 There's a relationship between the topics that are being discussed in this session in that my interest in wealth and Tony Shrocks and so on is not so much a fascination with the top end of the distribution as a concern about welfare and in a kind of an expanded view of vulnerability to poverty. Personal assets play a very important role in cushioning the blows that implanted people into poverty. Having personal assets is a form of self insurance and you know in Ethiopia for example herders will build up their herds of cattle when times are good and then these are run down when times are bad. In Indonesia people acquire quite frequently it's quite common to acquire gold as a personal asset when times are good and then use this when times are bad. Wealth and personal assets have many different dimensions of how they improve, can improve the lives of lower and middle class people. They provide empowerment and independence. If you're in a village in India and the local major landholder or somebody else is oppressing you in some way, if you have some financial assets and you can go and hire yourself a lawyer to fight back or a group of people can get together and hire a lawyer to fight back or take other steps to defend themselves then obviously they are significantly empowered relative to if they didn't have those assets. We see wealth as very important in those aspects. As Tony said yesterday the work that we've been doing estimating the global distribution of wealth started at wider and recently we've been doing estimates on an annual basis. They're published by the Credit Swiss Research Institute and the report comes out every year in October. So towards the end of October you should see press coverage of our 2014 report coming out and for this audience it's important to emphasize that we actually put out two publications every year about a month after the report comes out which is kind of a glossy publication that seemed at a very broad audience but is nevertheless, you know, we think professional, incredible and so on. About a month after that publication comes out called The Data Book which indeed has an awful lot of numbers in it but it also has a more rigorous presentation of the topics which are covered in the report. So for social science or academic audience The Data Book is very important so look for that as well. Okay so wealth here means real assets plus financial assets minus debts and as I say we're looking at the global distribution of wealth. We use official exchange rates. That does lead to issues of fluctuations in relative wealth in different countries and different regions and so on because of changes in official exchange rates so we supplement our estimates with other numbers where we take the average exchange rate over the period of time that we're studying and so we can say how things would be different if there had not been exchange rate fluctuations from one year to the next. In each report we've looked at special topics so here's the list. So first year we did gender dimensions and looked at wealth composition. The next year we looked at long-run trends in the level of wealth so we're looking at bubbles and busts and booms and so on and things like the wealth income ratio and how that's varied over time. The following year we did work on debt and inheritance and last year we did mobility including mobility among the Forbes billionaires and we took a critical look at wealth in the Eurozone because a very important development in 2012 was the release of the wealth survey data that had been prepared on a consistent basis for all the Eurozone countries. The quality of those surveys varied a little bit in some of the smaller countries was not very strong but that really is a major step forward in the development of wealth distribution data because of the comparability of definitions and methods across those 15 or so countries and this year we're looking at the inequality trend since the year 2000 so look for that in the report that's coming out in October. So some details about the methods. We look at the adult population. There are two aspects to our work. One is establishing wealth levels and the other is establishing the distribution of wealth within each country. I should probably put all my glasses here. So household balance sheets are actually part of the UN system of national accounts and these balance sheets are implemented for financial assets and for liabilities in 47 countries. This includes the richest countries, well most of the OECD countries. Developing countries are relatively few that have household balance sheets. Columbia has it, South Africa has it. It's a relatively short list but among the developed countries this is a standard form of data and a lot of work has gone into preparing it over the years and so what that means is that you can scale your estimate of the distribution of household wealth in these countries or 17 countries that have also in their household balance sheet they have non-financial assets so the US, the UK, Canada, Germany etc. So what we do and the technique was outlined in our article in the Economic Journal in 2011 is we use regression methods to look at the determinants of wealth in the countries that have the balance sheet data and then we extend this, do imputations for the countries that don't have the HBS data. You can do that in 120 countries. Of course you need the right hand side variables to perform the estimates. There are 216 countries in total, 39 of them as Tony was saying yesterday. We use region income class averages to get these levels. These are mostly very small countries. I think they add up to something like 4% of the global population. Now distribution of wealth, since we started our work the number of countries that have good wealth distribution surveys has expanded from 20 to 31 partly as a result of what was done in the Eurozone. But in the developing world, Chile, Thailand, a few countries have come online with this kind of data. China and India and Indonesia all have, while China and India have wealth surveys which are being conducted on an ongoing basis. Indonesia had a wealth survey in 1997 so it's getting a little bit out of date now. But when we started this work 10 years ago and discovered that there were these surveys in China and India we realized that we were in business and you could actually do something to estimate the world distribution of wealth. Without that it really wouldn't be possible but there you're bringing in a third of the world's population. And gaining valuable evidence that can be used to extend estimates to other developing countries as well. Okay, so just a little bit more information about estimating wealth levels. The balance sheet data, the statistical organizations, central banks and so on that prepare this can use essentially whatever data they have. So it's not just from wealth surveys, it's from other sources, it's from institutional sources, financial assets and debts and so on. This information is very reliable and credible. On the real asset side you can estimate the value of housing for example using a perpetual inventory, that's to be true for farm machinery and equipment and vehicles and so on. There are four countries where there's a survey but there's no household balance sheets, that's China, India, Chile and I guess one other, maybe Indonesia. Okay, so you might ask, well is there much difference between wealth and income levels across countries? In this graph we've got our estimate of wealth per capita versus GDP per capita in 2013. You can see that there is this fairly strong relationship, R squared is 83%. But there's some variations so you have, say if you look along the, look here we have a number of countries in this range between say $40,000 and $50,000 of GDP per capita and there is a significant amount of variation in terms of their wealth per capita. And this country up here I believe is Switzerland which is quite far off and deviates there. I believe this is Australia which is kind of smack on the line. Okay, so I'll say a few words about how we estimate our regression and which we use to do imputations. We have three different regressions actually. We've got financial assets and liabilities, the two of them come together so we have exactly the same number of observations for the particular countries. So we use the seemingly unrelated regression technique to estimate those two and then real assets are done separately. And the kind of thing that you're looking for here, actually you might ask why we use log consumption per capita as a regressor. Really we would like to use income, but income data are available for significantly fewer countries where we want to do the imputations and it turns out that if you run this regression using consumption rather than income for the countries where there is data you get very similar results and then you can do imputations for more countries so that's why we use consumption. This is in log form so if financial assets were going to go up 10% when income went up 10% this coefficient would be equal to 1, obviously it's actually higher than 1. And if we, I don't have the chart here for the real assets but you see that it's closer to 1 in the case of real assets. And the fundamental approach here is to put in the variables that the life cycle model would predict, would explain differences across countries. So this is growth rates, longevity, is predicted to have a positive effect on wealth in the life cycle model. In some of the regressions it turns out that the life LCM variables are not significant, we drop those that's why you don't see them all here. And then we have some other variables like the market capitalization rate which reflect financial conditions and so those are brought in in addition. You see here that the percentage of the urban population is a significant variable. Survey dummy, is this quite interesting, see this is a very large coefficient. This says if you get your financial assets estimate from survey, from survey source it's going to be 167% lower on average than if you get it from the household balance sheet holding everything else constant. So these countries in China, India where we have to rely on the survey data that would not give us a good estimate of wealth levels in those countries if we do not perform an adjustment. And so what we do is we use this coefficient to perform that adjustment to get a better idea of what the level of financial assets is in those countries. Here's what's been happening to the aggregate level of global wealth over the years that we've been considering. And you can see there's this big dip in 2008 and some recovery since then. And the extent of recovery has varied between different regions. We have global trends in wealth per adult. The top two lines are for net wealth. The lighter of the two, this one here, is at the constant exchange rates. The top one is in terms of US dollars. So broadly speaking these things are fairly similar but the growth in terms of USD from 2000 to 2013 is greater than in terms of constant exchange rates. So that reflects an appreciation on average of other currencies relative to the US dollar over that period. And these lower curves are for financial wealth, non-financial wealth. And so you can see for example from 2000 to 2007, I guess that's 2008, there was this convergence and so at this time they were of about equal importance, financial and non-financial assets as a result of the stock market crashes that occurred during the financial crisis. But then the two have separated again so the financial assets are a bit ahead of the non-financial. Here's a map of world wealth levels. This corresponds to the kind of thing that we're used to seeing so not too many surprises there. You can see that there are a few countries that are just white and this means that we don't have estimates for those countries due to lack of data. Okay so wealth levels across countries on average global wealth per adult has gone up from about 31,000 year 2000 to about 52,000 today. You can see there's not much increase in wealth per adult since the peak year just before the financial crisis. Country experience has varied and countries like well actually most of the countries that are just been selected here for illustrative purposes have rebounded fairly well from 2007 to 2013. One case where it's gone in the opposite direction is India and this is largely due to the depreciation of the rupee. Okay now we have 31 countries where there are wealth surveys. I'm going to show here and on the next slide is the top part of those distributions so these are shares of groups that are in the top 25% so there's another slide that I'm not showing would be the entire rest of the distribution and you might look at this and say oh well this is you know kind of sparse and so on. Well you don't need an enormous amount of detail in order to get a plausible estimate of the Lorenz curve. As Tony was explaining yesterday we used his ungrouping algorithm that fits the data points exactly and we've been quite a bit of testing of that and it gives fairly reliable results. Now as we know surveys are challenged in the upper tail of the distribution due to problems of sampling and non-sampling error and the non-sampling error takes the form of non-response differential non-response according to the level of wealth and also under reporting. So if you don't do anything to address those problems you get an implausibly low share of the top 1%. So here we see in Canada actually Canada does do something about what you can do about the upper tail problem is you have a special high income or high wealth sample and you oversample those people. So you get some kind of independent information. In the United States they do a really good job based on the use of income tax records that maintain a list of the people that they want to oversample and the approach works very well. And so other countries for example Spain which introduced a wealth survey about 10 years ago brought in the experts from the Federal Reserve Board to advise them about how to do this and these kind of methods are spreading a bit but they should be adopted in more places if there's a serious interest in using the survey to get an estimate of the distribution of wealth. So you see other countries with these relatively low numbers 15.7, 14.8 these are not really very credible and Japan I'm wondering even if that may actually be a typo here 4.3% it's kind of silly. Here are the remainder of the countries at the bottom here we have the United States you see they have an estimate of 34.1%. Another interesting aspect of their procedure is that they explicitly exclude the Forbes 400 and they've always done this. They say well the Forbes magazine tells us who these people are and estimates their wealth better than we can probably so they're just not in our population that we're sampling and those people have about 2% of the household wealth in the United States. So you would take this number 34% and you say okay well if I add on the Forbes 400 this is a tiny number of people so it wouldn't change the definition the top 1% appreciably you just add on 2% and then you've got a better estimate of what the share of the top 1% is and that's transparent and the Federal Reserve Board is quite up front about that. Okay so I could go through some of this detail about exactly how we estimate the shape of the wealth distribution Tony talked about this yesterday so I think I'll skip ahead and try and get on to the results more. We impute a top tail to the individual countries using the Pareto distributions or a common approach and the particular way that we do it is by taking the Forbes evidence about the billionaires and saying well that gives us a point on the Pareto curve and then basically we look at the so the philosophy is that these surveys are probably pretty good for about the bottom 95% of the population and then you want to adjust the upper tail so the oops where was I so here's your Pareto curve and so we know from a lot of past evidence from more reliable data sources like the state tax records that it does tend to become linear at some point and so the idea is to take the Pareto curve or a Forbes or Pareto curve which would do that and would line up with the number of billionaires so again as I say Tony explained this method yesterday. Okay here's a comparison of survey data and we got what we got from our estimate so you see for example in Canada I was saying well 15.5 is a bit low so I actually done quite a bit of fair amount of research over the years going back to the 1970s about the Canadian wealth distribution so I know that number is just too low and what we have here is 24.7 that is quite a bit lower than the US level that makes perfect sense. Canada still has a very quite a high rate of foreign ownership and so you could have think effectively it's as if God reached down his hand and took a bunch of our high net worth and ultra high net worth people and just moved them across the border so there that's one way of thinking about it anyway so that's a plausible number now the adjustments are very large for some countries India we're going from 15.7 to 48.7 and they just are a considerable number of very rich people in India this is also true in Indonesia but things are different for example in a country like Italy there's not a very large adjustment if I had more time I could go into the shares of the top 10% and so on but let's just look at the shares of the top 1% here. The Nordic countries are interesting we've got Sweden, Norway you can see that the adjustment that's required for the Nordic countries is relatively small Norway top 10% goes from 65.3 to 65.9% why is that? Well you don't have the non-response problem this data is coming from wealth tax records which go through adjustments to make them you know transform the valuation basis and so on but everybody's included right so it makes sense that the Nordic data is more reliable in the upper tail there are problems in the lower tail that I don't have time to go into at the moment okay so anyways you put this all together what you discover is wealth inequality is very high the world share of the top 10% in our data was 86% the top 1% at 46% in 2013 and the shares in some individual countries we've already seen up there the top 10% at 61% in China that's been rising over time 75% in the US more in some other countries overall the richest 2% of adults own more than half of global wealth which is a result that we've had since basically from the beginning of our work and received quite a bit of publicity. Global wealth genie is in here 0.905 Tony was saying yesterday we really shouldn't report these numbers to the third decimal place so I guess I should have rubbed that out so we're getting a number around 0.9 for comparison Franco Blanovich and his co-author Lackner in their very interesting recent paper have got the most recent numbers which are for 2008 are from 0.7 to 0.75 this is for income right so the wealth genie I mean the income genie globally is pretty high but the wealth genie is you know really high. Why is there that range there it's because they do top tail adjustment their 0.70 is using their standard methods and they look at the top tail adjustments and say that you could get this up as highest 0.75 that's kind of an upper bound for them. Wealth is more unequally distributed than income across countries so this is an interesting insight compared with income distribution differences between countries are relatively more important for wealth. Here is a chart that just gives you the numbers on the estimate of the global wealth distribution and you see you've got these very low shares at the bottom including a negative share for the bottom decimal. The global wealth pyramid which Tony again showed yesterday and let me say again this is a chart that Tony showed yesterday but I'll say a few more words about this when the Marcelo Neres was speaking yesterday he pointed out Brazil is kind of a microcosm of the world income distribution well here we have Latin America showing up as a bit of a microcosm of the world wealth distribution what this chart shows is it arranges the world wealth desiles from 1 to 10 and then it tells you where the people each desile are living so about 24% of the people in the bottom desile are in the Asia Pacific region which excludes China and India and then you have a large number from India almost nobody almost nobody in the bottom desile is from China a large number from Africa and so on what I was saying about Latin America you see the the height of their bit here relatively speaking is fairly constant although it tapers at the top right if it was perfectly constant that would say that they had the same share of every desile in the world wealth distribution so it really would be exactly a microcosm of the world distribution the other region which is approximately like that is Asia Pacific omitting China and India and then the shapes where the others are radically different India has this long long upper tail China has this great big blob here which has been moving to the right and you know you can kind of if that continues then we'll have dominance of China in the upper desiles of the wealth distribution before very long North America interestingly has look at some people right at the bottom okay and Europe again has got actually Europe's got even more representation in these lower desiles what's going on there well you know wealth is a funny thing right I mean having a lot of wealth is always a good thing but having zero wealth in a country where you have a highly developed welfare state you live in a rented apartment you don't own your own house the state has got a nice pension or your employer will actually be a state pensions relevant thing here you don't really need a lot of personal assets to necessarily to have a satisfactory life and you have young people that have big student debts and have mortgage debt and so on so you know you have to have to be a little bit careful right and so actually the Scandinavian countries Finland as well have quite low income inequality and they have quite high wealth inequality and people have their good papers have been done on this and essentially this is what people think is going on that you don't necessarily need to have a lot of wealth to do well in those countries this chart shows where the dollar millionaires are so about 39% of the millionaires in the world in the US Japan you know their economy has been stagnant for quite a while but they still are the second most important country in terms of world millionaires and then you have France, UK, Germany, Italy and then a bunch of other countries so although you know there's quite a bit of attention to increasing inequality in China despite the fact that this world's most populous country only has 3% of the millionaires so you know if we're thinking globally about wealth inequality and where the rich people are you shouldn't point your finger first at China it would be the US and Europe and Japan are the principal areas where the wealth here are located now there's some interesting contrast in terms of the structure of wealth and the relationship of wealth to income for the 2011 data I separated the countries by GDP and the aggregate wealth to GDP ratio for the bottom 80% of the countries which includes all the low and middle income countries in the world is basically it's about two for the 20% top income countries it's four okay so I was saying earlier that wealth differences are greater than income differences well you can see this is very dramatic right wealth to income ratio of two at the bottom and four at the top the ratio of financial to non-financial assets is interesting it's about two thirds for that bottom group and it's about one and a third so four thirds in other words twice as much which is what you would expect from these numbers actually here there was a very important scholar worked on wealth at Yale University many people recognizing Raymond Goldsmith and he published a number of important books and he thought this he called this the he called the next slide says the financial interrelations ratio and he thought that typically this financial interrelations ratio as he called it was about a half for developing countries and he expected that for developed countries it would converge on a value of about one so we've got numbers which are higher but the ratio between the two things is two to one ratio which is what he was expecting he talked about this in a book in 1985 and by the way in this book he also estimated what he called the planetary balance sheet so there's he had 17 data from 17 or 18 countries much of which he had built up himself through you know by dint of really hard labor you couldn't just go on the internet and collect the data in those days you actually had to physically go around to the different countries and get to know the statisticians and get their data and so on at any rate I think if Raymond Goldsmith could estimate the planetary balance sheet from 17 or 18 countries back then we shouldn't feel too shy about estimating it today on the basis of balance sheet data from 47 countries and the imputation methods that are used asset composition is certainly of great interest and it varies enormously between countries what we've got here is the sort of greenish bars over here are debt so this is taking your net worth below zero and on the right hand side we divide between financial and non-financial assets by the way should say that you know this doesn't seem very detailed well the reason is that different countries divide up their financial assets in different ways and so to make consistent comparisons across countries you have to have this fairly high level of aggregation so these are arranged from South Africa which has perhaps surprisingly a very high percentage of assets in financial form so it has very well developed life insurance and pension industry also property there is not very expensive if you go online and find out you'll discover that you can have a very nice house in either Sydney Australia or coastal city in South Africa and you can find ones that look pretty similar and the one in South Africa might be a quarter of the price of the one in Australia so that's something to do with that ratio US financial assets very important and then you go down well here look at France, France is also very high income country but the ratio of financial to non-financial assets is quite different than it is in the US or Japan so there are these big differences and we believe that there's a lot of room for research on things like that and anecdotally people will tell stories about why there are these differences between countries but it would be better to have more hard evidence here's the composition of financial assets what we're doing is just looking at countries that have household balance sheet data here, Columbia is the one developed, one of the few developed countries that has that and we've got a breakdown between currency and deposits so of course you see the Japanese are famous for their bank deposits and they have over 50% of their financial assets and currency and deposits and the Americans are famous for their participation in the stock market so you see they have this very long gray bar here which is the fraction of their financial assets which are in equities, Japan has a tiny amount in equities so somehow the finance gets through to the companies to some extent from the deposits to some extent from these other financial assets which include pensions but of course another thing that's going on in Japan is that there's very high government debt and so quite a bit of these currency, these deposits are some of them are in the Japanese post office which is major savings bank in Japan but this money, substantial part of it is not going to fund industry it's going to fund the government's debt anyway, a lot of variation there, very interesting so in one of our reports we looked at these long run trends and of course particularly with publication of Thomas Piketty's important book there's been increasing focus on these things you see so here's France, this is the wealth income ratio so back around 1900 as Piketty has told us it was high, 7 or 8 and then it slumped so badly in the 19, the interwar period and then again after the war and since then has been increasing it's very different from the pattern in the US it's one of these things that really there's a lot more research and thought so here the blue line is the US so the wealth income ratio went up in the Great Depression but for most of these years the ratio is between 4 and 5 and it's only very recently that it's once again gone up above that and when our 2014 numbers come up we'll see that this ratio is up again here's the, so the OECD has been publishing data on a consistent basis for quite a long time now for the G7 countries so we have this interesting picture of what's been happening to the wealth income ratio in those countries and you can see broadly from the 1980s onward there is this upward trend Japan is different, of course it had this enormous boom in the 1980s and then things have been kind of not doing very much since then but the other countries have had this upward trend so this is one of the reasons we think that why people are so interested in wealth these days is becoming relatively more important it's partly to do with the aging of the population but it's also because there's been this shift to the right in politics and there's much more emphasis on individual responsibility we're all supposed to save for our retirement and maybe for our future healthcare and for our children's education and so on and so in the developed world there's more emphasis on the importance of wealth and I think the same thing is, same emphasis certainly is showing in the developing world where due to the lack of social safety nets personal assets are actually I think more important but the long time series and the good data is only available for these high income countries here's what's been happening to debt actually if you look at how much, this is the debt to income ratio how much has it gone up, well go back to the 1980s the median was around about 0.6 or 0.65 and the median for these countries is now about 1.3 something like that and that percentage wise that increase is in line with the increase in the assets so it's one of these things like you an important result actually of our work is discovering that the debt to assets ratio is much lower in countries where debt is more of a concern some people are concerned about debt loads in countries like India but actually in aggregate terms the debt to assets ratio is much lower than it is in advanced countries you kind of have to be well off before banks will lend you money and you go to a country like Denmark that debts to gross assets ratio is 20% it's very high in a lot of developed countries it's not necessarily a problem but of course as we know can get out of hand if things are not handled appropriately okay so just wrapping up we've seen that I'd like to emphasize that wealth data are getting stronger all the time and we believe they're sufficient to estimate the global distribution if you're careful how you do it so sometimes people get discouraged about the quality of wealth data I think that's a really bad thing it's very unfortunate people hear while there's some problems with wealth data and then go away and think well we just forget about that the right response to try to make it better unfortunately a lot of people have been trying to do that and the quality of the data is improving and more countries have data so we have this high level of inequality wealth and income are imperfectly correlated across countries but the wealth differences are greater than the income differences top tail adjustment is really vital it makes a big difference to our picture of the upper tail of the wealth distribution and of course also the lower tail the more there is in the upper tail the less there is in the lower tail percentage wise so I think that's about it thanks very much