Alain Hecq, Joćo Victor Issler, and Sean Telg, "Mixed Causal-Noncausal Autoregressions with Exogenous Regressors", Journal of Applied Econometrics, Vol. 35, No. 3, 2020, pp. 328-343. The data used in the article are from the International Monetary Fund (IMF, for commodity prices) and Federal Reserve Bank of St. Louis (FRED, for exchange rate and industrial production index). They are attached in a .txt file combined_data, since these data are freely available but subject to revisions regularly. The data have monthly frequency and are reported from 1980M1 until 2019M4, which accounts for 472 observations. The first column indicates the time, while column 2-6 report the used commodity prices (Beverages, Raw Materials, Oil, Nickel and Copper, respectively), column 7-8 contains the explanatory variables (Exchange Rate, Industrial Production Index, respectively). Variables are in levels and need to be transformed to first differences to perform the data analysis in the paper. The file application.R can be used to reproduce results. Concerning the simulation studies: - Results in section 4.1 can be reproduced directly by using the R package MARX (use sim.marx function to simulate process and selection.lag function to determine lag order p) and the file model_selection.R; - Results in section 4.2 can be reproduced using the files identifiability.R and identifiability2.R; - Results in section 4.3 can be reproduced using the file forecast_simulations.R. All these files are ASCII files in DOS format. They are zipped in the file hit-files.zip. Unix/Linux users should use "unzip -a". There is also an online appendix containing graphs, extra simulations results and proofs: hit-appendix.pdf Please address any questions to: Sean Telg Department of Econometrics and Operations Research Vrije Universiteit Amsterdam De Boelelaan 1105 1081 HV Amsterdam The Netherlands j.m.a.telg [AT] vu.nl