 Thank you very much. So it's a it's a it's working progress. So please don't download anything from the web if you are interested, please rather Write to me and I'll send you some some work, which is a little bit more advanced. I have first Learn better how to use the things Okay, right. So it's about Multidimensional poverty like half of people in this in this conference I guess now it's very popular theme on this one will be a new method based on Priority information about Expenses so there there are many Now measure of multi-dimensional poverty as though the monetary based poverty measure are still the one which are Used in in practice in many cases on what I'm trying to do in these papers to explore An approach which is related to basic needs. There is a very large literature as I said Here are a few papers you can find Scores of all the ones on multi-dimensional poverty and this is a multi-dimensional poverty and this is are calculated by aggregating at the individual level Individual poverty or individual deprivation measures into scores and then aggregating these scores to reach a global poverty index which is called multidimensional this this is a big progress on on poverty analysis in in a sense that this rise of indicators multi-dimensional indicators as as as mobilized much better some information which was neglected on on diverse qualitative deprivation so So that's important to to to study these topics now there are issues with the multi-dimensional poverty and this is The the dimension of welfare the attributes is of often arbitrary On on some time. They are not very well Adapted to the countries which are which are used when you try to calculate them You find that in some countries that is almost nobody poor as though in practice There are poor people and there is a social policy on there and there is social work to do in this country so the choice of of the welfare dimension is an issue some people Discuss the fact that it's possible to aggregate heterogeneous indicators in in in the same indicators There is also the the question of using common variable like income rather than variables which would Directly describe some some welfare attributes or basic needs The needs the needs a measurement of needs is a problem because there is a lot of originating in household needs in in in actual data on in actual situations And on the way which are used to aggregate these various dimensions Are also often arbitrary. There's a huge quantity of papers now coming out with different way of calculating weights to aggregate individual deprivation Into an individual poverty index which would be aggregated for multi-dish national poverty the counting approach which has been developed by by James also is Important, but it has also its own Arbitraryness, so it's it's about trying to relax the the arbitrariness of of these indices by using responses from household to Spending priorities when you ask them To what would you spend a small additional sum of money which is a very easy question to include in a survey? You can use the answer for doing basically three things to identify the relevant deprivation the kind of welfare attributes that you would like to include in your analysis you can use also this answer to identify the poor as By defining them as people who are deprived in the top priorities instead of the orange of welfare attributes and show What is the problem to use all the welfare attributes and you can use also this priority data to Aggregate for each household The various deprivation and disease So that's that's the idea of this paper in in progress So it has many advantages first you you don't need to use common variables because you are going to be able to work with Stated needs or stated deprivation instead of necessarily income or the variables You can avoid the arbitrariness which arises in many stages of the construction of of this Poverty and this is or inequality and this is whatever you want to to study And it's also somehow easier empirically to elicit deprivation by asking people About their expenditure priorities rather than having very vague on general questions and By concentrating on the deprivation on safe deprivation information you can also avoid Many issues with needs a heterogeneity, right? So there are I'm not going to show any any any formula There are almost almost none But there are deprivation and this is they are called the eye for an individual eye and this is a vector So you have the I ji which is a deprivation in the in the attribute J Suffered by individual I and in the the literature is dominated by two approaches of of Identifying the poor the one is the intersection approach where the the poor person is is the person who is poor in all the selected welfare dimension and that leads empirically to Estimates of poverty which are ridiculously Small sometimes zero on the union approach Somebody is poor is defined as poor when in at least one of the selected dimension The person is is is deprived and this leads to exaggerated Estimate of of poverty often you have with with with the messy data that we use Half of the population who is considered as as poor in in even in in well of countries. Okay, so there are there are ways to to try to Go over this this this question for example using the count in deprivation like in In James work, right? so What I'm working on at the moment is how to develop an axiomatics for this kind of indicators Multi-dimensional poverty indicators, and you have series of axioms That many of you are familiar with some of these axioms you are familiar with you have technical axioms continuity normalization scale Invariance derivability population Invariant etc. These are more technical axioms to help you with with with with a mathematics or than than rather Normative content you have the decomposition axioms, which are sometimes replaced by paratome axioms, which will deliver the same kind of linear formula of kernel functions You have focus axioms which embody if you like what you Say when you are talking about poverty the fact that they are poor or poor people who are poor and people who are not poor and these axioms can be more or less strong in a multi-dimensional setting and you have transfer and correlation axiom Correlation between the different attributes of welfare, which tells you about some are awesome Subjacent utility properties of utility so There would be some convexity or substitutability or complementarity notion including in these actions and The the the three series of axioms the composition focus on transfer on correlation axioms are somehow I Would not say contradictory but difficult to reconcile because they they don't Proceed from the same Philosophy in a sense the transfer on correlation axioms. It's more They are more often inspired from a representation of individual utility properties On the other axioms are Building some constraint into utility consistent representation of poverty in in in in that case so there are already some difficulties in Reconciling these different kind of axioms on the axioms that I'm going to Add in this sitting are axioms to incorporate the information on priorities So axioms for the specification of the welfare attributes today. I'm just going to say okay The welfare attributes they are the one that the people mentioned in the answer section of the top priorities for identifying the poor I'm just going today to say okay Let's just count the number of top priorities and see what they are on aggregation of the deprivation to generate weights When you aggregate deprivation And this is at the Individual or at the at the global devour on the earth you see I just look at simple cases But each of these topics can be more or less technical or abstract depending at the level of generality You want to you want to place yourself? Right, so I Need to save time So identifying the poor the first thing to understand is that you don't need to use the same dimension to identify the pure and to measure the Intensity of poverty for somebody who has been identified as a poor Okay, and that is going to have some advantages in Generating some more interesting estimators to Know how many the deprivation you need to use for identifying the poor If you have data on priorities of people you can use this data Sometimes it's easy to see in the data or sometimes you can do some econometrics in this paper I'm going to use some truncated con data models on priorities on which it would be possible to Estimate some expected number of priorities and that would give you the kind of Criterion that people use in the in the count approach where they have to choose a threshold for the number of of Dimensions, okay, so it's possible to use these data Associated with some axioms to identify the the poor in a more Varied on efficient way. It's also possible to Generate non-arbitrary way. It's perhaps what I can do is to skip over the older slides because I'm sure I won't have much time and move Okay, there are axioms axioms. Let's look at simple pragmatic indicators In Seychelles, I'm going to show a few number for Seychelles. I tell you in advance in Seychelles I found that you just needed to look at two dimensions really to to to to describe the The the poor on their shelter and food in Seychelles the most important dimension on Poverty incidents in that case would be the union criterion except that this criterion is Not a priori. It's extracted from The data I didn't know in advance that food on shelter where the the important dimension before to look at the priority Data in in in this country. So now you can do a little bit more than that. You can if you want to build Amount of poverty indicators or multi-dimensional poverty indicators you can keep you see on the first line The formula to identify the poor with a union criterion based on these priorities Which has been considered as the top priorities by people and then you can complete it for example here by multiplying by a sum of weighted deprivation Indicators where the weights can be extracted from the priority data again So it's possible you see with these techniques once you introduce this kind of information to Generate a range of of indicators Which would Play the role that you you want them to play when you they want to define a multi-dimensional poverty here for example and a simple example of weight is the proportion of people who stated that the weight for the jth attribute is The proportion of people who stated that this is their first priority. This is a simple thing. Okay, let's forget about Math And it's not really so I'd rather skip it empirical application sechelle sechelle So it's technically considered as an African countries, but it's in a Indian Ocean and it's a middle-income country which is at the moment moving from a plan the economy to market-based economy with macro stabilization plan a few years ago structural reforms and the very Generous system of social transfer has been and is being Reform on on cut down Dramatically, so there is really a social problem and the poor are suffering a lot in in in this country So it's a good idea to pay attention to them at that time So I did a new survey new kind of survey in this country, which I called the living condition survey in which The household where what been surveyed in a household budget had been resurveyed and I was able to ask them additional question subjective Information that you find in many survey about their subject in being sold out But also also some simple question on spending priorities Let's okay. Thanks. Murray. So so okay. There are poor people in in in sechelle I estimated the poverty monetary rate at 17% where the poverty line is Reflects the opinion of people on the subsistence minima in sechelle So that's for monetary poverty with methods which are rather Standard and well-known and there are other needs here are the priorities Priorities when you ask people, what is your first priority? We give you a little bit more money. What do you spend it on? What do we learn by looking at that? Well, we find that food on shelter are More important, but there are other priorities water and electricity you see health And you have things which are not basic needs, you know holiday Okay, some people spend on holiday if you look at others you have a lot of people who when they have some additional money They are going to give it to somebody else. So it's very nice This is the kind of data that you can get in the in this question So five significant basic needs shelter food electricity water health education, which I didn't know clearly as though I've done the Estimation on on poverty shelter. I didn't know that shelter was so Important in the mind of people if you use the union criterion 42 percent of poor household in sechelle So clearly it's it doesn't make sense and it's not useful for social policy. You are not going to help half of the population So so that would not work and then you can cut down And if you if you cut down if you keep only the two main priorities and I'll give some argument why to keep the two main priority dimensions You end up only with eight percent People who are poor in the country You can do some econometrics, so it's very preliminary of to build my own estimator for this for this kind of data But already there the dependent variable is the number of priorities which have been stated by household Or you can regress on various Variable this is important because you want to condition There is no reason why the number of top priorities and on these top priorities should be the same for everywhere You could have different top priorities for rural and urban areas for example, so Conditioning having a model to condition is going to to to help you and from this kind of model you can look at expected number of priorities or If you want expected number of priorities for the monetary pool or for the real people or for different categories On what you find in Seychelles? It's the top line, which is about Let's just do one piece of poverty analysis with this kind of method You see I've not yet mobilized the very complicated Indicators which are likely to come out from from from the axiomatic analysis. This is at the moment very pragmatic Stuff and we have here a table where we have the it's it's a kind of poverty profile table You know you have the education of the the head on you look at the poverty poverty estimates So the the right column is the incidence of monetary poverty the usual one at count index What you find in that? it is Clearly decreasing with education what you find in almost all countries If you do this kind of table in almost all countries the level of monetary poverty decreases with the education of the head and Decreases a lot now with Multidimensional poverty the two columns are two different and this is a multi-dimensional poverty. You don't find it You don't find it so Where is that it because? The multidimensional poor in Seychelles in that sense they are the shelter food poor and if they are the shelter food poor even if you have a certain amount of Introduction of the other dimension of poverty this this focusing Coming from the data itself is going to dominate the analysis and they tell you that people who have low level of education No schooling primary education. They are less poor than people who have secondary education Why is that it's because in Seychelles you have social programs which give thanks Murray which give free accommodation To the people who have no secondary education So you're able to understand things that you could not see with the monetary poverty Analysis of course there are two different the notion of poverty and none of them is right or wrong but There is a potential to to to find some interesting features by using These approaches so that's all that's very preliminary So the idea is to use a Expense priority to improve on the methodology of multi-dimensional poverty and to to remove some arbitrary Choices in the in the in the in the current methods You need a special server, but it's really not costly and I've still to finish the estimation on the on the axiomatics on these people. Thank you. So any comment welcome