mattobham / reg_disc

Exploration of discontinuity regression for causal analysis

reg_disc

Exploration of discontinuity regression for causal analysis.

Hypothesis

This project is meant to explore causal effect modeling. I explore this topic by testing the hypothesis that banks with higher regulatory scrutiny need to hold more regulatory capital. I test this hypothesis by using Regression Discontinuity Design (RDD) on banks that are just over the $10BB in consolidated assets mark. The $10BB threshold requires banks to submit their financials to Federal Reserve stress tests, whereby the bank must forecast its losses to ensure it will remain well-capitalized under economically stressed scenarios.

DataSet Design

I create a dataset of banks around $10BB in consolidated assets. I chose to include banks between $8BB and $12BB, which resulted in a sample of 24 banks. The consolidated asset numbers for each bank is pulled from the March 2019 report on bank size. I used publicly-reported Common Equity Tier 1 ratios as a proxy for bank capitalization. I retrieved this data by manually searching through each of the 24 banks in the asset window above.

References

Lee, David S. and Lemieux, Thomas. 2010. "Regression Discontinuity Designs in Economics." Journal of Economic Literature, 48:281-355. Link