 Felly, mae'n meddwl. Mae'n meddwl fel yma yn 2013. Felly, nhw'n gweld ymellafol o'r cyfnoddau a'r cyfrannu mygwysig. Mae'r cyfrannu mewn gwirionedd yn ei bwysig ar hyn o'r tîm. Felly, mae'n fwrdd ar ystyried. Mae'n ddweud y cwrs rydyn ni'n meddwl ar y ddechrau ar gael a'r ddau ganddiadau o'r cyfrannu mewn gwirionedd. Mae'r ddweud yw'n meddwl bod yw'r cyfrannu yn gallu yn ymgyrch i'w rhan o'r eu gennych. Mae'r pethau yn ar-dwybr gan ymarfer. Mae'r gwaith, ac yn y gallu'n mynd o'r ysgolwyddiadau yn ymgyrch i'r rhan o'r cyffredinol. Mae'n gweithio ar gyfer gwaith gynhyrch yn ymgyrch, mae'n gweithio ar gyfer gael gweithio'r gwaith, a wneud yma y rhan o'r cyffredinol. I troi'r gynhwys eich ffostion o'r moddog ar gyfer y gweithio a gyfath o amserol yma yn olaf yn ei wneud dod o'n sefydliol ar gyfer yw'r mewn gweld. Er cofnodd yn golygu y gallu amddangos arol yr owniau, nad y gallw'n golygu y gallwch ychydig ac yn adroddodd yn eithaf wybodaeth y cyfrifpo i ymddangos yn gydag yn meddwl yn ei ddweud yn hynodd argyrchu yng Nghymru ar gweithi. Now what I'm going to do today is to talk about some of the preliminary findings from our more quantitative approach to try and unpick this observation that we've found from our qualitative work. The conceptual framework that is going to underpin our work, or the work that I'm presenting today, draws very strongly on the new economics of labourSupermigration and the discussion around potential motives for remittance. The idea that remittances might be motivated by a number of factors which the literature broadly groups into these three strands. So remittances might be purely altruistic or they might represent some sort of relationship between the family that could be called enlightened self-interest. So, for example, a co-insurance arrangement. I'll send money home in the hope that if I need support from my family, they'll support me later. o'r ddweud o'r ffordd, felly rhai gan ddweud â'r hynny yn y cyd-fodol o'r awrcaf o'r sorg o'r cyfwysgol oedd eich gweithio o'r ffordd i'r ffordd o'r cyd-fodol. Yn gyfwysig, yna yw'r gweithio gyda'r dyma a'i gŵl ei ddweud yng Nghymru oherwydd y ffordd o'r ddechrau oherwydd yna'r ffordd o'r ddweud o'r ddweud o'r ddweud o'r ddweud o'r auturism through enlightened self-interest. It's very hard for us to imagine what this might look like in a kind of quantitative way. And the evidence is broadly suggestive that exchange or co-insurance mechanisms dominate altruism. But of course, because this is a very difficult phenomenon to really understand, the evidence on this should be regarded as a little bit suggestive rather than conclusive. Mae'n cymaint ei wneud yylod yw gwahoddiadau yn y Llyfrgell Fyrlus yn ôl hwnnw. Dylai'r holl ddataeth yn ei wneud o'r rhannu, ac rydym yn ei ddweud o'r ddiddordebio'i fod yr roeddfyrddwr. Dadau er bod yn rhan o'r awlach ac o'r rhannu byddai ddau, mae'n ymhwmos ettonion cyfraeg o'r hynny, ac mae o'r brif yn hoffi o'r hoffi, Felly, unrhyw gael gyda ddim yn cael bydd yr arosgo a lawer o ddod y ddweud y ddweud yw'r ddweud o adrwydd 18egon o ganddwydol mewn cyfrifiadau. ac mae'n ddim yn cael ddigonio cymorth o'r ddweud o'r ddweud o'r ddweud o rhai o'r ddweud o'r ddweud o'r ddweud o'r ddweud o'r ddweud o'r ddweud o'r ddweud o'r ddweud o'r gweithio wneud yn Fietnaam. boedd am yr unigol Mni, Ili a'r iawn gweld y gwasanaeth a'r unigol sydd yn gweithio ystafell ni'n bod ychydig cael ysbyt yn gweithio gweithio gilydd mewn ly ef yn hawdd ar gyfer y dywed, nad oes yn gweithio gweithio gweithio gweithio gwych chi'n gweithio gweithio gweithio gweithio gwych, fel oddi'r cyflawn o'r cyflos yma sydd yn gweithio gweithio gweithio gweithio gwych ag well yn fawr carlwch gyda'i... other such institutions. In contrast, a study by Abrego looking at Salvadorian migrants in the U.S., has suggested that actually their families back home are better off if their migrant member is a woman, because the mother is more likely to be sending money home than the father. So, the evidence is a little bit mixed, and that's something we want to look at for the African context. Ac rwy'n clywed i eisiau i ddweud i ddweud i ddweud i modiwn Can we see any evidence in our data that maybe modives differ and can we or should we be ascribing this to different levels of altruism or is there something else going to go happening in this? So as an overview of the rest of my talk, I'm going to first of all spend a few minutes on data, on remittances and talking to you about what this data gap is under reporting, for example. I then going to introduce you very briefly to the surveys that we've been conducting and the Zimbabwe survey in particular. And then going to talk through how we model remittance decisions, talk you through a little bit on the technical aspect of our research, and then move on to results and discussion. I've kept this as quite a broad, hopefully accessible talk, I'm not going to be presenting lots and tables of regression output because I know this conference is trying to make our research more accessible to not just an economics audience. The one takeaway, I think, from this paper is that women, what we find is that although women, when we just look at the raw data it suggests that women do send less remittances home compared to men migrants. Once you control for characteristics of age, education, those kinds of characteristics, women are just as likely as men to be sending remittances home and to send the same amount. So we observe no difference in what women are sending home. What we do observe as a difference is how they choose to send their remittances home either in the form of cash or in kind, so goods that can be sent home like food and clothing. So that's the sort of takeaway and now I'll try and flesh that out. So first of all on the data gap, there's a number of elements to the data gap on remittances generally. We're all familiar with, we can understand very easily why estimates of cash remittances might be severely under reported because we're all familiar with the ways that people send money home to avoid paying official transaction costs and charges. Money is carried home by a number of ways and even cash transactions or financial transactions don't necessarily involve the direct transfer of cash across a border but can involve sophisticated arrangements where networks of people are involved paying bills for each other or transferring money on other people's behalf. A lot of this data is not reported by gender. So certainly the international remittances data, you won't see that broken down by gender. But the other aspect of the data gap is the data on in kind remittances. So these are goods and things like food, clothing, medicine, education supplies, those sorts of things that we also think are either not collected and when they are they're under reported. So although there are some surveys which I'll show you about that do collect this kind of data, they don't often report what the in kind remittances are. So we've done some work on these surveys but they also don't always capture the most common types of things that people are sending home and I'll elaborate on that in a few moments. So what one takeaway on this data gap for us is that we think that remittances might be underestimated by as much as between 10 and as much as 50% by not including in kind remittances. And the under reporting is particularly significant for women migrants. So let's look at some of the data on remittances that includes in kind remittances. This is data for two Pacific Islands and I'll speed up a bit I think. And this data for Fiji and Tonga suggests that in kind remittances are about 25% of total remittances. So there's some under reporting if we ignore the in kind. And then when we look at the African migration surveys, the World Bank African migration surveys, we can see again that remittances by women are much more likely to be in kind. So goods and these are significantly under reported and you can see by the height of the orange bar on the far right that the under reporting is more significant for women than it might be for men. This is the case of Kenya and Burkina Faso and we can see that the in kind remittances make up the gap once we include them in our estimate of total. These are our own calculations from the migration surveys. A slightly different scenario in Senegal and also in Nigeria where the in kind remittances don't make up the gap but we can still see the significant under reporting of remittances if we leave out in kind remittances. And then finally the South Africa survey which unfortunately doesn't report values and only shows what people are sending home rather than what the value of that is. But we can still see some evidence that women seem to exhibit some preference for sending goods home rather than cash. So let me move on to our research and what we're doing. So very briefly we've been collecting data in five countries in different regions of the world. We have a common approach to sampling and a common framework for key definitions and in the interest of time I will skip this now and I'll just say if you want to find out more about our data including if you want to use our data. It's all available at this web link which will be on these slides when you if you want to download the data at some point in the future would be very would be very happy for people to use our data. The Zimbabwe survey took place in 2015 around Easter time that year. We have data on around 1200 households from three different districts of the country. This comprises 18 villages. So we suggest that our data is representative of rural households in these three in these three districts but not nationally representative because our districts weren't chosen randomly. So therefore we don't claim national representation. We have from our households almost 1500 migrants amongst that. So our households are reporting on their migrants. So obviously there's issues around recall and information gaps between the migrant and the household. But we have data on nearly 1500 people very briefly on our sample size. So just two points to make on this. This table shows the number of households that fit into different categories of having migrants or not having migrants and shown for the three districts separately. So roughly 70% of our households have a migrant or at least one migrant. That percent is just slightly lower in Gwanda. So very close to the South African border because at the time we were doing our field work. South Africa was beginning a lot of repatriation of Zimbabwe migrants. So our field workers reported that households were very reluctant to tell us about their migrants because they were nervous that there may be repercussions for their migrants if they gave us too much information. So the percentage of households of migrants is a bit lower. But generally you can see that the two regions closer to South Africa have more international migrants than Hurungwe in the north which has fewer international ones. So let me return to remittances, in-kind remittances. So the World Bank Survey define in-kind remittances using that list on the left hand side of the screen. I've grouped them into broadly household appliances, business equipment, these sorts of things although that's my grouping. Obviously we could argue whether a sewing machine is a personal or a household or a business. But you can see these are quite large items that have significant economic value to them. But it doesn't include food or clothing. In another World Bank report looking at remittances from the Netherlands back to Surinam, they suggest that most of the parcels that are being sent home contain food or clothing. And the sort of larger items are much lower in frequency. So the African Migration Service are very, very useful resources. They are still underreporting on in-kind remittances if it's food and clothing that people are sending home more often. So when we designed our survey, we decided to focus on food and clothing and these smaller items. Resources for school, medicines, these sorts of things. And we can see that the bulk of in-kind remittances that people are sending home are indeed food and clothing. So these are the remittances that are sent home in the last 12 months. The cash value is reported by the household who is responding about the receipt of these remittances. And clearly they may be an issue about how you value in relative terms goods that might be purchased in South Africa compared to what their value is in Zimbabwe. That's something we're still looking at. But we can see that food and clothing are the most significant items being sent home. We can also see, using our data, that those in-kind remittances are proportionally much more significant for women. So again there's an underreporting if we only looked at cash. And we can see that if we look at the final graph on the right hand side that although the value of the in-kind remittances doesn't make up the gap completely it goes a substantial way to making up that gap. So on average it looks like women are sending less home. But of course our women migrants are not the same as our men migrants. They differ in a number of ways. They're younger, they're more likely to be international. More of our women migrants are in South Africa than staying within the country. So in a way you would think that maybe they'd be more likely to send remittances home by cash because they're having to send them across the border. But no, they're still sending them home in goods. Let me move on to our empirical approach. So we're estimating three different econometric models here. One on the incident, so just a very simple probability model of does a migrant send remittances home. We then model the amount, so the combined dollar value of cash and in-kind remittances. And then we look at the mix of remittances, so the percentage of total that is cash. I've put an equation up there which just puts that in mathematical terms for those of us who like to see an equation to understand a model. So we use gender, we use a range of other characteristics of the migrant and of the household that they've come from. We cluster our observations at the household level because of that unobserved heterogeneity across households and shared characteristics within households of different members. We also make a first attempt at tackling the selection bias that might occur in models two and three where we're looking at amounts and the mix because obviously this data is only available for people who are actually sending money home or goods home. So we use a Tobit model as our first attempt to deal with selection bias issues. I'm not going to put any tables of regression results up. The paper is linked already online so you're very welcome to read it in more detail. I thought I'd try and present the results in a slightly more accessible way. So not trying to put too many numbers on a screen. So this is what we observe from estimating the full model that I just showed you as an equation. So I've got all of the characteristics of the migrant and their household included. So when we control for all of these characteristics we find that there's no difference in the probability of sending remittances home between men and women. No differences in the amounts that they're sending home. But a statistically significant and not a particularly small difference in terms of magnitude in the mix. So we observe that after we control for characteristics that we think might relate to earning potential at the destination we find that women are just as likely to send money home, remittances home, to send similar amounts in value, but they are 12.5 percentage points less likely. Sorry, the percentage of cash that they sent home is 12.5 percentage points lower than it is for men. So that's quite a significant difference. So this suggests then that this idea that women remit less than men needs to be looked at not just for Zimbabwe, but for other countries as well. And we may be underestimating the value of remittances that are flowing between countries or between migrants and their households. I want to return to some of the ideas that I talked earlier about, about motives for remittances. I've got five minutes left. So I've summarised here some of the results we've got from our econometric work. And I've just pulled out some of the sort of themes that have emerged from that work. Let me first of all talk about the exchange hypothesis idea that I mentioned earlier. So this is the idea that migrants are sending remittances home in exchange for something that they're going to receive back from the family. So it might be inheritance. It may be part of a contractual arrangement about looking after children, for example. And there's some evidence that men are more likely to be following these exchange motives than women. So there's a number of points that I can direct you to. So, for example, if we look at our results on household wealth, so these are fixed, tangible assets that we have data on, we find that there is a positive correlation between how much men are sending home and the value of these assets that their households have. But we find no link for women. Another possible evidence for exchange is if we look at remittance decay. So if we look at the time away that the migrant has been from the household, we see that the longer the migrant has been away, the less likely they are to send money home. And also, the smaller the amount. But this only applies to men. There's no change over time, over that time profile for women. Another possible exchange or evidence for exchange might come in our results on dependent children. So we find that if men have dependent children left behind in the household, they send more remittances home. Whereas it doesn't seem to affect the amount of remittances that women are sending home. So there seems to be some evidence for exchange motives for our male migrants. But we don't seem to find any strong evidence for exchange for our women migrants. So that might lead us to think, well, maybe women are just so altruistic that they send money home regardless. I think that would be a jumping the gun a little bit. I think we would find it hard to think that women are more altruistic. Why should women be more altruistic than men? So we'd like to suggest some alternative interpretations of our results. So while we do find some evidence of exchange for men, we can't necessarily include that the absence of that evidence means altruism is the dominant motive for female remittances. And we think it's important to look at the context that remittances are taking place in. So this is what we plan to do in our follow-up work, is to try and understand institutions that might influence whether or not people send remittances home and might also influence what they send home, whether it's goods or cash. So, for example, in the Zimbabwean context, traditional inheritance practices favour men and sons and older sons compared to younger sons and also sons of higher order, higher status wives than lower status wives. So men may have a stronger incentive to send cash because that's going to be invested, hopefully, in the household assets, in the household business, which will be inherited in part, at least, by the man. There may also be an element of control. There may be gendered relationships within the household, which mean that male migrants have more control over what their remittances are used for, and that might not apply to such an extent to women. This might also reflect some differences in ethnicity and practices of inheritance that exist between different ethnic groups within our sample, which we're researching at the moment. This element of control might also reflect a preference for women to send things home in kind, so actual goods that have a specific purpose by a specific person in the family. So if you send baby clothes home, the only person that can use them unless they're sold is the baby in the household. Similarly, food or education supplies, those sorts of things are going to be consumed within the household. So there may be a control issue, and that's why women are preferring to send food and clothing. So we plan to try and unpack the household and look at relationships between migrants and other members in the family. The results might also reflect income sharing practices. There's a lot of literature that looks at the way incomes are not necessarily fully pooled within rural communities, but certainly shared to a certain extent, and a lot of pressure on families to contribute towards community level activities. I'm going to move on to just one last slide, if I may, on generational norms as well. So we observe in our data that older migrants are more likely to be sending money home, and we think this supports some of the observations we made in the field four years ago. Talking to older people, they seemed to have a stronger sense of responsibility towards their families talking about sending money home, whereas younger people seem to see their life, their future outside of Zimbabwe, so they had less incentive to send money home. I'll wrap now. Just with a couple of conclusions to take away, which I won't repeat them all. I'll move on to the last one about policy implications. So why does this matter? So one is that clearly if we ignore in-kind remittances, we're ignoring the contribution or part of the contribution that women make to rural economies and to household economies in rural areas. It raises questions about what might happen to remittance flows if, say, transaction charges come down. They are notoriously high in Africa. Will that change what people send home if the reason they're not sending cash is not because of transaction charges but because of other issues around institutions like inheritance, marriage, income pooling? It may have similar implications for who might contribute to a diaspora bond if, say, the projects that are funded by a diaspora bond don't necessarily concur with the priorities of the migrants. That's all. Thank you very much.