 As Andy announced, I will be presenting the case of Rwanda. I presented a similar presentation here five years ago and I was asked to present an update. With broad in the title, I refer to the fact that I will not only focus on the economic dimension but I will also touch on the political spheres. As you know, the debate on Rwanda is very polarized. There are scholars looking at the economics and they are very positive and then you have scholars looking at the politics and they are very negative. I will look at a bit of both and so there will be bits of positives and bits of negatives in my presentation. Now regarding the update, this can on the one hand be taken quite literally. I will present an update of the figures on poverty, growth, inequality and governance indicators. But I will also try to address some of the questions that I was unable to address five years ago and I think I can do so thanks to a new data set. I will talk a bit about horizontal inequalities and about political representation. And I will end my presentation with a cliffhanger in the form of a two-dimensional puzzle on Rwanda so to make sure that you're all curious and invites me back within these five years to present some answers to that puzzle. The easy part, growth, backline is what I presented in 2013. Redline is what I added for this presentation. You can see a continued pattern of growth. Rwanda is growing at 4 to 5% annually so actually 7% minus 2.5% population growth so 4 to 5% per capita growth and has landed in 2017 at a level of $1,900 per capita. Looking at poverty, it all depends. It depends on what poverty line we are using. If we use the national poverty line then we see that across the two latest integrated household service of 2010 and 2013 poverty declined with 6 percentage points. If instead we use the international poverty line of $1.9 which is higher than the national poverty line we find that poverty remained constant almost across that period. Now this contrast of course between the numbers is a theoretical possibility because the international poverty line is higher than the national poverty line. Nevertheless I should mention that the latest poverty estimates based on the new national poverty line have been heavily criticized and this critique came following a kind of discovery that the poverty line calculation drastically changed. So the method to calculate the poverty line drastically changed in the latest survey of 2013. In the previous surveys including the 2010 survey the poverty line calculation relied on revealed preference theory and so I used the prevailing culinary habits of poor-owned households to construct the food basket, minimum food basket. But in the latest survey the Iranian government deviated from the convention to use revealed preference theory and set a poverty line based on cheap and highly caloric food types. So they filled the food basket with cheap and high caloric food types such as cassava and sorghum. And so as a result the poverty line was set much lower and actually what you do when you use these two different poverty line calculations you're comparing the incomparable and basically overestimating the decrease in poverty. This point was first made by Philippe Reangens in 2015 on a blog on African arguments. And the point was backed up later by a number of anonymous researchers in another blog post. I'm not 100% sure who these anonymous researchers are but I'm pretty sure they do not work at the White House. They constructed a set of comparable poverty lines and they show that using these various comparable poverty lines to compare poverty changes between 2010 and 2013-14 what you actually find is an increase in poverty of around 6 percentage points. So the discussion on this is ongoing if you browse the internet you will find many other blog posts and calculations and people playing around with inflation and so on. But to say the least the poverty figures on Rwanda are highly contested. A data set that is much less contested is the one based on the DHS numbers. And so here you clearly see that there has been continued progress in the dimensions of asset wealth, education and health. In this table I highlighted the periods of progress in green and you see that also across the letter period. So the updated figures, the progress is continuing. For instance access to electricity, more than doubled, maternal mortality, more than halfed. So here the picture is rather too very positive. The DHS data can also be used to check whether there is a convergence across wealth quintiles regarding health and education. So in this table you see the ratio of a couple of health and educational indicators between the top and bottom asset quintiles. And so this ratio is decreasing also across the last two years indicating that the difference between top and bottom quintiles is becoming smaller and so there is convergence in human development. Education and health improvements are broadly shared and more and more so over time. So let me now move to three important questions. Here they are in the order of less to more controversial. Let me start with the first one. Can the poverty statistics be reconciled with contrasting findings from qualitative field work? There is a very large body of qualitative field work on Rwanda that is much more skeptical about economic development in Rwanda. Here's just one example. Ansoms and qualters, statistics versus livelihoods, questioning Rwanda's pathway out of poverty. These researchers find from their semi-structured focus group a much less rosy picture on economic development. And they try to explain the contrast with the conventional growth and poverty statistics. And one of the reasons they put forward is that because of social desirability households actually overreport their production in the standard conventional service run by the government. Why would households do that? The authors of this paper refer to the culture of performance criteria and contracts in Rwanda. And mentioned that there are even cases of households signing performance contracts and committing themselves to reaching a certain target. Much in line with the Emehigo contracts or performance contracts signed at different levels of local government. I will not forcibly reject this claim. It is theoretically possible, but the evidence for it is still thin, so I would still say that it's somewhat speculative. Another possibility to explain the contrast between the conventional statistics and what emerges from qualitative field work is that subjective and objective well-being measures can both be right, but they measure very different things. So this is one of the arguments I made five years ago that subjective well-being measures may capture relative income rather than absolute income. Recently I got the opportunity to work with a unique data set that includes an evaluative yet numerical measure of well-being reported by Rwandan households. And this data was collected by my colleague Berth Ingeleire, who is an anthropologist, and he interviewed 412 Rwandan households in 2007 and in 2011. And the households were asked to tell their life history, which is a typical method of anthropologist. But a bit atypical for an anthropologist was that Berth asked the households to systematically rank their economic situation through time. So we have a ranking of economic situation for each year of the life history of these 412 respondents. And so here I plotted this ranking on a graph for the period 1990-2010. And so two things are remarkable in this graph. The first thing is that in a period of a doubling of GDP per capita and a decline of poverty with 20 percentage points, the rankings indicate an economic situation that is stable, flat. And secondly, the ranking according to the respondents of the life stories in the post-genocide period is even lower than what they reported for the pre-genocide period. So very sharp differences compared to the conventional statistics, leading me to conclude that indeed these measures of well-being measure very different things. So using the subjective measures to question the reliability of objective measures per se is probably not a good idea. Nevertheless, the subjective measures are valuable because they reveal something about what anthropologists like to call the lived experiences of people. And these are arguably very important. Should they receive more weight in the policy debate? Yes, but I think there's still a long way to go in order to standardize their collection and validate them. So I don't think it's for tomorrow. What is interesting also about the life history data set is that it is actually one of the only data sets with recent information in which also the ethnic identity of the respondents is known. So we can dig a bit into the question of horizontal inequality, although it's perceived horizontal inequality because it is the economic situation as reported by the respondents themselves, their own ranking. Now looking at horizontal inequality in Havana is of course important, not only because of its history of ethnic violence, but also because the Tutsi minority dominates the high-level positions in government. This graph gives you the ethnic identity and gender of cabinet ministers throughout the period 1995-2015. And so you see that the Tutsi who constitute 15%, 10% to 15% of the population holds 60% of high-level cabinet positions. Among those Tutsis are Tutsi returnes that dominate. Tutsi returnes count for about 5% of the Rwandan population. So there are some concerns here and therefore it's important to look at horizontal inequality. However, maybe somewhat surprising or not, it all depends on what side you are in the Rwandan debate. When we look at the self-reported rankings of economic situation of Hutu and Tutsi, we don't see a lot of difference. So we don't see evidence for perceived increases in horizontal inequality. We do see that Tutsi returnes rank their economic situation a bit better than Tutsi survivors, but the difference is certainly not big. The trickiest question is left for last. Can development in Havana be sustained if the country continues to score very low on voice and accountability? So that should be severe, not several. Well, there is widespread consensus among Rwandan scholars, scholars of Rwanda, that there is severe societal and political repression in Rwanda. It is also reflected in a very low score on voice and accountability. This is a problem. Will this prove the accurate heel of the Rwandan success story? Some scholars would say yes. For instance, Rangians argues that the concentration of power that we see today, the power structures we see today, the governance style we see today, is very similar to the one that led to the genocide in the first place. A political economy analysis a la Asemogla and Robinson will predict that growth will run out of steam because the elite will try to protect its economic interest and so will not allow for creative destruction and innovation to take place. But equally bright scholars would say that it's not a threat to economic development. They categorize Rwanda in the box of a developmental state where there is interference indeed of the elite in the economic sector, but rents are centralized and used for broad-based and long-term development. Who will turn out to be right? Your guess is as good as mine. But maybe our guesses can be better informed if we know what ordinary Rwandans think about their political representation. So back to the life history data in which respondents were also asked to rank their perception of political representation over time. So what do we expect? Flat, declining, increasing. Your guess is not so good as mine now because I know the data, but it's increasing over time. So since the genocide we see a steady increase of self-reported rankings in political representation by ordinary Rwandan peasants. You see that the perceived political representation is markedly higher for Tutsi than for Hutu, and this is a reversal compared to the pre-genocide period when Hutu were in power. But the difference is actually shrinking over time and not increasing. So if you compare, for instance, with 2011 you see a much smaller gap in 2011. Political representation as perceived by Tutsi survives and returnes is almost similar. So here is the cliffhanger I promised. Two-dimensional puzzle on Havana. Conventional statistics say all is good on the economic front. Nothing is good, that's not true, but voice and accountability is flat on the political front. Life history data also sees this big contrast but upside down. Economic situation is flat, political representation is increasing. So there is still a lot that we don't know about this puzzle. It indicates that reality experiences and perceptions are complex and we still have to do some additional research. We will probably not solve the entire puzzle, but in the life histories there are lots of elements that can indicate some useful directions to start solving this puzzle. Every time a respondent changed its ranking over time in economic situation or in political representation, he or she was asked to explain the reason why. And we have thousands of narratives in the data explaining a change in political representation, explaining a change in economic situation. And here I give you already a flavor of the narratives that we will start analyzing in the coming months and years. These are narratives of change related to an increase in perceived political representation. I started being sure that the authorities would not kill me. There were no longer persons disappearing from the community. I was transferred from one prison to another. There we could find water to drink and wash ourselves. Clearly, the reference point is not the perfect liberal democracy, but the reference point is quite low. It's not very difficult to go upward, but that's only one conclusion that we draw from the narratives. There are also many narratives that point to real improvements in accountability, although accountability that runs upwards, not downward, and also real improvements in service delivery. So more to come on this front in the coming months and years. We will be systematically coding and analyzing these narratives and trying to solve these two-dimensional puzzler on Havana. And I leave you with the conclusions, because I think the alarm will go off. Thank you. You almost escaped the Tim Burton shock.