Were there any significant revisions of the methodology for computation of the HDI?
The modifications in methodology include the change in maxima for normalization of dimensional indices – previously they were equal to the observed maxima over the period since 1980, now they are fixed at 85 for life expectancy (LE), 15 years for mean years of schooling (MYS), 18 years for expected years of schooling (EYS), and $75,000 for GNI per capita (GNI pc). The previously used approach of ‘observed maxima’ was criticized mainly on the grounds that the HDI of the country should depend only on the country's own achievements, however when using the observed maxima the HDI also depended on other countries, on those whose values were used as maxima. For example the HDI of Brazil also depended on how long Japanese live, how well American adults were educated, and how high is GNI per capita in Qatar.
The other change is in the way the education indicators are aggregated. Previously used geometric aggregation was criticized on the grounds that a typical developing country has a (much) higher value of expected years of schooling than of mean years of schooling. By aggregating these two indicators with the geometric mean such a country is 'penalized' because of the difference, although the country is improving education level by having more children attending school at all levels. The use of the arithmetic mean provides an equal treatment to both indicators. These changes have a minimal impact on values and ranks.