 Onw'r danes fydd yn fawr, os ydym yn gweld o'r gweithio y fawr o'r gweithio, eu gwelwch yn gweld o'r gweithio, o'r gweithio o'r gweithio fynd a'r gweithio'r ymgyrchu. Fawr o'r gweithio'n gweithio er mwyn ffawr yn ymgyrchu'r tannig. Yn ymgyrch, mae'n ddweud, yn nodi'n gwahanol i ddweud o ddweud gwahanol sydd yn ymgyrch. ond we looked at what was available and we came to the conclusion that there was enough countries, enough background information that it was worth a stab. So we started to collect together information on balance sheets in different countries and what was available in terms of wealth distribution. And then we took the view that we could fill in the gaps for other countries in both of those areas and go for a construct a global wealth distribution. The basic idea is to begin with the solid evidence, the evidence we have on balance sheets, then add in countries that we have survey data for. Unfortunately, we have survey data, but this was back, we started looking at the year 2000. We had survey data available for China and for India and for Indonesia, which very much helped us fill in the developing world. We then used standard regression techniques to try to estimate wealth levels for the countries that we didn't have information for. And then there's a few countries at the end which tend to be quite small countries, very often little islands in various places, or countries which are not really connected with the global economy in places like at the time Myanmar, North Korea. So for those we just added an imputation value just based on sort of what their regional income level average was. So that gave us some information on the level of wealth in all countries of the world. On the distributional side we had, we only had data on about 20 countries where we had wealth data, wealth distribution data. But we do have income distribution data for a much wider variety of countries largely because of the WID dataset here. So the WID dataset gives us 180 countries. So we then looked at the wealth distribution data, compared it with the income distribution data for the 20 countries where we had both sets. And you can see very clearly that there's a relationship between income distribution and wealth distribution. Wealth is always more unequally distributed than income. So we used that sort of relationship to try to predict for the other countries where we had income data but not wealth data, what the wealth distribution was. It's a crude idea but on the whole it gives you some sort of ballpark figure for the wealth distribution. So then we put these two things together. We also, the other crucial thing, we have a little utility here which we developed at wider, which generates a synthetic sample from a distribution. So if we know something about the distribution, the diesel shares or some Lorenz curve points, we can generate a sample of observations which are consistent with that information. So we use this to generate a sample for each country, of course it's proportional to the population size. And that gives us, we're now generating a sample for the whole world and then we can process this and look at the inequality, the whole issue about what the levels are of wealth in different countries and what the distribution is, both within countries and across countries. So we've been doing that, that is what we started at wider and we came up with a study. This was the first ever attempt to look at the global wealth distribution. We came up with some quite startling figures. One of our core results was that the top 2% of wealth holders in the world owned half of global wealth. So we had a headline which was the top 2% owned half global wealth and when we did a press release this was our caption at the beginning and it took off and it just immediately drew people's attention. We had an enormous press reaction to this, it hit I think it was 500 newspapers around the world. It was at one point, it was the top story on the BBC website I believe and we had I believe that month 140,000 downloads of our paper from the wider website. The website actually crashed twice in one day because there was just so many people trying to download the data. I think that is indicative really of just how much interest there is in this topic and also how little information there is. So we were really filling in a gap there that I think was really very important. After that, since we had that study and it was published as part of the project, we've been developing it further in connection with Credit Suisse Research Institute putting out an annual wealth report. So we've taken the basic methodology and developed this year by year in various ways. Sometimes it's just a little bit of improvements, it's now fortunately there's more countries that have a balance sheet data so that's been a big plus. There's more countries with wealth distribution data so we've filled in quite a lot of the gaps and it's allowed us to perhaps make the other estimates with more confidence. The other thing that we've done since we've left here in particular is to pay more attention to the top tail of the wealth distribution. The problem with wealth distribution data is it often comes from surveys and there are some very serious problems trying to get accurate information on wealth distribution data in the top tail. It's really a challenge and if it's not done with a lot of care and attention then it can often just not really be very credible. The US does very well, I have to say in this, they pay a lot of attention. The real problems are first of all the question about sampling that very often the top wealth holders are not the ones who agree to take part in the survey. So you have a response bias which is going to reduce the number of top wealth holders. The second problem is people under report their wealth so you have these two combined problems. It's not really surprising but you have to then say how do we adjust for this? We've since developed various techniques for using information on the top, the rich list data which is available now for a lot of countries where you get lists of people and estimates of what their wealth is. So we graph this on to the distribution and try to re-estimate the top tail based on this other evidence. It's mainly now we're using the Forbes billionaire data, one reason for that is because it's done consistently across countries in the world and there's in many countries now enough billionaires to make that a credible exercise. So that's where we are heading, where we are up to now. We've developed this year on year, in fact we've now this year pushed it one stage further and we've got some really interesting results on inequality, wealth inequality back to the year 2000. So we've got 15 years of data on wealth inequality which will be outed this year as the 2014 wealth report.