Description

This paper accomplishes two goals: First, it proposes a way to compute monetary policy surprises in Chile based on a survey of financial market participants regularly conducted by Bloomberg. We argue this is the most suitable one among alternatives. Second, we use these monetary policy surprises as input in a Bayesian Vector Auto Regression analysis to estimate the effect of contractionary monetary policy. Output and inflation tend to fall while funding costs tend to increase. Expected inflation a has hump-shaped response and nominal exchange rates tend to depreciate instead of appreciating. We argue the latter two effects are consistent with an "information channel" embedded in monetary policy decisions.