Replication Files for “Divergent Effects of Aggregate and Local Uncertainty Shocks: Evidence from US Metropolitan Areas”
Description
These files are the replication files for “Divergent Effects of Aggregate and Local Uncertainty Shocks: Evidence from US Metropolitan Areas.” The Readme file provides instructions for replicating the main results of the paper using the files in the folders. The FAVAR-SVM folder contains MATLAB code and data for the stochastic volatility-in-mean factor-augmented vector autoregression model used to estimate aggregate uncertainty, local (metropolitan statistical area, MSA) uncertainty, and the responses of each MSA to aggregate and local uncertainty shocks. The Stata_crosssection folder includes code and data to estimate how the responses of the MSAs to the shocks depend on various economic factors.
Files
Steps to reproduce
Basic instructions for using the replication code are below, and more details are in the Readme.pdf file. I. Estimating the FAVAR-SVM Model: Step 1: Data Preparation: Set the MATLAB working directory to the Data subfolder. Run prepdata.m. This will load the MSA Economic Conditions Index data (“MSAdata”), macroeconomic data from the FRED-MD database (“FREDMD”), and regional economic indicators (“RegionaIndices”), and then transform the series to ensure stationarity. The prepared data is saved as Xdata.mat. Step 2: Estimation: Set the MATLAB working directory to the Benchmark subfolder. Run estimateFAVAR_baseline.m, which estimates the benchmark FAVAR model, and saves 20 output files in the results subfolder. Step 3: Generate results: Run getIRF_withIdioU_baseline.m once Step 2 is completed. The code generates all figures and tables under the baseline specification: Figures 1, 2, 3, 4, and A1, and Tables 1, 2, and 3. The code also saves the MSA responses in stata_outputdata.mat, which are later manually combined with other MSA-level data in a CSV file “OutputData” in the Stata_crosssection folder. II. Estimating the Cross-section Regression: Step 1: Import data from OutputData.csv, which includes the cumulative responses generated from FAVAR-SVM and MSA-level indicators used for the cross-sectional regression analysis. Step 2: Run CrossReg.do to generate all cross-sectional results, including Table 4 (baseline results) in the main text, and Tables A2, A3-A16 in the Appendix (robustness checks).
Institutions
- California State University Fullerton