A multivariate Bayesian analysis of policy rates: Fed and ECB timing and level decisions

Chiara Scotti, University of Pennsylvania


My research studies when and by how much the Federal Reserve Bank (Fed) and the European Central Bank (ECB) change their target interest rates. I use a nonlinear bivariate framework, which allows for elaborated dynamics and for the potential interaction between the two central banks. An open question in the monetary policy literature is whether open economy interest rate feedback rules should include exchange rates, in addition to inflation and output. From an empirical point of view, I explore the possibility that interdependence between central banks could play a role in describing interest rate decisions. Moreover, central bank behavior is one of the building blocks of dynamic equilibrium models, together with production and consumption. It is not clear that conventional linear interest rate feedback rules are sufficient for explaining the complexity of central banks' behaviors, especially in the presence of potential interdependence. I analyze the timing decision using a Bivariate Autoregressive Conditional Hazard (BACH) model, which converts each country's marginal hazard into a joint hazard probability through the use of a conditional discrete copula-type representation. Within this framework, I study whether the Fed and the ECB synchronize their policies and whether one follows the other's decisions. Conditional on the timing decision, I analyze the magnitude decision using a conditional bivariate ordered Probit model, and I investigate whether there exists correlation in the magnitudes of their interest rate changes, after controlling for traditional explanatory variables. Because the information available to policy makers in real time is crucial to studying monetary policy decisions, I construct and use a novel real-time data set. A Bayesian estimation approach is particularly well suited to the short data sample. I incorporate pre-sample information about the US and Germany into the prior, to better evaluate the existing data. This strategy is particularly effective in the estimation of the BACH model, which would generally need a very long sample. Institutional factors, such as FOMC and Governing Council meetings, as well as inflation rates, are important variables in determining timing decisions. Inflation rates are important factors for the magnitude decisions of both countries. Output turns out to have a major role in the US, confirming the idea that the ECB's primary objective is to maintain price stability. Empirical results support synchronization between the two central banks, emphasizing the September 2001 coordination attempt, and non zero correlation between the magnitude shocks, but do not support follower behaviors. The positive correlation seems to suggest that the interest rate feedback rules containing past interest rates, inflation, output, and exchange rates might not capture the interdependence in the level decisions.

Subject Area


Recommended Citation

Scotti, Chiara, "A multivariate Bayesian analysis of policy rates: Fed and ECB timing and level decisions" (2005). Dissertations available from ProQuest. AAI3179811.