 Thank you to Wyder for both the collaboration with ICTD and for organising this session. I'll talk a bit about this database and I think Mick will say a bit more afterwards. To go back a step, I'm going to take some credit and then I'm going to disassociate myself. Ymlaen am this data, can I come from my frustrations in about 2003, 45, trying to start doing work on tax, and finding that none of the sources from the world development indicators to the government financial statistics had the kind of coverage or quality of data that you felt you could do much with. Rwy'n gwisig os ychydig ychydig ar arnod dyfodwych o ynag'ch gweithioio a'r gweithio iawn, gan gydigio i gydigio 100%. Mae rytwm i'r gweldaddoddyn nhw ymddull yn achos y data rwy'n ddeall. Nid yw'r lightwch itit ag i'r eich gweithio a'r gweithio i'r registerdd. A rydw i'n gweithio ei chylet tiesiol. Dyn ni'n olygu yn y gweithio peirloedd yn y gweithio, yn y cyfreifol gynnal i'r gweithio i'r gweithio, yn unig gyfedd yn y gweithio. ac yn du cyflwyno i gyflwyno â'r rhai cylliddiadu ac bod y cyflopod CTID yn ei bapio yn ymwinell ar y prôl okaeth ymyryd ym ein gafor, gyda eu gastryd ymddangodol. Y cyflopod CTID yn ymddangodol i gyflwno ac mae'r ganddwyd wedi'u fod ymarfer ymddangodol ac ymddangodol Ysbryd Sen cylliddoedd yw ymddangodol cyflwno i gyflwyno i'r cyflopod CTID felly mae'r credit yn ymdweud â hyn. Ond Cyle Mcnab, y cyfnodd, yn ymdweud â'r ysgrifennu ac yn ymdweud â'r cwestiynau cyfnodd. Rwy'n gweithio i ddim yn gwneud ymdweud â'r cyfnodd, felly rwyf i gweithio'r cyfnodd ac mae'r cyfnodd yn gweithio'r cyfnodd a'r cyfnodd yn gweithio'r cyfnodd ac mae'r cyfnodd yn gweithio'r cyfnodd. Mae'r gweithio'r gweithio'r cyfnodd yn ardal o blod i fy n nagyon ac mae'n sg complain am y f lleistoedd mewn ddech arriveeth a ladrig o dyfynig o battu'i llond cael ei subod y dtyll written i bob tunnwyr ac efallai mae'r llyth wedi'u gwaith mae'r cyr supporter a'r Cyfnodd maen nhw yw stodd. Ac maen nhw'n com ond wedi fath gwn ni, hwn i nifer oed y rhy nesteul conflicts combineidgood o'r cyrnteil rydyn ni wir, ac rydyn ni'n cael video edrych ar bobl a rhywbeth ar wahyddoedd. The question is, they are looking at, but it is not necessarily transparently, and the decision that the researchers make is not necessarily in the public domain. So, their use is somewhat limited to use them as the basis going forward for replication and other research. Cymru this is a list of problems which I'm not going to spend my time going through. It's the kind of things that you would imagine, mainly coverage and comparability. But the particular issues around natural resource revenues and around GDPs series, different tax data sets tend to have to rely on different GDP series. Sometimes, they change GDP series within individual countries within the dataset, and this can introduce very large jumps if it's not transparent that it's happened or if people deal with rebasing of GDP in a bad way, then the ratios can go all over the place in a way that doesn't reflect what was actually happening. That's a problem that you find time, and again, Tony very kindly said he'd heard that this data set is better than the IMFs. I'm afraid that's not setting a high bar. The IMF data set has a series of issues particularly around this GDP series question where there are just jumps that nobody could reasonably look at and think this is what happened in this country, which is the result of just mechanistically applying an approach without thinking about what the numbers mean. To give just one example, this is a set of different series that we put together for Ghana and this is just taxed to GDP, using the GDP sources from the various series themselves. Keenan Mansour, which is researchers in the IMF's fiscal affairs department, the government financial statistics, the IMF's CR, I can't remember what that stands for, and the world development indicators. If you think about those individual series on their own, you're probably okay. If you start combining them, you've got a problem, and if you have a mechanistic approach that takes from one if it's there and if not takes from another without considering what that's going to do, then you've got a real problem and you're going to have jumps in the data that just don't make any sense. If you put in a common data, a common GDP series, and actually you feel quite reassured that all these data pretty much line up, not perfectly and not every time, and there are some remaining inexplicable issues in some of the data that I'll mention a bit later, but by and large you deal with a lot of issues just by having a common GDP series, something that seems a bit obvious but just isn't done in a consistent way. All right, so what do we do in the government revenue dataset, the GRD? The extended revenue classification, just to give you a sense of that, these pictures show the different revenue classifications that different sources use, and even at that distance you can see these are just completely different animals, so imposing a single classification is kind of important to get towards some kind of consistency and comparability. Other than combining existing sources, the bigger thing we do is to add in article 4 data that allows us to fill in a lot of gaps. The remaining issues are particularly around natural resources, social contributions, how you address federal states where things get different, but the GDP series is a big one. Finally, we did try at different times doing mechanistic approaches saying if this data exists take this, if not take this, if not this, and it just doesn't work. You don't manually go through each series for each country, you end up with things that just don't make sense, so there's no getting away from that. Now in some ways it would be cleaner to have a mechanical approach, so what we do is make sure that in the metadata every decision that's taken is expressed as clearly and transparently as possible, and if people want to take different decisions, obviously they're able to do that, but at least you're certain about where each individual data point comes from and what choice has been made. All right, there's the structure, which I will skip by. This just gives you a sense of how much broader the coverage is compared to the fiscal affairs data set in the IMF, the article 4 data on its own, the government financial statistics and the world development indicators. The government revenue data set is better than bad. I would say it's a lot better than bad, and I think it is by some distance better than any alternatives, but we also need to recognize that there are still serious issues, limitations. We don't originate any new data, so there is still significant missing data that just isn't in any of the sources we've tapped. There are some ways that some of that may be addressed over the next few years, but it will remain the case that where data doesn't exist there isn't going to be anything in the series. There are still questions with resource revenues, which are often particularly badly recorded in terms of whether they're taxed or not taxed and whether they're separated out at all, which can introduce real problems for particular countries. As I said, the variation across sources is often inexplicable. Sometimes we just can't see what it is that has led one series to have one number and another series to have another number, and the metadata around some of the series is insufficiently good to allow you to dig further into that, in some cases, there will just be things that we'll never know. Although, again, as people use this more and more, we are starting to get feedback that often highlights particular errors in one or other underlying series that we weren't aware of, so there's an ongoing process of improvement with user help. Very quickly, this is just one example of something that's been done with this data set that you couldn't have done before. This is from the World Investment Report of this year. Unkted researchers crunched the data down, had to make some assumptions and a few extrapolations, but I think it's pretty solid to actually work out what proportion of tax revenues in developing countries, multinationals are responsible for, and what the relative proportions of that are. It's not perfect, but it's actually quite an interesting picture which you couldn't have seen before this data set. One of the things it gives you is that the corporate income tax payments are something in the region of $220 billion a year, which is then interesting thinking about the other kind of things that the Tax Justice Network looks at, which is the extent of tax evasion and tax avoidance. I'm going to skip through this, because I think I should leave the time for others, but we have estimates of this scale, about $190 billion in revenue loss each year for global losses to individual tax evasion. That's Gabriel Zuckman's work, which is, I think, by and large, including by him, thought to be a low-end estimate. Jim Henry's work puts a rather higher figure on the total of undeclared wealth globally. Thinking about multinational companies, we have IMF and UNCTAD estimates of $100 billion in lost revenue for developing countries. That's UNCTAD's estimate of one particular form of multinational profit shifting, where FDI is directed through tax havens and through special-purpose entities. The IMF work is a much broader figure for tax belovers generally, and effectively the impact of havens on developing country revenues, and it's about twice that. Given that we've got the World Investment Report estimate, that puts those values at something from 50% to 100% of the revenues currently being obtained by developing countries, which allows us to scale it in a way that we just couldn't have done effectively before. The last thing to say, a neglect of this, and the progress that's being made in terms of national revenue data, there is also progress at the international level in terms of requiring multinationals to report their activities on a country-by-country basis, which allows you to see if it ever becomes public. Currently it's not, but it will allow you to see the extent to which profits are being declared in places where the economic activity is not taking place and how big an effect that is. There's also a set of measures around the exchange of tax information between jurisdictions and the identification of ultimate beneficial owners of companies, trusts and foundations and bank accounts that will eventually allow individual tax evasion to be cracked down. At the moment, however, most developing countries are shut out of that process, and that's the big problem now. It looks like it'll work for OECD countries, it may not work for others. In terms of country-by-country reporting, that's only going to go to the tax authority of the home country of multinationals, which may then exchange it with or provide it to developing country tax authorities or not. So it's kind of the least transparent transparency measure that they could have come up with. So there's an agenda there, again, around building the data to actually understand what's going on rather better. But at least at the moment, I'd say the glass is half full. The very last point I want to make is just that this is a plug for us. All of those measures, the automatic exchange of information, country-by-country reporting, identification of beneficial owners, this is what tax justice network came up with in 2003, 4, 5 when we started. Those were the three big things. Which, you know, the reaction to that was, you are crazy, none of this will ever happen. 10 years later, thanks to TJN's good work and possibly also a global financial crisis, that became the agenda of the G8 and the G20. That's why the glass is half full. And it's now right in the middle of the financing for development document. What we haven't got is delivery of any of these data measures for any serious number of developing countries yet or into the public for researchers. So we're halfway there, but we're not there yet. Thank you.