Description

This paper analyzes US Term Structure changes using linear factor models based on principal components analysis and the model of Diebold and Li. The analysis of factors time series could not reject the hypothesis of normality for changes in the first two factors that accounts for level and slope effects. This enables the assumption that factors follow correlated Ornstein-Uhlenbeck processes, and then construct 95% confidence ellipses that allow us to identify large movements that are interpreted as unanticipated by market participants. The results suggest the importance of the economic assessment released by the monetary authority, and the ability of agents to anticipate Fed’s actions over the sample period 1997:01 2005:04.