Procedure for Reproducing the Graphs and Tables from the Final Data for

“Africa is on Time”

By Maxim Pinkovskiy and Xavier Sala-i-Martin, Journal of Economic Growth

  1. There are three base data files:
  2. “resultsAfrica_new.dta”, which contains poverty rates, GDP per capita and inequality indices for all African countries in 1990-2011,
  3. “africa_countries.dta”, which contains classifications of African countries, which are partially reproduced in Table 1 of the paper.
  4. “ssagroupa_distributions.dta”, which contains the income distribution for all Group A Sub-Saharan African countries.
  5. There are 5 do-files to replicate the final results:
  6. “dummies_new.do”, which creates a data file “results_africa_brief_new.dta” that is useful in subsequent computations.
  7. “AfricaDistributionGraphs.do”, which replicates Figures 1-3 and Figure 10 of the paper (the figures with distributions)
  8. “AfricaPovIneqGraphs.do”, which replicates all other figures of the paper (the poverty and inequality time plots).
  9. “Table2.do”, which replicates Table 2 in the paper (the Group A table)
  10. “Table3.do”, which replicates Table 3 in the paper (the all-Africa table)
  11. To replicate the results,
  12. Merge the Data and Replication folders for Graphs and Tables.
  13. Run the file “dummies_new.do” first. Then run the other do-files in any order to replicate the graphs and tables that you wish.
  14. A note on naming conventions: the poverty and inequality series are named as

(measure)_(GDP type)_(inequality series type)_(extrapolation type)_(distribution type)_(adjustment type)

Where (measure) is a poverty rate or an inequality measure, GDP is World Bank (WB13), PWT 7 (7B), PWT 8 (8013), Maddison (M13), synthetic PWT (S13), or Alwyn Young-adjusted (WB34), inequality is 3 (the baseline), 3l (only CR (2010) surveys) or 3c (consumption inequality only), extrapolation type is near (conservative), line (linear) or pchi (baseline), the distribution type is normal, gamma or weibull, and the adjustment type is reg (baseline), gin (inequality from inverting the Gini) and adj (middle quintiles only misreporting adjustment).