Has UNICEF got it wrong on gender inequality?

inequality cartoonthe misleading statements on gender in its first version, but I still don’t think they’ve got it right, so with heavy heart, here comes a rebuttal……… The problems began when UNICEF stated in the original version of an otherwise excellent report on inequality (a critical issue that is too often ignored): ‘Unlike youth, income disparities do not appear to have a disproportionate, negative impact on women…… female populations, on the aggregate, face the same levels of income inequality as the population at large.’ But as the methodology annex makes clear, none of the techniques used in the paper make any attempt to lift the lid on what is going on within households. So the lack of difference between men and women is purely an artifice of the methodology, not a reflection of (an unknown) reality. We (and UNICEF) cannot tell whether they are wrong or right, because the data provides no means of doing so. One thing that bothers me though. As far as I can work out, the only factor that would register a significant difference in terms of gender would be a large number of women-headed households that are much poorer than the rest. The fact that this does not register suggests either that women-headed households are not on average poorer than others (which I doubt) or that they are not sufficiently numerous to make a major impact on the stats (more likely). But based on my knowledge of Latin America I would have expected both to be true, and so to see a gender difference even using this methodology – anyone got any ideas why that hasn’t happened? gender_cartoonAnyway, the new version of the report, posted last week, backpedals somewhat, but gets itself into a bit of a tangle: It still says ‘Unlike children and youth, using the same data and methodology, the distribution of income at the global level does not appear to have a disproportionate, negative impact on women’, but then, a couple of paras lower down, concedes ‘based on the available aggregate income data at the global level, it is not possible to identify the dispersion of income among household members.’ But surely, if the data you are using doesn’t distinguish between women and men within households, the finding is pretty meaningless, and the statement on gender inequality is misleading, right? It’s a shame to have to pick up on this, both because UNICEF is fully aware that the absence of gender disaggregated data is indeed a problem (see here for an example), and because the report is otherwise strong and well worth reading. This from the executive summary: ‘Using market exchange rates, the richest population quintile gets 83 percent of global income with just a single percentage point for those in the poorest quintile. While there is evidence of progress, it is too slow; we estimate that it would take more than 800 years for the bottom billion to achieve ten percent of global income under the current rate of change. Also disturbing is the prevalence of children and youth among the poorest income quintiles, as approximately 50 percent are below the $2/day international poverty line.’ But the gender section is unfortunate and potentially damaging. As the main ‘man bites dog’ surprise in the report, it has already attracted coverage, for example in this post on Global Dashboard by the ODI’s Claire Melamed. Claire’s post shows how a misleading bit of analysis can snowball, as she summarized UNICEF as saying ‘Inequality is not a gender problem’. I’m sure Claire would be the first to agree that when it comes to economics, gender inequality most clearly is a problem, and a big one at that – in assets, in finance, in access to training/extension. It’s also worth noting that income is a pretty hopeless way to try and understand what happens behind the front door of households – better to look at control of assets, consumption, or ‘time poverty’. Until someone takes responsibility for addressing the data gap, these kinds of confusions are only likely to continue. For starters, has anyone ever pulled together all the available studies on intra-household inequality in income, consumption and time use and found a plausible way to scale up to some general conclusions, however tentative? If not, who’s the best candidate to do so – a non-gender specialist outfit like the World Bank or keep it in the ghetto with UN Women, given that their latest report is so brilliant? Apologies for any discomfiture to colleagues at UNICEF, which does great work on this and many other issues, and they do of course have right of reply. [h/t Amanda Lundy] Update: See comments for a response from the authors]]>

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.


10 Responses to “Has UNICEF got it wrong on gender inequality?”
  1. Lauren McCarthy

    I would have thought by now such a powerhouse as UNICEF would have been able to identify that income is not the most important or accurate way of measuring inequality between social groups. Since we do not have the avaliable data, it is very dangerous to make sweeping statements that can then be misinterpreted by other groups.

  2. Isabel Ortiz and Matthew Cummins

    Has this blog post got it wrong? The title, to start, is incorrect and very misleading. UNICEF has for long worked for women’s equality – see, for instance: http://www.unicef.org/gender/index.html. The tone and content of this post are also surprising.
    Duncan Green’s blog is about a working paper, not about UNICEF policies or positions on gender equality. Like any paper of this type, it is a working document on a new area of research, and the findings and interpretations are those of the authors, not the official views of UNICEF.
    UNICEF supports a multidimensional approach to poverty, based not only on income poverty, but on other deprivations like access to health, education, water, food, shelter, information and others. While UNICEF has published widely on different disparities that affect children, women and poor households (see http://www.unicef.org/socialpolicy/index_43137.html), the working paper in reference – “Global Inequality: Beyond the Bottom Billion – A Rapid Review of Income Distribution in 141 Countries” – offers the latest analysis of income distribution across different levels.
    The paper’s ultimate objective is to serve as an advocacy tool to bring the issue of inequality to the fore of development discourse and to push for for urgent policy actions at national and international levels, especially in light of the ongoing effects of the global economic crisis that are likely exacerbating inequalities. This overarching message has been well received and widely discussed; see, for example:
    – Lawrence Haddad’s commentary, “Inequality needs to move up the development agenda”
    – Andy Sumner’s article, “What is really happening to inequality”
    As a common practice, UNICEF shares draft documents and working papers for review, and we welcome constructive comments on all of our publications. Several weeks ago, Duncan approached us about two paragraphs of the 65-page paper – the two paragraphs solely focused on women. This is a very small sub-section of the paper that was developed following a conversation with Michele Bachelet, Executive Director of UN Women, who questioned the validity of the widely touted statistic that “70 percent of the world’s poor are women” – Claire Melamed’s blog also picked up on this.
    Having developed a section on children and youth in our working paper, we thought about applying the same methodology to women. In doing so, we found that the global distribution of income has a much stronger impact on age than gender, largely reflecting higher fertility rates among poorer women. Clearly, other data sets and other methodologies will provide different results. As Duncan notes, there is an obvious need for disaggregated data to better understand income inequality on women at local and national levels, and we look forward to studies that give us meaningful results at the global level for policy advocacy.
    UNICEF is fighting hard to make inequality more visible: it is an issue which receives too little attention in global development policy debates and deserves far more prominence. To this end, development partners should be rallying together, not undermining each other’s efforts.

  3. Duncan

    Thanks for the prompt reply Isabel and Matthew. What you seem to be saying is
    1. UNICEF has done a lot of work on gender (absolutely)
    2. This paper is not an official UNICEF paper (fair point, thanks for clarifying)
    3. UNICEF takes a multidimensional approach to poverty (I would expect nothing less)
    4. This paper aims to increase attention to inequality (and does it well, as I say in my post)
    5. Why am I focussing on just two paragraphs?
    The answer to the last point is of course, that in my view the two paragraphs (now 3) wrongly downplay the importance of gender inequality. When revising the paper, wouldn’t it simply have been better to drop the section and say in the methodology annex that the lack of disaggregated data makes it all but impossible to discuss gender income inequality?
    Absolutely agree with Michele Bachelet on the need to stop using the 70% stat and even blogged on it in February 2010 see http://www.oxfamblogs.org/fp2p/?p=1797. But is it any better to go to the opposite extreme and suggest there is no gender inequality, when we simply can’t know either way?
    And yes of course we should be ‘rallying together and not undermining each other’s efforts’, but correcting each others’ mistakes is part of that. Interestingly, when I shared the draft post on the 70% stat, various colleagues said I should not publish, using that same argument – ‘this will undermine our work on gender’. I disagree – getting it right is crucial if we are to be credible and effective.

  4. Claire Melamed

    Duncan, you’re absolutely right that getting it right is crucial. But the arguments you’ve made don’t show that UNICEF is in fact wrong. A few points:
    1. The point about multidimensional inequality is a complete red herring. The paper sets out to investigate income inequality and doesn’t pretend to do anything else. I probably should have put ‘income’ ahead of ‘inequality’ in the title of my blog on this, but the UNICEF paper makes it absolutely clear that it is income inequality that is being discussed, and this is perfectly valid.
    2. Your point about female headed households brings home to me the problem with this whole debate. Many of us, myself included, have a strong belief that all aspects of poverty – income, assets, education, everything, are defined by gender inequalities. But most of our information on that is based on anecdotal or case study evidence not large data sets, because the data is generally pretty poor. So here comes some data – albeit imperfect – that sets out to test this belief empirically in relation to just one aspect of poverty – income – and it comes out with a surprising answer. I would argue that the first response to this should be to admit that our beliefs may be – though are not necessarily – wrong, and think about other ways of testing this. Not to immediately rush to trash the study. This seems unnecessarily defensive to me.
    3. I think the data is not quite as hopeless as you make out. For a start, if there were a strong correlation between female headed households and poverty, it would, as you say, have been picked up in areas where there are a lot of female headed households. Secondly, the proportion of men to women does vary a lot between households, so if women were systematically poorer, the households with a majority of women would come out in the lower quintiles and might show up, even if only as a small effect, in the data. So it’s not great, and there’s plenty of room for error, but I don’t think it tells us nothing at all.
    4. The key question here is how much of an individual’s income (or consumption, depending on what you are measuring) is determined by their gender, and how much by the overall income of the household within which they live. The UNICEF study suggests one thing, you are arguing that it’s the opposite. Is there anywhere where the data is good enough for these two hypotheses to be tested?
    5. I would hope that we all trust each other enough to realise that nothing in any part of this debate is intended to deny that women face specific and wholly unjust deprivations because of their gender. And where you and I do agree is in the need, as I said, to get this right.
    Duncan: Thanks Claire, I agree that the only place where the data being used by UNICEF looks able to shed any light on income inequality is on single headed households and that is an area I think we should pursue. There’s some interesting forthcoming World Bank research on Mali which suggests that only women headed households headed by widows are systematically poorer. But the problem with the UNICEF paper (and your original reporting of it) is that it claims to have something more general to say on gender inequality in general, when it doesn’t. In contrast, my post doesn’t claim to know the answer on income inequality (I say it may be true that it is not an issue, or it may not – we just can’t tell from this data).

  5. Michael L

    That’s right. If you hope for any credibility openers and honesty are your best bets.
    The politicians have been twisting data and words for too long. It’s time for some honesty without deception.

  6. Hi Duncan,
    Thanks for raising this.
    1. It is likely true that women are not disproportionately represented in poor households.
    2. This tells us nothing about intra-household distribution, as you note.
    3. This also tells us nothing about the impact of inequality on women. There are various ways of coming at this question. One is about income inequality–specifically, do women make less than men or, do women make less than men for doing the same job with the same qualifications? There is a host of data to confirm that they do face income inequality. Another way to come at this is to look at inequality in all its forms and see what impact this has on women. Take the US. As the wealthiest continue to dominate political processes that erode social protection programs, this burdens women disproportionately in a variety of ways because the disproportionate burden they bear for a variety of forms of caring work.
    4. Everyone is right that better data is needed, which is truly gender sensitive and capable of revealing gender disparities, that can resolve the question of whether women are disproportionately poorer. One project underway, which I work on, is attempting to do this exact work. (Your colleagues at Oxfam GB Southern Africa are major contributors as well). See http://www.genderpovertymeasure.org. There is not a lot there yet, but we’ll have more information up as the project proceeds.

  7. Picking up on the issue of improving measurement and gender indicators on poverty and income inequality, the Gender Affairs Division of the Economic Commission for Latin America and the Caribbean (ECLAC) has done a lot of work on producing and improving those over the last few years. This includes cross-country data for one important indicator which was already highlighted by Diane Elson in your 2010 blog (cited above by James): the percentage of women and men who have no access to personal income (i.e. income independent from that of other household members and hence an important indicator of women’s ‘voice’ and ‘exit’ options). There is a host of other indicators in their gender statistics system, not only on income poverty and inequality, but also related to reproductive rights, paid & unpaid work, violence, participation in political decision-making, etc. Many gender inequalities are difficult to capture statistically, but some of these may prove useful for complementing the picture and provide ideas for advancing gender-sensitive statistics in other regions. Here is the link: http://websie.eclac.cl/sisgen/ConsultaIntegrada.asp?idAplicacion=11

  8. Papa Seck

    Thanks Duncan for kick-starting this discussion as I think it is very important to have. Although the UNICEF paper (co-authored by a former colleague) attempts to shed light an important topic, I cannot help but to question the methodology, especially in regards to gender, youth etc. I think the paper suffers from two basic problems (aside from the data issues):
    1. The first one is the estimation method. Given that per capita income is used to classify countries by quintiles, the paper merely captures demography rather than poverty. For example, it is well established that countries entering their demographic transition tend to have a large youth population and tend to be poor. On gender, the demographic make-up of most countries is roughly 51-49 as the paper mentions and this is exactly what it is capturing. All of this has little to do with poverty—particularly women’s poverty—it is simply a confirmation of what is already long known to demographers.
    2. The second problem is something that is mentioned in the paper, namely when it asks (Page 10) “What do the extreme distortions in income distribution at the global level mean for different groups, such as the poor, children, women, or the middle classes”. The problem with this statement is that rather than recognizing that gender cuts across class, age etc., it considers women a distinct group. That alone contradicts any talk of inequality.
    As has been mentioned in this blog, any analysis of women’s poverty has to grapple with intra-household stuff. For example, according to our recently published report Progress of the World’s Women: In Pursuit of Justice , in some countries, women in rich and poor households alike have little or no say in vital household decisions such as expenditures; and in most regions of the world they are most likely to be educationally poor. To me, this simply indicates that women can be poor, even in rich households.
    Overall, I agree that measurement of gender and poverty is not even at a primitive stage currently and we need to applaud efforts to tackle it. However, I am also of the opinion that it may sometimes be better to simply acknowledge shortcomings rather than provide estimates that may not be grounded in reality. For instance, it probably took a stroke of someone’s pen to assert that 70% of the world’s poor are women, but it is taking much longer for the rest of us to debunk it. Although not on par with that Statistic, the Ortiz Cummins paper is also misleading.

  9. Angelica Sorel

    Much as I would like to join the calls for sex-disaggregated data, don’t we also need something a bit more basic here – some circumspection about the use of the concept of the “household”?
    What a “household” consists of in Lewisham, Lagos or Lahore is completely different, not only across these contexts but also within them. To assume – as economists have long done so – that household income is pooled is a fallacy that was long rumbled by sociologists and anthropologists, but has been oddly persistent. To assume female headed households are worse off than those headed by a man is one of those tired old normativities that I’ve never quite trusted because they just don’t seem to match with the many actually existing female headed households I’ve come across in the course of my work and life. In fact, I am very much looking forward to the study that puts some evidencial flesh on the hunch that for many women, especially women who have been in relationships with domineering and abusive men, life as a female headed household is blissfully happy. That’s why we need measures of wellbeing as well as the narrow kinds of measures that seem to be used in this study.
    So the problem here seems to be something rather more basic. It is about fundamental limitations in economic method and analysis – about the ways in which economic method uses assumptions, about the bluntness of the instruments used to measure, about the ways in which the great variety in human experience is flattened in order to measure and compare it through the production of economic data sets.
    I am not arguing against the use of numbers, but I think this is a case study in why some of the more sophisticated mixed-methods approaches that have been developing over the last couple of decades (including the use of participatory numbers) are so badly needed by agencies like UNICEF