Research Data for "The Effects of the EBA’s Stress Testing Framework of the European Banking System"
Description
We collect data for the 2006-2018 period from the Orbis Bank Focus (formerly known as Bankscope) database. This database provides data on the characteristics of the European Union (EU) banks. Furthermore, we employ the World Bank data on macroeconomic variables such as unemployment and GDP growth. In relation to the sample size, we include banks above the 10 billion euros threshold. This threshold is similar to the threshold used in the emerging banking literature examining the relationship between stress-tested and non-stress tested banks. We ensure that all the banks included in our sample are represented by their consolidated balance sheets for each banking group. This practice follows the EU stress testing framework methodology, whereby all participating banks must report their banking activities on the highest consolidation basis, including data of their respective subsidiaries. Finally, after cleaning the data and removing banks with insufficient data, we obtain a final sample of 282 banks. The number of stress-tested banks include 39 banks from 16 countries. All remaining banks in the sample are allocated to the non-stress tested banks group, which represents our control group for the difference-in-differences identification strategy. The primary data concerning the performance against the stress test are reported by the European Banking Authority (EBA), which publishes official documents such as methodologies and the results of the stress tests. The information is utilised for the second stage of the methodological approach, which helps construct the stress test exposure variables. All related documents are made available on the European Central Bank (ECB) webpage. In the first step, we analyse the effect of stress testing on bank lending. To this end, we examine the impact of the exercise on total loans, mortgage loans, corporate loans, consumer loans, and bank loans. In the second step, we contribute to the literature by examining the effect of the exercises on credit risk. To test this hypothesis, we examine the changes in credit risk due to stress testing and assess if there are potentially statistically significant results. Specifically, we use the non-performing loan ratio as a proxy for credit risk and assess the impact on banks that participate in the EU stress test. We find that stress-tested banks experience higher credit risk and reduce lending for specific loan types. In particular, due to country heterogeneities, we find that stress-tested banks in the GIIPS (Greece, Italy, Ireland, Portugal, Spain) region experience a greater contraction in lending. Our results also suggest that the elevated credit risk of highly-exposed stress-tested banks can be a driving factor of a reduction in bank lending. Consequently, prudential measures requiring banks to hold higher capital buffers are justified to contain credit risk shocks.