 Thank you so much for having me here. Over the last 20, there's been, I guess, the first thing to say is that according to monetary poverty estimates, and the bank calculates poverty in monetary terms, the estimate, the poverty line, it's a global poverty line, was set at $1.90 a day for all countries in the world several years ago. It was set that way in 1980. In using that metric of a poverty line, there's been steady but in even progress in reducing extreme poverty, the number of people or the proportion of people who live below that line. Over the last 25 years, this rate has dropped from more than 35% in 1990 to about 10% in 2015, which is the last year for which we have the bank has the data. And this is essentially the equivalent of moving more than a billion people out of extreme poverty. And that's actually the fastest rate of poverty reduction we've seen historically. But that 10% still equates to about 736 million people living in extreme poverty today. And I should also mention that the rate is particularly high in low-income countries and countries that are affected by conflict and fragility. And one of the trends that we see recently is the rate of poverty reduction has slowed. And that reflects a few factors, including falling commodity prices. We had a global economic crisis. If you listened to all the economic forecasters, there's some negative headwinds coming up as well. There's also been a changing regional composition of the global population of people who live in poverty. About half of the world's countries have reduced extreme poverty by below 3%. But the forecast suggests that the 3% target is probably not going to be reached by 2030, which is when we have the Millennium Development Goals. Poverty rates throughout sub-Saharan Africa are going to be unacceptably high. And essentially, this is the kind of work that institutions like the World Bank are working on. So I guess none of us can claim that if we care about poverty reduction, a path to 3% is not a success. So we need to go way beyond that. We begin with the question in this chapter, which I think is a really important question, and one that's very hard to answer, which is, how many women are poor? And there's a big statistic out there that I've been arguing against for people like Sharon, too, for many years. You probably have heard it's fact, not fiction, that 70% of the world's poor women. That's absolutely not empirically plausible. It's also not true. We think some bureaucrat at the UN actually made that up about 25 years ago. It's never been able. We've never been able to document that. And there's been a lot of academic, really good academic work that shows why that's difficult. But it is an important question for to ask, how many women are poor? How many children are there who live in poverty? And these may seem like very straightforward questions, but the answers are very difficult to actually arrive at. And the reason is because most measures of poverty refer to households, which are aggregates of the individuals who live within them. And individuals are classified as poor, according to the classification of the household within which they live. But my own research prior coming to the bank, as well as research by Sharon and others, show that actually this is a misguided measure for many reasons. And we know, we can say intuitively, although we can't precisely at this point quantify it, that women and children are more likely to be disproportionately affected by poverty than men. But there's considerable variation in heterogeneity across countries. So what I'm going to do today is to talk about different ways of approaching this issue of measurement. And I'm hoping maybe in the conversation we can get to the policy issues about what we do about it based on what we know. So one of the first things that I think is a newer finding from the work that the bank did, and that's reported in this report, and I'll show you some data in just a minute, is that women and children are disproportionately affected by poverty, particularly for women in their reproductive years. And I don't think we have pinpointed before when poverty bites. And this is because of social norms. It's because women often face very strong trade-offs between child-rearing and entering paid employment. And thank you so much. This tension is often most pronounced in the poorest groups and societies, as well as in the poorest countries. And I'll show you some data. It's also related to the fact that women's intra-household bargaining power and their poverty status is related to their position within the household. And one of the advantages of this data set that I'm going to talk to you about is that it has information on women in different types of households. For instance, in polygamous households, and we have information on whether women are a junior wife or a primary wife and what the relationship is of women to a principal male in a household, and that information is often difficult to get. So measuring consumption poverty and monetary policy, monetary poverty, which is what the bank does of individuals, is a very big challenge. It's a challenge because it requires information that we don't currently have in most of our data sets. The first is information on how total household resources are allocated among household members. And this has a lot of theoretical and practical challenges, which is probably very well known. So for instance, data on food consumption is really difficult to collect, particularly when households consume food together. There's no measures actually of, although we're doing some experimentation on this, on food consumed away from home. We don't necessarily know about issues about public goods, the public goods program problem. It's really hard to attribute how much of a, a lot of people live in a house and they all consume that house. So figuring out those benefits are really quite difficult. So that's another challenge. And then there's other challenges as well in terms of the use. Often when you use information that's collected at the household level, you have one of two assumptions. You usually assume that resources are equally consumed by all members of the household and that's really a bad assumption because we know that in fact there are different hierarchies of power within households and different individuals may have differential access to the resources. And the other way of getting to an individual measure is the use of something called equivalent scales, which is also problematic because it doesn't contain some subjective biases about the needs of different individuals in households and needs are not only caloric but they're also in a relative sense needs for important goods or resources. So that this household issue, the fact that data is collected at households has been a huge problem for many years. So let me tell you what the bank did and what we did. So we try to get around this problem in a few ways. There was a commission that was headed by a very distinguished economist who has passed away last year named Tony Atkinson. He headed something called the Poverty Commission and he recommended to the bank that we actually try to measure poverty at the individual level. And one of the ways that he suggested we do this is we calculate poverty rates of individuals who live in households that are classified as poor. So that's the first approach that we do and we look at households, individuals across their life cycle in households that are poor. The second way we look at individual level poverty is to actually go beyond what is the usual way of classifying households. How many times have you heard households headed by women or households headed by men? Man, the issue of headship is like the issue of households is very, very problematic and frankly I think it's a very sexist concept and you should never rely on headship particularly to compare data across countries because a head could be the eldest male of a household. It could be the decision maker. It could be somebody who makes decisions about earnings or other things. And actually the OECD countries have really evolved because they talk about respondents as opposed to heads but a lot of the surveys that particularly in low income and emerging market economies still use this concept of head. So I've never liked that concept and a few years ago I worked with some friends and colleagues from seven different countries and we came up with an alternative classification of households and we introduced that to this data set that I'm gonna tell you about. So an alternative classification methodology for households. And then finally we actually do go in with the households and try to get at the individual level data. So I'm gonna talk to you a little bit about the findings on these three dimensions. So for the first two types of the analysis we use a data set that's called the Global Monitoring Database. It's a global, and this is the main database that is the basis for all poverty estimates that you, that we hear about in the world that the World Bank derives. It's a global collection of harmonized household survey data and it includes data that was collected in 89 countries. It covers an estimated 84% of the population, about 655 million, so about 12 and a half percent of the sample by 2013 who lived at the international poverty line at that time which was $1.90 a day. We have really good regional coverage for most of the world regions. This database is updated as new surveys become available and all of the estimates that are gonna be reported here and in this forthcoming report use the purchasing power conversion factors from 2011. Now there's some questions about PPP conversions but we have to have some way to make the data across countries comparable. So the analysis that's in this forthcoming report uses the data from surveys in 2009 and as I said $1.90 per person. One of the things that the bank is doing is continuing to make improvements of this database which I think is important. And one of the things that was interesting to me when we started this exercise is that this database is, even though it's a household level database has actually never been exploited for the gender questions and so this is really a first attempt to do that. So what do we know generally about individual poverty in the world? Overall at the dollar 90 a day line we know that 655 million or about 12.5% of the population lived below the poverty line. And we know that poverty is concentrated in rural areas among those who have low levels of education, among children and one of the interesting things as we can do with this database is exploit marital status and so most of the adult poor are also married. And you can see from this table that actually the difference in the aggregate between men and women's poverty rate is really small. It's very, very minute between 49.7 and 50.3%. But as I mentioned more women in poverty are in their peak reproductive years and actually as some of the charts I'm gonna show you we can see that there's a lot of variation actually by marital status and age, education and labor status. So here's some of the figures. What do we know about individual poverty between men and women by different statuses? So we look first in the left panel which is education 15 above, marital status in the middle panel and employment in the panel at the right. And you can see that there are differences in many of the attributes to characteristics of women and men living in poverty. And this is basically the share by the way the share of poor men with no education versus the share of poor women with no education, the share of poor men and women who are married, widowed, living together and the share of males and females who have different employment statuses, paid workers, self-employed, unpaid workers and so forth. And you can see that of all poor women more have no education than is found among all poor men. So that's a first takeaway from these numbers. Women who are living in poverty are more likely to be widows or divorcees and that's actually something that we've known intuitively for a long time than poor men and they're also less likely to be never married. So that was an interesting finding. The marriage rates though are actually quite similar. And when you look at employment about half of all poor women are quote unquote inactive. Now I hate that phrasing, those are ILO typologies because we know that women are never inactive. They do all kinds of work. It's just that the work that they do is not counted. So if you look at this last panel on the right, you can see some really big differences between males and females, particularly in the self-employment category and in the paid worker category between women and men. So this gives you a picture of kind of the, by different disaggregations. And now let's look by age. These are gender poverty gaps by age group. And one of the striking things from these tables is that poverty rates among men and women actually decline with age but that relationship is not a linear relationship and the divergence actually increases as women and men reach adulthood. And that difference, and that's why the difference coincides with when people, men and women, but women in particular are at their peak reproductive years. And this is particularly related to issues in this time period of household formation, people getting married or living together and also related to when men and women go to work. So, and this is, I wanted to show you the regional view on this and I chose not to show all regions but I thought since we're in East Asia and the Pacific I would show the pictures relative to Latin America and the Caribbean. But let me just say a few things although I don't have the graphs for the other regions in Sub-Saharan Africa, women are statistically significantly poorer than men in the 20 to 39 age group and Sub-Saharan Africa largely drives the global aggregate results. In South Asia they have lower levels of poverty than Sub-Saharan Africa. The gender differences in poverty start very early in South Asia and they continue until about the age of 35. And then when you get to East Asia and Latin America you can see that in Latin America gender differences widen at about the age of 19 and they narrow down only after the age of 30 but in East Asia and the Pacific the differences are actually far more muted interested in hearing from you why you think that might be the case and the moment of the largest gap between males and females is actually narrow. So it's still concentrated in the reproductive age group 25 to 34 but East Asia does have a different pattern in the Pacific than different regions. So I mentioned that this data set allows us to look over the life cycle and this is a table that shows you the individual poverty rate by age group and marital status and when I say a life cycle approach what this means is the transition into adulthood from childhood and related changes in household formation so getting married, starting a family and this transition into adulthood reveals a lot of gender differences in poverty. What I think comes out of these numbers is clearly the negative impact of early marriage particularly for girls but also for boys but more pronounced for girls. The negative impact of divorce and separation and I think the thing that's really striking about this when you look at married women which is the third category in from the bottom when you look at married women according to household composition and presence of children the issue of having children or dependents really matters and so this is something that I think is quite important for policy makers who are really interested in targeting social protection or other approaches and I just thought I would show you the contrasting picture in terms of the life cycle approach for men and for men you also see the impact of early marriage but one thing that's striking is divorced men have much lower poverty rates than divorced women and older men who have never been married have really quite high poverty rates so that's something that I don't think we understood or realized before. Okay so I already showed you this picture I wanted to show you the sex and employment status. So let's turn now to this issue of going behind beyond the household, going beyond headship and let's figure out some other ways to look at gender and poverty. So what can we learn? So I mentioned a few years ago I was looking at a project at the impact of tax, different types of tax regimes and tax policies on men and women and we had to use household data because we didn't have individual level data and we came up with an innovative way of classifying households. One was what we call a demographic approach and the second was an economic approach and so these are two new household typologies and we think that they're better than the headship typology, why? Because they're much more neutral ways of looking at the household. So the demographic composition of the household is to take, to classify households by the sex and the number of adults who are the ages of 1864 who live in them. So you have households that have a single adult male or a single adult female, you have two adults of the opposite sex, you have multiple adult households and you have households with only seniors and households with only children. The other thing I wanted to say about why these are better classifications is they're more symmetrical. When you take the concept of headship you're often mixing up households that if you talk about a single female headed household it's usually a single female with kids but if you talk about a male headed household you're talking about usually there's adult women who live in those households along with children and you're mixing up the demographic and economic units. So for the economic composition of the household we look at the sex composition of the household and the number of earners above the age of 15. So households that have a single female earner, households that have a single male earner, two earners of the opposite sex, multiple earners and no earners and you can look at these households with and without children. And in the sample of the global monitoring database it's adult couple households with children and other adults is about 28% of the sample. Other adult combinations with children are about 7.2% of the sample and households that have one adult female with children is about 6% of the households. So the main takeaway of the database is that the most common type of households is an adult couple household with children followed by adult couple households with children and other adults. So the extended family type of household. And when you look at the economic household typology the most common type, and I know this is hard to see, oh actually it came out well on the slide. I was thinking nobody would be able to see this but the most common type is a single male earner household with other adults and dependents. So this is the extended family followed by households with no earners with adults. And that's pretty striking, households with no earners. Usually people have to earn something to be able to get by and dependents. And then the last, the smallest category is by multiple earners with dependents. So let me show you the pictures now of the demographic household typology. And so this is looking at poverty rates for households with different demographic compositions. And the first takeaway I think is that households with children are poor regardless of any other household composition. Two adult households, a couple with one male or one female are the largest share of poor households. And they're also overrepresented among the poor. The gap in the rates are largest for multiple adults, only male households and only female households. One of the things here was we cannot attribute causality. We don't know whether households are poor because of the composition or is it the poorest household that takes in more people to make ends meet or is it the reverse of that? So again, this is just the data but it's no attribution of causality. When you look at the economic household typology in poverty, it's a similar kind of story at least for those households that have dependents. And when we look at this, it's households with no earners of course are the poorest. That makes intuitive sense. Households with dependents on a single earner regardless of that earner's sex are the household that's most likely to be poor. And single earner households with dependents have much higher poverty levels and are much more likely to be overrepresented among the poor. And one of the other things that comes, I think through these numbers is that households with female unpaid family workers are in most cases much more vulnerable to poverty than others. So we can actually look at, when we say single female earners, we can actually look at the type of economic activity that they're in and it's the unpaid worker that really matters. So I wanted to show you these household demographic types. The percent of poor households by region and the takeaway here is that the regional variations are masked in the aggregate. Female adult households with children among the poor in Latin America and the Caribbean are really striking because it's compared to other regions. These households are more likely to be poor in sub-Saharan Africa than in other regions. And the high, in East Asia and the Pacific, in East Asia and the Pacific, Europe and Central Asia and South Asia, a non-negligible share of poor households are too adult households with children. So there's quite a bit of regional variation when you look at this. So in terms of the household types. And the picture is somewhat similar when you look at the household economic type. Although I have to say let's go back and look. You can see in East Asia and the Pacific, adult couple with children on the demographic slide. It's about 32.6%. And here you can see it's the female earners with children. They're only 5.5% but with children at 77.4% in East Asia and the Pacific. So, but the picture is somewhat similar. Non-negligible share of poor households are those with two adults and dependents. Okay, so these are these typologies. So I think they're richer than the household type and that's something that we could look at in terms of targeting. And finally, let's now look inside the household because people live in households and what can we learn about poverty and gender. And this is some new very experimental work that bank economists did. It's much more exploratory. And the idea here is to look at inequalities within households. This data what I'm gonna show you here is derived from surveys from seven different countries that the bank teams collected. And I think one of the main messages that we have for policymakers like ourselves is that it's really important to go within the household because we need to unpack the idea that poverty in households can be actually hidden. So again, aggregated poverty rates by sex are quite similar at the global level. But within the household, there's a lot of inequalities and we have case studies that I'm just gonna mention from Bangladesh, from Senegal, from Malawi that show in terms of consumption poverty that women, not entirely across the board, but women are slightly more likely to be consumption poor within households than men, but children are actually consumption, more consumption poor than others. I'm gonna show you the pictures from China. There's large inequalities in consumption across sex over the life cycle. The goods that we looked at were food, goods, alcohol, tobacco, some of the non-food items and we looked at clothing and other goods that are assignable to males and females. And in China, where intra-household and gender dynamics have evolved significantly over the last 20 years, given China's economic and demographic change, in 1991 extended food consumption was really twice as high among men as among women. That was a really interesting finding, but by 2009, this ratio had actually risen to 2.3, 2.3. And the gender gap in consumption is largely accounted for and this actually reinforces the findings that we had in this tax project I mentioned, is largely accounted for by four items, coffee and alcohol and tobacco and guess which goods are male, alcohol and tobacco and coffee, more men consume coffee than new women and tea. So these are goods that are consumed disproportionately by men, they may reflect social norms, they may reflect addictive properties of some of these goods and I think one of the things that is interesting to note is that excluding these four items gives a much narrow measure of consumption poverty within households. Okay, so the final thing that the bank did is not only did it do a multi-dimensional poverty measure generally for households, we commissioned Stefan Klassen who some of you may know and Rahul Lahoti who's at the Indian Institute of Management in Bangalore to do some work for us on the individual level of multi-dimensional poverty and this approach reveals a number of gender deprivations in those domains that tend to disadvantage women. So if you look at the panel on the left on education you can see that women are more likely to, so at the household approach by the way, the standard multi-dimensional poverty measure works is on education you're classified as poor if any member of the household doesn't have education, but here we do it somewhat differently because we're looking at the individual education rates of men and women. So in the panel on the left we see that education deprivation tends to disadvantage women because it shows that more women are more likely than men to live in a household where no adult has completed primary school and these gender differences in education are very muted under a household approach and they're really amplified when you bring the individual data to bear. So we looked at five countries, Ecuador, Iraq, Mexico, Tanzania and Indonesia and women are much more likely to be deprived in education than men if the deprivations are derived or measured across individuals. So that's the first finding. On the nutrition side it's a lot more muted and I think right now we're also doing some work on what's called the human capital project. I think the nutrition measures are much harder to get at and they don't, when you look at things like childhood stunting rates and body mass index it's really difficult to piece together kind of a gender gap story. So I wanna give you the picture overall of the gender gaps in individual multi-dimensional poverty across the three countries. So this is overall looking at education, looking at monetary policy and looking at the nutrition scores and you can see that multi-dimensional poverty is more prevalent among women than among men in all of the countries that we looked at with the largest gender gap in Iraq. This is not a surprise given what the Iraq has gone through and the rates are about 54% versus 38%. There's also a multi-dimensional poverty gap found in India also not a surprise. The one thing that I would say from these tables the last three tables is that I would argue that and Sharon would probably agree with me that this still represents an underestimation of the true extent of gender inequality in multi-dimensional poverty. And the reason is because few data sets actually permit estimation of resource allocation across individuals and we fall back on very unsatisfactory assumptions and some really important missing domains include things like access to services, public services whether they're education or health or social protection like cash transfers. Most data sets miss exposure to violence that could be intimate partner violence for females. It could be public violence for males and that's really important for understanding poverty dynamics. They also miss time poverty, the fact that men and women have very different allocations of their time if you look at time use surveys. And it also is missing issues of socioeconomic dimensions for instance stress or anxiety or some of the attitude issues. So where does this leave us in terms of how do we get to answering the question of how many women are poor? What do we do from here? We've been working with a set of economists in the UK and France particularly to try to adapt different types of household models, household bargaining models to see the share of women, the share of men who are poor for different consumption items. That for me is a little bit unsatisfactory because of the assumptions, the very strong assumptions that are involved and sometimes the models are so complicated you have to really unpack the assumptions and get through the math to understand what it's telling you. And then the other issue is data collection going back to what Amy said. Can we actually go inside the household and collect key information from individuals themselves on their status on what they earn, what assets they own and control, how they spend their time, whether they participate in entrepreneurship and those kinds of questions. And currently our household surveys in very few countries are actually designed to go in and interview multiple members of a household on all of these domains that reflect individual level welfare. We're trying to do this actually in a set of pilots in five different countries of developing a module that will actually ask at least two members of the opposite sex in households about this kind of information. But I have to say in the communities that I work in right now there's still a little bit of skepticism that it's actually gonna make a big difference so we have to actually prove that when you go into the household and you collect this information because it makes it more expensive you're gonna have to show that it really matters. But I do think that given the work that has been done that DFAT has funded that researchers here and elsewhere have done that for some domains of information going into the household and like asset ownership going into the household and collecting information or data directly from individuals does make a difference and we know for instance that things like the rate of asset ownership who owns what in a household whether it's land housing, businesses and so forth is really greatly underestimated when we rely on proxy respondents. So I'm gonna stop here. I just wanna say that for those of you who are budding researchers I think this is a really interesting and exciting research area. The fact that you have the mainstream the fact that individual level poverty management has come into the mainstream in institutions like bank like the World Bank is probably something at least that I celebrate given the outsize influence of the bank with finance ministers and the fact that the bank can translate this work into real social policy and economic policy is really promising. And I'd be happy to talk about other things about the relationship between gender and inequality and poverty but I thought I would talk about where I see some of the innovations and measurements. So thank you.