 This is a joint work with Nils Rambrusse-Mellar and Nils Tarp. And the overriding question was, how does aid work in the macroeconomy? There are many different conclusions based on the use of essentially the same publicly available databases. And if you use the same data and you get different conclusions, then I think there's just one possibility that the differences have to be due to the choice of econometric methods or models. And we have in particular looked at the effect of say different assumptions of exogeneity and endogeneity on the results. And this kind, the endogeneity and exogeneity has already been discussed a lot. I think it has a slightly different role to play when you have time series analysis, but the effect is exactly the same. If you assume exogeneity and you have endogeneity, then you can easily get to zero effect. But if you can disentangle them, then you can see it actually that there is a positive effect of aid on growth, for example. We have also looked at data transformations in time series analysis. Again, data transformations can be very risky. And most studies are based on per capita aid, per capita income, for example. And if population is actually, say, changing in a non-homogeneous way over the sample period. And with this I mean now that the proportion of, say, adults, old people and children changes significantly over this period. And for the data we are analyzing, the period is actually from 1960 to 2010, not 2007 in this case. So it's a very long period and you can expect that the composition of population has changed quite dramatically over this period. If you then divide by a population, you will get an additional effect on the parameter explaining how aid influences growth, which has nothing to do with aid. It has just to do with the change of population. So you also have to be very careful there. Then of course we are doing it in a system approach, meaning that we have a system of five equations, which is different to the usual approach, which is a single equation approach. And the reason why we use the system approach is that it actually, then you can address the endogeneity and the exogeneity questions. You can actually address most of the problems that possibly harm the existing analysis reported in the literature. It is a very broad approach. So the purpose of the study is then to offer an econometrically coherent and transparent picture of how aid has worked in Sub-Saharan Africa, which is of course one of the poorest countries or areas of the world. And we would also like to assess previous results in the literature within our very broad econometric framework. And also to address the widespread misuse of statistical insignificance as an argument for aid in effectiveness. And Fin already mentioned this part, so but it is equally relevant in the time series literature. The econometric approach is based on the so-called Cointegrated VAR model, which is a fairly complex method, and I will not, I will spare you from all details. You will not appreciate them anyway. And it is a system approach. So we have five equations in our system and the specification is generally based on a broad statistical characterization of the data. And so we start with a very, very broad characteristic representation of the data and then we are testing all kind of theoretically motivated hypothesis so that in the end we end up with a much more narrow model but this model is completely consistent with the information in the data. We do not impose any restrictions which in a sense violate the information in the data. And that means by this kind of procedure it means that this analysis will provide what I call broad confidence intervals within which the empirically relevant claims should fall. And then you can say, well, all models provide confidence bands and they do but they provide confidence bands within the context of that specific model and the choices being made. Whereas this is such a broad approach so when we get a confidence band it is a confidence band for that empirical reality not just because the choice of model. So some of the summary results. So, let's see now, the first column. We have tried to distinguish between these two. Eight has a long run effect on the macroeconomy that is the first column or eight does not have a long run effect on the macroeconomy. And you can see, oops, that was not. And you can see that in the first column we have actually 20 plus seven of the countries indicating that eight actually has a long run effect. Then we also distinguish between whether say the macroeconomy has affected the eight. This is, I have to point here. This is now the two groups where we say it's not the direction of causality is not just from eight to growth but also from growth to eight. And the lower one says that there is no link from macroeconomy to eight. And it means that for the first group here that is what we call that we have endogeneity between the eight and the macroeconomic variables. And the second one here, that is where we would say we have actually exogeneity, eight influences the macro growth and the macro variables but not the other way around. And I think in almost all the studies I have seen at least, this is more or less what is assumed but there's only seven countries that actually satisfy this assumption. Whereas most of them are actually in this group. Then we have the next group where we say eight does not have a long run effect on the macroeconomy. And there we have the first one, first group where we have seven countries where we would say eight doesn't push the macroeconomic variables or GDP growth and so on but actually the macroeconomy affects eight. That could have been more or less what Channing showed. No, it was Sam with eight and education. For example, that was a typical example of this case. Whereas then we have two countries left where we say eight and the macro variables are essentially unrelated. And this is a very, in a sense a very puzzling outcome and because it is, because you would expect eight to have some effect. So we have been looking more closely in more detail on these countries by including the effect of an open economy, real exchange rate and also including the effect of inflation which has not been in this part, in this study. And then when we do that, the puzzling results disappear and what we find is that eight is actually effective. So in a sense, most, you can see already here, most countries fall into the group which is saying that just eight works. And for the few countries where it doesn't seem to work, we can actually find out that it is, it is actually, we can reverse that outcome by including more variables. Of course this study is based on the effect of eight on GDP growth, on investment, on government consumption and private consumption. And of course there are more variables, we all know that. And if you include more, then of course you will also get more detailed results and more precise results. So the major conclusions is that eight has had a positive long run effect on the key macro variables for the vast majority of countries. And only in three out of 36 countries did we find a negative effect of eight on GDP or investment, either or. And because in a sense that negative effect actually also, it was one in one of the countries which we have made a more detailed analysis on and that effect disappeared in those countries. So essentially I think the claim that eight can even be harmful, I think there is absolutely no evidence for that in this study. And then we also find the results of our analysis study show that the transmission of eight on the macroeconomy has been quite heterogeneous and we think that the country specific approach is therefore vital. Then there are also some econometric conclusions, we asked the question in the beginning, does the choice of method matters and it seems to matter. And it seems also critical to distinguish between the effect of eight in the short run and in the long run. And I think it is also extremely important to use a system approach in order to get a correct estimate of the effect of eight in time series analysis. And the third thing is that it is absolutely essential to account for changes in political government, wars, conditionalities, major reforms and extraordinary effects such as droughts and floods. If you don't do it, you will bias your results.