Banking net income and macroeconomics, from multicollinearity to Granger causality using US data

We select six macroeconomic variables and study their relation with (aggregate) net banking income. The aggregate net banking income was reconstructed from US banking sector authorities' data. Usefulness may be twofold, it provides aggregate insight and the methodology can be replicated at bank institution level. We use standard tools such as linear regression analysis (to study multicollinearity) and Granger causality. The obtained results suggest a highly changing relation between all variables in time and an increase of causality and feedback relations after the 2008 crisis.