Does inequality lead to credit growth? Testing the Rajan hypothesis using state-level data