This paper analyzes the impact of news on several Italian financial variables, paying particular attention to the effect on the conditional volatility of these variables. The analysis spans a period of great financial and political turbulence in Italy, including the rapid succession of three governments. News releases (articles on political and economic events collected daily from both the Italian and international economic press) are classified as unscheduled (mostly political) and scheduled (i.e. economic and monetary statistics whose announcement is expected by market participants). The analysis is divided into two phases: first, we estimate the impact of each single political and economic news item on asset price changes and their conditional variance; second, those items that are identified as significant in the first stage are then aggregated into six dummies according to their nature and origin and employed as exogenous variables in a trivariate Garch scheme. Results show that i) news affects both the first and the second moment of the daily changes in the analyzed variables; ii) there is a significant regime shift of the unconditional variance of the analyzed variables across the three different governments; iii) the conditional variances display a significant — albeit rather small — seasonal dayweek pattern; iv) contrary to the conventional view, the impact of news on the conditional variance is more pronounced for exchange rates than for Italian long-term interest rates.
Published in 2002 in: Economic Modelling, v. 19, 4, pp. 611-639
