 Hello everyone. My name is Beatrice Habak. I'm a master's candidate and I also am a research assistant for the University of Texas inequality project which is also called UTIP. The paper I'm presenting today summarizes an update and revision of the pay and income inequality measures calculated by the University of Texas inequality project which was started in 1999. The data set, which is known as UTIP Unido, consists of measures of cross sector industrial pay inequality computed using the between group component of T-statistic across a consistent set of industrial sectors. In total it has 3,871 observations over 149 countries covering the years 1963 to 2008. My work with the project has been specifically with the estimated household income inequality or E-high data set which was also extended as part of this update. The data set consists of estimates of the genicoe efficient using gross household income. These are derived using a regression of the original high quality genicoe efficient from the Dininger and Squire data set published in 1996 against the UTIP Unido measures and there are also controls for the share of manufacturing and total employment and dummy variables for income or expenditure reused whether the unit of analysis is the household or the person and whether the income concept is gross or a net of taxes and transfers. Ellen, final here can I point with this? That is the log of the tile measure and as you can see it's clear that the tiling index is a strong indicator of the geni measure. So the coverage as I mentioned is the same as for the UTIP Unido measures and first I'd like to show you some maps that highlight the coverage of the E-high estimates by decade. So here we have the 1960s. The red hues represent values of the genicoe efficient below 35 with countries with the lowest inequality in pink and then the blue hues represent values above 35 with the highest inequality in dark blue. So here you can see Canada and the US with values in the low 30s on par with some of the Scandinavian countries and with Australia. And then here in the 1970s you see increases in inequality in the US and India and we have more data coverage with China appearing on the map. Here in the 80s we see increasing inequality in Canada and more coverage in South America and Africa. In the 90s we start to see rising inequality in South America and Africa. India, Pakistan, much of Eastern Europe and Australia. And finally in the 2000s most French users are still seeing a rising trend in inequality. So these maps give you an idea of the coverage of the E-high measures and how accurately they compare to worldwide trends. And my specific project focused on seeing how the E-high measures compared to other measures based on national household surveys. So for a set of about 40 or so countries I looked at any reputable genie coefficients I could find from international organizations, national statistical institutes and individual researchers and many people here will know because I emailed you a lot so this is the fruit of that work. So I coded the measures based on what type of income concept was used and the unit of analysis and then I made charts for each country. So black represents our measure green is for market income, blue is for gross income red is for net income and yellow is for consumption. And then there were some cases where I just wasn't able to find the information about the income concepts so those are in purple. In addition dotted lines represent measures of personal income and solid lines represent household income and then the very light dotted lines represent the SWID estimates by Frederick Seult. So our main finding is that our measures are quite consistent with the existing literature with a few exceptions which I'll mention later on. So this is Canada. One of the first things you notice is the sheer breadth of the measures. So for example in the year 1998 right here you can see market income inequality is measured at right about 52 points and then disposable income inequality is measured at about 29 points. And that's for just the same year. So it's pretty broad. There's a similar trend for all of the advanced social democracies in the series and there are quite a few studies out there that say that the most advanced countries have very unequal primary distributions and that these are offset by redistribution policies. However if you look at the original UTIP UNITO data which looks specifically at pay inequality it seems that the Nordic and North European countries are actually the most egalitarian in their primary structures. So how do we explain the paradox? Well our answer is that in advanced welfare states very high market income inequality must be due to the presence of households with zero market income and presumably in countries with strong public pension systems it's possible for many elderly couples and single individuals of all ages to form households on non-market income while these households would be very rare in countries where market income is necessary for survival. One of the other things you notice here is that the measures or our measure falls generally above net income below those for market income and around those for gross income which is a pretty common pattern. So here is Denmark. You can kind of see the similar pattern. Here's France. This is one of the countries where the data wasn't very continuous so there are a lot of markers rather than lines. Here's Germany. I included measures for East West and Unified Germany. And here you can see we're right on par with the gross household income measures. Here's Greece and Italy. You can see most of the measures are based on net income here. Here's Japan. Here you can see that the measures at the bottom seem to be very different from the other ones. You've got net and market income and gross all below most of the other measures and that's due to the survey choice of the authors and they told me specifically that because of the sample restrictions they wouldn't be comparable but I wanted to include them just to show how much survey choice really does matter when we're making inferences about the measures. Here we have the Netherlands. As I mentioned before the very high market income inequality is what would be expected. Here's Spain and Sweden. Again the high market income inequality. Here's the UK which is a very well documented country and then here's the US and this is actually a case where our measures are problematic because it misses the capital income that a lot of the other sources pick up. This next set of countries, I should mention that my colleague Alexandra Melinoska collected the data for these eastern European countries. This set of countries has a narrow but still distinct difference between market income inequality and disposable income inequality. One explanation for this could be that these countries don't have welfare states as developed as those in northern Europe but neither do they have the very high level of pay inequality you see in Latin America and Africa or a lot of sub-Saharan African countries. This seems to support the interpretation about the high levels of market income inequality I mentioned before. That being said though a lot of the market income measures are very erratic and variable and they can't really be said to be trustful indicators of changing economic conditions. Here's Hungary, Poland, Russia and Ukraine. Once we start moving away from the long industrialized countries we encounter other issues with the data. Sometimes there are less independent sources of data like the case of Mexico where all of my sources of data come from the same household survey conducted by the National Institute of Statistics and Geography. So the variances present in this data are all a result of variable definition, sampling and data construction differences. At other times you'll see a lot more purple because I wasn't able to find income definitions. And we also see that for these countries there's much less distinction between market gross and disposable income measures. On average they tend to overlook and overlap and look jumble. So this supports the general idea that market and disposable income are partially determined by the structure of the welfare state. And for India most of the other data out there tends to be consumption based, but the one measure published by the LIS right here is only about two genie points off from where we are, so that's a good sign. We also have a few cases where the E high estimate tends to be problematic and in these cases our estimate is generally too low. So these include Brazil, South Africa, Colombia and Thailand. So here's Brazil. We don't really know why our estimate is off here but we have a few guesses. So one is that Brazil and South Africa are countries with particularly well known divides and Brazil it's north, south, in South Africa it's black and white. Another reason could be just idiosyncratic measurement in these countries and here's South Africa. So here I just have some summary statistics on the E high and some more. This chart shows the global mean values and as you can see it picks up the widespread rise in inequality in the 80s and in the 90s which is pretty common across other measures too. So to offer some conclusions from the charts we can see that the literature on income inequality generally is very messy in advanced countries the difference between market gross and disposable income is usually very large. Our number is usually quite close to the other gross income estimates and generally between the market and disposable income estimates which means that the underlying industrial pay inequality measures on which the E higher base are a pretty good instrument for income inequality generally with some exceptions such as the problem of the capital income in the US and the level estimates for a few large middle income countries like Brazil and South Africa. So if you found this interesting if you're interested in using the data please visit our website and we also welcome your comments and feedback. Thank you.