At the beginning of 2013, the UN’s High Level Panel on the Post-2015 Development Agenda called for a “data revolution for sustainable development”. It contended that harnessing new technologies for the collection of more and better data related to development would make it easier to target the neediest and to track the progress of development efforts.
In order to better understand the role of data in development work, OpenCanada asked Sabina Alkire, the Director of the Oxford Poverty and Human Development Initiative (OPHI), about how to better understand measures of poverty and the impact of foreign aid. Alkire, who was named as one of Foreign Policy Magazine’s Top 100 Global Thinkers in 2011, played a key role in developing the UNDP’s Multidimensional Poverty Index(MPI). Rather than measuring poverty in terms of income, the MPI measures poverty in terms of ten indicators related to three dimensions: health, living standards, and education.
A recurring question in Canadian debates about foreign aid is whether or not aid actually works. This is largely due to the lack of consensus over how to measure success. How does the work of the Oxford Policy and Human Initiative (OPHI) speak to this question?
OPHI has been developing measures of poverty that more accurately reflect the experience of poverty that poor people have – which is often having different kinds of deprivations at the same time. One of our instruments for doing this has been an international Multidimensional Poverty Index that the UNDP Human Development Report publishes. This is a crude index. It doesn’t reflect the nuances of poverty in many different countries, but it does enable us to look at how poverty has changed over time.
What we have done has been to, first, develop this index and compute it for over 100 countries, and disaggregate it by sub-national regions, and also look at trends over time. So, for example, in 2013 when we launched the index, we looked at 22 developing countries containing 2 billion people. We looked at the rate at which they had reduced multidimensional poverty and how they had done so. We identified key successes, such as Nepal or Rwanda or Bangladesh, and countries that had reduced poverty much more slowly, like Madagascar or Senegal. At present we have not directly linked these trends in poverty with aid flows in a country – it’s complicated to do that well and we’ve not had the time. But what is clear is that our methodology can be used to evaluate development success – indeed agencies could design their own MPI that included indicators they were working on in particular, and use it to monitor progress.
You highlight Nepal, Rwanda, and Bangladesh as successes from the 2013 report. Do these countries share any characteristics that made them more successful than others in decreasing multidimensional poverty?
We are doing in-depth policy studies at the moment of these low-income countries. And what we can say is that multidimensional poverty reflects growth less directly than income poverty reduction does, but is accelerated by policies and even local activism of various kinds. For example, Nepal, which was our top among these 22 countries, had a baseline taken just before they ended the peace accord in 2006, and updated the survey 2011. It’s been a rocky time politically with many, many changes. Yet multidimensional poverty decreased dramatically. It’s in part because the baseline was taken at a very painful time for the country and it could only get better. But Nepal did draw our attention to the fact that it’s not only national governance and social expenditures which reduce multidimensional poverty, but also activities at a local level by civil society and non-governmental organizations. Non-governmental organizations in Bangladesh have played a vital role, and an active civil society that exercises its right to information has too. So, there’s no clear simplistic story.
How do you think we should define and assess the efficacy of aid?
I would follow Angus Deaton in saying that there are multiple ways of evaluating the effectiveness of aid. At the moment, randomized control trials are popular and rightly so, yet these (a) are quite costly and (b) simply cannot be used in many circumstances. In these other circumstances, what we need are forms of evaluation that use both qualitative and quantitative information. So, a more holistic toolkit is needed that has RCTs and also the insights and observations of participants, and contextual notes.
It is also essential to spot where aid has been misused or misallocated. For example, where aid is used for political reasons but not for poverty reduction or broader development gains; or where the costs are ginormous, mostly spent on activities that could be done at a much lower cost, perhaps with a bit more engagement. Conversely, you can spot high-impact investments in long-term activities that are designed so as to be self-correcting, and empower participants to detect and counteract corruption.
In a sense, the question is not really ‘What is the gap that we are trying to close with our aid?’ – which will differ by context – but ‘Where can we do things that other actors in this current context cannot do and where our actions will also catalyze a sustainable response that people value?’
Do you think the Millennium Development Goals (MDGs) have been effective?
Certainly – in some ways. Yet who does not yearn for more progress? As you know from the many studies, the MDGs often served to galvanize public interest and action. By having measurable targets, they focused attention on progress towards transparent outputs that were incomplete yet moved the agenda. Now, the question is ‘What will replace them?’ We clearly need an agenda that will galvanize political will within countries and that is both animated and feasible, and does eradicate the worst kinds of human suffering.
How can measures help this time?
OPHI has been privileged to launch, with the Governments of Mexico and Colombia, a Multidimensional Poverty Peer Network that now has 22 governments exploring national multidimensional poverty indices to complement or supplement their monetary poverty measure. Most of the emphasis is on having better national indicators of multidimensional poverty that reflect their own context and priorities. Mexico, Colombia, and Bhutan already have official national statistics on multidimensional poverty that use a methodology developed at OPHI. For example, the Government of Colombia uses indicators from its national development plan. Not all of them reflect functionings, but they do reflect things that the government has promised to its people. So in that context, it’s useful to have a measure which annually monitors institutional outputs to meet stated national goals.
The national demand is driven by the need to complement income poverty measures with measures that make visible progress in non-income deprivations such as in health, schooling, housing, assets, work, electricity, water, and sanitation. Measures using this methodology create powerful incentives for change, because they are more informative than income poverty measures alone. They show the interconnected deprivations that trap each person or household in poverty, and can be analysed easily and accurately – both by individual indicators, and by regions or social groups.
In addition to better national metrics this network, together with members of the Organisation for Economic Cooperation and Development (OECD) and some regional bodies, including the Southern African Development Community (SADC) and the Economic Commission for Latin America and the Caribbean (ECLAC), decided to advocate for a multidimensional poverty index (MPI 2.0) in a post-2015 context. They felt that a global and internationally comparable headline of multidimensional poverty would really add value to a list of individual indicators. One of the things OPHI is working on is a proposal for there to be an indicator – complementing the $1.25-a-day indicator – of multidimensional poverty that gives an at-a-glance figure of how people’s lives are going in other dimensions.
So, you’ve mentioned that multidimensional measurement should be prioritized in the post-2015 agenda. What other issues do you think are important?
The “data revolution” that the High Level Panel advocated is fundamental. It potentially costs less than some of the analysis that ‘update’ MDG estimations without new data, to guess where a country now is. In my mind, it would be a much greater service to the poor in these countries to have timely information about their lives made available through regular surveys. We have data on the stock market every hour, on the labour force every quarter, on inflation, GDP, and so on. But data on malnutrition, housing, electricity, education, etc. is updated maybe every five years. What would help is a brief multi-topic survey, that includes key indicators on multidimensional poverty, violence, and work, and allows space for nationally chosen questions. This could be implemented by national statistics agencies or an international body, as countries decide.
In an age, where, in so many domains, we are flooded by information, it’s a travesty that we don’t have up-to-date information on key dimensions of poverty, to design high impact policies and celebrate policy success. This must change. People will say, ‘Well, you work on measurement. Of course, you want data – but that’s not a core priority.’ Well, I think poverty reduction programs could be made much more cost-effective if they were informed by regular, timely data. To give just one example of many, India, the country with the highest number of malnourished people and high absolute rates of child stunting, has no national data on malnutrition since 2006.
How can the Multidimensional Poverty Index, or similar measures, be used to decide on priorities for development programs?
The current global MPI contains ten indicators related to three dimensions: health, education, and living standards. In the MPI country briefings, you see not only the percentage of people who are multidimensionally poor – meaning they are deprived in one-third of the dimensions at the same time – but also the composition of their poverty and the intensity of their poverty – the average percentage of deprivations that a person suffers at the same time. And you can do this by country, state, or ethnic group.
This information is being used already and can be used further to inform aid allocation, for example by focusing on the poorest groups. The MPI complements a monetary measure and say, ‘Well, if this many people are monetarily poor, how many of them are multidimensionally poor?’ In some cases, it’s similar; in others, it’s very different. In fact, we’ve been surprised by the extent of difference between the measures. In some cases, like Ethiopia, the MPI values are quite different from the income poverty levels.
The MPI serves as a single figure that draws together, in quite a natural way, many of the core MDG indicators that are doubtlessly essential – identifying households that have experienced child mortality, malnutrition, children not attending school, and poor housing and services. I think aid and poverty assessments of countries need to be informed by both measures: monetary and multidimensional poverty.
So, what kind of gaps do you see in Ethiopia?
Well, approximately 90% of the population is poor according to the MPI; whereas, by an income assessment, less than 40% of people are poor. These are really quite divergent assessments as to the magnitude of poverty in Ethiopia. It basically raises questions, questions which people with an in-depth understanding of the country can address.
Drilling down, 46% of multidimensional poverty in Ethiopia comes from basic standard of living indicators. For example, 79% of people are poor and don’t have electricity; 82% lack sanitation; 83% live on a dirt floor; 87% cook with wood charcoal or dung. These are very tangible deprivations that can be addressed.
Hopefully, the multidimensional poverty index can give people a bit of a steer about the deprivations which people experience together in different contexts. The MPI is not one number, but a high-resolution lens. If you open the tables in our data bank, you see the single number being broken down in many different and informative ways.
Would you say then that we can use the MPI to determine how we should allocate scarce resources to target the most common deprivations?
Well, that’s one way. But, if you think of the first key message in The MDGs at Mid-point – a 50-country study on accelerating progress that the UNDP released in 2010 – they found that successful countries had addressed different deprivations together because they’re interconnected. That is something that the MPI enables you to do. It enables you to see which people suffer multiple things at the same time, and what these are.
In my mind, one of the real advances of this work (that was not possible earlier, due to slower computers and/or lack of data), is starting from each household, as OPHI’s methodology does, looking at each person’s deprivation profile, and developing a rigorous measure from these. It is incredibly information-rich.
All in all the MPI seems to reflect a pretty dramatic shift in how we think about measuring poverty. What inspires your multidimensional approach?
A multidimensional poverty approach for me, personally, is very much inspired by the work of Amartya Sen, which argues that we should measure poverty according to the aspects of the lives of poor people which are going well or which are not going well. To do so, we need indicators on what Sen would call functionings – things like being well-nourished, being well-educated, being sheltered from the elements, having need for work. Functionings reflect what people value and have reason to value.
Sometimes we have relatively good indicators. For example, for malnutrition using body mass index and micronutrients or anemia or other indicators. Other times, we make do with quite insufficient indicators. For example, you and I might have the same number of years of schooling, but you might have a much, much better education because you paid attention, whereas I goofed off all of my 12 years of schooling. So, years of schooling is an imperfect proxy for the functioning of being well-educated. But sometimes that is as good as we can get.
Focusing on functionings helps us to clarify what the indicators are capturing and where they fall short, as in the example of years of schooling. The capability approach also sees poor people as agents, whose visions and values can and should shape activities. So measurement is but one part of a full response.
Jean Drèze and Amartya Sen’s recent book, An Uncertain Glory: India and its contradictions, is a beautifully written yet hard-hitting text. It embodies the patient creative balanced analysis and policies needed to confront unacceptable kinds of human suffering. But in the case of poverty they do not advocate patience. Indeed they rail against it, pointing out that patience was cited by Ambrose Bierce in The Devil’s dictionary as ‘a minor form of despair, disguised as a virtue.’ Their impatience also inspires.