Business Cycle Synchronization and Asymmetry in the Eurozone

Description

This code and data replicate the results in the Manuscript ECMODE-D-23-00367R2 "Business Cycle Synchronization and Asymmetry in the European Union" written by Vladimir Arčabić, Irina B. Panovska and Josip Tica. The ReadMe File provides the instructions for replicating the results. The user will need RATS version 10.0 and Stata version 17 to replicate the main results. Other than that, R version 4.1.3 and Gauss version 19 or later are required to replicate results regarding principal component, individual nonlinear output gaps, and the Phillips curve coherent output gap and its weights. It will take up to 15 minutes to replicate the main results. Data sources: Quarterly seasonally adjusted data for real GDP, inflation, employment, and the unemployment rate are obtained from the Eurostat and OECD databases. Data and sources are explained in detail in the paper in section 3.1 Data. The raw data are saved in Excel files Data - GDP.xlsx, Data - Employment.xlsx, Data - Unemployment.xlsx, and Data - Inflation.xlsx in Linear gaps estimation folder and in the file Data_GDP_R1.xlsx in the Nonlinear gaps estimation folder. All data is publicly available and obtained from EUROSTAT and OECD. The replication files are organized in five folders: Linear gaps estimation, Nonlinear gaps estimation, Replication of Figures 3, 4 and 10, Replication of Figures 1-2, 5-8, 13 and Tables 1 and 2, Replication of Figures 9, 11 and 12

Steps to reproduce

The replication les are organized in five folders. "Linear gaps estimation" folder: Linear gaps.RPF estimates the four linear gaps. The included procedure named regfilt.SRC is required to be in the working directory or in RATS's Procedures and Examples folder. The code will make graphs of all linear gaps and store them in le gaps.RAT. "Nonlinear gaps estimation" : bb_code.gss estimates the BB gaps, ham_ms code.gss estimates the Hamilton Markov-switching model. Both scripts take as input an excel file data GDP_R1.xlsx with GDP data for all EU countries and ie_gni.txt for Irish GNI. The user has to enter the country code for the desired country in line 69 in the bb_code.gss  in line 55 in ham ms_code.gss. The example is set to run for Austria (AT). The code saves the estimated output gaps as Excel files. "Replication of Figures 1-2, 5-8, 13 and Tables 1 and 2" folder: The code Main file for replicating Figures 1-2, 5-8, 13 and Table 1.RPF is the RATS program that replicates the main results in Figures 1, 2, 5, 6, 7, 8, 13, and Table 1. The program is set to use the PC coherent output gap. To change the definition of the output gap measure, the user should change 360. The user can choose between "cycleMAOG" (simple average), "cyclepmapi" (PC coherent), "cycHP" (HP), "cyclePC" (unrestricted Principal Components with possibly negative loadings), and "cyclePCA1 " (PCA with restricted non-negative loadings which summing up to 1). Table 2 can be replicated using the Stata do file main file to replicate Table 2.do. The folder also contains subfolder for the PCA estimation. The R program "pca using restrictions.R" in the subfolder "PCA robustness check" estimates the principal components using restricted factor loadings for the robustness check (red line) in Figure 2. The code exports the first principal components when the loadings are unrestricted and when the loadings are restricted (version used in the paper). "Replication of Figures 3, 4 and 10" folder: R code for weights to maximize correlation inflation R1.R calculates weights that maximize the correlation between the weighted output gap and inflation in the main and extended sample. It uses "all_gaps_8gaps_extended_sample in.xlsx" with the estimated output gaps for the extended sample from Section 4.4. To reproduce the benchmark results from Section 4.2, replace the toggle "full sample=1" with "full sample=0" in Line 29. The code saves the estimated weights and the averaged output gap. The results for full sample=1 reproduce the weights from Figure 10 and for full sample=0 reproduce the weights from Figure 3. Figures 3 , 4 and 10 are created in the Excel files "Figures 3 and 4.xlsx" and "Figure 10 replicated.xlsx" using the weights estimated with the R code. "Replication of Figures 9, 11 and 12" folder: main file for replicating Figure 9-12.RPF is the RATS program file that imports extended datasets covering Covid period and replicates gures 9 to 12.

Institutions

Categories

Funding

Licence