The analysis considers the role of non-performing loans (NPLs) for bank lending rates on newly granted loans. It is based on euro area data.
The focus is on an effect caused by the stock of NPLs that extends beyond losses that banks have already incorporated into their reported capital positions.
The paper assesses the channels through which such an effect occurs most importantly whether it runs through banks' idiosyncratic funding costs.
File 0 contains a description of the data used for the analysis. It does not contain actual data as most data used for the analysis is confidential.
The file contains the names of the Stata-dta-Files in which the datasets are stored. These Stata-dta-Files are the starting point for the data processing which is activated by
the code in the subsequent Stata-do-Files.
Files 1-3 contain the code for processing SNL and Bankscope / Orbis data. This data includes the banking group level data for the analysis (most importantly NPL /
regulatory capital data).
File 1 contains the code for the processing of SNL data.
File 2 contains the code for the of the Bankscope / Orbis data.
File 3 contains the code for merging SNL and Bankscope / Orbis data.
Files 4-6 contain the code for processing the CSDB data which includes data on the cost of bond funding on the banking group level,
iBSI / iMIR data which includes data on lending rates and lending volumes on the single bank level and the macroeconomic data.
File 4 contains the code for the processing of the CSDB data. Note that this data is initially on the single security level and is processed such, that
information on costs of bond funding on the banking group level is retrieved.
File 5 contains the code for the processing of the iBSI / iMIR data.
File 6 contains the code for the processing of the macroeconomic variables.
File 7 contains the code for merging all datasets.
File 8 contains the code for producing the descriptive statistics in Section 3 of the paper.
File 9 contains the code for the estimation of Equations 1 and 3 of the paper.
File 10 contains the code for the estimation of Equations 1 and 3 with random samples (Appendix D of the paper).
File 11 contains the code for estimations with loan growth as dependent variable (Section 5.2 of the paper).
Files 12 and 13 contain code for the data processing and estimation of Equation 2 on the banking group level.