Oscar Jorda and Massimiliano Marcellino, "Path Forecasting", Journal of Applied Econometrics, Vol. 25, No. 4, 2010, pp. 635-662. There are several zip files. The contents of each are described separately. All included files are ASCII files in DOS format. Unix/Linux users should use "unzip -a". 1. The file jm-sw.zip contains the files required to replicate Figure 3 in the paper, that is, the forecasts generated with the Stock and Watson (2001) VAR. There are three files: - jae_sw.g: main file - jae_mc.src: file that contains all the procedures (you should not need to change this file). Note that this file is used for the Monte Carlos as well and contains procedures that are not used here. - swqdata.csv: this file contains the data for the VAR By default, the file plots 8-period-ahead, out-of-sample forecasts based on the SW(2001) VAR. There are several options that you can modify to suit your particular application. Specifically, - You can automate the lag length selection using AIC, SIC or AICC (see Hurvich and Tsai, 1993). - You can use direct forecasts (local projections) instead of the VAR to produce the forecasts. - You can isolate the effect of parameter estimation uncertainty from uncertainty generated by the arrival of shocks. - You can generate impulse responses from the VAR. The output is set to plot the forecasts and marginal, Bonferroni, and Scheffe bands for a confidence level (1-alpha). Some of the options require a bit of programming. We have tried to add comments to make clear what is happening at each step so that you can modify the code to suit your needs. 2. The file jm-figs.zip contains the code required to replicate figure 4 in the paper and includes the files: - jae_fig4.g: main file - jae_lp.src: file with the procedures - Massi207.csv: data file Running the code as is should replicate figure 4 in the paper. Notice that figure 4 was "beautified" in EViews. This file also contains jae_fig5.g, Gauss code to generate figure 5. 3. The file jm_mc_sw.zip contains the files required to replicate the Monte Carlo simulations based on Stock and Watson's (2001) VAR. The files are: -jae_mc.g: modify this file accordingly to cover all the cases -jae_mc.src: this file contains all the procedures and you should not need to modify it. It also contains two directories with output files. 4. The file jm_mc_ar.zip contains the code for the Monte Carlo simulations for the AR(1) Model. The files included are: - ar_jae_mc.g: change the options here - jae_mc.src: same file used in the other simulations. You should not need to change it. It also contains a directory with output files.