• Run SR and MR analysis also for inflation

General structure comparison sections:

• number of countries for which lower RMSE (every issue)
• share of countries by geographical region
• magnitude of average difference
• all forecast errors evolution by issue
• recessions table (forecast errors during recessions and booms)
• FOR THE APPENDIX: individual table with difference RMSE
• FOR THE APPENDIX: individual table with list of countries
• ONLY FOR WB (maybe EC as well?): figures and tables aid and bias

To do (on figures and tables already worked on):

Comparison consensus:

• add figures evolution bias
• aggregate figure of number of countries with better accuracy
• magnitudes rmse difference
• order variables in the first table about recessions
• remove IMF forecasts in second table with best forecaster

Comparison EC:

• produce all four figures for bias
• figure magnitudes
• table with recessions

Comparison WB:

• update material from note

Solves problem of loading assigning name

• Individual Consensus forecasters (distribution recessions and collinearity with other forecasts)
• Program adjustment at first revision - comparison with Consensus
• Preliminary results on Covid-19 forecasts (using January 2021 WEO as actual)

• Missing the script with cleaning of consensus data

• produce individual tables bias for all horizons (spread out by H=0 and H=1)
• round numbers bias in tables
• graph share of countries with bias by region
• table with magnitudes biases by region
• tables recession by geographic group

• correct geo group name in share of biases by geographical group plot
• elongate legend in share plots
• set pre-defined export dimension of figures

• Change chunk of code (section 1.1) where setting parameters to retrieve actuals latests value after raw data update

• top optimistic and pessimistic errors during Covid-19 crisis

Need to see with different colours recessions, recoveries, trough in the distribution plot. Otherwise confusing

Share of countries with inefficient incorporation of spillovers

• Problem converting imf codes in country names with the country.code package in the wrangle_weo_forecasts function, the package produces NA for some small states (e.g. Tuvalu)

• after 2011 dummy and interaction with amount quota

• check and correct bibliography
• write conclusion and abstract
• format tables appendix

• type of programs, movement of distribution with reviews

• ensure main script runs everything and generates all plots, tables etc. (all packages present and ordering correct)