Gianni Amisano

Overview

Gianni Amisano is an econometrician at the Università di Brescia. Known for Bayesian VAR models with time-varying parameters, cointegration forecasting in a BVAR context, and marginal likelihood computation methods.

Key Contributions / Features

• Amisano and Serati (1999) — "Forecasting Cointegrated Series with BVAR Models" (Journal of Forecasting 18): proposed the IP-BECM model — a Bayesian ECM with informative priors on the factor loadings $\alpha$; showed that flat priors on $\alpha$ over-weight ECM correction terms relative to short-run dynamics; demonstrated best forecasting performance across all horizons on Italian macroeconomic data.
• Amisano and Federico (2004) — "Alternative Time-Varying Parameter Specifications for Bayesian VAR Models": compared four specifications for the state equation error covariance $\Omega$ in VAR-TVP models; established the Kronecker structure $\Omega = \rho(R \otimes Q_k)$ as the best practical specification for Euro area inflation forecasting; compared Chib, Gelfand-Dey, and Laplace marginal likelihood methods.
• Amisano and Serati (2003) — "Time Varying Parameters BVAR Models for Inflation Forecasting" (mimeo): developed simulation methods for TVP mechanisms with non-constant variation intensity; introduced and analyzed mean-reversion in VAR-TVP coefficient dynamics.

Related

• Cointegration
• Time-Varying Parameter VAR
• Gibbs Sampler
• Marginal Data Density
• Vector Autoregression