Syndicate content

A debate on multidimensional poverty indices

Gabriel Demombynes's picture

At Duncan Green’s blog, there is a fascinating back-and-forth on the UN’s new Multidimensional Poverty Index (MPI) between its co-creator, Sabina Alkire, and the World Bank’s Martin Ravallion. This is very much a live debate in development circles. The MPI is a descendant of the earlier Human Development Index and is similar to the various Unsatisfied Basic Needs indices long used in many countries.

I agree wholeheartedly with Martin’s critique, but Sabina does offer a spirited (and highly hyperlinked) defense. Martin’s emphasizes two points: 1) what’s the point of aggregating a bunch of indicators into a single index? and 2) the choice of weights for such an index is inherently problematic.

No development economist would disagree that poverty is multidimensional. Our notion of poverty includes measurable consumption along with things like health, education, and access to infrastructure, as well as less tangible aspects like rights and opportunities. But the question is, should we try to squeeze all those measures together into one super sausage of an indicator, or should we just consider them separately?

 

Sabina’s response illustrates the problem with the all-in-one measure:  to explain the contrasting values of the MPI for Kenyan Somalis and Kenyan Masai, she immediately goes to discussing the differences in the index’s components: child mortality, school attendance, etc.  But if we want to compare the welfare of these two groups, why not start out by looking directly at the components, rather than mashing them all together and then pulling them apart again?

Sabina argues that the added value of the MPI is that it captures the overlap between its various components. But if a quarter of the children in a country are malnourished and a quarter lack access to clean water, to what extent are we talking about the same children? In practice, the correlation between such measures is likely to be high, and the best way to examine the overlap would be to consider it directly, e.g. tabulating child malnourishment vs. access to clean water.

The more powerful critique of the MPI is that there is no solid basis for the choice of index weights. When I teach poverty measurement, I give my students as a homework assignment the task of constructing their own country-level welfare indices, using whatever formula and data they like. They end up with formulas like: the square root of the child mortality rate, minus half the murder rate, plus one third the number of kilos of ice cream consumption per capita. Who am I to say that this is an adequate welfare measure? (What, you don’t like ice cream?) Unsurprisingly, country rankings vary greatly across the different measures my students concoct.
 
Sabina’s response to the problem that the weights are arbitrary is that she’s not claiming the final word: “we strongly encourage countries to develop national measures having richer dimensions,” and “critical scrutiny” of the weights is welcomed. The point, however, is not that some other set of weights would be better. There is simply no “right” way to come up with such an index, so the index will ultimately reflect the preferences of its designers.

I do have a bit of sympathy for one argument for the MPI: it could encourage public discussion on the fact that poverty is multidimensional. The general effect of launching a composite measure, such as the Corruption Perception Index or the Commitment to Development Index, is to focus attention on the subject matter. We’ll probably hear more people talking about multidimensional poverty in the next few months as a consequence of the MPI.

Still, my very practical worry is that the new push for multidimensional poverty indices will soak up much of the oxygen around poverty work. 

Here in Kenya, the Kenya National Bureau of Statistics has given us a wealth of data to work with to measure and think about poverty. We have a number of Demographic and Health Surveys, the 2009 Population and Housing Census which will be published in the next few months, the 2005-06 Kenya Integrated Household Budget Survey (KIHBS), and a new round of the KIHBS scheduled for next year. Rather than using all this data to calculate the MPI and variations, I’d prefer to focus on, say, understanding how Kenya has achieved a stunning drop in child mortality since 2003 (more on this in a future post.)
 

Comments

Submitted by Gonzalo Hernández on
There will always be a debate on the way we measure things, especially poverty. I agree that the choice of weights, cutoffs, dimensions and the aggregation issue will always be on debate. In the case of Mexico, the "solution" of the problem (weights, cutoffs..) was simplified for the following reasons: The need to construct a Multidimensional Poverty Measure was a mandate by Congress (the Law of Social Development was launched in 2004). The same Law indicated CONEVAL (an institution created especifically to measure poverty in Mexico) which dimension should be included. The approach CONEVAL undertook for the methodology was based on social rights (following again what the Law and the Constitution mandate). Constructing a poverty measure based on the Mexican regulations, especially the social rights approach, made us take the following decisions: Since all social rights are important, then all social dimensio ns (education, housing, social protection, health, food) must have the same weight. There are various Mexican laws (education, health) which indicate the basic access to all Mexicans, therefore the cutoffs where those indicated in the Mexican regulations. For example, the Educational Law says that the basic and compulsory educational level es Secondary; thus we took that cutoff. In summary: we constructed a technically solid methodology (we hope), but since this was not only an academic exercise but a measure which should be used by the Mexican State, many decisions (dimensions, weights, cutoffs, the general approach) were embedded in the Mexican regulations, which werer decided by Congress (elected by the Mexican population). You can check the CONEVAL's new (2009) multidimensional poverty measurement in www.coneval.gob.mx

Submitted by Gabriel on
Gonzalo, Thanks for your comment and the link to CONEVAL's site. I am still left wondering what an index like Mexico's is for. For any attempt to understand why one area is higher or lower in the index, you'll have to look at the components. But then why not just skip the index and look at the components to begin with?

Dear all Millennium Development Goals (MDGs) are the prime agenda for the international community to contribute towards development primarily towards poverty reduction. However, there are mismatch in the understanding and in investing as most of the factors are interdependent and crosscutting. Intellectual elites get biased towards their discipline and provocate as supreme over others. The rest do the same. In fact the poors are in the confusing spiral to decide where to go and what to do. If we all are sincere and committed let us work for a common agenda and direction. PAUDYAL, Dhruba P.

Submitted by Paul Anand on
As Director of the Capabilities Measurement Project, which draws on contributions from the Open University, LSE, Oxford and elsewhere I have to say that this is an interesting and exciting debate. Our focus has been on seeing what might happen when the capabilities approach is applied to welfare analysis in high income countries and the development of data that really does measurement freedoms in a way that is consistent with the methodological conventions that apply to sample surveys. The survey instruments we have developed over the five years or more are highly multi-dimensional and we have used them tolook at deprivation using latent class and regression techniques and find that the multi-dimensional is often an interesting and valuable complement to money based measured. For one thing, we are beginning to develop a sense of what opportunities and welfare outcomes are particularly comprised by low income, and what types of people are most affected. In a lower income country context, it would of be wrong to let policy of the income hook and our implementation of Sen's ideas do not overly stress indexation for the reason that Martin outlines. Given that we have invested in developing data that is highly multi-dimensional, we want to apply analysis tools that make use of this dimensional information.

Submitted by Ugo Gentilini on
Dear Shanta, I really enjoyed the debate – as a policy adviser, I tend to concur that composite indicators are often less informative than individual ones. The arbitrary selection of indicators (and their weights), for example, is challenging to defend given the ample availability of relevant measures. And the overall costs (especially from missing information) that stem from merging multiple indicators often largely outweigh the benefits. Yet, they remain highly attractive to the broad public and rapidly capture media visibility. I have to confess, however, that I’m not entirely anti-composite indexes. For instance, Prof. Patrick Webb and I published a paper a couple of years ago (Food Policy 33(6): 521-532) on tracking progress on MDG-1 on poverty and hunger. We did so by presenting an indicator we called Poverty and Hunger Index (PHI) and analyzed it over time and against its single components. The PHI illustrated country performance in attaining MDG-1 in a single number, while at the same time showing that progress in one dimension (e.g. income poverty) didn’t automatically translate into improvements in others (e.g. children underweight), and vice versa. Differently from other composite indicators, however, we didn’t really engage in a process of index selection – we simply picked the five poverty and hunger measures agreed by the UN members in 2000. (Yet, one could question the reasons for which the UN selected those five indexes…). The PHI, of course, maintains the pros and cons of all multidimensional indices, but since the MDG summit in September is approaching thought it was useful to share. Thanks again for the great debate! Ugo

I appreciate all the thought that is being brought to bear on the methodology that Sabina Alkire and I have proposed, and which has recently been implemented for cross country comparisons in the MPI by Sabina and my student Maria Emma Santos. The new technology builds upon the FGT or P-alpha indices that Erik Thorbecke, Joel Greer and I introduced more than 25 years ago, and that are now quite standard for measuring poverty in a single variable (such as income, consumption, or even calories as applied in our original paper using the 1975 Integrated Survey in Kenya). Moreover, the multidimensional methodology reduces to the traditional FGT measures when zero weight is placed on non-income variables. We think that our more general approach might usefully complement the traditional income or consumption based approach, and there are many who share this view. I thought I would make a few points of clarification. The Alkire-Foster poverty measure is not so closely related to the UNDP’'s Human Development Index (HDI). The HDI aggregates country-wide average achievements into an overall indicator of human development, and is analogous to an aggregate welfare measure for a country. It requires many assumptions on the underlying variables to construct a meaningful measure (in the sense of Roberts, Measurement Theory, 1979). It satisfies a range of axioms appropriate to its aims. The Alkire-Foster methodology yields a poverty measure in the sense of Sen (Econometrica, 1976). It first identifies who is poor and then aggregates to obtain an overall measure of poverty. The measure takes into account several dimensions and their dimensional cutoffs in identifying who is multiply deprived and hence who is poor. It likewise uses information on these deprivations to evaluate the extent of poverty for a person, a group of person, or for a country. It satisfies a different collection of axioms appropriate to its aims. The MPI uses the simplest of our measures, the adjusted headcount ratio, to aggregate across deprivations (and not achievements). It can thus be applied to qualitative or categorical variables so long as one can ascertain what is meant by being deprived. Few assumptions must be made on the underlying variables to ensure that the measure is meaningful. In cases where individual deprivations are not equally important, weights are used to reflect the importance of one deprivation relative to another. Dr. Gonzalo Hernandez has indicated above how his team at CONEVAL in Mexico was able to decide upon weights and cutoffs for the Mexican methodology, which is related in structure to the MPI methodology. I have some responses to the two questions revisited in the above commentary. 1. Why a single measure of poverty? Because that is the goal of Sen'’s (1976) defining contribution, which all of our work in poverty measurement builds upon. If there is a superior conceptual framework for poverty measurement, I would be interested in seeing it. 2. Are weights inherently problematic? Perhaps. But once one adopts the position that poverty is multidimensional, this comes with the territory. The selection of weights may differ depending on the context. For a cross-country exercise, where there are fewer normative guideposts, the selection may naturally tend to the focal point of equal weights for distinct dimensions, with equal weights for variables within dimensions. For within-country comparisons, the decision becomes explicitly normative and can be informed by a range of studies. The underlying choice is not dissimilar to the setting of a poverty line in income poverty. One response to this challenge might be to place full weight on a single dimension and zero weight on all the rest - which as I mention above is the solution implicit in the traditional income or consumption poverty measure. But this choice of weights is increasingly coming under scrutiny and is hard to reconcile with the belief that poverty is fundamentally multidimensional. A point that has been missed in the discussion, but is quite apparent to experts, is that each methodology employed to measure poverty has its own list of arbitrary choices. A discussion of weights in the MPI should be contrasted with, say, the analogous choices involved in cross-country comparisons of consumption poverty. How does economic theory tell us to pick the particular bundle of goods for determining a poverty line or for setting appropriate PPP values? Does it recommend a uniquely best equivalence scale? Isn’t the “market price” of a particular category of goods an aggregate itself, and not the marginal price that a decision maker faces in theory? How exactly should market prices be altered where there are market imperfections or missing markets? And what theorem in economics allows us to use consumption to make interpersonal welfare comparisons? These are familiar challenges. Most of us are willing to suspend disbelief in order to move forward. We think a similar willingness is justified in order to augment the informational basis of poverty measurement. James E. Foster The George Washington University and OPHI

James Foster says that once one adopts the position that poverty is multidimensional, adding up the multiple dimensions “comes with the territory.” Is that right? We all agree that poverty is not just about lack of command over commodities. The issue is NOT whether poverty is multidimensional but whether one should add up the multiple dimensions using weights that seem to drop from the sky, with barely a hint of justification, and weights that may well be deemed inappropriate in every specific country for which this MPI is measured. For example, it is incredibly difficult to say (as the MPI does implicitly) that a child’s life is worth so much in terms of material goods. I would want to be very careful about making such a judgment and building it into global poverty comparisons such as done by the MPI. Thankfully, for many policy purposes we do not need to do so. But when we do we should not take on the task so lightly. James and other commentators have pointed to the fact that consumption-poverty measurement also requires assumptions. That is not in dispute. However, I can’t imagine that anyone seriously thinks that adding up expenditures across commodities (or incomes by their sources) to measure economic welfare is as problematic as adding up child deaths and consumption deprivations. We know from economic theory under exactly what conditions the market price-weighted aggregate of consumption, appropriately normalized, measures economic welfare. We have much less idea about adding up child deaths and consumption-poverty. Adding up the multiple dimensions using weights with no obvious foundation does not “come with the territory.” We can be multidimensional about poverty without creating a one-dimensional index. In fact international poverty measurement has long been multidimensional, as has policy making for fighting poverty at country level. We can do much better for sure, particularly on the measurement of each dimension, but I remain unconvinced that creating this “multidimensional poverty index” has got us any further.

Submitted by Anonymous on
This is a great conversation. If I can make a request, I think it would help clarify the issues if some commenter would expand on why the more nuanced approach to poverty measurement that Martin Ravallion describes is necessarily better than targeting a very imperfect weighted composite. What I mean is that of course each of these approaches will be prone to error, and I'm not sure about the context of the decisions. Poverty policies are very politicized - which of these approaches is more immune to political pressure (and is that a good thing)? Who is making these carefully calculated, multidimensional poverty assessments? Does it depend on the context which of these approaches we prefer?

Submitted by Anonymous on
I'd like to make two comments. First, there is a large, highly advanced, but almost entirely overlooked literature on multidimensional poverty measurement. I encourage those who are interested to look at the work of Maasoumi, Silber, Kakwani and many others. Secondly, the information theoretic approaches (by Maasoumi et al) provide a unique empirical criteria to compute the otherwise "arbitrary" weights. Unfortunately, the Bank has so far focused on the mainstream literature.

Thank you everybody for your comments. I have done a roundup of the debate in a new post. Let's continue the debate in the link below: http://blogs.worldbank.org/africacan/the-multidimensional-poverty-index-debate-rounds-2-3-4

Submitted by quotation template on
Lack of basic human needs is called poverty. It is multidimensional term and includes lack of clean water, lack of health facilities and clothing/shelter etc. In actual it is not lack of these things, it is inability to buy these things due to non availablity of funds. It is a great effort indeed, to make people understand "poverty".

Poverty is not at all any curse. It is totally the lack of one's awareness. Plenty is everybody's birth right. Any sort of help from any corner, never makes the poor plenty. Mere ever long dreaming for better future will never become a reality. It is only the right Knowledge of awareness along with total determination, to overcome one's poverty, will only make the poor, plenty. Here is an assured hope for the poor to make their lifelong dream, easily, early, a reality. Now anybody can dare to challenge the poverty.

On 11 Th July every year world’s celebrates World population day. On the world’s population day, here are amazing free and effortless ways to build your heaven with single child family’s 70 benefits, as well as reduce the burden of world’s population. Hunger, scarcity of water, unsolved illiteracy, ever increasing violence, regional disputes, unhappiness, competition, pollution ….. for these unlimited problems the route cause is ever-increasing world population. Controlling population is not only the responsibility of the concerned individual governments, but also the prime strength rather than member’s strength. Here are amazing, unlimited awareness tips for the newlyweds to build their undreamt heaven and absolutely free on earth with smallest family i.e. single child family adoption. BENEFIT’S TO THE CHILD: You can bring up the single child in the best way as per your dreams. It will prevent the child going on in wrong ways, because of families’ total concentration. If something happens to one of the young parents, the surviving parent can bring up the child with least burden. In the eventuality of the death of both the parents any relative can take care of the lonely child. There will be no necessity of seats reservation in colleges, jobs reservation for various categories as everybody gets enough opportunities due to less population. No unemployment problem. If the single child cannot get a good job, then he can pursue his parents’/in-laws’ profession. The eligible single child will be of high demand for marriage alliances with choicest offers. Child mortality rate drastically comes down, because of optimum care & caution. Parents neglecting non-performing child, will not arise. BENEFITS TO THE FAMILY: Your family will never face hunger. Your generations will not suffer illiteracy. In the alarming scarcity of water situation you will have the least water problem. Family can enjoy maximum privacy, silence as well as maximum leisure. The entire family members enjoy smooth and happy life the satisfaction of their dreams. Parents can become free early from their family responsibilities. Tension will be absolutely nil in the family, therefore the family members will have peaceful and long life. Single child family is the sweetest family, as there will be less family disputes. It will reduce stress, strain, worries, and botherations in many ways. Single child family enjoys special status in the society. In case of any eventuality, remarriage of the surviving parent is easily possible. Even if, something happens to both the parent, surviving single child responsibility can be taken by any relative. Dowry related problems like disputes, harassment, deaths; couple separation will be enormously reduced. BENEFITS TO THE SOCIETY: Because of the population control, pollution reduces and you can breathe fresh air and live long. Drastically reduces the birth of disabled children. It will result in physically, emotionally and mentally stronger society, as the single child concept will encourage breast –feeding of babies. As the human values grow up the crime rate comes down. Alarmingly lowering of female birth ratio will come back to normal. Because of literacy, intellectual, creative, scientific society takes shape. Philanthropic activities will increase. Many social problems like child labour, child marriage, begging etc will be eliminated. It will route out hunger/poverty/humiliation/stigma related to mass family suicides. Poverty/hunger related forced prostitution/sex trade will reduce. CAUTION: SINGLE CHILD FAMILY BUILDS ALL-ROUND PROSPEROUS HEAVEN FREE. BOON TO ALL, PARTICULARLY FOR THE POOR. TWO CHILDREN FAMILY JUST MAINTAINS STATUSCO.MORE THAN TWO CHILDREN FAMILY VOLUNTARILY TROUBLES INVITED FAMILY. SEE THAT UNWANTED CHILD FIRST OF ALL NOT CONCEIVED INSTEAD OF GOING FOR ABORTION LATER. For full details of 70 benefits see the Awareness Mission website: www.srimission.org

Submitted by Amin Beg on
At at lower spatial level, we have been surveying socio-economic trends and income poverty measures for over 18 years at sub-national levels. While we are agree poverty is multidimensional, but the debate continues whether poverty is spatial or territorial or non-territorial. Second challenge is climate change and incidences and intensity of natural and man-made disaster events have created more vulnerabilities, and is exposing traditionally rich and poor, socially excluded and active families and communities to same kind of vulnerabilities in a single territory. How to measure such vulnerabilities, where the 'have-nots' with no assets, income base, and education and health services, the 'rich' who lost everything to a disaster, are now starting a fresh. New groups who are most vulnerable and poor are those women, children, unemployed, unskilled youth, and communities with no political voice and representation, and areas that are trapped in ongoing crisis situation.

Add new comment