I enjoyed this paper very much and believe it usefully adds to our stock of knowledge about what it takes to end extreme poverty. It is a powerful rejoinder to the erroneous claims I’ve heard from some quarters that the Bank’s 3% global poverty target is all but guaranteed to be met and is therefore lacking in ambition.
A key question is how far business as usual will take us. This informs our judgment of how ambitious the 3% target is, or in other words, how much bending of the arc of history is required.
As your bar graph shows, the difference in results between Martin’s benchmarking exercise (which results in the 3% target being met) and your benchmarking exercise (which results in a global poverty rate of 8.6% in 2030) is nearly all accounted for by your use of country-specific historical growth rates to project future growth.
Looking more closely, two regions - South Asia and sub-Saharan Africa - have experienced lagging (survey-based, per capita consumption) growth rates which account for most of the difference between the two exercises.
South Asia is dominated by India, but as you point out in the paper, slow consumption growth recorded between 2002 and 2010 can be explained by the 2009/10 survey occurring at the height of a severe drought. If the newer 2011/12 survey data point is used, South Asia looks more or less on course to meet the 3% goal and this change alone would bring your global benchmark down 1.3 percentage points to 7.3%.
As for Sub-Saharan Africa, you find that per capita consumption has only grown at 2.1 percent a year between 2002 and 2010. This may surprise some readers as 2002 is the year when Africa’s economic growth took off. The truth, however, is we don’t know nearly enough about what has happened to household consumption in this period. Less than half of the region’s countries have undertaken two surveys - the minimum required to discern a trend - since the start of the century.
The reliance on historical survey-based growth estimates to project future growth is one of the virtues of your approach to benchmarking global poverty, as it avoids the messy process of combining irreconcilable data from national accounts and surveys. However, as the points above suggest, it is also a weakness given the paucity of surveys.
One separate point: while I agree that the way in which the benefits of growth are distributed will be crucial to meeting the 3% target, I disagree that past distributional trends at a country level are a meaningful guide to future trends. Forecasting distributional trends may be even harder than forecasting poverty.