 Good morning, Michael. Thank you for being here for us to have this really interesting conversation. I think we are really interested about what you have to tell us. And I would like to start linking our conversation with the conversation Pia had yesterday with Oli. They were discussing about how there was this translation from European countries in the micro simulation models to developing countries. And I would like to ask you how was this process with South Africa? How was the idea of creating the Samod and how was the discussions were involved? How was the relation with the government to collect data, etc.? Sure. Okay. Yes. And I was at the session with Oli so I can continue to make connections. I think Samod specifically really arose out of the commission that was set up by the African National Congress, the new South African government, the post-democratic South African government to review social security. And it was chaired by a woman called Vivian Taylor and contained various experts in social security from within South Africa. But they were also looking for international experts. And as I'm half South African, and at least I have a dual nationality of South African and English and had been working in South Africa, albeit briefly at that point, Vivian approached me to be one of the panel of international experts. And so that involved discussing the future of social security basically in the country to move away from a racialized social security system under the apartheid government to a deracialized system. And I recognized that I was not the only international expert that could come from the UK. So I brought in a number of other people, including Tony Atkinson and John Hills. And I was at Oxford at the time, Oxford University, and I actually brought the Taylor Commission to the UK. And we had many discussions there and continued those discussions and deliberations in South Africa. And when the Taylor Commission wound down, the National Department of Social Development, which is responsible for social security, was charged with taking forward the recommendations and developing a research agenda in order to work out the best way of moving social security in the country forward. And the then head of social security, the person called Fazir A. Macawani approached us at Oxford, and we started a dialogue with Fazir A. and later with the DG, Busse, Madeleine Saylor, in developing a whole program of work, looking at social security, poverty and inequality in the country. And that program of work we set about getting funding for from the British government from Difford, Southern Africa actually. And by 2003, I guess, we'd had negotiated a partnership between the National Department of Social Development, ourselves at Oxford, and colleagues from the University of Quasuna Natal in South Africa, a seven-year multi-million pound program. So I had many, many projects in it. But one of the projects that seemed like a no-brainer to us was to have some kind of tax-benefit micro-simulation model on the agenda. And there were many, many projects, as you can imagine, over a seven-year multi-million pound program that we at Oxford were responsible for even. And the micro-simulation project was kind of not one of the earlier projects. I think it actually kicked in, I'm thinking the timeline, 2006, I guess, really, because the actual program of work was signed off in October 2003. And then about three years in, we started to look at the tax-benefit micro-simulation modeling. And kind of we thought briefly in terms of constructing our own hard-wired model, either using some programming language or other or stator or something like that. But, you know, I'd got a history going back and knowing Holly and being in touch with well, PolyMod, first of all, which was predecessor to EuroMod when Holly was at Cambridge. And then Holly and I had a longer, so to speak, fireside chat about micro-simulation when we were both of us examining a doctoral thesis in 1998. So I got kind of like the quite into micro-simulation was in my thoughts at that time, although I had been dabbling earlier myself in it. But so anyway, I did approach Holly, and I'm sure it was not by a fireside, but probably almost certainly over a cup of coffee somewhere, because we bumped into each other quite frequently in terms of other meetings and so forth. And so I sounded her out as to what she thought would be the possibility of taking EuroMod beyond the countries in Europe, because the software had never been used outside of Europe. And she thought, well, let's give it a go. So basically in partnership, by this time, Holly had gone and moved to Essex. And I think Holly and colleagues at Essex and ourselves then set about constructing SA-Mod. And I suppose it took, just thinking of the timeline, I suppose it took a year, more than almost a couple of years really, to get it all together. Because there were big challenges on the data side. And there are also big challenges on the modeling stuff side, as far as we were concerned, because we were not really novices, not not, of course, for Holly and team. But the data particularly was an issue. We thought that South Africa would be quite a good country anyway as a developing country to start with. Because of its apartheid history, it had got a well-developed civil service, it had got a well-developed statistics bureau in the form of statistics of Africa, and relatively well-funded compared to other countries in sub-Saharan Africa. But even so, though the data were really quite high quality, they didn't ask the kinds of questions that you needed necessarily for a tax benefit micro simulation model. So we had to kind of cobble together a data set initially, which involved the income and expenditure survey, but also the labour force survey. We had to do kind of things like propensity score matching and so forth to get a single data set. So it took longer. And as Holly said yesterday, data prep is all really the quality of the data. The model is only as good as the data that underpins it. So we did a lot of work on the data. And it went through an iteration of 1.0. And then I think we managed to get to the point where 1.1, which we felt happy with, was launched in January 2008 in South Africa. And we had a kind of seminar with key academics and so forth. But Holly and I also, because Holly came to South Africa for this, Holly and I went to Parliament and presented it to what's called the social cluster of directors general. And director generals are the kind of head honchos of each department. They're political appointees, but they're not ministers, but they're the head of each department. The social cluster covers not just social development, which is social security, but also health, education, someone from the Office of the Presidency that's responsible for social sector, someone from the Treasury responsible for social sector, etc. And that went well, fortunately. So we've got quite a lot of buying from the South African government at an early stage. And because this was all a partnership with the South African government, this whole program, which was called actually SAKED, which is Strengthening Analytical Capacity for Evidence-Based Decision Making. So it's quite a well-known program within South Africa. And it actually was signed off by the Treasury, National Treasury. So everyone knew about SAKED and was watching for good outputs. But yeah, so we got buy-in really from the social cluster, which is a big deal, because it does mean that it had got credibility at an early stage, I guess. And then an enormous support from within the department, particularly by this time, the Director-General was who, Madam Saylor, just supported it. Zola Squire, the Minister for Social Development, who was a far-thinking and very, very eminent member of the AMC, supported it wholeheartedly. So we got kind of high-level support and buy-in. And by 2013-14, I guess, we got to the point where we looking elsewhere, really. So we were actually approached by UNICEF Namibia to produce a model for Namibia, which became Namot. Again, we did it with Essex. Namibia's infrastructure is much less good than South Africa, so it felt like a different kind of challenge. And then I think we were talking about moving beyond Namibia to the Southern African Development community, which is, I think, 15, 17 countries across Southern Africa. And we'd already ticked off South Africa and Namibia. So Essex and our sales, Holly and our sales, basically, we're talking then about, and she referred to this, about seeking funding to produce a SADIC mod, so to speak. Yeah. And in fact, then along came UNIWIDER. So Yuka approached Holly and the rest is history. But he did also ask me how I got involved in all of this. And that's, I hope, might be an interesting story for people. In that, I've always been interested in computers and computer software because my first degree was in science, biology, actually. But then after that, I became a lawyer. And towards the end of my stint as a lawyer, in the 1970s, basically, I became a welfare rights lawyer and became a welfare rights lawyer that was then kind of affiliated to a community project. Sorry, I was a welfare rights lawyer in a community project, which was affiliated to the University of Oxford. And I was actually employed by Oxford. And as part of that, I started with a colleague a sociologist at Oxford, developing a benefit calculator for individual claimants so that you could, a computer program at the time, in 1981, 82, 83, time when computers were not very popular, there weren't many home computers around. It wasn't the mainframe. But we developed a model that would calculate an individual level benefit. So it kind of was like, if you will, the benefit policies in a Euro mode, South Mod regime. So it's an arithmetic calculator, basically. And then, leading from that, we set up at the university, a welfare, what do we call it? I think welfare benefits research group. And we were particularly concerned that there was a massive social security reforms happening in the UK, which were to be implemented in 1988. So we were concerned to have been a lot of press about how people would lose as a result of these reforms, claimants. So we set about trying to model who would be gainers and who would be losers in certain aspects of that system, using administrative data. So my very first micro simulation was not a micro simulation that used servo data, but was one that used administrative data. So what we were able to do with initially with a couple of local authorities, Oxford City Council and Oldham Borough Council, Oldham is in the Midlands, Oxford is technically in the Southeast. And we managed to get extracts in 1987, eight time of the whole of their computerised housing benefits and council tax benefits systems. And we got them on big mag tapes and carried them to the computing centre and they were anonymised records, but they were individual records, which had information on people's earnings, other sources of income, etc. and rent, mortgage, whatever. And enabled us to model, first of all, changes in housing benefit, but also changes eligibility for other benefits, such as family credit, which was an in-work benefit. So we used the entire databases and constructed a model in a statistical analysis package called PSTAT. PSTAT probably still exists, but stands for Princeton Statistical Package or whatever. But PSTAT was the, what our computing service called a thinking person's SPSS, but SPSS, I don't think PSTAT did. So there clearly weren't many buyers for that particular slogan. Anyway, we did it in PSTAT and we did it on terminal, we didn't do it with screens like we do things now. We had a teletype terminal, we had to type in the commands and it printed out, and you had to write your whole program without being able to see it on the screen. I tell you, modeling like that at Hollywood have told a similar story. This mainframe modeling is not fun, but we were able, because we could only do it on a mainframe. PCs weren't capable of handling tens of thousands of records, which is what we were looking at in those days. They had like, not even a megabyte RAM, they used to have like 250K, 56K of RAM was a bigger computer. You certainly couldn't process these administrative data sets and so it had to be done in that way. So that was really how I got really thinking about micro-simulation back in the 80s, but not really branding it as micro-simulation, if you know what I mean. So I think that's really the whole story. Yeah, this is really interesting. I think it's completely different from how most of the people end up in micro-simulation models. And also the history about how Samod was created, because I did my homework before this session, and when you search for Samod, you find documents from 2007, but it seems that the discussion started many years before. So it's really interesting to listen from you about it. So while our audience is formulating the questions, I would like to ask you another question. Can you tell us a little bit about the relation between the activities developed in the Samod with the real practice? I mean, what I mean is how the activities the Samod development become policies or which policies were requested to be simulated and in fact were implemented or the discussion went ahead. Okay, yeah. Well, it's very hard to actually point to a particular simulation and say, no, that actually became policy because it hasn't quite worked out like that. But it's also true that probably most of the simulations that we've undertaken with Samod have been in a context that's been request from government as part of a reserve project that's been either separately commissioned by the South African government, one part or other of it, or has been commissioned indirectly through a research program that the government departments had or whatever. So we've done that kind of thing and I can talk about some of those simulations. It's also been used in a much more routine way, which is to estimate budgets and even forecast. So regularly, the South African Social Security Agency, Sasa, who's responsible for paying out grants, needs to know how much money they're going to need to pay out in years to come. So actually the basic model, we use it and re-weight the output regularly for Sasa to give projections. And so it's kind of used in that routine way and for in-house expirations, etc. But there have been a whole range of specific requests from government, starting really with sort of one-off specific benefits that didn't exist type requests, just I think testing the water like a youth benefit. Because the social security system that was around in the early 2000s after Taylor Commission had a big hole in it, which is basically that children were progressively covered, older people were covered from 60 onwards, people with a disability were covered, but non-disabled people from 18 to 59 weren't covered at all. So there's some earlier explorations as to kind of unemployment grants for such people. We also at the beginning, even child support grant, which is a child grant, didn't go right up to 18. So one of our early simulations, which is kind of fairly directly relevant to a policy change, was actually modeling, I think we got involved when they were wanting to extend it. It started off by being just powerful up to age six and then it got extended to naught 15 year olds and then 16 and 17 year olds were the last group and that by that time we got our model. So we were actually asked before that 60 and 17 year old extension to child support grant, we were asked to model that. And subsequently on child support grant, we've done work for DSD modeling, converting it from a means testing grant to a universal grant and how to fund it through the tax system. For the National Treasury and Department of Social Development, we've been involved in modeling a universal old age grant because it's again the current one still means tested. And yeah, and we carry grants and so forth. But most recently, I think we've been involved in work with DSD, looking at the social relief of distress COVID grant, and also the possibility of a basic income grant because the whole thing about the pandemic, even though way back in 2000, the Taylor Commission recommended a basic income grant for South Africa. And Zola Square was always incredibly supportive of that. He with our support would never get it through Parliament. But it's now actually on the agenda and we're doing work currently with the South African government on that. Really interesting. Yeah, I think I would like to hear about it because one of the great benefits that we receive from being academics is that we are not related to the policymaking. And I think it is the work in the micro simulation is really related to the policymaking and to this reward. So we still have some, we have still have two, three minutes and I will ask a question Pia wrote in the chat. Basically, what Pia is saying is that you have a big motivation of working with young scholars and government officials. So she basically asked about how these relations of engaging with young scholars change with time in some way. How does engagement with young scholars happen? And change over time. Sorry, I didn't. And also change over time. So how this happened and how changed with time? Well, initially the model was very much an in-house model really for the Department of Social Development. It really didn't involve more open access. We however didn't negotiate with the Department of Social Development that we needed to get it out there more. And providing it was used by educational institutions or not-for-profit institutions, then I think we were all, because it was kind of a jointly owned model, S6 ourselves and Department of Social Development were really happy for it to go out there. And so we have in our various, as a non-academic institution, we haven't had direct scholars ourselves, but we've kind of intersected with other universities, particularly in South Africa, who had access to the model and supported them indirectly rather than directly. But we've always, the one thing we have always insisted though is that people have some kind of training in Euro-mod or specifically SA-mod before they can run away and play with the model simply because we don't want it to, you know, force claims to be made on the basis of people who are inadequately trained. Okay. Thank you very much, Michael. I think this was a very interesting talk. I'm really glad that this was recorded, so people will be able to watch this later. Yeah, so thank you very much. Thank you, everyone.