 So, and the idea is that some share of the population that is at a disadvantage migrates in response to the rural urban breakdown of population that is advantaged. So there's a kind of differential in, in perception at least in terms of economic conditions in this case, but it's broader than that. So if a start curve is rural urban divide, the more people will affect it and the more migration there will be. So that's the bottom line, right? The model in spirit is compatible with the hairs to their approach, this notion that people migrate because of an expectation of improving their life. The paper has kind of a model that is developed based on a physical analogy. I'm not going to get into that for this presentation, but it is essentially a model that has a certain level of structure and can be interpreted. Now, so the basics of the approach. So the basic premise of the approach is that there is a cut-off income level separating the poor from the non-poor when it comes to the incentive to migrate. So if you are below the moderate poverty line, you're in that category that would like to be above the moderate poverty line. And so it depends the breakdown of rural urban population in these two categories. So we will be operating with shares of the national population that are above or below the poverty line, both rural and urban areas, and dealing with net migration rates between rural and urban areas. So the rest is best explained graphically. So I'll try to walk you through a little bit of the thinking behind this paper. So here we have on the x-axis a percent of urban population in the total population, on the y-axis a percent of the rural population in the total population. And we start by mapping the share of people below the moderate poverty line. So this is for a hypothetical country. You have 15 percent of the people are urban poor and 50 percent are rural poor. So this diagonal line is the eyes of poverty line. So essentially, if you migrate in the short term, you're just moving along this line. And you're migrating per se does not change your income level. So the act of migration just moves you along this line. So if you're going from rural to urban, you'll be moving down that diagonal. Now we plot the share of a population that is above that poverty level, and there are two, I mean the rural and urban breakdown. So this schematic representation is one where the higher income population is mostly urban, not very rural, right? So you have that essentially in urban areas income, the likelihood of being poor in urban areas is lower than that in rural areas based on this breakdown. Now this diagonal line is the total population line. So you're adding these two vectors and essentially you get whatever combination you have over time, you end up on that diagonal when you add up these two vectors, right? So the, and this is the rural urban breakdown of the country. So in this case you have that the country is 45 percent urban and 55 percent rural, right? Now the basic, and here what we're trying to do is, I'm not sure how well people can see from the back, but the basic analogy that we're trying to make is that the more these two vectors are aligned, the lower the incentive to migrate. And so for example if you're, if the situation is like this where you have the same kind of distribution among the poor between rural and urban, but among the better off, it's more along that diagonal, the effect there is that essentially your likelihood of being, if you're poor moving to an urban area, the likelihood of being poor is exactly the same, you know, kind of abstracting from the fact that it depends on skills and a number of different things, but in a purely kind of statistical manner if you want, right? So this notion that if you move from a rural area to an urban area, you're going to improve your likelihood, the expectation is pretty low. Whereas in a situation like this, in fact the proportion of people that are poor in urban areas is lower. So what matters is this angle theta and the size of, you know, how many poor you have, are distributed between rural and urban areas and similarly in terms of those above the poverty line. So migration as I mentioned just moves you along this diagonal. So it just changes the rural-urban distribution of the poor among the rich if nobody has migrated, then that vector is unchanged. Now what we see is that for example the angle theta, it's kind of the effect of migration if you think of an economic equilibrium perspective in terms of wages, it's narrowing that gap and effectively it becomes less appealing to move to urban areas as the more people you have that are moving there but haven't yet shifted out of poverty. So you're increasing the number of urban poor. And now however in the longer term it's clearly more complicated in that, right? I mean there are dynamics of development that make it more difficult to say exactly what's going to happen. So for example in the longer term you will have that the urban, the people who are above the poverty line in urban areas will in fact absorb the rural people who migrated. So you're moving off of that isopoverty diagonal line here and so you're changing the nature of the incentives to migrate or not. So and here also in the longer term you have another kind of complication which is there's a natural rate of urban increase linked to differentials in mortality rates between rural areas and urban areas. So you have to factor in once you go into the empirical side, you have to factor that in. But essentially you have a component that is due to natural migration, natural growth rates, differentials between urban areas and rural areas and then something that is urbanization due to migration. Now up on the top in summary is what comes out of the model. You can estimate the migration rate as essentially the cross product of those two vectors. And in terms of the magnitude of the size of the population that is below the poverty line, the magnitude of the size of the population above the poverty line H. And then this angle that shows how much they express a difference. The sign of the angle theta, basically that angle, if the angle is zero then it means that the incentive to migrate is zero because your chances of exiting poverty are basically very low. If it's a big angle then it says that really an urban condition gives you that possibility to move out of poverty. So as I said, I mean for this kind of big report that is meant to try and be a little bit global, we needed something where we had data. And in fact for this below income in rural and urban areas and above the poverty line in rural and urban areas, we have that data thanks to EFA who kindly shared it last year for a report that we did and then the angle between those two vector diseases to calculate. So in fact we can think of this measure as an incentive to migrate. And the parameter A, which you can then estimate, represents the propensity to migrate. So let's say a larger L, meaning larger proportional population that is poor, means that they will try to improve their livelihoods potentially by migrating. A larger H implies a certain capacity to absorb people coming in, meaning that there is a breadth of wealth so that you can try and exit poverty, possibly by going to urban areas if the distribution is in that direction. And as I mentioned the theta and the sine of theta means unequal distributions of poor and non-poor between rural areas and urban areas. So at the core I mean this is a little bit, it's a push-pull narrative without saying what's push and what's pull. It's all determined endogenously, it's very basic, but it does capture this nuance of the differentials. So putting real data to the graphical approach, this is data for China. Up here is China in 1990 and so a large part of the population was below the moderate poverty line of $3.10 a day. And a lot of it was a large number that were rural, 70%, 15% were urban. And then you fast forward to 2011, here at the bottom, period of exceptional development in China. You have a much smaller share of poor and still predominantly rural. And I think what's interesting is in fact this incentive to migrate that we calculate in terms of that angle and how it interacts hasn't changed that much. So you've had a lot of migration, but this incentive to migrate has not gone away, right? It hasn't balanced essentially what is the labor market. Now looking at another example, these graphs we can do for 70 countries pretty much, right? And so this is instead the case of India. Similarly, starting in the 1990s, going to 2012, and we see that this differential in terms of distribution between rural and urban areas, poor and non-poor, it's much less stark. So for example, for China, what we see is that huge developments, huge changes, and the incentive to migrate kind of increased even though the rural part of the population was decreasing. And then slightly to decrease later as the population in rural areas is shrinking and the actual base of poverty in China is shrinking. In India, much lower incentive to migrate, we saw that it's less stark the difference between rural and urban areas in terms of distribution of the poor and non-poor. And but similarly, I mean this incentive to migrate goes up and then levels off and goes down, but still above what it was in 1994. And so despite very different development paths that China and India had, this incentive to migrate evolved in a similar way, right? So I think this is an interesting aspect of this measure that there are many ways in which the interaction between rural and urban areas can then affect the incentive to migrate. So but going from the incentives to the actual flows can estimate the propensity to migrate. So based on the fact that we have the data for kind of rural and urban poor and non-poor distribution, we can do that. The propensity to migrate essentially captures some cultural norms, bears to women migrating for educational purposes, the age profile of the population. It's younger people tend to have a higher propensity to migrate. So this is a very kind of preliminary empirical application using UN days of population data and DHS information on fertility. We basically calculated the migrant shares as a share of total population growth that is not due to natural population growth. So accounting for these differentials in growth between rural and urban areas. So this is a kind of very preliminary results. I just showed the results at the kind of continental level. You know, I mean, we can we can go below this, but it's that they're so preliminary that I think there wasn't much point in in presenting a boatload of results. But essentially what we have is for Asian countries and Latin American countries, we have significance in the in the co-efficient of this propensity to migrate different values, being in Latin American countries that has a higher propensity towards towards migration. The R square in Latin America is quite high. Actually, when one does a scatter plot, it's it's quite it is indeed quite quite linear. Part of it is that I mean, the propensity to migrate really only makes sense at the country level. Right. I mean, it's it is. So the issues that you can't really estimate at a country level. But for example, for the Asian countries, if you do just Southeast Asian countries, the R squared is is higher. It's like 0.45. I mean, which considering that you're only regressing on one variable, it's I was surprised. And for Sub-Saharan Africa, I mean, we get what would be I would say a counterintuitive result. I mean, there are two possibilities here. One is, well, you're doing it over a continent doesn't really make sense to talk about propensity to migrate. I mean, West Africa is very different from East Africa is very different from South Africa. So that's one aspect. The other is that you you could also be capturing the fact that being below the poverty line in Africa may also mean being much below the poverty line, meaning that you're in extreme poverty as opposed to another another contents where being below the moderate poverty line, you're still close to that. So you might not have the resources to to migrate. So you have a lower propensity. So clearly, as I said, I mean, the propensity to migrate should be estimated at the country level or at least in homogeneous regions. So we're just starting off the constraint here is that we have all the data of poverty, but we're still working on on kind of sorting out the migration rates by country. So we only have a small subset of countries for now. And also the paper in the paper, we extend the approach to also not just the poverty, but also access to education and health services. So there's a kind of formal derivation where it shows that you can you can do these as independent things in the estimation and independent, independent variables. With or without interaction terms, depending on the assumptions, wherever somebody migrates, somebody migrates for education. If that means that they're not migrating for economic reasons, then you don't have any interaction terms otherwise. You have a bunch of interaction terms that you have to estimate, which can be problematic given the limited size of the data set. So in terms of the advantages, I think I haven't gone into the details, but essentially the the parameters do have a structural relationship to the drivers. I mean, you're trying to separate out the the incentive linked to the economic economic conditions or access to services. Differential between role in urban areas from the actual propensity, which is tied to age profiles and cultural norms. And it tries to capture this kind of push-pull in a bit of a more continuous fashion. So it's in that tradition of the push-pull narrative, but I think it quantifies that in a way that is relatively simple. It can be mapped out for a large number of countries just to see what the incentive to migrate would be. I mean, that number is very easy to calculate once you have the data. As an extension, I mean, this can be extended beyond here. I just segmented into poor and non-poor. One could use quintiles, for example, and see, you can assume that the poorest of the poor don't have the resources to move, but the next up are looking at those in the fourth quintile, and you can still use this approach. It can be extended to thinking about migration from urban areas to rural areas and developed countries. I mean, it could introduce amenities in this approach. In terms of caveats, and here I'm nearly concluding, we have that there are essentially three sources of potential errors in estimating this model. One is while models of mis-specification, in this case, particularly omitted variables, since we just did, you know, very preliminary of what we had. But even if we factor in access to health and access to education, there could still be issues of mis-specification. One aspect is also having to define a threshold to distinguish between advantage and disadvantage in terms of access, be it to services, income, or education. So it may be that the threshold you choose is not reflective of the driver, right, in terms of what makes that migration decision that triggers it. And I think one aspect which is, I think, unresolvable is this notion of disentangling the natural growth rates we can do, but this reclassification of rural areas into urban areas, which is happening at considerable speed in many developing countries, that's something that we can't really address, at least for now. We don't have that capacity, and apparently it is quite important. And that is ending up in our migration estimate, migration number for a variable. And finally, I mean, we assume the propensity to migrate as a fixed parameter to be estimated, but in fact, it may not be stationary, right? So I mean, it can be affected by laws restricting rural urban migration. One can think of the Hukou system in China. Here, I mean, it's quite straightforward to separate out the propensity to migrate from migration costs. So it's something that I'm not too concerned about, but something that we haven't done yet. So to conclude, this is very much work in progress, driven by this need to do a global report on a relatively short time frame, where the focus is intended to be on rural migration. Different aspects, this would be only one small part of the report. We're interested in the feasibility of the approach presented here, this notion of introducing a measure of incentives to migrate, and possible other sources of data to improve the estimation, and suggestions on moving forward. Most welcome. Thank you.