Title: Economic policy uncertainty with Ada-net
Authors: Niels Gillmann - ifo Institute and TU Dresden (Germany) [presenting]
Ostap Okhrin - Technische Universitaet Dresden (Germany)
Abstract: A local vector autoregressive model is developed. It allows us to estimate time-varying multivariate models without requiring the model parameters to change at every point in time. The estimation is done by a number of locally homogenous intervals and thereby identifying structural breaks. The local intervals are determined in a sequential testing procedure. This approach is especially suited for short time series since our approach usually results in only a few changes in the coefficients over time. We illustrate the method with simulations and a real data application. Using monthly Economic Policy Uncertainty data for five countries over 20 years, we show that uncertainty is connected across the world in a network. Furthermore, there seem to be three major breaks in our sample. Namely, the Global Financial Crisis, the European sovereign debt crisis and the election of Donald Trump as president of the USA.