Going Missing: Filling the Gaps in African GDP Data

Going Missing: Filling the Gaps in African GDP Data
Photo: Freedom Square, Accra, Ghana, June 26, 2011 (photo by Wikimedia user ryansworld licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license).
On Nov. 5, 2010, Ghana Statistical Service, the country’s government statistics office, announced that it was revising its gross domestic product (GDP) estimates upward, and as a result Ghana’s GDP per capita almost doubled. The country was upgraded in an instant from a low- to lower-middle-income country. A sense of bewilderment and confusion arose in the development community. When did Ghana really become a middle-income country? What about comparisons with other countries? Shanta Devarajan, the World Bank’s chief economist for Africa, struck a dramatic tone in an address to a conference organized by Statistics South Africa, calling the state of data collection on the continent “Africa’s statistical tragedy.” The real tragedy was that we did not know how little we knew about income and growth in Africa. Observers and analysts have for too long taken the numbers at face value, but there is a large gap between the economic realities on the African continent and the numbers that purport to describe it. Since the publication of my book “Poor Numbers: How We Are Misled by African Development Statistics and What to Do About It,” a healthy debate on the meaning of development statistics in the African context has taken place. Predictably, the emphasis in the media has been on the politics of the numbers. Inspired by the old phrase “lies, damned lies and statistics,” commentators and headlines have focused on the dark forces that tamper with numbers and consciously mislead the public discourse on development Africa. My main message is much simpler: We know less than we would like to think we do about growth and development in Africa based on the official numbers. The problem starts with the basic input: information. The fact of the matter is that the great majority of economic transactions in the rural agricultural sector and in medium- and small-scale urban businesses go unrecorded. Ghana’s huge jump in GDP in 2010 is a symptom of this gap between what we think we know and what we actually know about income and growth in Africa. What Happened in Ghana? How could a country like Ghana be among the poorest in the world one day, and find itself among aspiring middle-income countries the next? To answer this question, one needs to understand some basic facts about national accounting, which yields the aggregate we refer to as GDP. GDP is calculated as a sum of the “value added” in the production of goods and services. The reality is that all GDP measures are an approximation, but in most African economies the statistical offices simply do not have all the information, time or resources needed to generate a new aggregate each year, and there is only very limited data available. Usually, the statistical office picks a “benchmark year” when it has more information on the economy than is normally available, such as data from a household, agricultural or industrial survey. The information from these survey instruments is added to other administrative data to form a new GDP estimate. This total is then weighted by sectors, and thereafter other indicators and proxies are used to calculate or guess at new annual figures. One variation of the usual method is to assume that food production grows in line with rural population growth, that the informal and unrecorded urban sector grows in line with the recorded service sectors and that construction grows in line with cement imports. The estimates are made using a bit of qualified guesswork and brave assumptions. This means that the benchmark estimation is important. Sectors that were important in the benchmark year will continue to appear important in following years, even if structural changes have occurred subsequently, while sectors that were unimportant or didn’t exist in the benchmark year will barely have an impact on GDP. The data sources and the use of proxies are also set in the benchmark year, so even when new sources of information become available, national accountants may be unable to add them. When the benchmark year is out of date, the GDP series becomes unreliable. The IMF statistical division recommends a change of base year every five years—in Ghana’s case, the benchmark year prior to the 2010 revision had been 1993. Quite obviously, the structure of Ghana’s economy has changed radically since then. In 2010, the GDP of Ghana was recalculated using new data sources using 2006 as a benchmark, which is what caused the big jump in GDP. It turned out that since 1993, almost half of the Ghanaian economy had gone missing from the official count. Now, Ghana is one of most closely watched economies on the African continent, yet the result of the revision took most if not all observers by great surprise. What then should we make of the reported numbers from other African economies? One country whose numbers warrant consideration is Nigeria. Like Ghana, Nigeria has recently announced a political goal of reaching middle-income status. Since the fall of 2011, news of an upcoming revision of GDP statistics underway in Abuja has circulated widely, but the revision is not yet complete. It has also been reported that the revision may lead to a doubling of per capita income in Nigeria. The current benchmark year for the national accounts in Nigeria is 1990, meaning Nigeria’s GDP estimates are even more outdated than Ghana’s were prior to that country’s revision, and therefore it is very likely that Nigeria’s upward revisions will be as large as expected. These GDP revisions are of game-changing proportions, and have direct implications for what we think of African economic development. The revision in Ghana certainly invites a rethink about the relationship between numbers and reality, but the Nigerian revision promises to change the economic picture of sub-Saharan Africa quite significantly; indeed, it will amount to a redrawing of the economic map of the region. According to the most recent World Development Indicators, the total GDP of Nigeria in current U.S. dollars was more than $200 billion in 2010. Nigerian GDP, according to the unrevised numbers, already accounts for 18 percent of the total GDP of sub-Saharan Africa, which was measured to be about $1.2 trillion in 2011. Assuming Nigeria’s revised GDP is double the current figure, as media reports and private communication from the International Monetary Fund (IMF) and Nigerian Bureau of Statistics indicate is likely, the revision by itself will mean that GDP for the whole region will increase by more than 15 percent. The value of the increase adds up to no less than 40 economies of Malawi’s size. The knowledge that there are currently 40 “Malawis” unaccounted for in the Nigerian economy should raise a few eyebrows, and leaves statements about trends in growth and poverty on the African continent in deep uncertainty. How Good Are the Numbers? My book presents a study of the production and use of African economic development statistics. I emphasize that this is not just a matter of technical accuracy—the arbitrariness of the quantification process produces observations with very large errors and high levels of uncertainty. At the same time, this “numbers game” has taken on a dangerously misleading air of accuracy, and the resulting figures are used to make critical decisions that allocate scarce resources. International development actors are making judgments based on erroneous statistics. Governments are not able to make informed decisions because existing data are too weak or the data they need do not exist. In “Poor Numbers,” I surveyed the status of GDP statistics in sub-Saharan Africa, and in particular collected information on the methods and data used to compile national accounts. Of 37 countries I surveyed in 2011, I showed that only 10 countries had a benchmark year that was less than 10 years old. I further showed that seven countries had a base year that was more than two decades old, and that there were only 6 countries that followed the IMF’s advice to have a benchmark year that was 5 years old or newer, meaning 2006 at the time. In response to this survey, the African Development Bank (AfDB) commissioned a study, published in June 2013, that provided information on the same variables. The AfDB attempted to get a response from all 54 member countries, and received a response from 44 of them. In the survey of base years, the AfDB reports results from 34 countries, of which 9 countries met the five-year rule with a base year of 2007 or later. Another report that set out to replicate my study of GDP statistics in “Poor Numbers” was published in the IMF’s 2013 Regional Economic Outlook for Sub-Saharan Africa in May. According to that survey of 45 countries, only four countries met the so-called five-year rule. It is symptomatic of the state of knowledge on African statistics that two key players like the IMF and African Development Bank published conflicting metadata on African statistics within a month of one another. Despite these discrepancies and disagreements on the number of very recently updated GDP estimates, both reports confirm that many countries use very outdated base years in calculating their GDP statistics. The African Development Bank reports that 19 countries have base years older than 10 years old, including eight with base years more than 20 years old. In the 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. But metadata such as benchmark years is just scratching the surface. A country’s GDP estimates are only as good as the data on which they are based. As the African Development Bank reports, there is a lack of data on industrial production, and very few countries conduct regular surveys of agriculture. And while half of African countries have undertaken surveys focusing on the informal sector over the past decade, regular annual data is still based on guesses and very partial information on the economy. Statistical Capacity in Africa: Rise and Fall and Rise? In “Poor Numbers,” I describe how the statistical capacity of African states was greatly expanded in the late colonial and early postcolonial period, but was greatly impaired during the economic crisis of the 1970s. Statistical offices were neglected in the decades of liberal policy reform—the period of “structural adjustment” in the 1980s and 1990s— that followed. In retrospect it may be puzzling that the 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 actually 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. First, the decline in economic growth in Africa in the 1980s was overstated. Second, for many economies, such as Tanzania and Zambia, the upward swing in the 1990s was overstated. The marked improvement we see in the GDP time series in the mid-1990s was driven by expanding the estimates for the informal sector—thus it was statistical growth, not real growth. Third, and similarly, a lot of the recent rapid growth we are now recording in Africa is in fact statistical growth deriving from adding the informal sector and the service sector to the old estimates. Certainly, much of the recent apparent growth derives from appropriately recorded growth in external trade, but exactly how this growth relates to the domestic economy, and more generally to economic development such as poverty reduction, remains guesswork. The contrast between the statistical offices and the central banks in the region is striking. While statistical offices are located in rundown offices, often with limited computer facilities, the central banks of African countries are located in new high-rise buildings with modern facilities. Positions in the central banks command higher salaries and prestige than those in the statistical offices, and central bank employees are in a better position both symbolically and physically to provide timely and useful advice to policymakers. A similar lift is needed for statistical offices. The current development agenda is set by the Millennium Development Goals of the United Nations. This has led to some statistical capacity-building in a number of countries, while in others there have been perverse effects when statistical capacity is diverted to data collection that serves the monitoring of particular donor targets. The Millennium Development Goals have generally meant that there has been more data available for measurement of social development, but the data needed for economic governance is either not supplied or of very questionable quality. 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. 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 toward social statistics. Moreover, the funds that are made available to statistical offices are generally ad hoc funds that support data collection for a donor-funded project. In practice, many statistical offices operate as data-collection agencies for hire, not as offices that provide objective information needed for day-to-day politics or policy planning. This means that donors distort data production rather than expanding statistical capacity, While resources for manpower and infrastructure are stretched thin. The problem here is lack of coordination: Many countries have national strategies for improving their statistical capabilities, but, more often than not, donors break with these plans' priorities and demand the data they need, thus adding to the fragility of statistical offices under increasing pressure. What to Do About It? Since the publication of “Poor Numbers” the debate has continued, and many useful and practical steps toward the production of more reliable data have been suggested. But recent events have also shown that an open and transparent debate on the reliability of statistics may be difficult to conduct. Discussing economic statistics and GDP estimates of African economies is clearly important, but it’s also sensitive. In Bill Gates’ review of “Poor Numbers,” he wrote, “It is clear to me that we need to devote greater resources to getting basic GDP numbers right. As Jerven argues, national statistics offices across Africa need more support so that they can obtain and report timelier and more accurate data.” At face value, this should be wonderful news for statistical offices, but that was not how it was initially received. Some members of the African statistical community responded by trying to block me from attending conferences and trying to control or stifle the debate on African statistics. Now that some of the dust has settled, there are good indications that partners will unite on a common agenda for improving the data needed for economic governance in Africa. Improving the data collection in Africa is not only a question of the amount of funding. It requires more than simply increased resources—the main problem is the incentives and the political economy surrounding the provision of statistics as well as the global governance of the demand for data. There are a lot of diverging interests among international organizations, and sometimes gaining support for one organization’s particular program or agenda comes before actually doing the job of improving measurements. Those who thought that there was a data revolution taking place, with the world united toward achieving a better measurement of development, might be disappointed. Lack of firm facts leaves convenient room for negotiation of the numbers when it is needed. Currently the data, if they are available, are not timely or of the quality needed, and that hampers any serious agenda for economic governance or development planning. But as anyone who has read “Poor Numbers” will know, the big story in the book is governance by ignorance. As I have pointed out, this stands in striking contrast to the demand for data in the development community. The most extreme version of “evidence-based policy” comes from those who suggest that we tie financial rewards directly to statistical evidence of success. The trend is that donors are increasingly demanding monitoring and data in return for funding. With the huge gaps in data combined with clear incentives to distort the results, the risk is that the outcome will be “policy-driven evidence” rather than evidence-based policy. At present there is no coherent global strategy for improving the provision of data for development. This is related to the general problem of accountability in development. Sometimes ignorance is bliss, for both the donor community and local political leaders. Putting a coherent global strategy for statistical capacity building in place is important, and such a standard must be geared toward solving local problems. For the business community and investment banks, the availability of reliable statistics is crucial for future growth in Africa. It is indeed a real tragedy that the statistical capacities of sub-Saharan African economies are in such a poor state, but it is a reality: African development statistics tell us less than we would like to think they do about income, poverty and growth in sub-Saharan Africa, even as governments, international organizations and independent analysts need these development statistics to track and monitor efforts at improving living conditions on the African continent. One of the most urgent challenges in African economic development is to devise a strategy for improving statistical capacity, as the current system causes more confusion than enlightenment. Morten Jerven is associate professor at Simon Fraser University in Vancouver, Canada. An economic historian, he has published widely on patterns of African economic development, including a recent book, “Poor Numbers: How We Are Misled by African Development Statistics and What to Do about It,” published by Cornell University Press in 2013. Jerven’s second book, “Economic Growth and Measurement Reconsidered in Botswana, Kenya, Tanzania, and Zambia, 1965-1995,” will be published by Oxford University Press in 2014.

Keep reading for free!

Get instant access to the rest of this article by submitting your email address below. You'll also get access to three articles of your choice each month and our free newsletter:

Or, Subscribe now to get full access.

Already a subscriber? Log in here .

What you’ll get with an All-Access subscription to World Politics Review:

A WPR subscription is like no other resource — it’s like having a personal curator and expert analyst of global affairs news. Subscribe now, and you’ll get:

  • Immediate and instant access to the full searchable library of tens of thousands of articles.
  • Daily articles with original analysis, written by leading topic experts, delivered to you every weekday.
  • Regular in-depth articles with deep dives into important issues and countries.
  • The Daily Review email, with our take on the day’s most important news, the latest WPR analysis, what’s on our radar, and more.
  • The Weekly Review email, with quick summaries of the week’s most important coverage, and what’s to come.
  • Completely ad-free reading.

And all of this is available to you when you subscribe today.

More World Politics Review