Association for Public Policy Analysis & Management

This paper studies the effect of central bank communication on monetary policy predictability. Using modern text mining techniques, I extract the informational content from the set of monthly interest rate decision statements, published by the Bank of Israel from 2006 to 2016. I construct a dictionary based sentiment measure, and estimate a Taylor type forward looking reaction function. I examine whether the sentiment measure can provide a clear signal about the future direction of monetary policy. I find that, in combination with standard macroeconomic variables, i.e. output gap and inflation, the sentiment measure improves short-run predictability of policy interest rate.