James Foster: ‘Multidimensional poverty measurement has followed an interesting path’

5 June, 2018

James Foster is a co-creator of the Foster-Greer-Thorbecke class of measures, one of the most commonly used methodologies for estimating income poverty. He is also the co-author of the Alkire-Foster methodology, a method for measuring multidimensional poverty that has been adopted by the United Nations Development Programme as well as a number of countries.

How did you get involved in multidimensional poverty measures?

My recent interest in measuring multidimensional poverty traces back to work on chronic poverty, or income poverty experienced over many periods of time. My approach began by counting the number of periods in which a person is poor: If the number is high enough, then the person is considered to be chronically poor.

I presented a paper at Manchester and Sabina Alkire, who was in the audience, said to me afterwards: “You know, this is relevant to multidimensional poverty”. I had failed to see the connection, and, in fact, replied “I don’t think so”. But she was quite insistent and invited me to come to Oxford to discuss it. This I did in December 2006, and, after about five hours, she had convinced me that it was very much applicable. In fact, it became the core of our new multidimensional approach (i.e. if you are deprived in enough dimensions you are considered poor, analogous to how I had identified the chronically poor). I had never made the connection nor had I even invested time in considering multidimensional poverty, since I thought it would be too difficult to bring all the dimensions together. It turned out that the counting approach in deprivation space provided the key. It just needed the guidance of someone who had thought a great deal about multidimensional poverty – Sabina Alkire – to make it come to fruition.

The acceptance of multidimensional poverty measurement has taken different paths, with countries adopting these measures while strong debate was still taking place within academia and some international organizations.Why do you think this happened?

Academics and others who deal with poverty measurement, using, say, an income-based approach, have one emphasis: How defensible are their measurement choices within the context of the broader literature? Their notion of what is defensible includes what is able to be justified ‘scientifically’ or ‘objectively’. Multidimensional poverty measurement is not well fitted for this kind of abstract academic discussion because of its normative approach that goes beyond purely scientific and objective considerations.

We try to be explicit as to how the normative context intervenes in the choice of dimensions, in the cutoffs, in the values that are given to each dimensional deprivation, and the ultimate poverty cutoff (which is very much analogous to an ordinary poverty line). These normative aspects can be seen as problematic by those who would like a scientific justification for filling in the essential details.

Yet the whole point of the technology [the Alkire-Foster method] is to translate the underlying values of a group of people, such as a country, into a measurement tool that allows them to monitor their own level of poverty, ultimately reducing poverty as they have defined it. Thus, the fit for country-specific applications is extremely good. You have the possibility of a committee coming together within a country and working diligently to combine the best academic work and the best notions of what it is that people in the country value, what the most important policy issues are, how much of an achievement is needed before someone’s status is no longer seen as deprived, etc.

These are local matters for a country (or perhaps a city or region) to determine, so it was natural that people saw almost immediately that this tool could be used effectively at the country level. They understood its potential for wrapping an entire development plan around a single measure, which can monitor the poor’s progress in all the dimensions simultaneously. The approach is tailor-made for policy and targeting needs. Countries recognised this pretty quickly though it obviously helped that OPHI was so active in engaging countries that were curious about multidimensional poverty.

Yet the whole point of the technology [the Alkire-Foster method] is to translate the underlying values of a group of people, such as a country, into a measurement tool that allows them to monitor their own level of poverty, ultimately reducing poverty as they have defined it.

I think that is the explanation: On one side, academics and people who cover global poverty hope to fill in the details in a way that is viewed as academically sound and scientifically based, as opposed to countries who are more concerned that their measure should reflect the underlying values within the country – which is what national poverty measurement should be about.

What do you think the future is for multidimensional poverty measurement? Will it become a more popular approach than income-based approaches? Will it coexist with income poverty measurement?

Multidimensional poverty measurement has followed an interesting path. First, you have one big example in Mexico that combines income with other dimensions to obtain an overall multidimensional poverty index. If you look at OPHI’s global MPI, it measures deprivations directly across countries in as health, education and living standards indicators. Then you have a whole array of national indices (as in Colombia or Chile) that are purely outside of the income-based poverty approach – which initially I was surprised to see happen.

However, it is not an unexpected development because people are comfortable looking at incomebased poverty and they are also comfortable looking at multidimensional poverty as an idea apart from income. Theoretically, a natural joining of the two could happen, either by merging the two approaches (a person is poor if they are either income or multidimensionally poor) or by intersecting the two, leading to something like what we have in Mexico (a person is poor if they are both income and multidimensionally poor). But at present they are mainly kept separate for very good data reasons.

On the academic side or the international organization side, I would expect that interest in the multidimensional sphere would create theoretical questions along the same lines as we have seen in income-based approaches. For example, if you are a country with a multidimensional poverty measure, how do you most efficiently lower poverty? Or put differently, suppose you have certain tools and a certain budget: How do you maximise the decrease in poverty? That kind of optimal policy exercise is something that I would expect to see more often as academics and international organizations study the issue more thoroughly. I think that people are beginning to see that multidimensional poverty can be usefully analysed in many of the same ways as income poverty, although results may be affected by the presence of multiple dimensions.


Published in Dimensions Special Issue – June 2018


James Foster MPI Multidimensional Poverty