Prof Morten Jerven
Address to be delivered at “Inauguration Meeting of the Continental Steering Committee (CSC) for the African Project on the Implementation of the 2008 SNA”[1]
There is a growing demand for economic statistics on the progress of African economies. This demand for data will be met. The key question is how these data by will supplied. Who will supply these data? What quality of data will be supplied, and how will the demand for data affect the governance of the central data suppliers – and key among them – the national statistical office. This meeting is specifically addressing the future of the provision of GDP statistics, and how to improve the data needed for economic governance of African economies.
In my book, Poor Numbers: How We Are Misled by African Development Statistics and What to Do about It, I provide a scholarly and objective analysis of historical and contemporary problems confronting statisticians in Africa. The book argues that one of the most urgent challenges in African economic development is to devise a strategy for improving statistical capacity. At the center of such efforts is the statistical office, which has long been neglected in debates on development in the African region.
As many in this room would recognize, the statistical office has not been at the center of attention for institutional reform, and it has less access to resources than other of its related institutions. In Poor Numbers I advocate a historical perspective on the statistical office in order to understand some of its current problems (Jerven, 2013, 5-6).
The statistical capacity of African states was greatly expanded in the late colonial and early postcolonial period, but it was greatly impaired during the economic crisis of the 1970s. The importance the statistical offices was neglected in the decades of policy reform that followed—the period of “structural adjustment” in the 1980s and 1990s. In retrospect it may be puzzling that the International Monetary Fund (IMF) and the World Bank embarked on growth oriented reforms without ensuring that there were reasonable baseline estimates that could plausibly establish whether the economies were growing or stagnating. For statistical offices, structural adjustment meant having to account for more with less: Informal and unrecorded markets were growing, while public spending was curtailed. As a result, our knowledge about the economic effects of structural adjustment is limited. More generally, the economic growth time series, or the cumulative record of annual growth between 1960 and today, for African economies does not appropriately capture changes in economic development.
The contrast to the Central Banks in the region is striking. The central banks have not only relocated to new high-rise buildings, but they can also offer higher wages to attract qualified staff, and have a predictable long term stream of revenue (Jerven, 2013, p.113).
Perhaps what is most striking is the contrast of this situation with that of another pivotal stakeholder in the economic policy process with much greater resource endowments: central banks. While statistical offices are located in rundown offices, often with limited computer facilities, as is the situation in Ghana, Malawi, Nigeria, Kenya, Tanzania, and Zambia, the central banks of these countries are located in new high-rise buildings with all of the modern facilities. Employment positions in the central banks command higher salaries and prestige, and central bank employees are in a better position both symbolically and physically to provide timely and useful advice to policymakers.
As should be clear, this question is not only a comparison between statistical offices and other institutions. As has been well documented in Poor Numbers, there has been a clear shift in priority away from the collection of some of the basic data needed in the compilation of national accounts, and a shift towards social statistics (Chen, Fonteneau, Jütting, and Klasen, 2013). This means that declarations such as ‘Africa’s Statistical Tragedy’ may seem unfair or misleading (Devarajan, 2013). It may be too early to declare the ‘African Statistical Renaissance’ (Kiregyera, 2013). It is of course true that there has been a growth in output of numbers from statistical offices, in response to the growing demand. But the progress has been uneven.
I believe that the recent debates rightly draw the attention to the key problem – the lack of data needed for economic governance. However, the debates in the international development community have so far focused more on issues of comparability of these statistics, than the provision of statistics. The first question is what I call the knowledge problem. In this paper I lay out some of the problems of getting a correct diagnosis of the knowledge problem. The metadata is lacking, and as we will see, existing fact-finding studies on the state of economic statistics in Africa, while agreeing about the general problem, find quite different results on the details of the problem. I call for a transparent exercise that assesses what the gaps in the knowledge actually are. In any debate on ‘Africa’, the problem is always that it disguises the heterogeneity of capabilities, outcomes and problems in individual countries. The heterogeneity does of course not only vary by country – some areas of statistics are in better state than others. Moving ahead one needs to needs to not only take into account what is known and what is not known. It is crucial that we address the issue of economic statistics with a holistic approach to the statistical systems of each country.
This takes us to what to do about it. There has been less focus on what I call the governance problem. This means dealing explicitly with tradeoffs – how expensive are the different indicators, and how does the provision of one number, rather than another, affect the general goal of the statistical office – which should be to provide reasonably objective evidence on those variables that are of interest of the public. Statistics is a public good. There are well established theorems why the supply of public goods is a public matter – and therefore most of times and places, but not always, pertains to the state. There are also well-established theorems that warn us that there is an inherent risk of the tragedy of the commons in provision of public goods.
There is considerable evidence on the current provision of statistics in Africa that these very typical problems are emerging. Indeed, the analysis conducted in Poor Numbers not only documented that statistical offices had been completely incapacitated – and this did not only relate to extreme cases, such in periods of civil war. The dominant pattern in the 1970s and 1980s has been that IMF and World Bank started to publish numbers before they were ready at the statistical offices. This still happens. The UNECA background document here documents that the mean time for completing GDP numbers is one and a half years (UNECA, 2013, p.4). The mean is not the most important here, and nor does this say anything directly about the quality of the estimates. It has been the case in history, and not only because of timeliness, that even Central Banks and Governments in the same country do not use the data provided by the statistical office. As I document in Poor Numbers this is a real threat in some countries. As I also document in Poor Numbers it is already the case that the World Development Indicators report different data than the ones published by statistical offices. This can run into a vicious cycle. According to the StatCap website, the rationale for the program is that “many national statistical systems are caught in a vicious cycle where inadequate resources restrain output and undermine the quality of statistics, while the poor quality of statistics leads to lower demand and hence fewer resources” (Jerven, 2013, p.103).
This question of governance is twofold. One pertains to how to regulate global demand for African statistics. This is what PARIS21 and the National Statistical Development Strategies are trying to co-ordinate. There are other efforts, described eloquently in the background document to this meeting (UNECA, 2013), and in many contributions such as the address delivered by Ben Kiregyera to the meetings on African statistics in Vancouver (2013). My interpretation is that international coordination of donors will not work without credible rewards and punishments for breaking with agreed upon priorities. This would require a complete rethink of the data agenda in post-2015 MDG. The other question is the far more difficult one – to think about how to regulate supply. Statistics is a public good, but sometimes there are private rewards in providing it. This means two things for funding of statistical offices. First, funding would need to be regular and long term, and second relatedly, rewards to employees needs to be based on base salaries and not per diems that currently create pervert incentives in provision of statistics.
Getting the diagnosis right: beyond ‘Tragedy’ and ‘Renaissance’
How much do we know about income and growth in Sub-Saharan Africa? In Poor Numbers, I attempted to survey the status of GDP statistics all countries in Sub-Saharan Africa, and in particular to collect information on the methods and data used to compile national accounts. The table I compiled in 2011 contained information on 37 countries, and showed that only 10 countries had a base year that was less than ten years old (Jerven, 2013, p24-25). I further showed that seven countries had a base year that was more than two decades old (1990 or older), and that there was only 6 countries (Burundi, Ghana, Mauritius, Niger, Rwanda and Seychelles) that followed the advice of the IMF to have a base year that was 5 years or newer (2006 or newer).[2]
In response to this survey, the African Development Bank, commissioned a study, published in 2013, which provided information on the same variables (AfDB, 2013a).[3] The AfDB attempted to get a response from all 54 member countries, and received a response from 44 of them.[4] In the survey of base years, the AfDB report results from only 34 countries (as compared to 37 in Jerven, 2013), and reports: “the base years now being used for constant price estimates in 34 countries. Only nine – Cape Verde, Egypt, Ethiopia, Djibouti, Guinea, Malawi, São Tomé & Príncipe, Togo, and Zimbabwe – have base years that meet the five-year rule (i.e. 2007 or later). Nineteen countries have base years that are at least ten years old, and eight (Benin, Central African Republic, Comoros, Congo Republic, Madagascar, Mali, Nigeria, and Sudan) use base years that are more than 20 years old.”
The second report in 2013 that set out to replicate the collection of some metadata on GDP statistics, in response to the attention brought to the importance of base years in Poor Numbers, was published in the IMF’s 2013 Regional Economic Outlook for Sub-Saharan Africa in May (2013, p.6). According to their survey of 45 countries, summarized in a table only four countries meet the so-called five-year rule (with base year from 2007 or newer) – Cape Verde, Malawi, Mauritius and South Sudan.[5] Recall that Jerven (2013) found in 2011 that seven countries met this criterion, with base years of 2006 or later. If, for the sake of comparison, we would consider base years 2006 or later as meeting the five-year rule, thus including among others Ghana, you will find that according to the IMF, 11 countries fall within this range in 2013. How come the IMF and AfDB differ so widely in their information? Well, the two institutions agree upon the base year of Cape Verde and Malawi being recent, but the AfDB either did not get information on Mauritius or missed it in their count. Egypt and Djibouti are not included in the IMF table, whereas the countries that AfDB reports as having a base year from within the last 5 years include: Ethiopia (whose base year is 2000 according to IMF); Guinea (whose base year is 2003 according to IMF); Sao Tome and Principe (whose base year is 1996 according to IMF); Togo (whose base year is 2000 according to IMF) and finally also Zimbabwe (whose base year is 2000 according to IMF).
The IMF concludes from their report (2013, p.4) that their “median base year is around the year 2000, which, although now 13 years ago, is more recent than had been suggested by Jerven (2013)”. In Poor Numbers I purposefully did not report a mean or median year, because I am not sure if it is a useful statistic. The samples reported in Poor Numbers and by the AfDB are both positively biased. We do not have responses from countries that are in more economic and political distress, which we would expect, all other things being equal, affects the timeliness of economic statistics negatively. For the record, in my sample the mean and the median base year is 1999 and 2001 –my book paints a similar if not more positive picture compared to the data reported in the IMF table.[6] Meanwhile, the AfDB (2013a, p.5), conclude that: “Overall, the situation with regard to GDP is not nearly as bad as has recently been suggested”. It is not clear what this conclusion is based on, but in the same executive summary it is noted that (2013, p.5):
A country’s GDP estimates are only as good as the data on which they are based. Although industrial production is believed to be rising sharply in most countries, nearly one-fifth of the respondent countries had not conducted an industry survey since 2000. Even fewer countries conduct regular surveys or censuses of agriculture, despite its criticality to the food security situation in the continent. What is equally surprising is that Algeria, the Democratic Republic of the Congo, and Nigeria, which are three very large countries, have not carried out a population census in the last 20 years. On the other hand, almost all the 44 respondent countries have carried out at least one household survey of income/expenditure since 2000, more than two-thirds have conducted a household labor force survey, and half have undertaken one or more special surveys focusing on the informal sector.
This summary mirrors the picture painted in Poor Numbers, where the main trend since the 1990s was observed to be a low priority for industrial and agricultural statistics, and a high priority for household budget surveys. It should be noted though that the AfDB paints an overly bleak picture of the population census taking in Africa. Nigeria did conduct a population census in 2006 (as described in Jerven, 2013, p.56-61).
Despite these discrepancies and disagreements on the number of very recently updated GDP estimates, in both reports they are outnumbered by countries using very outdated base years. The African Development Bank reports that 19 countries have base years older than ten years old, including eight with base years greater than 20 years old. In IMF’s larger sample, one finds 28 countries with base years more than 10 years old, while 13 countries are still using base years more than 20 years old. Since the IMF data has the best coverage (but if we trust the AfDB they may have missed a few base year revisions) the information on current base years is reported in Table 1. [7]
Table 1: Base years and planned revision in SSA
Country
Base Year
Planned Revision
Years Btw Revisions
Angola
1987
2002 (2013)
15
Burundi
1996
2005 (n/a)
10
Benin
1985
1999 (2014)
14
Burkina Faso
2006
Botswana
2006
10 (1996-06)
Central African Republic
1985
2005 (2014)
20
Cote D’Ivoire
1996
Cameroon
2000
DRC
1987
2002 (2014)
15
Republic of the Congo
1990
2005 (2013)
15
Comoros
1999
2007 (2013)
17
Cape Verde
2007
28 (1980-07)
Eritrea
2004
Not compiled after 2005
Ethiopia
2000/01
2010/11 (2013)
10
Gabon
2001
Ghana
2006
13 (1993-06)
Guinea
2003
2006 (2013)
3
Gambia
2004
28 (1976/77-2004)
Guinea-Bissau
2005
19
Equatorial Guinea
1985
2007 (2013)
22
Kenya
2001
2009 (2013)
8
Liberia
1992
2008 (2015)
16
Lesotho
2004
2013 (2015/16)
10
Madagascar
1984
Mali
1987
1997 (2013)
10
Mozambique
2003
2009 (2013)
6
Mauritius
2007
2012 (2015)
5
Malawi
2009
2014
5 (2002-07)
Namibia
2004
2009(2013)
6
Niger
2006
19
Nigeria
1990
2010 (2013)
not known
Rwanda
2006
2011 (2013)
5
Senegal
1999
2010 (2014)
11
Sierra Leone
2006
5 (2001-06)
South Sudan
2009
Sao Tome and Principe
1996
2008 (na)
12
Swaziland
1985
2011 (2014)
Seychelles
2006
Chad
1995
2005(2014)
10
Togo
2000
22
Tanzania
2001
2007
6
Uganda
2002
2009/10 (2013)
8
South Africa
2005
2010 (2014)
5
Zambia
1994
2011 (2013)
Zimbabwe
1990
Source: International Monetary Fund 2013; 21
To argue about how many countries do or do not have a five-year-old base year, or calculate means and medians on the basis of surveys conducted in different years will only reach conclusions of temporal validity. It follows from basic probability that if a group of 54 countries randomly update their base year every 20 years or so, that you would in any given year have a handful of countries that have a base year within the past five years. Another key question has been the size of the revisions. Table 2 provides one overview.
Table 2: Impact of rebasing GDP in African countries at current prices
Country
Old base year
New base year
% difference between GDP old base and new base
1. Botswana
1993/1994
2006
-10
2. Burundi
1996
2005
40.3
3. Cape Verde
1980
2007
13.7
4. Chad
1995
2005
6.6
5. DRC
2000
2005
66.4
6. Egypt
2001/2002
2006/2007
8.9
7. Ethiopia
1999/2000
2010/2011
-1
8. Ghana
1993
2006
62.8
9. Lesotho
1995
2004
-4.4
10. Morocco
1988
1998
11.7
11. Mauritius
1992
1999
1.2
12. Niger
1987
2006
2.5
13. Rwanda
2006
10
14. Sierra Leone
2001
2006
25.6
15. Tanzania
2001
2007
10
16. Tunisia
1990
1997
9.8
17. Uganda
1997/1998
2002
10.5
18. South Africa
1993
1998
13.7
Source: Kiregyera 2013, 13.
The AfDB report makes the mundane point that there are also other countries outside of SSA that have GDP revisions. Fair enough, GDP revisions are not an issue particular to African countries. However, I think there are a few things worth bearing mind. Ben Kiregyera used this table to argue that the Ghana revision was atypical. Before we agree with his diagnosis, let me remind you of the following. First of all, the table above tells us something about the revisions – but it does not tell us whether those actual revisions were exhaustive. The peer review of the Ghana revision conducted by the AfDB concludes that GDP in Ghana still does not give a full coverage of the informal economy (AfDB, 2013b, p.5). Secondly, the table does not include the revisions not yet undertaken. Nigeria is still on a base year from 1990. Thirdly, those revisions in that table are actually quite big. There are a few that are not – Ethiopia, Lesotho, Botswana, Niger and Mauritius. I do know too little of the current situation in Niger and Ethiopia, but I am inclined to think that those – Lesotho, Botswana, and Mauritius – are outliers. They certainly are on GDP per capita terms. So, I am still inclined to think of Ghana as not so extreme – but is in a group of large revisions – DRC, Burundi, Sierra Leone and still not done: Nigeria. Finally, the table does of course leave out several known revisions. Malawi, the Gambia and Guinea Bissau. I think it is useful to put the Ghana revision in comparative perspective – however I should remind you that the argument put forward here, and in Poor Numbers, is that the essential part is not about GDP revisions – it is about basic data availability. This is the refrain, not only in Poor Numbers, but also in the report issued by the African Development Bank.
The African Development Bank survey noted, “one-fifth of the respondent countries had not conducted an industry survey since 2000. Even fewer countries conduct regular surveys or censuses of agriculture, despite its criticality to the food security situation in the continent” (2013a, p.5). The peer review of the Ghana revision makes it clear that there are no data to account for the informal sector; “The informal sector is not currently included in Ghana’s GDP. (When and if information becomes available to allow the GSS to include value added by informal producers a further substantial upward revision is probable).” (AfDB, 2013b, p.4).
Therefore, it is perhaps a bit worrying that the background document to this meeting has such big emphasis on methods. It is reported: Among all the African countries, 12% of them are still using the 1968 version of the SNA and 88% are using the 1993 version of the SNA, while the latest version is the 2008 SNA (UNECA, 2013, p.1). But if one knows the basic data gaps at the national accounts divisions in the region, it is clear that this cannot be correct. 1993 SNA, and let alone 2008 SNA, is too data demanding to fully complete. I warn against a superficial adherence to methods, when the real problems are access to data and the resources to analyze the data. Again, the AfDB finds a different pattern when they ask more specifically whether the components of 1993 SNA are adhered to. They conclude, “Only Ethiopia and the Seychelles reported implementing all five of the new SNA 1993 features listed” (AfDB, 2013a, p.14). This does not square well with the information in the UNECA background document just quoted. The AfDB report elaborates: “When national accountants are asked which system they are following, their answers will depend on how far they have progressed in adopting the new features. At some point they may feel that they have adopted a new SNA system even if their accounts still retain features of an earlier one. Implementation of a new system is a continuum and not a fixed point (2013a, p.14)”. The base years and methods in use is just a symptom. The final estimates of growth and income are no better than the primary data they are based upon.
Conclusion: Why (and How) We Need to Invest in African Development Statistics
I think we have a unique moment and opportunity to invest in African development statistics. Not to act now would be foolish. The advocacy campaign should draw attention to the knowledge problem in development statistics in Africa and beyond. We need to be transparent about data weaknesses and GDP revisions. Maintaining credibility is of course of key importance, and I think that at this point in time embracing data problems is the way ahead, rather than seeking to relativize them or put them in a rosier light. This means that we need a better diagnosis. An investment in African development statistics must be guided by a reliable baseline of the problems. The problems of reaching agreement on the methods and base years in reports published by the IMF, UNECA and AfDB highlights that one of the main problems is gaps in the metadata; we still do not know how much we know. A first recommendation is a transparent exercise that specifically seeks to map gaps in knowledge across the region.
The vicious circle described by StatCap is a real threat: “many national statistical systems are caught in a vicious cycle where inadequate resources restrain output and undermine the quality of statistics, while the poor quality of statistics leads to lower demand and hence fewer resources” (Jerven, 2013, p.103). This threat is not handled by contesting the diagnosis, or handling the problem as a publicity problem. We have a knowledge problem, and the long-term future of the statistical offices as a provider of statistics for economic governance is at stake.
To engage with the problem we need to look wider than at GDP and National Accounts, but seek to engage in global debates on data for development, and take a holistic view of the statistical office. As I previously discussed in the introduction, this will have to start by co-ordination of global demand for data.
The Millennium Development Goals have increased political pressure on the statistical offices. Evaluations of progress toward the targets are based on statistical reports, and donors make funding dependent on completion of these quantitative targets. The positive outcome of this process is that more attention is being given to the statistical office, compared to the neglect of statistical offices during the period of structural adjustment. Currently, more funds are being made available to statistical offices than
in the previous three decades, but this is being done in an uncoordinated fashion. Typically, support has been ad hoc and has been directly linked to particular donor-funded projects. In this way, donors distort data production instead of building up statistical capacity. This stretches current staff and infrastructure resources. Statistical officers are richly remunerated with per diem allowances when they are engaged in data collection in the field, but this leaves fewer people and resources for analysis and dissemination back in the statistical offices (Jerven, 2013, p.105).
There is a need for a shift in emphasis.
The monitoring of specific projects should be tempered by a realistic assessment of the capacity of the statistical office to deliver information on the basis of which national leaders can confidently govern. The Millennium Development Goals agenda is committing the same mistakes that were committed at independence, during structural adjustment, and during the recent era of poverty reduction. In each of this eras, targets and policies to reach those targets were identified, but less thought was given to where the information should come from that would measure progress. It might be useful to turn an important development question around. Rather than asking what kind of development we should target, perhaps the question should be What kind of development are we able to monitor? (Jerven, 2013, p.106).
More specifically, we need to re-think the MDG and other donor agendas for data and do a cost benefit analysis – what are the costs of providing these data – and what is the opportunity cost of providing these data? The opportunity cost is often ignored. Local demand for data needs to come into focus. A statistical office is only sustainable if it serves local needs for information. Statistics is a public good, and we need a good open debate on how to supply this public good.
As I have pointed out above there needs to be a shift away from focus on methods towards actual data availability.
I would argue that ambitions should be tempered in international development statistics. The international standardization of measurement of economic development has led to a procedural bias. There has been a tendency to aim for high adherence to procedures instead of focusing on the content of the measures. Development measures should be taken as a starting point in local data availability, and statisticians should refrain from reporting aggregate measures that appear to be based on data but in fact are very feeble projections or guesses. This means that it is necessary to shift the focus away from formulas, standards, handbooks, and software. What matters are what numbers are available and how good those numbers are. Comparability across time and space needs to start with the basic input of knowledge, not with the system in which this information is organized. (Jerven, 2013, p.107).
In order to provide useful data for economic governance to local policy makers there has be an emphasis on timely data.
In data collection there is a tendency to aim for high validity—an emphasis on full coverage of the economy rather than reliability of the data. Thus there is a preference for aggregating data, conducting censuses rather than surveys, and making estimates of levels rather than measuring change. These preferences come at the expense of frequent survey data that tell data users something useful about changes. In practice, this means that funding is available for large one-off data collection projects. Both statistical offices and donors share this preference: the statistical office gets access to per diem funding for data collection, and the donors fulfill globally demanded standards of statistical sophistication. (Jerven, 2013, p.107).
While recent quotes from Bill Gates (2013) and others on the importance of investing in economic statistics are encouraging, and I genuinely think that is a unique chance to put the statistical offices and data need for economic governance higher on the agenda, I will end on a word of caution. I am worried about some trends in the international community. The much noted demands for evidence-based policy has put the cart in front of the horse. In doing so, some data have been supplied that has not had the quality or reliability needed. One perverse reaction from some analysts and organizations may be to circumvent the statistical offices all together. Larger and irregular surveys with standardized data collection needs that do not serve the needs of local users of data, such as Ministries of Finance and Central Banks, are becoming more popular. There is also a trend, particularly in the scholarly community to turn to using passive, big data. Such as measuring growth from outer space, using satellite data on light emissions or other alternatives (Henderson, Storeygard, and Weil, 2012).
I think the key for survival for the statistical office is to re-find its comparative advantage as an analyzer and disseminator of data. The statistical office should work as a data broker. Here the national accounts divisions have, or used to have, a key competence. To calculate GDP one needs to be an omnivore of all types of statistics and data generated by other divisions on the statistical offices, ministries and other key stakeholders such as businesses and infrastructure providers. The capacity to passively and actively collect these data and to analyze and disseminate them – acting like a data broker – should be the key activity, and not field based data collection. The incentives have to be aligned with such a strategy, and that means shifting financial away from rewarding out of office activities (per diems) and towards rewarding collation, analysis and dissemination (salaries).
This is the time for African statisticians to remind the development community that while stakeholders are demanding evidence for policy, we must not forego the opportunity to invest in accountability. I think it is a mistake to think of data as a technocratic search for facts – it has to be viewed as an exercise in building institutions. For all this talk of ‘institutions matter’ and ‘governance’ in development circles, there has been a surprising gap in analyzing the statistical office. The statistical offices and their ability to provide timely high quality data on their economies have long been neglected. This is a chance to set these past mistakes right, and to move beyond a statistical tragedy, to stake a path towards a statistical renaissance in Africa.
Reference List
African Development Bank (2013a). Situational Analysis of Economic Statistics in Africa: Special Focus on GDP Measurement. Statistics Department. Retrieved from http://www.afdb.org/fileadmin/uploads/afdb/Documents/Publications/Economic%20Brief%20-%20Situational%20Analysis%20of%20the%20Reliability%20of%20Economic%20Statistics%20in%20Africa-%20Special%20Focus%20on%20GDP%20Measurement.pdf
African Development Bank (2013b). Peer Review of National Accounts – The Case of Ghana. Statistics Department. Retrieved from http://www.afdb.org/fileadmin/uploads/afdb/Documents/Publications/Economic%20Brief%20-%20Peer%20Review%20of%20National%20Accounts%20-%20The%20case%20of%20Ghana.pdf
Chen, S., Fonteneau, F., Jütting, J. and S. Klasen (2013). Towards a Post-2015 Framework that Counts: Aligning Global Monitoring Demand with National Statistical Capacity Development. Paris21 Discussion Paper Series, 1. Retrieved from http://mortenjerven.com/wp-content/uploads/2013/04/Panel-8-Jutting.pdf
Devarajan, S. (2013). Africa’s Statistical Tragedy. Review of Income and Wealth, 59: S9-S15. Retrieved from: http://onlinelibrary.wiley.com/doi/10.1111/roiw.12013/abstract
Gates, B. (2013). The Problem with Poor Countries’ GDP. Project Syndicate, May 6, 2013. Retrieved from http://www.project-syndicate.org/commentary/poor-countries-need-more-accurate-gdp-data-by-bill-gates
Henderson, J.V., Storeygard, A. and D.N. Weil (2012). Measuring Growth From Outer Space. American Economic Review, 102(2): 994-1028.
International Monetary Fund (2013). Sub-Saharan Africa: Building Momentum in a Multi-Speed World. World Economic and Financial Surveys. Regional Economic Outlook. http://www.imf.org/external/pubs/ft/reo/2013/afr/eng/sreo0513.pdf
Jerven, M. (2013). Poor Numbers: How We Are Misled by African Development Statistics and What to Do about It. New York: Cornell University Press.
Kiregyera, B. (2013). The Dawning of a Statistical Renaissance in Africa. Paper delivered at African Economic Development: Measuring Success and Failure, School for International Studies, Simon Fraser University, Vancouver, Canada, 18 to 20 April 2013. Retrieved from http://mortenjerven.com/wp-content/uploads/2013/04/AED_Panel_8-Kiregyera.pdf
UNECA (2013). Statistics for Good Economic Governance, Regional Integration, and Sustainable Development in Africa – African Project on the Implementation of the 2008 System of National Accounts: Phase I. Retrieved from http://mortenjerven.com/wp-content/uploads/2013/09/ProDoc-on-SNA_2013-09-03.pdf
[1] This address was cancelled.
[2] Only Mauritius had a base year from 2007, the rest were from 2006, so by the time the book and survey was published (2013) these were all out of date according to the strict IMF criterion.
[3] In the introduction of the report: “But are Africa’s statistics as bad as they are being portrayed by some critics?
In attempting to answer this question, the African Development Bank in March 2013 decided to undertake a survey to assess the reliability of GDP data, including the availability of survey data, price indices, and base years for constant price GDP” (AfDB, 2013a: 6).
[4] In difference to Jerven 2013, and IMF 2013, the AfDB 2013a also cover North Africa. The non responding countries in the AfDB survey were Angola, Burundi, Eritrea, Gabon, The Gambia, Liberia, Libya, Sierra Leone, Somalia, South Africa and South Sudan.
[5] Cape Verde and Malawi have updated their base years since I did the research for Poor Numbers, and I did not have information on South Sudan, because they had not yet made their first estimates.
[6] In fact according to mean and median it is identical. The median is 2001 in the IMF table as well.
[7] According to information submitted to me from Burundi, it has a base year from 2006, not 1996 as reported in Table 2. Furthermore, Madagascar has, according to information submitted to me from UNECA, a base year from 1995, and the statistical office is currently preparing for SNA 2008 (and presumably a rebasing) for 2016. For Mali I reported 1997 Base year (compared to 1987 as reported here) based on information submitted to me from Mali. The provenance of the information in the IMF and AfDB is not detailed in their reports, whereas it is described how all the data was retrieved for Poor Numbers in the appendix, pp. 123-137.