This repository contains files used in the paper "Monetary Policy Shocks and Skill Differences in the U.S. Labor Market" (2020).

This paper studies the effect of monetary policy shocks on different demographic groups in the U.S. labor market. I look at the effect of a contractionary monetary policy shock on unemployment rates of high and low-skill workers, finding that the low-skill group is more sensitive to these shocks than the high-skill. Further breaking the skill groups down by gender and race, I find that female workers and non-White workers, regardless of their skill type, are more adversely affected by these shocks. Results suggest monetary policy shocks clearly have heterogeneous effects in the labor market, which should be taken into consideration while implementing monetary policy.

Paper available at https://sites.google.com/view/prithachaudhuri/research.

• RRshocks_single_equation.m: This Matlab file describes how I created the monetary policy shocks following Romer and Romer (2000) "A New Measure of Monetary Policy Shocks: Derivation and Implications". The data used to create this shock measure is included in the file RomerandRomerDataAppendix in the data folder. This m file also describes the single equation regressions carried out in the paper.
• compute_irf.m: This m file describes how to compute impulse response functions for the single equation regressions carried out by Romer and Romer (2004).
• create_cps_data.R: In this script I describe how to clean and merge the Current Population Surve (CPS) MORG files. These files are available at the NBER website in .dta format, starting from year 1979. The MORG files have undergone some revision in some years, changing the name of some varibles or revising their categories. These need to be kept in mind while merging the data. During the cleaning process I also categorize the population as high and low-skill using the education information in the data. Finally, after merging the data for the years 1979-2018 I calculate some important labor market indicators such as the unemployment rate (used in the paper), labor force participation rate, hours worked and hourly wages for the high and low-skill population.
• run_final_vars.R: This script describes the main analysis in the paper. I use the monetary policy shock calculated earlier, along with data on GDP growth, CPI inflation and unemployment rates for high and low-skilled workers calculated from the CPS, in a Vector Autoregression framework. In the first set of regressions that act as the benchmark case I only consider high and low-skilled workers, estimate a VAR model and compute the impulse response functions of a monetary policy shock. The impulse response functions show the dynamics of each macroeconomic outcome over a period to a one-time change in monetary policy. The second set of regressions I break the high and low-skilled population up by gender and in the third set of regressions I break them up by race.