Methods Economics Seminar: Identification Verification for Structural Vector Autoregressions with Sparse Heterogeneous Markov-Switching Heteroskedasticity

How do you know your structural VAR is actually identified? Tomasz Woźniak proposes Sparse Heterogeneous Markov-Switching Heteroskedasticity, where each structural shock’s conditional variance follows its own Markov process. The model allows more regimes than the data generates, leaving some empty across the sample.

A new distribution for normalised conditional variances simplifies Gibbs sampling and identification checks. The model improves the precision of structural parameter estimates, reliably tests homoskedasticity, and forecasts on par with Stochastic Volatility. Applied to a well-known macro-financial system, it identifies the US monetary policy shock through heteroskedasticity, producing estimates consistent with the literature.

Speaker: Tomasz Woźniak — econometrician developing methodology for empirical macroeconomic analysis

When: September 22, 13:30
Format: online via Zoom