This summer school is designed to gain an insight into frontier research on Bayesian Structural Vector Autoregression (Bayesian SVARs). Within this context its focus is structural identification of economic shocks at business cycle frequency rather than forecasting or longer-term analysis. Bayesian SVARs are frequently employed to empirically assess the effects of monetary policy, oil market, labour market and uncertainty shocks. The summer school is divided into four parts:
- Reduced form estimation: We will learn which prior distributions can be chosen to estimate Bayesian VARs and how they can be implemented, both using existing toolboxes and building our own codes bottom-up.
- Structural Identification: We will introduce three approaches to identify SVAR models using economic reasoning: (i) short run restrictions, (ii) sign restrictions, and (iii) proxy variables.
- Posterior Sampling Techniques: In a stylised setting we will learn techniques to sample from non-standard distribution, a method often required in Bayesian SVARs. In a second step we will assess various diagnostic tools for these samplers.
- Advanced Topics: We introduce the following advanced topics: (i) Smooth Transition VARs, (ii) Local Projections, (iii) Factor-augmented VARs, (iv) statistical identification via heteroskedasticity and non-Gaussianity.
The summer school will take place in person. It combines lectures, workshops and coding exercises in Matlab.