 I've been running the International Centre for Tax and Development for the last five years, and in the course of that time I've actually changed my mind on several issues about taxation. The first one I have radically changed my mind, but certainly to some extent changed my mind on, is that I have realised just how poor both the volume and the quality of data on almost any issue around taxation is especially in developing countries. The gap between the very wide range of theories, propositions about tax and the actual data that we have available becomes more and more striking. The one big contribution I think we've made is as Alex has just explained the National Revenue data set, I really should say that this is very much Alex's baby. I mean, if Alex really started this and pushed it, if it hadn't been for him, it wouldn't have happened. So, Alex again, thank you very much. Kyle has made a major contribution to this. My colleague Wilson Pritchard who's not here has made a major contribution. You might have thought from what Alex said that we have somehow persuaded wider to adopt a very badly behaved teenager here. But I think Tony you'll agree that I think you came into this with eyes wide open as it were and you realise that these data are better than bad but there's still a long way to go and I very much hope you're able to improve their behaviour enormously in the next few years. I'm very pleased you're taking this over. What I'm going to do briefly, Cary, is talk about a few other issues about data in taxation and relation to developing countries with some reference to what I think are interesting research areas and it's probably not a coincidence that I'm going to talk mainly about some things in which my centre has been involved but I'll try not to make it too focused on our work. The first thing I would say is that it is very clear that one of the very big data gaps in relation to developing countries is data or least publicly available data about almost anything in relation to the extractive sector and especially to the mining sector. I will just say here briefly if anyone wishes to talk about it later I'd like to do that. One of the things we hope to do in the future is to make a major push over obtaining and making publicly available data over extractive projects and extractive taxation in Africa in particular. There are a number of publications I have here that I'll put that available later. Some indication of this is a working paper we've just published by what do we know about mineral resource rent sharing in Africa that I will make available later. Another growth area is the use by researchers of what we call administrative data that's to say the data available with the revenue authorities themselves. Using administrative data for research purposes is becoming fairly common in the OECD countries still not very widespread but it's becoming fairly common. Fairly rare so far in sub-Saharan Africa but it is becoming just about possible raises many challenges in terms of actually making data usable for research purposes. But just for example some Ethiopian colleagues have been using administrative data from the Ethiopian Revenue and Customs Agency to look at the impact of introducing electronic tills for registering VAT and found the data usable just about and found a very positive effect of introducing electronic tills. Another area which has been a very big growth area has been in the, sorry not a growth area I think is potentially a growth area is real experiments cooperating with revenue agencies to look at different ways for them to approach and engage with taxpayers and see what is the most effective in terms of generating compliance. Relatively technical routine staff fairly common now in the OECD countries very rare in developing countries certainly in Africa. We have a rather large project with the Rwanda Revenue Authority and I think the one very important lesson from this is that you have to have a very good revenue authority in order to do these experiments because to actually organize in such a way that at a certain day of the month you send different messages to taxpayers or different letters to taxpayers does require a very effective organization. The Rwanda Revenue Authority have been very good at this but it's been very challenging and it's only because the commissioner general personally is backing this that I think so far we have managed to make the first round of those experiments. Another quite big growth area especially in political science recently has been simulation experiments sitting people in laboratory like conditions and asking questions about tax paying and setting up games in which they play essentially tax paying games with the idea of getting some sense of their underlying attitudes and likely behavior in relation to tax paying. Now I feel very ambiguous about this because on the one hand it's very interesting and one of the couple of names from political science are very prominent Laura Paylor and Lucy Martin. I've done very interesting work. Lucy Martin has run a whole range of experiments in Uganda recently that essentially seem to show that where taxpayers believe that tax money has been misused by political leaders they seem much more likely predisposed to take risks to punish those political leaders than if they believe the money had come from some other source not from taxpayers. It's very interesting. It certainly conforms to all our kind of that's what you would expect to find in a sense so it's not surprising. I am actually a bit suspicious or wary of how far one can actually push these kind of experiments because ultimately you're always in an artificial experimental situation. You're not actually standing in front of the president and telling him you stole my money and therefore which is a difficult different situation. But there will be a big growth in that kind of activity in future. But I have another concern about that kind of experiment which I think is slightly more important for the political economy which is that most of those experiments are based on a model a political model of the tax relationship that is a model of between a state and a taxpayer with the assumption that there's one political authority and a fair degree of homogeneity among taxpayers. Now that is a very problematic assumption. I don't know if any of you have tried to use the afrobarometer data on taxation from the last round my centre actually funded quite a lot of extra questions on taxation from afrobarometer and so you'll get much richer data there than previously. I'm not entirely sure in retrospect we are right to do that because the problem with afrobarometer is they go and they talk to a very large sample of people in African countries and ask people questions about tax paying. Did you pay your taxes? What do you think about paying taxes? Without ever asking what kinds of tax people might pay or should be eligible to pay and you might say in practice there is no such thing as a taxpayer in Africa. There are a very diverse range of different people who have very different potential relationships to tax authorities. To characterise very crudely I would say that very few people in Africa pay in the sense of directly take from their pocket and hand over any direct taxes to central states. People certainly pay indirectly via customs duties and VAT and excises etc but the very few are actually in the position of handing over money to central states. Central states are funded very heavily through either taxes on corporations or taxes raised by corporations, corporate income tax, VAT etc. Most people in Africa are probably and this is a guess more likely, well I know it's not a guess, I know they are more likely than to pay to central state to pay to an incredibly wide variety of sub state and non state actors and all kinds of other people who are somewhere in between that. It is a very very confusing kind of situation. Some of my colleagues are currently doing a very large scale survey in Congo on what we call informal taxation and in order to actually get data on what people pay they have 400 separate categories of tax, 400 different taxes. That's partly because they differ so much from locality to locality that you can't standardise. But in that kind of situation I think those sort of political economy models of the idea that taxation and governance are sort of directly linked in this relationship between the taxpayer and the state are problematic. That relates to another respect in which I have I think probably changed my mind on tax issues to some extent. I used to be a great advocate of the what you might call the social contract model or the good governance model of taxation which says that the social contract model or the good governance model of taxation which says that the more governments raise tax on their citizens the more likely you will be to get good governance. I still think there is a lot of truth in that proposition but we have so far found no evidence to contradict that proposition but the actual political relationships of taxation in Africa are really so complicated that it is very hard to test that proposition very clearly. Amongst the various things here there is something by me called the tax and the governance dividend which represents a row back from my previous very militant position on this and says well I'm not quite so sure after all that this is quite that simple. One other issue I would just like to mention briefly this is also in a sense not where I've changed my mind I think it's more complicated. I've been a great advocate of property taxes for the same reason that almost all economists in the world are great advocates of property taxes because they generally thought to be incredibly good taxes and I still believe that but we know property taxes are grossly underused and I now better appreciate one reason why they are grossly underused in developing countries is because there is an intimate relationship between property taxation and land rights and land registration and I'm not going to say any more but there is an excellent working paper by Tom Goodfellow on land rights and urban taxes in Qigali and in Addis that just illustrates how incredibly complex these relationships are and often not very clear to the actors themselves what kind of categories they are actually using. I want to just conclude to go back to the I mean most of the things I've raised now are if you like questions things to which we don't know the answers but I want to conclude by going back to the national revenue data set which has been available in a research usable form for well over a year now and one thing that some of my colleagues did early on was to test the proposition in which I used to believe very strongly I should say which is that aid decreased the domestic revenue mobilisation effort I used to believe that because it seems so sort of logically and intuitively obvious well it doesn't there are still publications from some colleagues in IMF that claim that there is a question of why they claim that but I'm simply going to say it's because they're using bad data and there are a couple of our working papers that are going for later publication on the whole question of foreign aid and domestic taxation one by Paul Clist, the other one by Oliver Morrissey who's now in another meeting Wilson Pritchard, Samantha Torrance both which look on the basis of cross-national statistical analysis at the connection between domestic resource mobilisation and aid and the answer seems to be rather firmly that there is no general depressing effect of aid on domestic resource mobilisation which is hopefully a very good news at least from the point of view of people who work in the aid business thank you