This code replicates the empirical results of the paper Causal Inference for High-Dimensional Time Series by Cordoni and Sancetta, publihsed in Journal of Econometrics: https://kwnsfk27.r.eu-west-1.awstrack.me/L0/https:%2F%2Fauthors.elsevier.com%2Fsd%2Farticle%2FS0304-4076(24)00253-7/1/010201936e10fd09-1e212bb5-a4d1-4487-9356-d43016268d46-000000/GOLNGNVh_f5yuJuPeZzeietVYbk=402.
It is written so that it can be modfiied and applied to general data-sets. The file CausalTimeSeries_fx.R contains generic functions loaded by the two scripts script_MACRO_application.R and script_HFT_application.R.
script_MACRO_application.R replicates the results of the macro application. It uses the datafile data.csv.
script_HFT_application.R replicates the high frequency results. It uses the datafile Xall_volume_time.csv. The additional files HFT_application_PC_result.RData and lambda_tau_cv_hft_empirical.RData contain intermediate results. The file is too big to be committed (45MB), but can be dowloaded from here: https://www.dropbox.com/scl/fi/5pmk5dq9jjiuxrbim31io/HFT_application_PC_result.RData?rlkey=rz647hrs56dvwjwynb6n64yos&dl=0
These two files will be automatically generated if flag_first_run in the script script_HFT_application.R is set to true.