Nobel laureate Amartya Sen’s argument that poverty is more than a simple lack of income
is widely accepted in theory. But according to a new study by Professor Jean-Yves Duclos
and colleagues, published in the October 2006 Economic Journal, it is time for it to be
properly implemented through poverty comparisons that take account of deprivation in a
range of different dimensions as well as correlation between the dimensions.
Most people think of poverty as a lack of income and for most of the twentieth century,
economists interested in poverty accepted that intuitive idea. Beginning in the 1980s,
however, Amartya Sen challenged economists, ethicists and social theorists to think of
poverty more broadly as an inability to be or do certain intrinsically important things, which
Sen called ‘functionings’.
While a lack of income certainly limits our ability to function in certain ways, it is not the only
thing that may limit our functionings. Political repression, physical handicaps or a lack of
social services like schools or hospitals all limit us as well.
An important implication of Sen's argument is that poverty is ‘multidimensional’, involving
more than a simple lack of income. Sen won a Nobel Prize for his ideas, which are now
widely accepted among economists, in theory. In practice, empirical research on poverty
continues to study lack of income, with very few exceptions.
Duclos and his colleagues think that this should change, so that empirical studies of poverty
are aligned more closely with Sen's poverty theory. Their study shows that it is feasible to
make multidimensional poverty comparisons, and it provides examples in which these
comparisons differ from standard poverty comparisons based on incomes alone.
The research team’s methods are more general than two apparently intuitive methods. The
first is the approach taken by the United Nations' Human Development Index (HDI). This is
a weighted average of three dimensions of well-being: incomes, life expectancy and
literacy. But the weights are arbitrary, and changing them may well change the poverty
estimates. The methods proposed in the new study are valid for any set of weights.
The second approach is to compare each dimension of well-being ‘one-at-a-time’. For
example, we can check to see if income poverty is lower in country A than country B; then
compare life expectancy and then literacy. If country A shows less poverty in each
dimension than country B, then we can conclude that poverty is lower… or can we?
As it turns out, a genuinely multidimensional poverty comparison should take into account
not only the individual dimensions ‘one-at-a-time’, but also the correlation between the
various dimensions. This is because, ethically, it is worse for someone who is poor in one
dimension, say health, to be poor in a second dimension, say education, than it is for
someone who is not ‘health poor’.
As an example, the researchers find that the correlation of consumption and health status
for children in Uganda is higher in urban than rural areas. As a result, even though
consumption is higher in urban areas and health problems are fewer, it is not necessarily
the case that poverty is lower in urban areas, because those that have low consumption in
urban areas are more likely to also suffer health problems compared with rural areas.
This higher correlation means that at least some poverty measures – those that put
greatest emphasis on multiple deprivations – will judge that poverty is higher in urban
areas, despite the comparisons based on one variable.
ENDS
Notes for editors: ‘Robust Multidimensional Poverty Comparisons’ by Jean-Yves Duclos,
David Sahn and Stephen Younger is published in the October 2006 issue of the Economic
Journal.
Jean-Yves Duclos is at Universite Laval. David Sahn and Stephen Younger are at Cornell
University.
For further information: contact Romesh Vaitilingam on 07768-661095 (email:
romesh@compuserve.com); or the authors via email: Jean-Yves Duclos (Jean-
Yves.Duclos@ecn.ulaval.ca), David Sahn (des16@cornell.edu) and Stephen D. Younger
(sdy1@cornell.edu).