How should a post-2015 agreement measure poverty? Vote for your preferred methodology

The blog’s been insufficiently techie of late, so step forward ODI’s Emma Samman with a piece + poll on measurement. Maybe the start of a ‘Friday geek ‘ series?Emma S Headshot

Some one in five people today still cannot provide for their most basic needs, progress on Millennium Development Goal (MDG) 1 (to halve extreme poverty and hunger) notwithstanding. The High-Level Panel report affirms that ‘eradicating extreme poverty from the face of the earth by 2030’ should be at the core of a post-2015 agreement: ‘This is something that leaders have promised time and again throughout history. Today it can actually be done.’ The World Bank has endorsed this viewpoint, as have David Cameron, Barack Obama and The Economist, alongside several NGOs.

But is the goal ambitious enough – in terms of who it targets, and how? We’re exploring these issues as part of Development Progress, a four year project that aims to explore what’s working in development and why. We asked several experts to make proposals as to how to measure poverty in a post-2015 agreement. Their contributions show some consensus, but also several areas of contention.

The definition of poverty

Who, among the following, would you consider to be poor?

  • A rural smallholder in Tanzania who earns $0.75 a day selling his produce, and cannot provide enough food, clothing and medicine for his family.
  • A woman living in London’s Tower Hamlets who earns the minimum wage, putting her slightly above the poverty line, but who feeds her children via a food bank, for which the family is stigmatised.
  • A young man living in New Delhi with earnings that put him comfortably above the Indian poverty line but below the US poverty line.
  • A slum dweller in San Salvador who has an income slightly above the poverty line, but is illiterate and lives in condemned housing in a violent neighbourhood.

Measuring poverty cartoonOur contributors agree that a poverty measure must identify people who cannot fulfil their basic needs defined globally, i.e. absolute deprivation. Several contributions – Martin Ravallion, Stephan Klasen and Lant Pritchett – make clear that poverty is relative as well as absolute, and that a societal reference point is needed. People should be able to live not only free from starvation and disease, but in accordance with social norms – what Adam Smith labelled centuries ago the ability to appear in public without shame.

For Ravallion and Klasen the appropriate reference point is the society in which a person lives, while for Pritchett, it should also be global. Pritchett seeks to capture absolute poverty in rich and poor countries alike by proposing multiple poverty lines, including a $1.25 a day measure – attuned to the world’s poorest countries – and a $12.50 a day measure, attuned to the richest. A plurality of measures would also deflect criticism that current poverty lines exclude a large number of people who, if not actually destitute, might be hovering precariously around that line.

Sabina Alkire’s proposal is rooted in absolute deprivation. Indeed, her multidimensional MPI 2.0, an updated MPI, is the only measure that combines the poverty ‘headcount’ – i.e. the share of people that are poor in several dimensions – with the ‘depth’ of deprivation – the number of deprivations, on average, that poor people experience.

Amanda Lenhardt argues that, regardless of the measure, there is a need to capture the circumstances of particular groups by monitoring ‘below the averages’ – e.g. the bottom quintile, decile or ventile of the population and different social groups.

One major strand of debate arises between advocates of an income poverty measure (Ravallion, Pritchett, Klasen) and those of a complementary multidimensional ‘MPI 2.0’ index (Alkire). Pointing to little correlation between measures of extreme income poverty and other types of deprivation, Alkire argues for also focusing directly on multiple dimensions of illbeing – for instance, the lack of adequate housing, improved sanitation, education, and, in extreme cases, the likelihood of survival. Klasen argues that such a measure may not be needed, as deprivations in education, health and other dimensions of wellbeing will be captured in an MDG ‘dashboard’. The debate would appear to hinge on how important it is to identify those people who are experiencing numerous deprivations at the same time.

How to construct an income poverty measure?

The $1.25 measure, on which the MDG target is based, is the average poverty line among the world’s 15 poorest countries, denominated in internationalmeasuring poverty 2 dollars using Purchasing Power Parity (PPP), an adjustment designed to compare purchasing power across countries and over time. One dollar (PPP) in Madagascar should, in principle, have the same value as one dollar in Indonesia. Ravallion’s ‘weakly relative’ poverty lines are derived from national poverty lines converted into international (PPP) prices, and Pritchett’s $12.50 line is also in international dollars.

But some argue that PPP conversions don’t work very well, causing uncertainties about how many poor people there are and where they live. Part of the problem lies in how international exchange rates are computed: the common denominator is a basket of goods including items that don’t feature highly in the lives of extremely poor people (e.g., wine from France’s Bordeaux region). Methodological updates and the need to project prices ‘backwards’ are also issues. Stephan Klasen advocates, in place of PPP measures, working toward ‘internationally coordinated national poverty measurement’, based on the basic needs of the poor.

None of these issues are easy to address, which might well be why the experts still haven’t cracked it. Some are technical questions to do with valuation and exchange rates, and monitoring. But others are much more fundamental questions about what we consider to be just societies.

Isn’t how we define poverty too important to be left to technical experts? After all, it gets to the heart of who we consider to be excluded, how we try to tackle deprivation, and how ambitious a new global framework should be.

Now it’s your turn…

Which of these poverty measures do you prefer as the basis for a post-2015 agreement?

  • $1.25 a day poverty line alongside a ‘weakly relative’ measure
  • National poverty lines constructed in an internationally consistent manner
  • Multiple poverty lines: $1.25 a day, national, $12.50 a day
  • Share of the bottom quintile, decile and ventile of the income distribution relative to the average
  • Multidimensional Poverty Index (MPI 2.0)

And as a special favour, you’re allowed to vote for more than one option (max of 3)

Subscribe to our Newsletter

You can unsubscribe at any time by clicking the link in the footer of our emails. For information about our privacy practices, please see our .

We use MailChimp as our marketing platform. By subscribing, you acknowledge that your information will be transferred to MailChimp for processing. Learn more about MailChimp's privacy practices here.


11 Responses to “How should a post-2015 agreement measure poverty? Vote for your preferred methodology”
  1. Tony Harrison

    $12.50 a day should become the new global poverty line. Then there should be a ‘extreme wealth’ line above which you are not allowed to go. Say, $100 a day?! Let’s have some actual ambition, development community!

  2. A thought provoking piece.A few thoughts…

    Firstly, people living on less than $1.25 need to be brought above the line. Then we can then move on by the raising the bar higher by tackling people below $2.25 a day and so on, but we have to start somewhere if we are serious about ending absolute poverty. No one anywhere, should be left behind or left in a cycle of poverty from which it is impossible emerge without assistance.

    To end poverty we need to know who the poor are, where they are and what their needs are. (Laurence Chancy’s recent report for us at Development initiatives is a useful contribution to this debate Unfortunately, at the moment the ability to do this is limited and as a result we are making decisions about addressing the needs of the poor based on bad data. We need to:

    • acknowledge that counting the number of people living in poverty worldwide is difficult. It incurs technical, resource, coordination and institutional challenges.
    • recognise that the availability and quality of poverty data has improved dramatically over the past two decades. At the same time, however, certain weaknesses within the data have crystallized.
    • recognise there is scope to significantly improve the quality and usefulness of poverty data through a combination of immediate reforms and longer term investments.

  3. One weakness I see with all of these approaches (and I do acknowledge that ANY approach will have weaknesses) is that, although some attempt to account for differing cost of living in different countries, none seem to be tuned in well to differing costs of living for different households within the same country, or even different individuals in the same household.

    For example, a person with mobility impairments may need a wheelchair or other mobility devices before they can even go anywhere to independently purchase goods. Or, go to their job to earn the money they need to purchase goods. In developed countries, a good quality wheelchair can be upwards of $2000 or $3000 USD–for some wheelchair users, this could be easily their third largest expense after their home and car.

    A wheelchair user also may need more space in their home than others simply to turn around–not a problem if you’re able to afford an 800 square foot apartment but more of an issue if you’re poor enough that you’re looking at homes where there will be barely enough room for everyone to stretch on the floor to sleep.

    Quite simply, disability can have a signicant impact on the amount of income you need to continue surviving and meeting your minimal needs from day to day. Amartya Sen has commented on this. Many medical conditions can impact these also: in countries where consumption of wheat products is commonplace (eg, US), breads and other items made to meet the needs of people with celiac disease are often much more expensive, which means a poor person with celiac disease may have more difficulty purchasing enough food to meet their nutritional needs. But most measures of poverty do not account for these kinds of individual variations.

    The World Health Organization and the World Bank report that 15% of the world population–one out of every seven people on Earth–is a person with a disability (or disabilities). That’s at least 1 in 7 people who may have more difficulty affording a hypothetical standardized “basket of goods” than their income level would suggest. And that is not even accounting for other variations among individuals and households. Women of a certain age need materials for managing menustruation that men (and older women/younger girls) do not. A fast growing, muscular teen boy who engages in hard physical labor every day needs more food than an infant or an elderly woman. And so forth.

    I do not have enough knowledge of higher maths or methodology to propose a concrete suggestion for how to account for these variations. But I would love to see the experts at least begin debate on how they could go about it.

  4. Ryan Edwards

    Great summary post. And I really enjoy your blog, Duncan.

    Your survey results are interesting so far, but not unexpected. It is clear that the current MPI dimensions are important and I am quite a fan of this work, but from scanning the links and some other docs online it remains unclear what the proposed ‘2.0 changes’ are, and how the issues with weightings and the natural tendency to not focus on each important policy dimension – but rather tout headline figures – will be managed.

    For this reason, if ‘zero goals’ are to remain the flavor of the day, I think I prefer the old fashioned simple extreme poverty monetary consumption and income measures simply because they are most practical (and probably reliable) to work with, coupled separate goals related to each MPI dimension and other aspects of human development. It is a risky business picking one headline variable (or a composite for that matter) and I don’t think it is a great idea to have to choose between various HD dimensions when we know they are all important.

    Also, for the last issue, is there a better method than PPP conversions yet (when working with $$)? I’d also be interested to learn more about how domestic lines are constructed in an internationally consistent manner (guessing its in the other posts though!)

    Thanks and look forward to reading all the other posts on this.


    PS. This link below provides a short and interesting commentary from when this years HDR was released, which may be of interest:

  5. Stuart

    It is not looking too likely that there will be a goal on inequality. This makes it especially important that the poverty goal includes a relative element. Looking at the income share and income growth of the bottom 5 or 10% is a reasonable way of doing this, in addition to national and/or international absolute poverty lines.

    The MPI just confuses things, when there are likely to be separate goals on education, health and housing. There could be a separate supplementary goal on reducing the extent of overlapping deprivation in terms of the other goals, but really, what would that add? It is also much harder to communicate when (for instance) education is seen as a ‘dimension’ of poverty rather than as an issue in its own right; it introduces a whole new batch of counter-intuitive jargon.

  6. Dennis

    I believe Stuart has a very interesting point. I am certainly in favour of a multidimensional notion of poverty, in general. But when asking the question in relation to post 2015, I am less convinced. Perhaps it would be important to consider most of the proposed post 2015 goals as part and partial of poverty, and then have one specific on income or consumption. I do think that income cannot be left out, but it certainly is not the only dimension of poverty. So probably the question is not the right one to ask.

    The point on inequality is also a bit confusing to me. There is much talk about the fact that a goal on inequality is missing. I do agree there is not much explicit on equality, except for the fact that the HLP call for each goal to be only achieved when everybody has achieved it. But that is not inequality, that is focussing on the bottom group to also make it. I would be in favour of having inequality measures as part of each of the goals, as a way of looking at the extent to which the goal is really being achieved. And then inequality of course can be measured in many different ways, and definitely not only for income.

  7. nadine malapira

    Definitely there would be post development programmes to be launched in order to address some goals that were not fully realized over the past years. Hopefully, more feasible projects would be funded and supported by the government, NGOs and international aid agencies; hence the collaborative efforts, willingness and initiative of the people to be assisted.
    There must be a comprehensive analysis of poverty measurement standards and levels across regions as developing and poor countries do not experience the same things and scenarios all the time. A slum dweller in Africa lives totally different from a similar slum dweller let’s say in South Asia or in Latin America. With the given specifications, particular projects modeled and patterned under the needs of different locus must address the problems more effectively.

  8. Andrew Dorward

    Another approach might be to look at cost of minimum needs as a % of income for the lowest income decile. Minimum needs could be cost per person of 80% of basic calorie needs from the cheapest staple, housing, energy ….
    There are of course challenges and shortcomings with such an approach (as with all approaches, as in previous comments) but if prices and needs (eg for energy) were locally determined this could provide an important focus on minimum needs and effects of price shocks (which are important for food and may become more important for other expenditure categories). The measure could be built up over time, incorporating more expenditure categories over time (eg food is probably the most important and could be addressed initially with a bit of effort).

  9. Emma Samman

    Very thought provoking comments, and poll results. The MPI lead suggests support for a focus on overlapping deprivations, though votes in favour of an income poverty measure are split across several alternatives. Surprising that the proposals to include $1.25 measures received relatively little support – it may be useful to think more concretely about how Klasen’s proposal to set poverty lines consistently across countries would be advanced. Interesting point on disability – and the more general issue that food baskets are not tailored toward any specific needs. One route may be to take factors like disability into account when constructing national poverty profiles – much as they are often adjusted for household size and age composition – though better data on disability are needed. And important comment on data availability – recent advances but great scope for improvement. The final contribution to our blog series, by Emmanuel Letouzé, touches on this issue, focusing on the potential for Big Data to enrich poverty measurement (see

  10. For development purposes, I find poverty not really the best term to use.

    I would like to promote a more vertical and multidimensional approach to development. On income poverty, equity and an absolute bottom line are important. On education, six years of primary education, and access to secondary education should be the minimum. Access to basic health services and human rights another one.

    It is exactly putting all the separate aspects in a blender and working from one number that makes the whole debate fuzzy and without any practical use.

    We should talk about income poverty to use an income poverty indicator. Otherwise we speak about education poverty and use education indicators. We should definitely never use aggregated indicators: we end up with too many degrees of freedom inour equation to have any meaningful policy guidance.

Leave a Reply

Your email address will not be published. Required fields are marked *