Everett Grant and Julieta Yung, "The Double-Edge Sword of Global Integration: Robustness, Fragility & Contagion in the International Firm Network", Journal of Applied Econometrics, Vol. 36, No. 6, 2021, pp. 760-783. All files are stored in gy-files.zip. It contains a mix of binary and ASCII files (in DOS format), so Unix/Linux users need to be careful. Data The data used in the article are from Bloomberg, Markit, and Global Financial Data and can be retrieved with the appropriate licensing agreements. > Bloomberg Data. First, we included the daily equity closing price for each firm from January 1st, 1981 through September 30th, 2016, with all market values converted to U.S. dollars at market exchange rates, removed exchange traded funds, and took the remaining 1,182 equity securities in the union of these annual sets as our sample (representing the top 1% of global firms by equity market capitalization). If there were multiple tickers for a single firm then we first filtered by keeping only the securities marked as primary, and if there was still more than one ticker, we took the one with the best data coverage. From firm quarterly statement data, we collected EBITDA (profit), revenue (Sales/Revenue/Turnover), and total firm equity. We also included the annual return on equity (RoE) over each calendar year. From these we calculated log changes as well as distress/crisis state indicator variables for the worst 10% performance within our sample by each measure. The headquarters country, currency of equity issuance, and industry firm meta-data was also retrieved. > Markit Data. We gathered daily 5-year CDS spreads from October 2013 through November 2015. > Global Financial Data. We collected the 3-month TED spread; S&P 500 Total Return Index; and the CBOE S&P 500 Volatility Index (VIX). We then calculated the log changes of each variable and indicators for the worst 10% of outcomes over monthly, quarterly and annual tenures. For more details on the data collection process and summary statistics, refer to the accompanying Internet Appendix, Section A. Estimating the Network Edges Refer to Internet Appendix, Section B for a detailed outline of the algorithm used, including corresponding R Packages and parameter specifications. General Instructions Run SpringPlotsNetworkAnalysis.R to reproduce network figures in R Studio. Run RyFRegressionTables.do to reproduce regression tables in Stata. Reproduce Figures 2 and 3 1. Run main file SpringPlotsNetworkAnalysis.R, which calls on ForceAtlas2.R (stored under data). This file imports the data that contains all the firms (network nodes) and the weights that describe how they are interconnected (network edges) and displays the networks using spring plot algorithms. See the paper for details on network estimation and interpretation. The file was last run with R studio Version 1.1.456. You will need to install the necessary packages first. The entire code takes 5-10 minutes to run, depending on processing speed. 2. Data files are stored in the data folder. Con_Long*dates*_lag1_horizon1.csv and NetwEdges_Long*dates*_lag1_horizon1.csv contain the nodes and network edges as estimated with one-step-ahead impulse response functions from a vector autoregression of order one using adaptive elastic net for different estimation periods denoted by *dates*. NetwNodes_Long.txt contains firm attributes for determining the color of the nodes. 3. Output files are stored in the output folder, where each file is a network figure. Eps files can be opened with Adobe to see each figure in pdf format. Figure 2. Global Network in Different Periods by Industry a) SpringPlots_Long1992_1996_lag1_horizon1_GIRFAbs_ForceAtlas2_IndShort.eps b) SpringPlots_Long1997_2001_lag1_horizon1_GIRFAbs_ForceAtlas2_IndShort.eps c) SpringPlots_Long2002_2006_lag1_horizon1_GIRFAbs_ForceAtlas2_IndShort.eps d) SpringPlots_Long2007_2011_lag1_horizon1_GIRFAbs_ForceAtlas2_IndShort.eps e) SpringPlots_Long2012_2016_lag1_horizon1_GIRFAbs_ForceAtlas2_IndShort.eps Figure 3. Global Network in Different Periods by Currency a) SpringPlots_Long1992_1996_lag1_horizon1_GIRFAbs_ForceAtlas2_FXRegion.eps b) SpringPlots_Long1997_2001_lag1_horizon1_GIRFAbs_ForceAtlas2_FXRegion.eps c) SpringPlots_Long2002_2006_lag1_horizon1_GIRFAbs_ForceAtlas2_FXRegion.eps d) SpringPlots_Long2007_2011_lag1_horizon1_GIRFAbs_ForceAtlas2_FXRegion.eps e) SpringPlots_Long2012_2016_lag1_horizon1_GIRFAbs_ForceAtlas2_FXRegion.eps Reproduce Tables 2 and 3 Run the main file RyFRegressionTables.do, which calls on three data files (RegressionDataMonthly.DTA , RegressionDataQuarterly.DTA , RegressionDataAnnual.DTA) stored in the data folder. The tables in XML format are stored in the output folder (table2a, table2c, table3aMonthly, table3aQuarterly, table3aAnnual). For details on the regressions and variable definitions, refer to the paper. The file was last run with Stata/SE 15.1. The entire code takes about 1 minute to run.