Machine Learning of Optimal Interest Rate Policy
Machine Learning of Optimal Interest Rate Policy
In this research project we aim to algorithmically determine the optimal interest rate policy of central banks by means of realistic model-based simulations of the economy. To carry out these simulations, the macroeconomic model by Riedler (2017), developed at the ZEW, is extended by a realistic modelling of the heterogeneous consumption and saving behavior of households. Employing modern methods of machine learning, we algorithmically determine the “optimal” interest rate policy: During a great number of simulation runs, the central bank “learns“ the respective optimal interest setting strategy for reaching a certain target figure. This approach allows to realistically evaluate the proposals for an optimal interest rate policy that are currently being discussed in the academic and the public sphere, such as the “Taylor Rule”.