 The Sarah just told you about the beginning of marriage, and I'm going to tell you about the end of marriage. Marital shocks and women's welfare in Africa. So this talk draws on the wider paper, which is about Senegal, as well as a more general Africa-wide paper that I've worked on. I think it's useful to start to just to get ideas into place. It's useful to study this graph a bit. And here you see proportions of men and women by current marital status and controlling for their age, or rather, at different ages. And you see immediately how different the patterns are for the men on the left and women on the right. And so men basically spend much more of their lives married from their early 30s to about their early 80s, more than 80% of men are married. If you turn to the women in the green line, again, marriage, it's completely different. Here you see that marriage peaks at 80% around 30 and drops below 80% at 40. And then it's kind of precipitously drops, while widowhood rises inexorably for women. So women spend much more of their lives in widowhood than do men. So a big, big difference between the marital experience for men and women. So there are a few factors that underlie these patterns. First, marital dissolution is extremely frequent. Women marry young and typically much older men. They live longer lives. And of course, many of these countries have conflict in Africa, HIV-AIDS. I know today that more women are dying of HIV-AIDS. But in general, there's a lot of death and changes like that. The other issue is that men are much more likely to remarry than women. So very high remarriage rates for men. And this is aided by polygamy, which is still legal in 25 countries. But even where it's not legal, there's a lot of polygamy that occurs. And for women, remarriage rates vary much, much more. They're typical in some places for women under 40, they're still in their reproductive age. And then in some countries, it's very frowned upon to get remarried. So the key issue, I mean, all these sort of the pattern of marriage and so on has concerned me for quite some years. And I've all just wondered whether this is something that we should be concerned about, that social policies should be concerned about. Let me say, as all the other talks have said already, that here too, the issue is correlation, not causality. Causal is very unclear. We are certainly not claiming anywhere here that it's random that some women experience marital dissolution. And really, for policy, what matters is the correlation. And so the key question, I think, of all this work is, are ever-widowed, ever-divorced African women worse off than otherwise similar women that are still in their first union? And that's essentially the question I've been trying to work on for eight years or so. So why even think that this is an issue? Well, I think that historically, widows have been identified as generally disadvantaged groups, certainly in today's rich world. If you think back on early social protection policies, social insurance and so on, the poor lives in England, widows were often the targeted groups or the groups that benefit, the group that benefited the most from those early policies. There's been a lot less policy attention in developing countries. I think the one big exception to that is India, where there was work some 20 years ago, quite a lot of work, that found that widows were particularly discriminated against and destitute. And I think India is very relevant here because it's quite similar to Africa in the sense that there's very weak economic and legal equality between the genders and that women are extremely dependent on their husbands. Now, divorce is a more complicated issue, in part because women can also instigate it. And so there's much more mixed evidence of negative economic and other consequences around the world. What about for Africa? Well, we all know for one that there's a huge amount of work on gender inequalities in Africa, right? In access to land, to assets, in legal rights in terms of human capital and so on. So that's there. Now, otherwise, marital dissolution has been generally understudied, I'd say. There's a number of papers on specific countries which are certainly relevant. For example, widow-headed households have been found to be particularly poor amongst all households in Zimbabwe, Uganda, and Mali. A recent paper by Anderson and Ray argues that lack of a husband accounts for some 35% of excess adult female mortality in Africa. There's a number of case studies showing that in specific countries, it's widows of men that die of illness that are most vulnerable to losing land. There are two recent papers showing a finding that evidence of responses to the risks of marital ruptures with strong implications for broader societal welfare. This is, I think, very important and really an area to do more work in. I'm going quickly here because I don't have a lot of time. And then divorce, I think, again, because women can ask for divorce as well as men. There's a sparse literature that suggests that the consequences are sometimes positive and sometimes very negative for women. And it's just a more complicated thing to look at. The other evidence comes from short village studies or a lot of anecdotal evidence from legal human rights and sociology literatures and public opinion surveys. For example, in Nigeria, about you are widows worse off than other women. And high proportion of people say yes. There's a lot on dispossession, land grabbing, a lot of issues. Again, very anecdotal, very specific case studies. What's really, really missing is more large sample population representative empirical research on the consequences of marital dissolution. Now, the other issue is that we know that public safety nets and insurance mechanisms are very underdeveloped in Africa. I'd say one exception may be southern Africa, where there are grants to elderly widows or elderly women. Now, it could be argued that there's private support that helps. There's a large literature on informal solidarity and risk sharing institutions for Africa. I think having studied that literature, there's very little said about gender at all in that literature and really nothing about marital dissolution. So that's an open question. A question showed that child fostering is a tool for risk sharing. And there's certainly a lot of child fostering going on. And then in our Senegal paper, I think what we would like to argue is that the leverage and the tradition of polygamy that both are quite strong in many countries still provides a kind of safety net for widows and divorced women. Now, so there's very little evidence, I'd argue. And even if causal attribution isn't the aim, it's not a very straightforward topic to look at. And I'd like to highlight two key issues. One is prevalence. We don't know the prevalence of divorce and widowhood. Essentially, household surveys, even the DHS, most of the DHS, collect current marital status. So the group of married women contains all ex-widows, ex-divorses. It's not just married women in their first union. And we just don't know, just don't know how prevalent it is. Micro studies, including our Senegal study, which was designed very much to look at these kinds of issues, shows that about 19% to 22% of ever-married women in Senegal have experienced widowhood, 13% to 17% divorce. Those are very high numbers. And 7% of ever-married women have had more than one breakup and often up to three or more. The Senegal study also suggests that disadvantage for women in their offspring persists through remarriage. And another study that I've worked on in Firmali shows the same. So the good thing is that the DHS had a very short phase from about 93 to early 20s, where they asked this question about how did your last union, if it broke up, how did it end? And so we know whether a woman is an ex-widow or an ex-divorcy. And so all my research, the overall research, not the Senegal, but has focused very much on the 20 countries that we have that data for. And this is only for women 15 to 49 years old. OK, so prevalence. What is the prevalence? Well, in the first column there, that's the information from 29 DHS surveys, the most recent, where in the household roster for the DHS, you get current marital status for all women, 15 and over. And we see that about one in 10 women in Africa, in 29 countries at least, are widows. Now, if you compare that to the next two columns, where it's the data from the 20 surveys where we have full marital history, or not totally full, but fuller, we see that current widowhood in amongst the 15 to 49 is about 3%. But amongst that young group, you already have 5.3% of women who are ever widowed. So already in that group, you're getting many, many women. In the divorced group, it's a very different pattern. In that, you have 14.5% ever widowed in the very, very young group of women. And that's really because divorce is happening very young, and women are getting remarried. Now, the other hurdle to studying this is poverty. We, of course, do not have any measures of individual poverty. Everything is household-based, and we just assume that all the individuals within the household have the same living standards. And so anybody who's disadvantaged remains essentially invisible in our data sources. The one important dimension of individual welfare that we do have access to is nutritional status. And that's what I've been using in a lot of my work. Again, it's available only for women 15 to 49 from the DHS. In Senegal, we have a kind of individualized measure of consumption that we can use. It's very nice, but I don't have much time to describe it. OK, so new evidence on well-being. So I'm starting with the Africa-wide study. And here, essentially, what are we doing? We're pooling DHS data for the 20 countries. And so it's a data set of individual women. We're regressing nutritional status. We use various measures here. I'm going to focus on under nutrition. And we regress that on marital status, dummies for the detailed marital status, and a wide array of individual and household level characteristics, and in either country or household fixed effects. Now, so the question really is to see whether ever-widowed, ever-divorced women are nutritionally deprived relative to otherwise similar women in their first union. So that's really the question we're posing. OK, so we find that current widows and current divorces have statistically worse nutritional status indicators on average. So this first result is across all of Africa, if you compare an African widow compared to an average, similar married once woman, being a widow is associated with a 1.8 percentage points increase in the likelihood of being underweight. It's generally higher in urban areas. So for example, on average, it's 2.4 percentage points difference. Being a divorcee raises the probability of underweight by 1.4 percentage points. Now, OK, these are very highly statistically significant effects. They seem quite small. They are quite small. One reason is that we're averaging over very heterogeneous countries, and I'll show you that in a second. But one way to interpret these numbers is really to compare with the coefficients we get on the impact of education on nutrition. And essentially, the overall gain of not becoming a widow is equivalent to about six to seven years of schooling and four to six years of schooling for divorces. So these are substantial impacts. I have to say that those regressions include the wealth index of the household, which could well attenuate the impact of education. If you take out the wealth index, it's still four to five years of education for widows. Now, so still overall, these impacts are relatively small. Next, what we do is essentially estimate these impacts for specific countries. So everything is at the country level. We're comparing an average widow in the country to a similar married once woman in that country. And so here I've just plotted, and I'm just going to focus on widowhood because the paper looks at widowhood, ex-widows, ex-divorses, divorces with country and household fixed effects. And there's just so many results here. I'm just focusing on this. So the blue is country fixed effects. The red is household fixed effects. These are, so for example, in Ethiopia, controlling for household fixed effects, the probability of being undernourished goes up by 15 percentage points. Often, and I'm not showing you here the urban rural distinction, and often it's quite incredibly high in urban areas. What you see is a lot of variance across these countries. So some countries with very strong impacts that are typically higher with household fixed effects. And then some countries where there's actually advantages to being a widow. Now, one of the issues here is in some of the countries where things are very bad, there's high HIV AIDS incidence prevalence. Let me show you that here. You see that the prevalence of HIV AIDS varies a lot across countries, but it varies a lot also across marital status. I haven't seen this before, but it's quite striking, I think, to look at this. And you see that in most places, widows have higher HIV incidence than other women. And married once women have the lowest. But there's a variance in terms of the ranking. OK, so for 14 of our 20 countries, we have tests for HIV status. We control for that. And here are the impacts of widowhood on underweight for those same countries, the 14 for which we have the tests. And you see, I mean, basically the estimates go down a bit, but they really persist in those same places. OK, so there are big conditional differences in nutritional status for widows Africa-wide. But the extent of disadvantage varies. And we get relatively worse outcomes in non-West and especially East Africa. Significant negative associations with nutrition, and these are often in areas that are most affected by HIV. Yet, our findings are robust to controlling for HIV. Turning quickly to the Senegal study, here we find that poor women are much more vulnerable to both having a marital dissolution and to getting remarried. And remarriage is typically as a lower-ranked wife in a polygamous household, whether your marriage ended in divorce or widowhood. It does not compensate fully for the economic loss that we find associated with widowhood or divorce. And remarriage is also least effective in ensuring consumption levels for the most vulnerable widows who tend to remarry in levered marriages. So levered marriages for those who don't know is when you remarry in the lineage of your husband, your dead husband, and mostly a brother or maybe another relative if there's no brother. Levered marriages totally survives in Senegal. You see it a lot. But it comes for women with the lowest consumption levels. And it's really the only option for very poor widows and happens in poor lineages. And for those women, those poor widows basically who don't have housing, they don't have somewhere to go. Or there's that, they just don't have the means to not get remarried immediately. So they do it. And the other reason is that there's enormous social pressure from the lineage to keep the children. And so if the widow wants to stay with the children, she gets remarried under the levered. But we find that these marriages act as a poverty trap for those who can't refuse to remarry. When women can afford to, because they have a job, because they have an education, a house, for example, they really typically choose not to remarry. And we find that they maintain a level of consumption comparable to married once women. One more second. So in summary, marital dissolution is a frequent shock. It's also a severe shock. We find that these women are often significantly worse off. And so are the children. But again, it's not universal. There's large regional and cross-country differences. And the lower welfare persists through remarriage. I just want to finish with this quote from Jean Dresd from 1990 about India, but I think it's just very relevant here. Combating the neglect of widows and divorces must be seen as an integral part of the broader struggle against gender inequalities. And I really think that's a quote that applies to Africa. Thank you.