 It is a pleasure to come here and present this paper, which is building on the research conducted for my recent book, Poor Numbers, how we are misled by African Development Statistics and what to do about it, but also responding a bit further to the kind of ongoing discussions on arising from that debate on Africa Statistical Tragedy and concurrently debate on Africa Rising and some of the issues I'll be dealing with today is what kind of data do we have to interpret growth inequality and poverty living standards across the past two decades? Do we have any alternatives? And what can we do about it in terms of investing in better data for development in the future? First I should give a full disclosure that some people have recently drawn my character in considerable doubt. The directors of statistics in Zambia concludes in a public statement that it's clear from the asymmetric information that he had collected that Mr. Gervin had some hidden agenda which leaves us to conclude that he was probably a hired gun meant to discredit African National Accountants and eventually create work and room for more European based technical assistance missions. Now you can trust me or not, but I assure you this is not the case. I'm not a hired gun and for those who have read the book if there is one thing I do not advocate it is those policies that would create more room for per DMs for European consultants. I actually describe that as part of the very problem. And I should also say I think the director of statistics gives me way too much credit. I have not single handedly ruined African statistics. The credit is due elsewhere. Pauline Ola did also make a statement to the press yesterday saying that Morton Gervin will hijack the African statistical development program unless he stopped in his tracks. I was supposed to on Sunday I was supposed to board a plane to Addis Ababa to speak give the opening statement at the United Nations Economic Commission for Africa to give to give the opening statement there in order to introduce the African working group on national accounts in order to see what we can do about investing in development statistics. My paper is why did we need to invest in economic statistics getting the diagnosis right before boarding the plane. I got a phone call from UNEKA ordering me not to board the plane and and also informing me that Pauline Ola had given a ultimatum that he would withdraw from all working groups unless my speech got cancelled. So it's a pleasure to come here and talk to you with change of plans. You'll find the news stories media out there and any kind of support can give me in this moment much appreciated unless you happen to think I am a hired gun from European powers. The book that has created all this commotion is poor numbers how we mislead by African development statistics and what to do about it. And for those of you who are intrigued by this debate I encourage you to read the book rather than the reviews by Pauline Ola and actually judge for yourself and you'll find out that this is a huge misunderstanding. My main contribution of the book as we'll talk about today as well is to first state that they do have a serious knowledge problem. We know less than we would like to think about economic growth in Africa. I do write a unique story in it's I'm an economic historian. So this is the the craft of a historian writing the history of the national statistical offices in the region. How were African economies measured since colonial times as an independence. The statistical offices are new new types of demands new kind of tasks new kind of administrative survey data available coming with development plans so far I trace what I where I think if there is an African statistical tragedy it goes parallel with the type of decline in economic growth. The economic shocks that face the African economies but also statistical offices in the 1980s where also the statistical office have faced new demands from the growing from from structural adjustment programs liberalization the challenge of recording informal markets. Finally I talk about how the poverty reduction agenda has put new demands for new types of data and the millennium development goal agenda which is our current agenda where we're asking statistical offices in the region to collect data on eight goals 18 targets and 48 indicators and thereby diverting resources but I'll talk about that my main my main contribution about this book is to to elevate the discussion we have about African development statistics and to get around doing better in investing in statistics. Today I'll talk about GDP in Africa diagnosing the knowledge problem look a little bit on proxies and I'll look into some kind of ways in which some very brave papers have tried to fill some data gaps and then thinking about what we should how how happy we are about that. As we know there has been there's been a very clear symptom of a problem expressed when gamma revised their GDP and in November 2010 overnight GDP almost doubled in the GDP per capita terms as the economy was accounted change their base year from 1993 to 2006 in in in 2010 they changed the base year. It also meant that Ghana before the revision they were a poor country after the revision they were a lower income a middle income country you know this is good news but there is a knowledge problem that does emerge where did this extra Ghanaian economy come from how do we now compare Ghana 2006 to Ghana 2005 where the series have not been backcasted how do we compare Nigeria to Ghana to other countries and it has spread a lot of reactions you know Todd Moss at the Center of Global Development exclaimed boy we really don't know anything before he moved on to to note that if this could happen in Ghana overnight without observers really knowing about it what then about other countries that are less closely observed and the Sumner and Charles County took the good news in this and said it proves that Ghana escaped the poverty trap there is no bottom billion they went out of it overnight United Nations Development Programme said this is a statistical illusion the millennium development goals indicator did not change overnight meanwhile Shanta Dravarjan after having heard about the revision and read my paper on this declared Africa's statistical tragedy so that's where we're getting at and it turns out news after this is pointing to the direction that Ghana is not a singular case the base year currently being used for accounting for economic growth in Nigeria is from 1990 that means that we do it's now approaching quarter of a century since we had reasonably accurate GDP statistics of what is probably the biggest economy in Sub-Saharan Africa and I'm saying probably because one of the big stakes that comes out of this is that if the rebasing is as big as they have announced it will be before it is ready I might add and it would imply that it might jump out jump South Africa GDP in Africa in total increases by at least 15 percent and it should be a bit striking that there is currently existing about 40 Malawi's inside of Nigeria which is not accounted for which I will talk about effects how we think about long-term trends in not only growth but also poverty and inequality so what do we know about income and growth in Sub-Saharan Africa while basing from from how we can download the data from the World Bank database you would think that you know quite a lot because there you find data from every year from every country from 1960 to 2012 all looking downloadable as functional equivalents now we know right now we know that we cannot possibly compare Ghana GDP per capita with Nigeria GDP per capita that would simply be misleading and tell us nothing about what is really going on and then this is also at stands with the fact that some countries have not in 2012 according to you NECA report the lag time is on average two years so it means that most countries are now working on their 2010 estimates so how come we are now already publishing estimates for 2012 another thing which I talked about when Ghana did publish a new series it base year new series in 2006 it was not backcast so we have a break series we don't know where this extra income come from we have reasonable accurate picture in 1993 we have reasonable picture of 2006 it's anyone's guess what went on in between and so how do they come up with these numbers or more specifically where does the international databases get their data from well some of you might know this but I hope at least now everybody knows if they come from statistical services for you have to go to the statistical offices so even though you download a GDP statistic from the World Bank the observation is not better than the one at the local level so we have to use this is a distinction between primary and secondary data you have to use the skills of a historian or the questions of a journalist who made this observation under what conditions was the observation made how good is that observation is there any reasons why I would think it is biased in any direction and I think we've forgotten to do so here's the Ghana statistical services where I conducted my interview with Duncan my head of macroeconomic statistics in Ghana before the revision here's the informal sector right here it's the peanut sales woman there where I used to buy my lunch I asked Duncan you know is she covered in the new estimates he said it's just peanuts but it turns out when you do adopt these new service sectors it does really make a quite a difference here's the Nigerian living standard survey still in bags in Abuja Nigeria they're still inputting in the data I mean still awaiting the new results so that's what I've been doing that's the basis of the book is to ask how are what kind of sources do you have what kind of data are do you have do you have labor data you have a survey of the informal sector I don't my legwork on this in Ghana Nigeria Kenya Uganda Tanzania Sambia what's one in Malawi in addition I did a survey of all these countries and my question has been how has these countries measured their economic growth and so forth now in the book I give also a historical thing so I'm using our archival work to tell how this is going on from 1950 onwards today I'll focus in on on the last two decades and particularly painting a accurate picture of how much do we know so the basic kind of things as you already started to realize is to think about how long you know this is a survey done me published in the book I did ask a couple of questions like when did you last publish an estimate so what's the lag time between and some countries in 2010 a Ghana had a 2009 date they were up to date yeah but other countries did not have some countries you simply have no information on it whatsoever that might have been because there is no GDP data like in Equatorial Guinea and or it might be that they were too busy doing other things like in Ivory Coast and Congo understandably with the civil unrest at the time and so forth the other important thing is of course the base year now we have learned that the reason why Ghana's economy was took so it's a big jump is that they have now a new base year for 2006 whereas you know Nigeria is at 1990 so they these data are not comparable and and there is other estimates which I rather not you can engage with that table as well and the good thing is I did my table in 2010 in response to the publication of the book not only is the book peer reviewed and fact checked and all possible ways but I also had the honor of having my book being replicated the study of the book has been replicated by both the IMF and the African Development Bank in recently released reports and this is the IMF data and the report by the African Development Bank is not called poor numbers it's called the reliability of African GDP estimates and so forth but we are there to summarize that the my survey is published I had the sample of 37 countries in comparison AFDB managed to secure the response from 34 countries probably were in a bit of a hurry IMF managed to get this kind of information from 45 countries the IMF recommendation was to have a base year every fifth year seven countries met this in 2011 according to my study according to the latest IMF only four meets this and what I have to be reports that nine do you can that means that they have different information the mean base year is 2000 in all the studies so means that we are 13 years behind according to the African Development Bank eight countries only according to IMF 13 countries have base years that are more than two decades old this is not only a problem then of comparable GDP data as you see there's also a problem of comparable metadata so it's manifesto a little bit about how little we know that we cannot AFDB and IMF cannot agree upon the sources and methods not even baseline estimates in the same reports published in the same month this year implications for the growth evidence as I detailed in my paper and my book and so any ranking of course according to GDP is going to be misleading any statement of growth over the short and medium term might be picking up statistical growth changes its methods as well as actual real real productivity growth which we would like to think so there's a mitch match between how we often interpret the GDP data and what is actually going on and very recent growth data for a different couple of reasons is likely to be very overestimated clearly in the case of Ghana although you try to smoothen this out you still have the revision effect with you and and it's also true that the GDP per capita of many countries are now underestimated as we saw Nigeria is richer than we think but that also means that currently Nigeria is growing faster than you think for a couple of reasons because you have a lot of statistical growth if your GDP measure is not exhaustive for instance so what are the alternatives which scholars have been using this is okay that is rainfall by Miguel et al to discuss in the paper one thing is to do is to try to control for GDP measures using rainfall for a lot I'm as I discuss I think that's a too narrow type of measure we are often interested in the stuff that is not fluctuating with rainfall but rather that which fluctuates with policy and therefore you get the very opposite kind of GDP measure than you would like and it's also this thing that it's kind of weird using GDP data and controlling from rainfall when you know that that's how they estimated the agricultural output in the first place if you know how crop forecasting works then you know that often rainfall data is the one that you use to estimate those very output data to begin with so you're kind of redoing what you already did but Miguel didn't know that assets young et al as young not at all he did this all by himself young in 2012 you journal political economy does yeah yeah yeah but still people are suggesting that maybe you could use trends in rainfall and so forth well okay that's fine I agree as well the instrumental process quite different yeah clearly but that's a way of getting along you know that you've they changes in GDP that are not explained by measurement you know so it's a way of so instrument is rather how to the thanks for the clarification assets data used by young and others also has different kinds of problems they are stocks not flow and other people have shown class net dollar shown that this is not really working out as a proxy luminosity many people are suggesting that we should measure economic growth from outer space yes that's has some types of advantages no bias perhaps except if there is legislation about light outside lighting that is but there is also the problem if you are an economic historian like me you have to you can't get data from the 1980s then because you don't have satellite imagery unless you're able to travel as fast as you can capture light recapture light which is a bit hard to do what I'm arguing in the paper here is that none of them give a predictable outcome and all science steps the issue of seeing like a state measurement is not only knowledge it's governance accountability and policy circles that's well displayed by I don't think a World Bank for instance will issue a which is the lightest country in the world kind of table at the time soon so filling the data gaps how is people dealing with we don't making brave assumptions between it so I'll just take a two minutes to go through one very much famous paper is African Party is falling much faster than you think they claim since 1995 African Party has been falling steadily they then use you know the wider data and the PVT data and calculate elasticities and transient equality on this the problem is that they have only 188 118 service so that there is you know a lot of the actual years in this data set is of course made up and not only that countries which they are actually claiming they even in the paper they have a figure of data on Angola and Nigeria but they do not have data of that invider as you would know the people who know it the same thing happens with when the rise of the middle class they think it increased from 11 to 15 percent again they have a method of using a synthetic panel and they only have 84 observations and the short story is that these 35 countries are not randomly picked and we are we are missing the big countries where we're missing the poor countries so therefore to go from these observations when you have as much as above thousand gaps from actually having annual series we're actually making very very bold guesswork indeed conclusion our current estimates are doubly biased we know less about poor countries and we know less about poor country poor people in those poor economies this knowledge problem stands in a striking contrast with the demand for numbers in development community everyone wants measurable impact of development policy the Africa rising debate has increased is even further there is even now a demand for statistics from from investment banks etc etc so there's an increasing demand for statistics I argued that numbers matter and evaluation of Africa's development must begin and then with a careful evaluation of the growth and income evidence without such analysis one runs the risk of reporting statistical fiction the poor numbers are too important to be dismissed as just that thank you very much thank you