Resumen de Essays on volatility networks and uncertainty
This thesis empirically investigates different aspects of time-varying volatility. Chapter 1 estimates a large TVP-FAVAR and recovers a dynamic directed network of connections between European stock volatilities. We propose an ad-hoc estimation methodology that is shown to outperform both standard approaches and competing models. Chapter 2 focuses on tracking dynamic connectedness between US sectoral volatilities using Generalized Forecast Error Variance Decompositions with a Bayesian model. As opposed to estimates obtained with rolling windows, we allow parameters to vary in a more flexible way. We show that there exists a stable relationship between the network structure and the volatility regimes in place at a given time. Chapter 3 estimates the unexpected time-varying volatility component of fiscal budgets in Italy. We show that periods of higher unexpected fiscal volatility are likely to be recessionary. Expansionary policies are effective only when not accompanied by increases in uncertainty.
