Macroeconomic Effects of Federal Reserve's Unconventional Monetary Policy: A Computational Linguistics Approach

I propose a new approach to estimating the Federal Reserve’s unconventional monetary policy after the Great Recession. During the 2009-2015 U.S. zero lower bound period, the Federal Reserve resorted to unconventional policies like forward guidance and large-scale asset purchases. I quantify these policies by creating an unconventional monetary policy index using computational linguistics and machine learning on the FOMC meeting minutes. This index provides information on how much of the discussion during each meeting was pertaining to the large-scale asset purchase program and forward guidance. A positive shock to this measure can be interpreted as more discussion about this topic in particular meeting. Using such shocks as instruments in a in a factor-augmented vector auto-regression framework, I find that the Fed’s unconventional policy had its desired effects of stimulating the economy, by increasing output and reducing the unemployment rate. I also find significant, albeit less, effects on the financial markets.