The dataset used in the analysis also includes stock market prices for all other banks and firms in the home and host countries that are used to construct default risk control variables in the regressions. We use daily stock market information from Compustat Global for international banks and firms and stock market information from CRSP for U.S. banks and firms. Bank-level variables are constructed from Bankscope, a commercial database of banks’ financial statements produced by Bureau Van Dijk. Since we are interested in how bank characteristics affect the correlation between parents’ and their foreign subsidiaries’ default risks, and since there have been significant changes to bank balance sheets during the crisis, we construct and use bank-level variables measured prior to the crisis (as of December 2006). For each bank, we calculate relative bank size (bank assets to total system assets), capital ratio (regulatory capital to risk-weighted assets), equity ratio (equity to total assets), provisions (loan loss provisions divided by total loans), deposit funding (deposits divided by total funding), profitability (net income divided by total assets), and liquidity (liquid assets divided by total assets). We winsorize all financial variables at the 1st and 99th percentile level of their distributions to reduce the influence of outliers and potential data errors.