Identifying chronically deprived countries: results from cluster analysis, CPRC Working Paper No. 70


This paper provides a classification of non-OECD countries on the basis of the level of, and the change in, their average welfare over recent decades. The method used is that of cluster analysis. The main finding is that four welfare indicators - GDP per capita, child mortality, fertility and under-nourishment - all show strong evidence of three distinct country clusters: a 'chronically deprived' group, characterised by low initial levels of welfare and relatively slow rates of progress; a 'good performers' group, consisting of countries with similarly low initial levels of welfare but faster rates of progress; and a group of 'others', consisting of countries with higher initial levels of welfare and moderate rates of progress.

On the basis of these results, approximately 30 countries are defined as chronically deprived 'across-the-board', i.e. with all four welfare indicators showing the three-cluster pattern (or at least all of those four for which data are available). These account for approximately 10% of the total population of non-OECD countries, but close to 20% of the total number of people living in less than $1-a-day poverty, and more than 30% of the total number of child deaths. A further 45 or so countries are defined as 'partially' chronically deprived, i.e. with one but not all of the four indicators showing the three-cluster pattern. They account for over 50% of total $1-a-day poverty and 55% of total child deaths, compared with approximately 40% of total population.

Various factors increase the probability of being a chronically deprived country. Although it is difficult to separate out the effects of these different influences, the results of multivariate regression analysis suggest that geography (e.g. location in tropics, distance from major markets) and external conditions (e.g. shifts in the terms of trade) play the largest role. Furthermore, the most recent rates of progress in chronically deprived countries continue to lag significantly behind rates of progress in all other non-OECD countries. This suggests that whatever has been driving the differences in performance between chronically deprived and other countries, it has been persisting in recent years.

As regards policy implications, one of the main issues is the amount of aid the different groups receive. The across-the-board chronically deprived countries currently receive around 30% of total aid. Although this is close to their shares of total poverty and total child mortality, one could make a case for their share of aid being even larger, given the natural disadvantages they appear to face. The partially chronically deprived countries are more clearly under-aided: their share of total aid is also currently around 30%, compared with their shares of total $1-a-day poverty and child deaths of over 50%. Whether donors and multilateral agencies are pursuing the right aid strategies in the different country groups is something which needs to be considered further.

This paper was commissioned as a background paper to the 2nd Chronic Poverty Report and was funded by AusAID.


Identifying chronically deprivedcountries: results from cluster analysis, CPRC Working Paper No. 70, IDPM/Chronic Poverty Research Centre (CPRC), Manchester, UK, ISBN 1-904049-69-9, 47 pp. + 4 annexes

Published 1 January 2007