CGD & APHRC (2014) identify six key challenges to timely and accurate African data by the continent’s statistics agencies viz. (1) Lack of legal & functional independence (2) inadequate budgets (3) lack of autonomy (4) misaligned incentives (5) relative dominance of donor priorities over national priorities and (6) limited access and usability of data.
| National Strategies for Development of Statistics | |
| Status (May 2019) | Number of African countries |
| No strategy | 1 |
| Completed, awaiting adoption | 5 |
| Implementation | 37 |
| Expired | 11 |
| Total | 54 |
| Source: MIFAGR (2019), PARIS21 | |
They find only 12 of the 54 member countries of the African Union have legally and functionally independent statistical agencies. On the challenge of inadequate budgets, only two countries viewed themselves as adequately resourced. Quite palpably, African statistical agencies do not as yet still enjoy the importance they deserve in the priority lists of governments. And even relatively meagre budgetary allocations are often cut. In some cases, international donors and aid agencies fill this gap to some extent. Unfortunately, the divergence of goals and interests among donors and governments can be problematic.
| Divergent interests of donors and African governments | |||
| Size | Scope | Frequency | |
| Governments | Large-sample surveys | Fewer key indicators | High |
| Donors | Small-sample surveys | More key indicators | Low |
| Source: CGD & APHRC (2014) | |||
The downsides of misaligned incentives are easily discernible. Political considerations, when unfavourable but accurate data could cost elections or cause rifts between communities or tribes, sometime lead to suppression, for instance. Donors’ orientation towards performance targets tends to also push some governments towards gaming official statistics to attract more aid.
There is also the knotty issue of access and usability. It is a pervasive practice amongst African statistical agencies to almost literally hide data. “Only 56 percent of the microdata from household surveys conducted between 2000 and 2014 are available to the public”, for instance (CGD & APHRC, 2014).
There are sometimes justifiable reasons like budgetary & staff constraints for this reality. A fear of being second-guessed is also another motivation for this seeming lack of transparency. That said, more African statistical agencies now publish increasingly more data online. But in general, the data hoarding practice remains prevalent.
It is important to point out that unique circumstances sometime necessitate unintentional measurement inaccuracies. For instance, the COVID-19 pandemic is forcing statistical agencies around the world to adopt innovative approaches to data collection. In France, on-the-ground data collection and observations of behaviour were literally impossible as citizens were stuck at home. Instead, France relied on credit card data and other alternative measures to compute the relevant statistics.
Similarly, with face-to-face data collection constrained owing to a countrywide lockdown, Statistics South Africa will conduct its labour force survey for Q2-2020 via telephonic data collection. These forced innovations bring to the fore the numerous alternative measures that are available to firms and investors to gauge the accuracy of published official data, especially in Africa where accurate and reliable data are ordinarily a constraint.
More specifically, African statistical agencies released inflation data for March and April 2020 showing only moderate effects of what were palpable and pervasive price hikes & gouging and hoarding of essential goods and services, owing to COVID-19-induced lockdowns. Ordinarily, these unfair practices should have pushed up headline inflation rates significantly. They did not. Put simply, they are likely inaccurate and unreliable. While this is not a uniquely African problem, the probability that the palpably flawed statistics would be revised later by many of the continent’s authorities, as is typically the case for advanced economies, is very slim.
Advanced economies similarly released economic statistics for these months that likely failed the accuracy and reliability tests. Take the case of consumer price inflation. With most economies around the world in one form of lockdown or the other, a significant portion of their respective consumer baskets inadvertently misrepresent current realities. Services, which were literally all shut down, should probably not have representations in the consumer price index (CPI) basket for those months, for instance.
While statistical authorities are not necessarily to blame, since a temporary change in the CPI weightings may not be optimal or even feasible regardless, their retention, with heavy weights in some cases, means the headline figures are unduly weighed to the downside. Some statistical agencies have owned up to the problem. For example, India chose not to publish inflation data for April 2020 owing to not being able to conduct the requisite fieldwork because of a nationwide lockdown.
Some African countries similarly approached the challenge cautiously. As a number of statistical agencies had already compiled the relevant economic data for March 2020 and completed scheduled and ongoing surveys before the authorities-initiated lockdowns, they were probably able to publish statistics for the month with reasonable quality assurance. In April 2020, however, when most African countries were already in lockdown, it is highly unlikely that data acquisition and production would have been similarly encompassing and detailed enough to assure a reasonable level of accuracy and reliability. Regardless, some African countries went ahead to publish inflation data for the month of April 2020. Others chose to delay publication.
The case of South Africa, which not only chose to postpone the publication of its April 2020 inflation data to late June but published changes to how it would calculate the consumer price index owing to the COVID-19 restrictions, is exemplary. Stats SA highlighted the price collection issues it was facing and the innovations towards overcoming them. As it was only able to collect 12 percent of the weight of the CPI basket directly, it relied on online prices for 20 percent, imputed 26.5 percent, and carried forward the 41.5 percent that were not due for collection in the month using the standard method.
Bear in mind that even in normal times, these weighting-led errors occur; and in supposedly advanced economies at that. For instance, there is a raging debate about the weightings of the Euro-Area CPI basket of goods and services, which some suggest should show higher inflation if weighted more accurately. The underlying reasons for this misspecification are not of interest for our inquiry. What is, is that sometimes-statistical errors, by African agencies or otherwise, arise not from sloppiness but simple mis-classifications. Thus, some of the identified deficiencies in African statistics should be put in proper context.
Edited version of article was first published by the NTU-SBF Centre for African Studies of Nanyang Business School, Singapore. References are in the original article.
