In this study, we empirically assess the impact of capital regulations on capital adequacy ratios, portfolio risk levels and cost efficiency for U.S. banks. Using a large panel data of U.S. banks between 2001-2016, we first estimate the model using two-step generalized method of moments (GMM) estimators. After obtaining residuals from the regressions, we propose a method to construct the network based on clustering of these residuals. The residuals capture the unobserved heterogeneity that goes beyond systematic factors and banks' business decisions that impact its level of capital, risk and cost efficiency and thus represent unobserved network heterogeneity across banks. We then re-estimate the model in a spatial error framework. The comparisons of Fixed Effects, GMM Fixed Effect models with spatial fixed effects models provide clear evidence of the existence of unobserved spatial effects in the interbank network. We find a stricter capital requirement causes banks to reduce investments in risk-weighted assets, but at the same time, increase holdings of non-performing loans, suggesting the unintended effects of higher capital requirements on credit risks. We also find the amount of capital buffers has an important impact on banks' management practices even when regulatory capital requirements are not binding.
Download full working paper (1006.62 KB)