Output Structure

The forecasting outcomes are stored in the folder named OUTPUT, where the forecasting comparison chart, the impulse response plot, and the variance decomposition plot are stored in the subfolder Charts, and individual model forecasts can be found in the subfolder USMODELS.

Note

In addition to forecasting outcomes, there are some results directly derived from Dynare, such as the model solution, and they can be found in the subfolders of the MODELS folder.

Folders

Graphs saved in Charts

The charts and plots generated by the platform are all stored in the subfolder Charts, which include forecasting comparison charts, impulse response plots based on the Bayesian IRF, and historical variance decomposition plots. These charts are saved in PNG-format.

Model forecasts saved in USMODELS

The Excel spreadsheets storing individual forecasting results are saved in the subfolder USMODELS. The syntax of the subfolder name in this folder is as follows:

ModelName_DataType_EstimationType_(spfnc)_EstimationMethod

  • The ModelName denotes the name of the DSGE model.
  • The DataType is the type of data used for estimation, which can be RT for real data, or DT revised data.
  • The EstimationType summarizes the type of estimation, which can be EW for expanding windows, or RW for rolling windows.
  • If SPF nowcast data are used in the estimation, then a string spfnc will appear in the name of the subfolder, otherwise it will not.
  • The EstimationMethod shows the estimation algorithm used, which can be modeforecast or MHforecast.

For example, if we estimate the Smets and Wouters (2007) model using real time data-set with expanding windows, and we choose to augment the data by including SPF nowcast, and if we choose to employ the Metropolis-Hasting algorithm in our estimation, then the derived forecasting results will be stored in a folder named as US_SW07_RT_EW_spfnc_MHforecast.

Inside these subfolders, you will find Excel spreadsheets that store model forecasts. The names of these spreadsheets include a string to present the vintages. In each spreadsheet, the forecasting outcomes are stored in different sheets.

Note

Only the forecast of three observables are in these Excel spreadsheets, which are the annualized real GDP growth rate, the annualized GDP deflator, and the Federal funds rate in annualized terms.

The median sheet and the mean sheet contain the median and mean forecasts respectively. The first line of each sheet are the names of the observables. Starting from the second line, you find the forecasts of the observables. Again, if SPF is not conditioned on the first forecast will be nowcast, and if SPF is conditioned upon the first forecast will be the next period forecast.

Output from Bayesian VARs and Metropolis-Hastings estimations have another sheet named distribution. On this sheet, the deciles of the forecast distribution are stored.

Adding model forecasts

Should you wish to include your own model forecasts into the platform for comparison purposes only, please save it under another subfolder, and name the subfolder as well as the spreadsheet following the conventions above.