar results.