1. See, for example, the study by Pinkovskiy and Sala-i-Martin (2010) and some criticism and counter-criticism at http://blogs.worldbank.org/africacan/is-african-poverty-falling and http://www.salaimartin.com/academics-and-books/65-altres/552-response-to-martin-ravallion-and-the-world-bank.html.
  2. On the positive effect of economic growth on poverty reduction, see also Dollar and Kraay (2002).
  3. As an illustration of the limited data even the most influential studies were based on ECA (1999) and Ali and Thorbecke (2000) as well as Fosu (2008 and 2010) used a sample of only 16 African countries, with one data point each for urban and rural poverty.
  4. The results in Table 4.1 are based on a model of poverty specified as follows: growth in the poverty rate is modelled as a negative function of growth in income and a positive function of the growth in inequality, initial inequality, measured by the Gini coefficient, as well as the ratio of the poverty line to income. The reduction in poverty is specified as depending on the interaction between income growth and the initial level of inequality, income growth and the ratio of the poverty line to income, growth in inequality and the initial level of inequality, and growth in inequality and the ratio of the poverty line to income (Fosu, 2011).The extent to which these factors are important is empirically determined and represented by the income and inequality elasticities of poverty (Table 4.1). These elasticities are defined as the ratio of the percentage change in poverty measured as P0 to the percentage change in income per capita expressed in 2005 PPPs, and as the ratio of the percentage change in poverty to the percentage change in inequality measured by the Gini coefficient. The data are from the World Bank (2010a). 
  5. These elasticities were estimated using a limited number of household surveys at country level, so they should be treated as only indicative of the strength of the relationship between economic growth, inequality and poverty reduction. Also, given that the timeframe for which there are data on poverty differs from country to country, comparisons of these elasticities across countries should be made with caution.
  6. See also World Bank (2010b).
  7. Growth of non-mineral commodity exporting countries has also been strong but volatile.
  8. For a detailed discussion of the origins of the HDI and the theory underlying it, see Fukuda-Parr and Kumar (2003).
  9. Dimensional indicators were rescaled into indices using the following formulae: (actual value-minimum value)/(maximum value-minimum value), where maximum and minimum values were goalposts chosen for each indicator. For example, the maximum goalpost for adult literacy rate was 100, with a minimum goalpost of zero. Minimum values were considered to be “subsistence” values or “natural” zeros. In HDR 2010, minimum values for life expectancy are 20 years; zero years for both education variables and USD 163 for per capita gross national income (UNDP, 2010).
  10. These methodological changes have raised a passionate debate in the literature (see, for example, aidwatchers.com/2010/12/page/2/ and http://hdr.undp.org/en/humandev/lets-talk-hd/2010-12a/
  11. The derivation of the IHDI is based on the Atkinson (1970) family of inequality measures. For more technical details, see Technical Note 2 in UNDP (2010).
  12. Maternal mortality ratio and adolescent fertility rate are coded as “na” (not applicable) for the male gender. See Technical Note 3 in UNDP (2010) for details on the averaging of the sub-indices.