| Abstract |
The purpose of this paper is to study the role of diversification of geographic risk as one of the motives for bank expansion after the Riegle-Neal Act in 1994. We propose and estimate a dynamic game of banks' decisions to operate branches in local markets (i.e., US counties) where new entry can be either through merger or de novo branching. A key feature of the model is that a bank's decision of where to operate branches is modeled as a portfolio choice between risky assets. The returns to a bank's investments in a local geographic market (e.g., loans to local business and households) have a risk component that is specific of the local market. A bank is concerned with both the aggregate expected return and the aggregate risk of its portfolio of geographic markets. This concern with aggregate risk implies that a bank's branching decision has a particular type of network effect: i.e., the contribution of a local market to the bank's aggregate risk depends on the correlation between its return and the returns of other local markets in the portfolio. To estimate this model, we construct a unique dataset that combines information from four different sources: (1) branch and deposits data from the FDIC database; (2) information on mergers/acquisitions from the Chicago Fed; (3) a detailed and comprehensive description of the timing of adoption of the Riegle-Neal Act and other restrictions on geographic expansion in different states; and (4) Census Bureau county-level data on employment, salaries, and firms' revenue, which we use to construct measures of county-level risk as well correlations between county-level risks. The period of analysis is 1994-2006. |