 I'm going to be speaking about a book project conducted with Guillermo Cruces, David Home, Mariana Villellas. The book is entitled, I'll show you the cover of it, Growth, Employment, and Poverty in Latin America. It was published last year based on a number of wider working papers and of those working papers, most of them are country-specific studies. And then we have one lengthy cross-country working paper which became a number of chapters in the book. The work is motivated by a previous book of mine entitled Working Hard, Working Poor. And Working Hard, Working Poor was an attempt to bring together some of the things we know from around the developing world about how people are working. And one of the things we learned from that is that people, in fact, are working. Their unemployment in the developing world is very low. The issue is what I like to call the employment problem in Latin America and Africa and Asia, which is that people, although they are working, earn so little for each hour of work that even if they work very long hours, they end up being poor. Hence the title, Working Hard, Working Poor. And so of concern to me moving forward from this previous book was the question of what can be done to improve the material well-being of people like these who you see on the cover of the book. That is not just them, but people like them, literally billions of people who are fallen into the category of being low-income people. And in particular, would economic growth help improve their material well-being over time? I personally am interested in the entire developing world, but to do a comprehensive study of all of the countries in the world, developing countries of the world, would be an impossibly large task. Happily in Latin America, there is the SEDLOC database, which we just heard about from Carlos. And these are harmonized data sets that are based on 150 household surveys, 5 million households, 18 million people. And they have been harmonized in the sense that the data are in comparable format. And so once you decide to study one country and write the code for it, you can study all of the countries within the database. And this was a great opportunity to learn what had happened in Latin America. Well, I'll tell you what my hypotheses were. And the good news is that my hypotheses turned out to be wrong. My hypothesis was Latin America, as you also heard just now from Carlos, is way up there with Africa as being tied for the lead in terms of inequality within the region. And what worried me, and what worries other people a lot, is the possibility that in a high inequality region, what would happen when economic growth takes place is that those who have the economic power and who have the political power at the top of the distribution would capture large shares of the gains of economic growth. And those who are in the middle class and lower classes might capture very little. In fact, I was extremely pessimistic because of my own country, the United States. And in the United States, we have had the remarkable record, exceptionally bad, that economic growth has taken place. Wages, on average, that is mean wages in real terms, have grown in the United States year after year after year. And yet, median wages and the wages, median wages of workers, median wages of full-time, full-year workers, I should say, labor earnings, not hourly wages, that the middle has stagnated, the lower part of the distribution has stagnated. You have to get on up to about the top 10% before you start seeing large earnings increases. So my hypothesis for Latin America was that what we would find there would be that there would be gains in earnings concentrated at the top. Well, you heard from Carlos just now that that was not what happened in Latin America. In our book, that's what we also found did not happen in Latin America. So from that point of view, of course, it depends on one's value judgments. To me, that's very good news, that working people actually benefited and benefited a lot. I'm not going to be talking about inequality of labor earnings as much as I am going to be talking about the changes that took place in labor markets. So what we have here are the three questions on growth employment and poverty in Latin America. Has economic growth resulted in economic development, by which I mean people attaining higher material standards of living? Has it resulted in those higher material standards of living via improved labor market conditions in our period of analysis, which is the 2000s? Literally starting in the year 2000 and going up to what was most recently available at the time we began our study, which in most countries was data from 2012, 2013. Were these improvements halted or reversed since during and since the Great Recession of 2008? And how do the rate and character of economic growth, the changes in employment and earnings indicators, and changes in poverty and inequality indicators relate to each other? And so the analytical framework, I'm hesitant to call it an analytical framework, but anyhow, a schema that helps me think about this is what I've called here a policy action. Actually, it might be better to call it some kind of impetus like economic growth or external events or something else. And if growth or other economic events or other events affect outcomes of people, there has to be some transmission channel by which growth, if it takes place, gets transmitted to people in the form of higher earnings and better labor market opportunities. If that transmission channel is blocked, we could end up with what the United States has, which average workers not gaining in earnings and being pretty angry about it. And I'm not going to make any further political hints than what I just did. The other major channel for linking growth and other external, other events to poverty and other outcome variables would be social programs. And so the question then that we focus on is what happens via the labor market, via the employment channel. And so we use the SEDLOC database. And from that database, we use a number of indicators for 16 Latin American countries. They're grouped. One is as an indicator of labor market conditions, the unemployment rate is not a good indicator of changes in labor market conditions, but it's the one that people usually start with. We have a number of indicators of the type of work that people are doing, what occupations they're in, what their employment status is, which sectors do they work in, what educational levels do they have, and are they registered with the social security system, which is a somewhat country-specific definition such as, in Argentina, are you registered with the pension system or not? And we use those as other indicators of employment and earnings. We look at real monthly labor earnings on average for workers as a whole and for broken down by different demographic and economic groupings. We look at the poverty rates and how they changed over time. We have data on the Gini coefficient. I'm not going to say anything more about it than other than the fact that it is included as one of our indicators. And so our analysis started by taking these 16 indicators. For some of them, like labor market earnings, higher labor market earnings are generally regarded as a good thing. Higher unemployment is generally regarded as a bad thing. And so what we did is that for the ones that are where an increase is judged negatively, we just simply put a minus sign in front of it in order to be able to take, then, an index that we begin with, which is of the 16 indicators, how many of them changed in the welfare-increasing direction. So we have, then, looking at the overall period, we have data on these changing indicators for Latin America as a whole and then for countries within Latin America. Starting with Latin America as a whole, these are the changes that we found. I'm colorblind, so I'm not going to refer to the blue or the red or whatever, because I'm not sure that they are blue or red. But the one that goes up is GDP per capita. And so economic growth took place in the region as a whole. Of the various indicators that I just mentioned to you, the unemployment rate went down in Latin America since the year 2000. Whoops, what happened here? There. Mean labor earnings grew in Latin America. The $4 a day poverty headcount, which is the lowest poverty line used in a middle income region like Latin America, fell substantially. When we look country by country and see what happened to economic growth, some rapidly growing countries, some slow growing countries, others in the middle, but the ones in the middle did better so that overall Latin America had more economic growth, than did OECD countries or than did the United States. If we look at the changes in labor market indicators, and this is that summary index, just take the 16 indicators, did each of those indicators increase in a country? For each of those indicators, did it increase in the country or not? This is the percentage of improving indicators. You see that if we take the 5 eighths mark, which is so 10 indicators out of 16, all countries, except for Honduras, had increases in that number of indicators. If we go up to 75%, most of the countries had those indicators improving as well. So in general, Latin America was improving in its labor market indicators and helping to achieve solutions to its employment problem. We then performed cross-country analysis of the link between growth and employment and poverty and asked the question, did the countries with higher rates of economic growth experience larger improvements in labor market indicators? The answer is a very mixed one. So the percentage of indicators that we're improving related to the growth rate of GDP per capita, yes, there's an upward sloping line there, but the R squared is only 0.1. So clearly what this is saying is that more than economic growth is involved in looking at what is happening with respect to the indicators as a whole. If we look at the growth of labor earnings, this is quite remarkable. This is an effective 0. And so it says that labor earnings are increasing, but the increase is not linked very closely to the growth rate of GDP per capita, nor is the poverty reduction linked closely to the growth rate of GDP per capita. What the change in growth is linked more closely to is the share of workers registered with the social security system, what some people would call formalization taking place. And so we find that there are these links, but they're not very tight when we look across countries. So that takes us beyond economic growth. And what we wanted to do there was to see if there was anything that was tied in closely that is variables other than economic growth tied into the changes in employment or the changes in poverty. The answer was not really that there are country-specific things that are going on. And it really is a country-specific story. It's a story of this country decided to do this. And it decided to do something, partly for political reasons, partly for social reasons. And there are these other factors at work. So let's see. Don't need that. OK, next question was, what is the link between employment and poverty? And are the larger improvements in employment and earnings indicators associated with larger reductions in poverty? So this is then the question. If poverty is going down, to what extent is it linked to employment and labor market variables? And if so, which ones? And so these are the correlations between the change in poverty over the observation period, 2000 up to approximately 2012, linked to each of the indicators one by one. And so the variable that is most closely linked to poverty reduction with a correlation coefficient of 0.88, which is extremely high, is the increase in mean labor earnings. What that's saying is that the poverty rate was falling in Latin American countries, in large part because of workers earning more for the work they do. Now what else might have been responsible for the fall in poverty within the labor market? It might have been unemployment, but the correlation between the change in poverty and the unemployment rate, the change in unemployment, was much lower. Then there are these things in the middle. These correlations are pretty substantial. And so the change in the types of jobs that people are working in do have a role to play, but not as much as the fact that they're earning more wherever it is they're working in the labor market. And that's what we found there. OK, so this is, I think, the single most important empirical finding we have. And that is if we look across the countries that the correlation between the R-square, between the change in labor earnings and the change in poverty, R-square to 0.78 is enormous, I think. And again, what it's saying then is that poverty is being reduced in these Latin American countries through labor market earnings. And the larger is the increase in labor market earnings, the greater is the poverty reduction. In other words, the labor markets are transmitting the economic growth that's taking place through higher earnings to lower poverty. And that's my summary. All right, there are lower correlations, lower associations with change in poverty and other variables like the growth of wage and salary employment. It is a pretty substantial correlation, but not as high as with earnings. And so across countries then what we have found is that there are some pretty strong associations, some lesser ones, but labor market earnings and poverty is the strongest link we find. OK, we have growth incidence curves. And the growth incidence curves depict what happens then within the income distribution. We constructed standard growth incidence curves in which the vertical axis is the percentage change in incomes in each of the 10 deciles. Here are some selected growth incidence curves. And you see that a number of them are downward sloping. Downward sloping growth incidence curve indicates that the highest percentage changes were taking place in the bottom end of the income distribution in a number of countries, but not in all of them. And so it's a whole other study. And Carlos just reported on such a study of why some of these patterns are what they are. And so I just wanted to show you that in our research. We were concerned not only with mean changes in labor earnings, but also the changes across the income distribution. I'll skip this part on the crisis. The crisis was still a hot issue when we began this research four years ago. And now people have more or less forgotten about it. So we'll move on beyond that. And final remark that I want to make. I don't need to summarize, I think. This has been a short enough presentation, so you've gotten our main messages. But what I do want to say is that what this does is that it augments Bergen-John's triangle. So Bergen-John had put forward a poverty growth inequality triangle in which he said, do the results of his study imply that growth has no significant impact on, he says, distribution, I would say, inequality. Bergen-John says, certainly not. The results mean that there's too much country specificity in the way growth may affect distribution for any generalization to be possible. And our results are fully consistent with that. He goes on, indeed, case studies show that distributional changes in a given country have much to do with the pace and structural features of economic growth in the period under analysis. And what our research adds to that is to say we agree with everything he said, and we would add that these case studies indicate that the pattern linking growth and poverty is mediated through labor markets. And so this leads to a call for more labor market research in Latin America, but also studies like this in other parts of the world, because we don't know what the story is for Asia, for Africa. I would love to know it if somebody did it, but I think I've burnt myself out in doing this kind of research for other countries. That said, we're about to hear from one African country right now, and so I will step down.