Syndicate content

Should inequality be reflected in the new international development goals?

Adam Wagstaff's picture

The last few months have been a busy time for inequality. And over the last few days the poor thing got busier still. Inequality is now dancing on two stages. It must be really quite dizzy.

We need an inequality goal. No we don’t. Yes we do

One of the two stages is the post-2015 development goals. At some point, someone seems to have decided that reducing inequality needs to be an explicit commitment in the post-2105 goals. The UN System Task Team on the Post-2015 UN Development Agenda wrote a report on inequality and argued that “addressing inequalities is in everyone’s best interest.” Another report by Claire Melamed of Britain’s Overseas Development Institute argued that “equity, or inequality, needs to be somehow integrated into any new framework.” Last week a group of 90 academics wrote an open letter to the High Level Panel on the Post 2015 Development Agenda demanding that inequality be put at the heart of any new framework.

But not everyone concurs. Last December Stephan Klasen, Professor of Development Economics at the University of Göttingen, wrote a blog post on The Broker entitled “No, we don't need an MDG for inequality.” And a couple of weeks ago Martin Ravallion, former World Bank Research Director and now Professor of Economics at Georgetown University, wrote a post on the same site entitled “Let’s avoid creating a dog’s breakfast of MDGs” in which he too argued that a goal for inequality isn’t needed.

Inequality has been dancing recently on another stage too: the World Bank. At his first Annual Meetings in Tokyo last year the Bank’s new president, Jim Yong Kim, said: “The World Bank Group’s mission is to end poverty and build shared prosperity. That’s why I’ve asked the institution to come up with a bottom line in the form of ambitious targets for these two goals.” The speech signaled two changes for the Bank – a bolder poverty goal involving the ending of poverty, not just its reduction; and a shift away from a pure poverty focus toward a broader mission that embraces inequality.

Here too there has been debate. Last week The Guardian carried a piece claiming slippage by the Bank on the shared prosperity part of its new mission on the grounds it looks set to track "shared prosperity" by looking at the income growth only of the bottom 40%. “Bank's promotion of 'shared prosperity' fails to tackle inequalities and growing gaps between rich and poor, critics warn” ran the article’s headline. (For the record, the leaked documents The Guardian saw are merely drafts – the monitoring framework isn’t actually set in stone.)

Unpacking the debate

One thing to clarify upfront: neither Prof Klasen nor Prof Ravallion is an anti-egalitarian; far from it. Their point is that given the commitment to reducing poverty, an inequality goal is superfluous. Economic growth that leaves inequality unaffected reduces poverty by less than growth combined with reduced inequality. So if the poverty goal is ambitious enough, reaching it will necessarily require that countries reduce inequality. In fact, it’s possible to write down a mathematical relationship between changes in poverty on the one hand and economic growth and changes in inequality on the other. For a given set of economic growth forecasts, a particular poverty target tells you how inequality has to change to reach it. You could easily inadvertently set an inequality target that isn’t consistent with your poverty target. In short, you could quite easily make for yourself a nice dog’s breakfast. Which, unless you have a hungry dog, isn’t what you want to do.

This argument sheds some light on the “shared prosperity” issue too. The goal of “ending poverty” is a very ambitious one, and its achievement will likely require that countries reduce income inequality. It’s possible in other words that ending poverty will necessarily require a greater “sharing of prosperity” than there is right now. I guess the question then becomes: How far will inequality need to fall, given current growth forecasts, for poverty to be “ended”? With very strong growth, the reduction in inequality might not be “enough” for an egalitarian.

Perhaps what would be useful for both agendas are some numbers showing how far inequality would likely have to fall if poverty is to be ended under different economic growth forecasts. That way we could see whether a separate inequality goal is needed.

What does all this mean for human development goals?

This discussion raises the question of whether inequality targets are needed for non-income goals such as education and health. Klasen argues they’re not needed. In fact he says the case for not having an inequality goal is “even clearer” in the case of the non-income goals.

“Universal primary education or massive reductions in mortality cannot be achieved without reducing inequality in health and education. The rich, healthy, and educated are already doing so well on mortality and education indicators that further improvements for them will simply not even come close to achieving the large reductions called for by the MDGs.  Declining income, education and health inequality is thus a critical means to achieving the MDGs… a separate inequality goal seems redundant.”

Here it’s worth being clear what we mean by inequality. Do we mean “pure” inequality – the differences across people in education and health outcomes? Or do we mean inequalities in education and health between the poor and the rest of the population?

If we mean the former, for many indicators the population average and inequality move hand-in-hand. For a binary indicator, the Gini coefficient equals one minus the mean. So as the mean rises towards one (e.g. every child is immunized), inequality falls towards 0. But as the mean falls towards zero (e.g. no child dies before their first birthday), the Gini coefficient rises toward 1.

I’ve blogged elsewhere about inequalities in education and health outcomes in the context of the Human Opportunities Index. I argued that the HOI doesn’t in practice capture all the “morally relevant” inequality in their data, since it classifies as fair too much of the inequality in education and health outcomes. But I also agree that looking at “pure” or “total” inequality isn’t helpful. There will always be some inequality caused by random events that a government can do little if anything to eliminate. So the Gini probably isn’t the right thing to look at.

One part of inequality we can all agree matters is the part associated with family circumstances. After all, most – if not all – of us would agree with the statement that inequalities in education and health that are systematically associated with a child’s family circumstances are unfair. So instead of measuring total inequality, we would measure differences between in education and health outcomes between, say, the poorest 40% and the rest of the population.

Do we need a separate target for this type of inequality? Klasen says No because progress over time will necessarily involve the poor catching up.

We can test Klasen’s hypothesis using a new dataset from Caryn Bredenkamp and her team at the World Bank (full disclosure: I’m a member of the team). The maps below give a sneak preview of the findings of the team’s ongoing analytic work on the MDGs. In the maps I’ve compared – for each indicator – the annual growth among the poorest 40% of the population with the annual growth among the richest 60% of the population. (I calculated the growth rates using only the first and last available surveys.) Green means the outcome is improving faster for the poorest 40% than for the richest 60%; red means the opposite. The deeper the green, the faster the poor are catching up; the deeper the red, the faster the rich are pulling even further ahead. (By the way just in case you were wondering, green on the mortality and underweight maps means mortality and underweight are falling faster among the poorest 40%.)

What’s striking is that in each map there are several red countries – and not just in Africa. In just over half of the countries, the poorest 40% have actually been making slower progress than the richest 60% on under-five mortality and child malnutrition. This is quite shocking. It’s perhaps even more shocking that in around one quarter of countries the poor are making slower progress on indicators like antenatal care, safe deliveries, and immunization, all of which can be influenced fairly directly by governments, NGOs, and donors.
For health, at least, an inequality goal is needed

So from these maps it seems that it’s not actually true that progress at the population level will automatically entail faster progress among the poor. If inequalities in education and health outcomes across the income distribution matter, and if we want to see “prosperity” in its broadest sense shared, it looks like we really do need an explicit goal that captures inequality. It seems inequality will be busy for a while yet.


Submitted by Pauline Rose on
Thank you for this very informative blog. The same logic holds for education as you have presented for health. Although in principle we have had a goal on education for ALL for the past 2 decades and progress has been made, attention has inevitably focused on the low-hanging fruit - with a risk that those who have been left behind will continue to be forgotten if the next set of goals moves to higher ambitions on learning and secondary education. Unless we have a goal that tracks progress for the poorest and richest (and other indicators of disadvantage) on education access and learning, gaps are likely to remain when we reach the next deadline for goals.

Thanks, Pauline. It would indeed be interesting to see these maps for education. My guess is we'd see a somewhat similar story. And yes I think the logic for an equality angle to a development goal applies equally to education. AW.

Submitted by Toomas Palu on
Adam - thanks for continuing to push the debate To me the current aggregate MDG targets completely missed the distributional aspects, and this needs to be corrected in the post-MDG era. But not as a separate target but embedded into measurement whatever the new targets end up being. If it will be UHC in health, then let's measure its impact on distribution of UHC outcomes. Best regards

Thanks, Toomas, for continuing to engage and encourage! I agree wholeheartedly with your logic. One could imagine a target for the population, which could be a level or rate of change, and a stipulation that the poorest 40% achieve say 80% of the population target or improve at say twice the speed of the richest 60%. AW.

Submitted by Helen Abadzi on
The inequality starts on day 1 of grade 1. The better off go on to learn reading and the rest don't. Many drop out illiterate in grades 1-3. And illiteracy does contribute to early dropout, according to various studies. Reading has been taught in low-income countries through complicated ways tailored to the English spelling and the high-resource schools of the "west". In local languages and by using the many relevant findings of cognitive neuroscience, nearly all can learn basic reading in the first semester of grade 1 (about 100 days, in all but about 7 languages). Decoding is only a start, but students thus will not be reading 0 words per minute. A concrete and measurable follow-up goal would be the attainment of automaticity at most by the end of grade 2 (45-60 wpm across languages and scripts). The big problem with this goal is "face value appeal": it's considered trivial by those who do not understand how automaticity develops. So while many donors talk about the profundities of reading to learn, they have little interest in the road that leads most efficiently to that state.

Great overview piece. My feeling is that UN consultation on post-2015 inequalities has already moved us beyond the 'absolute ambition sufficiency' view (that is, the view that meeting a sufficiently ambitious absolute target will by definition address inequality). Aside from additional and valuable evidence of the type you present here (looking forward to seeing the final paper!), there is a more fundamental issue. The absolute ambition sufficiency view treats inequality, as you suggest, as simply a part of a mathematical identity: so that addressing inequality is necessary, or otherwise, insofar only as it has an impact on policymakers’ ability to improve the absolute outcomes at the bottom of the distribution. The synthesis report of the inequality consultation, in contrast, is explicit in stating that inequality should be addressed in response to the damage suffered by society, and all within in, so that a targeting response, or indeed one only concerned with absolute outcomes, is by definition insufficient. The crux of this is of course whether one considers inequality to be intrinsically damaging, or only instrumentally so. The evidence across a range of development outcomes points strongly to the former, and hence the absolute ambition sufficiency view seems ill-suited. In terms of measurement, I’d agree with the view that the Gini coefficient is unhelpful (it hides more than it reveals in too many cases, by its oversensitivity to the middle of the distribution) and that we should be concerned with the ratio of outcomes between those at the bottom of the distribution and others. As Andy Sumner and I have written, however, we would propose the Palma: the ratio of outcomes of the top 10% to the bottom 40%: "For the record, the leaked documents The Guardian saw are merely drafts – the monitoring framework isn’t actually set in stone." I'm hoping that we can read your aside to mean that there is space yet for the Palma in the Bank's framework. And per the inequality consultation, we should hope to see in the post-2015 framework not only a goal on economic and gender inequality but also a set of inequality targets and indicators throughout all the goals (reflecting inequality faced by people living with disability, urban-rural and regional inequalities, ethnolinguistic inequalities, age-based inequalities and gender, as well as income/wealth).

Thanks, Alex. As I say, with respect to the poverty, inequality and growth issue, it would be interesting to see how far inequality would need to fall to reach the poverty goal for different economic growth forecasts. Then stakeholders would be in a position to have a debate about how much extra inequality-reduction they'd like to see beyond what's necessary to see poverty "ended". I see the issue of which measure we use as a second-order one. As for education and health (and for that matter other non-income goals too), my personal view is that targets should reflect not just how the population as a whole fares but also how different groups (especially the poor) fare compared to the rest of the population. The need for that is pretty clear from my maps. AW.

Adam, You have made a case for modifying the old aggregate MDGs, to assure that they properly reflect inequalities in the relevant dimensions. Doing this in a sensible way would not create a "dog's breakfast" of conceptually muddled indicators--quite the opposite. Essentially one would be adding indicators for the health and education of poor people. I would go further and propose that we replace the present aggregate indicators by those for the poor. And we have the data needed. But don't you agree that simply adding one or more ad hoc aggregate "inequality" indices (as is being proposed by some observers)does risk creating such a muddle? Martin Ravallion

Thanks, Martin. My sense is that people are concerned about how the poor progress both in absolute terms and relative to the rest of the population. Suppose child mortality is falling at 5% per year among the poor. I think people feel differently depending on whether mortality is falling at 5% per year among the rest of the population or at 10% per year. If they do, we could set a target for the population and another target for the poorest (say) 40%. These could be in terms of levels, or rates of growth. In both cases, they need to be consistent with one another and with the starting values. Or we could set a target for the mean and the concentration index, again making sure the targets are consistent with one another and with the initial values. In the income case, it seems to me that an egalitarian could reasonably argue they’d like to see (within-country) inequality fall by more than the amount required to end poverty conditional on one’s growth assumption. And I suppose one could find this minimum inequality reduction for a variety of different inequality measures. My sense is your concern is partly about picking an inequality target from the air (it needs to be consistent with the poverty target, income growth assumptions, and initial values) and partly about being careful to choose a sensible inequality measure (hence your quotation marks). Adam

Submitted by Simona K. on
You say that we are going to measure health and education and income inequality within countries and recommend for having UN targets for reduction of inequalities WITHIN countries. In other words, we will study and track for example farmers in Niger and whether the farmer with one hoe is growing less poor than the farmer with a hoe and a wheelbarrow, or whether the girl with no shoes is finishing 3 years of schooling , up from 2 (a 50% increase!) while her compatriot with shoes and a bicycle only goes from finishing 7 to 8 years? What about inequalities regardless of nationality? There is much more room for reducing inequality between the poor and middle class in poor countries and those in high-income countries. The one world and relevant reference group does not stop at country borders. instead of handwinging over inequality within countries, would it not be better to have Free migration target? This is fastest way to reduce inequality we concerned about. Or is it just pretending concern? United Nations really united?

Submitted by Claire H Melamed on
Adam, interesting blog - two comments. 1. For the record, while I think that inequality/equity (I sometimes think the difference between these two ideas is overstated, but that's another issue) is important for post-2015, that doesn't mean I think an inequality goal is the answer. I'm pretty agnostic on that point. I think a goal that's well specified - e.g. on gender equality, where the dimension of inequality that is relevant is quite clear, and where targets could be set in specific areas like violence, access to services, income etc, might be helpful. I'm less sure about a general inequality goal, as I think there's a risk of it being quite non-specific and therefore unhelpful, though I can also see the political point of having something there to focus minds and attention on disparities. So on that issue I am genuinely uncertain. 2. But I do think that addressing inequality in some form is essential for any level of aspiration for post-2015. Improving health and education outcomes, eradicating extreme income poverty, tackling crime and violence (which comes out as a high priority for poor people in the MY World survey), would all be much easier if a new agreement contained some incentives (in the form of a goal, targets or indicators, or, most likely, some combination of these) to tackle horizontal inequalities. DHS data suggest that around two-thirds of those still living in extreme poverty are from an ethnic minority in their own country. It is likely that given poor data this is just the tip of the horizontal inequalities iceberg. If the aspiration is to end extreme poverty, I would very much doubt if that can be done without attention to the different dimensions of inequality that form a barrier to progress for some groups. An agreement that had incentives to uncover, tackle and measure progress on these barriers would be a more effective agreement.

Submitted by kevin watkins on
Thanks Adam for initiating a fascinating debate. It strikes me that there are several sub-texts running through the discussion. Are we concerned about inequality as a barrier, or at least a brake, holding back progress towards absolute goals? Or do we see some inequalities beyond a certain threshold as a problem in and of themselves? And whatever position we take on these two questions, what's the point of bringing the equity onto the already overcowded MDG agenda. I suspect all of us view equity as a development issue (your blog has a selection bias!). To me the big question - and this is the point that Clare raises - is whether or not it makes sense to bring equity onto the MDG agenda. My view is that it does, for two reasons. First, inequalities are clearly holding back progress towards the current MDG goals. You make the point nicely on health. As Pauline Rose comments, the situation in education is similar though the binding point is lower (universal primary education). The stalled progress towards universal enrolment has alot to do with failure to reach the marginalised. Raising the ceiling to encompass secondary schooling and quality would magnify the effect. Stefan Klasen is right, of course, that progress towards an absolute goals narrows inequality in relation to that goal. But this is a minimalist equity requirement. Including equity goals as interime targets for reaching the absolute - stepping-stone targets if you like - would have a number of benefits. It would require governments to report on what is happening to inequalities. And it would turn the policy spotlight on how best to reach, say, the last 20/30 per cent, depending on the country. I very much support Clare's point on avoiding general goals. Government's love them because there's mnothing quite like signing up for bold principles with no tangible commitments - witness the fate of MDG 8. But I think in an number of areas - health, education, water and sanitation - we do can articulate simple and practical goals. In fact, Bangladesh already has equity goals built into the national education strategy (wealth gaps in attendance and learning differentials across districts). The tricky one is income. As Martin and others have documented, enhanced equity is good for poverty reduction at any given rate of growth. But I think we need more deabte on which wealth ratios to adopt. Maybe Alex is right on 'the Palma' - but I'm not convinced. Last but not least, I'm glad that inequality is having an energetic moment. You say she is dancing on several tables - and that's a good thing. There are are a lot of people out there that see inequality as the cause of our time. Social movements have combating inequality have brough a new dynamism and energy to political mobilization. And, boy, do we need a dose of energy in the MDG debate. Leaving aside the technicalities, post-2015 targets that don't include equity are likely to look anachronistic; and they'll hold out little appeal for constituencies that could revive international development campaigning. So let's keep Inequalitina dancing a while longer!

Submitted by Mickael Hoelman on
Adam ! Yes, you have really made a case. Inequality not just income, but also gender and opportunity ! Either our current models of growth has successful enough with MDGs? Recent Ban Ki-Moon Op-ed did mentioned inequalities are growing. He said diplomatically that "to many are still being left behind". In most developing countries like Indonesia we then face what i call as Groverty; a sophisticated and "promising" growth with also an increasing poverty (or thin decline). Unfortunately the last round HLP-EP Bali wasn't really bold enough in answering this, by leaving/ disclaim access to justice and reluctant to move beyond GDP. Mickael B. Hoelman/ Indonesia

Add new comment