 Thanks for the organizers of the conference for inviting me here, as I said earlier my first time here was in 2001 as a PhD student, so my hair was darker. That's the main difference and thanks for the organizers of the panel for such a great amount of work that I that I read and that that I have to comment on so Just a little bit on my background right now. I come from a research center that standardizes household survey data for that in America with the World Bank and other institutions and 15 years ago we used to do all these micro simulation models the bourguignon Ferrera Lustig type of model. So this is really literature I'm familiar with and the Argentine government had for some time their capacity to do benefit incidents analysis a bit like the CQ approach a bit of Nora But that capacity unfortunately was destroyed a few years ago in the previous government when we also had the Intervention falsification of statistics. So right now I am in the government and I Do a lot of things but my main pet project was to reinstate that capacity and so we're gonna be borrowing A lot and I mean I think we don't call it copying in in academia, so But definitely I I relate to a lot of the issues you raised because I've been doing stuff on on family allowances And we have the problem of the random seed and what we want to have etc. So all these I relate but Really, I think these whole Euro mode and and children Models are are much more organized and what we did and and we're really Looking at what you are doing and incorporating your work. So thank you for that So in terms of praising the asterisk mod model, I'm gonna explain that in a minute I think having a unified and coherent framework for analyzing taxes and benefit systems and comparing them between countries and Between time within countries is amazing. I think it's a great For descriptive analysis and also for exploring policy options And I think that the case you presented is is the perfect example of why we want to have something like this And and then I was over lunch I was I was chatting with with dr. Levy of IDB and I think just just having as It's described on your paper having people from the social that I mean It's another name but from social development ministries and and and treasure ministries sitting together and looking at and tax authorities and looking at Taxes and benefits all together is I don't know how much this project is but that's worth It's many times worth it. It's it's costs. It's that's great And and it's very uncommon in developing countries. We did a tax reform a year ago in Argentina and it was very hard for We were a bunch of academic nerds trying to say no But if you change the income tax you also have to change the tax allowance because the bracket overlapped Etc. And it's very hard for policy makers in developing countries at least or middle-income countries to incorporate that and so What I read in your your testimony there is Something we should really have You should really show I think as a great achievement of a project beyond the the very interesting results So I had to to think about wrapping this up and and and all of these were very different papers in a sense and so Pia prompted me to try to to to to wrap it up and I'm very thankful for that So there's a euro mod. I call them asterisks mod because there's Latin mod and south mod etc. But it's a family of mod models What do we have in in this well what we've seen in this session is a series of outputs So the benchmark is the six models from African countries. That's something we we look in comparison We also had the counterfactual analysis that the Ecuador versus Columbia match in terms of social protection policies, etc Which I think is a very Very interesting and and original analysis and my favorite. I'm sorry. I shouldn't be picking favorites But I relate a lot to your efforts Applying this to the implementation of actual policy changes In Zambia and I'm going to talk about the the Zambian case because I think it's it's it gives us a lot of food for thought And in the way it's done and it's I think it's the way this kind of analysis should be done In terms of analysis analyzing the impact But also the options for financing and also the second round effects of the options because the taxes are not neutral So usually if we have something in developing countries, we have some kind of Butcher's exercise so we have a million kids and three dollars each and it's three million dollars and so these analysis I think was really useful And I'm very glad that I that I had to see it and then we have another layer Which is the layer of the inputs, right? And does the old joke about making sausages and looking at the process, etc But we have to look at the process and look at what goes into the machine I think in in fortunately for us in in in middle income Lower upper middle income countries in Latin America. We have We rely a lot on income data and we don't have as much consumption or expenditure data And so for these exercises, it's an advantage for measuring poverty. It's much much worse But so this is how I made sense of this. Is that okay? It's a plus I've been practicing a lot of PowerPoint skills. It's I'm very I'm very proud of it One thing that that that I shouldn't that I also want to highlight this is from from lectures in public economics by Rush Chetty It's how in main economic journals we've been There's been a growth in the use of admin data and I think What the last paper in fact I had a South Africa version, but it's it's still the same Is a very good example of how we can exploit and use more this administrative data. So I'm going Paper by payroll shameless plug. I am now an editor of journal of economic inequality I know I don't know if you've heard of it. Your submissions are welcome. I Had to say So looking at these individual papers The South Africa I'm sorry called it South Africa because I had mostly the South Africa exercise I think has a very comprehensive and careful analysis of several Alternative data sources and there are many insights there for one on the reporting measurement error Missing data imputation And how to deal with them like consistency adjustments grossing up also the tax collection the number of taxpayers which is also a kind of a difficult balancing act but I think at the base is a combination of Serving and admin data and it's something that we need to work more on and so that's why I think this work is is welcome We did this for Argentina. This is we took the the main house of survey and also data from like person tiles of Formal income and pensions etc. So we corrected for under development just to show you the left This is for formal salaried workers. So the whole distribution shifts to the right This is For beneficiaries of family allowances We have to impute and also we randomize who receives this because we have an under reporting in a survey And so we run it 50 times or 20 or 30 we do all the same things and this is the result We inequality increases a lot that the survey really on the reports income for higher incomes But this is just my favorite graph What we do is we try to see the re-ranking of households When we do the whole correction and so what we have here is which households Move up in the percentile the ones that move down and the ones that stay the same So that that that gives us a flavor of what kind of Changes are going on on the six African countries paper. I know For I mean for a living for 10 years. I harmonized household survey data. So and that's already hard So harmonizing the data and tax system models is quite an accomplishment. I I'm humbled As I said income versus consumption expenditure is not such an issue for us, but a General comment on all these papers Is that I think we should clarify a bit more Which part of the exercises are Accounting with no disrespect my father was an accountant. I have no problem with accounting at all and and what is micro simulation in the sense of incorporating? behavioral changes because when we simulate Large changes for marginal changes. There's no problem But when we change a lot income taxes when we change transfers, etc. You might expect some kind of non-trivial Behavioral response and I think that's something that that maybe should Should be included in the future not in this one, but in the whole family of Southmond Models in terms of assumptions and robustness for two of the countries All of the employee all of the sorry employees are assumed to be formal if I got this right and that that might introduce some some bias there maybe on what you've shown so I Don't want 56 tables by quintile by six countries That's too much, but maybe having some robustness to those assumptions. It would be useful and perhaps using more Figures crafts charts, etc. So for the work we did in Argentina. We chose not to go the Euro mode route We didn't go the CQ lucid route. We just did it on our own, but we have basically the same we start with a Market income adjusted and net transfers Direct taxes indirect taxes all of these rings of bell and then we have the deciles genie, etc so We tried to show the the whole effect of the state action and Here we have Taxes and social insurance contributions all the transfers all the direct tax indirect taxes that set all the And all the goods what we have and I think that's something we have the luxury to have because we're working With a lot of detail into one country is that we monetized health all education levels and also we included a lot of fuel and energy subsidies which I'm not sure in the African the six African countries as you say you have some limitations in what what you can monetize or not but at least in Argentina these are very Very important items in terms of government expenditure and So these are just taxes by diesel and expenditure by diesel on the Colombian Ecuador paper is it's a very Interesting exercise this counterfactual. It's very well published already in a very prestigious journal So I don't have much to add in that sense, but I really like that the composition exercise and I think what would you suggest it? That maybe you can do more of these the composition exercises with the other south mud models is Really great if you can build these into the capabilities of the models. How much I have Okay, good Now Colombia is richer than Ecuador so There's an issue of absolute versus Relative distribution and the level of benefits, but but I think that's already addressed in the paper my only point here is that Taking the the social insurance and tax and benefit system of Ecuador and throwing it across the the the border To Colombia will imply non-marginal changes In in people's behavior, and so that that might have an effect there Finally the micros amont so our I'm a bit Envious here that our next step is to build a micro simulation model for 14 and this this is exactly The type of analysis that we try to to provide to senior ministers and then the president when we have to make changes to social To cash transfer programs So here we have the expansion of a social protection program that looks Into its direct distributional impact, but also That's we have a treasured persons here We look at the cost and financing issues and also to the second round Distributional consequences of these taxes, okay, which are might be again non-trivial And so that's why I think this this idea of the joint work between social protection tax authorities and tax and expenditure authorities and and the joint assessments of both tracks is really a lot of Value added here for for policymaking or for the analysis of this type of Initiatives so Again Behavioral reactions there's an issue of the amounts if during dollars or not that I had never my ignorance But I had never heard of the currency. So I I didn't have a sense of how much that was There's only one point as you know perfectly well the devil is usually in the implementation detail, okay, so I was shocked to see that children zero to two were not covered and I think this is the main The my main take home of these papers how to cover them and the impact that would have now It's very easy for us with the whole survey to simulate how How conditions would change if we give this group some money It's not that easy for the social protection or social insurance authorities to reach them and and ensure that Coverage etc. So that's that's where I think We need to be a bit more careful when we say what the impact would be maybe some assumptions about imperfect Takeover incomplete take not take over take up. Thank you take over. I want to take over Zambia now that they have this model and Then the tax and financing of proposals section was very short, but you talked more about it So probably you're going it's it's in the making So keep on these excellent work where we all benefit from it There's one bullet point which is not here. Which is but is what you mentioned Confidence intervals. We need a bit more on that. We need we need to show I mean, we know how how much we we might be we might be wrong here So using more admin data Having more consistency work between admin and survey data robustness of results to alternative assumptions Adding as much as we can in terms of in-kind service provision and Next step probably agents behavior reactions for major and non non-marginal policies changes. Thank you very much