Departmental Working Papers, Rutgers University, Department of Economics 2018-02
We develop a novel technique to decompose banks' ratio of nonperforming loans to total loans into three components: first, a minimum ratio that represents best-practice lending given the volume and composition of a bank's loans, the average contractual interest rate charged on these loans, and market conditions such as the average GDP growth rate and market concentration; second, a ratio, the difference between the bank's observed ratio of nonperforming loans, adjusted for statistical noise, and the best-practice minimum ratio, that represents the bank's proficiency at loan making; third, a statistical noise. The best-practice ratio of nonperforming loans, the ratio a bank would experience if it were fully efficient at credit-risk evaluation and loan monitoring, represents the inherent credit risk of the loan portfolio and is estimated by a stochastic frontier technique. We apply the technique to 2013 data on top-tier U.S. bank holding companies which we divide into five size groups. The largest banks with consolidated assets exceeding $250 billion experience the highest ratio of nonperformance among the five groups. Moreover, the inherent credit risk of their lending is the highest among the five groups. On the other hand, their inefficiency at lending is one of the lowest among the five. Thus, the high ratio of nonperformance of the largest financial institutions appears to result from lending to riskier borrowers, not inefficiency at lending. Small community banks under $1 billion also exhibit higher inherent credit risk than all other size groups except the largest banks. In contrast, their loan-making inefficiency is highest among the five size groups. Restricting the sample to publicly traded bank holding companies and gauging financial performance by market value, we find the ratio of nonperforming loans to total loans is on average negatively related to financial performance except at the largest banks. When nonperformance, adjusted for statistical noise, is decomposed into inherent credit risk and lending inefficiency, taking more inherent credit risk enhances market value at many more large banks while lending inefficiency is negatively related to market value at all banks. Market discipline appears to reward riskier lending at large banks and discourage lending inefficiency at all banks.