Abstract:We revise the solutions of the dynamic optimization model of Elyasiani, Kopecky, and Van Hoose (1995), and show that the optimal policy function for the case of adjustment costs and no portfolio separation is a vector autoregressive model of order 1, and not of order 2. In addition, the empirical approach of Elyasiani, Kopecky, and Van Hoose (1995) is incorrect. The optimal policy functions for the no portfolio separation case have to be estimated with a panel vector autoregression model and not with a simultaneous equation approach. We also describe how the portfolio separation hypothesis can be tested empirically based on the correct policy functions.

Costs of Adjustment, Portfolio Separation, and the Dynamic Behavior of Bank Loans and Deposits – Revised