ABSTRACT
A handbook on African statistics states that national accounting practices in African countries ‘focus their attention heavily on the main tables, especially the gross domestic product (GDP), and the international agencies reinforce this bias by requesting national statistics offices to provide data for the aggregates long before the preparation is defensible, resulting in figures that are little better than random numbers’ (Kpedekpo and Arya, 1981: 208, authors’ emphasis). Three decades earlier one of the pioneers of development studies warned about the potential pitfalls of producing statistics on developing countries claiming that ‘in the hands of authorities, such international comparisons may yield correlations which throw light on the circumstances of economic progress, and they tell us something about the relative inefficiencies and standard of living, but they are widely abused. Do they
not on the whole mislead more than they instruct, causing a net reduction in human knowledge?’ (Seers, 1952: 160). Despite these concerns very little research has been undertaken on the reliability of
official African statistics. Limited work was carried out under the auspices of OECD on the measurement of the non-monetary economy (Blades, 1975) and on GDP level estimates (Blades, 1980). However:
the GDP per capita growth rates published by developing countries have never been examined for their reliability [and] it seems unlikely that in developing countries GDP real growth rates have errors of less than 3 per cent attached to them. An estimated year-to-year increase of 3 per cent may mean anything from no growth at all to an increase of 6 per cent. (Blades, 1980: 72)
The issue of data quality is best approached by examining whether the data are valid and/or reliable (Ariyo, 1996). The first question is whether national income is correctly measured. There is generally an element of under coverage in all national accounts, but in African countries this is a problem of larger importance, especially given the magnitude of non-monetary transactions and own-production in the large and important rural sector (van Arkadie, 1971/1972). Furthermore, both in urban and rural areas all types of economic transactions are often not recorded because of the combined effect of the state’s lack of capacity of record keeping and the small scale and informality of these transactions (MacGaffey, 1991). The question is whether this element of mismeasurement of national income is consistent through time and space, that is, whether the measure is reliable. This is not likely to be the case. There have been important changes in national accounting practices, and most importantly the resources available to national statistical offices. Initial estimates after independence did not generally include, or included only very modest estimates for, the unrecorded or non-monetary economy. These were improved when series were rebased in the 1970s. Structural adjustment, the growth of the importance of the urban informal sector and a general shortage of resources in the state administration created problems in the 1980s and 1990s. Moreover there was significant variation across countries with regards to the relative strength of the state administrations and the extent of collapse and decline in the 1980s. Some countries have now implemented informal sector surveys, others have not. In some countries state and parastatal activity was very important. Some economies rely on a diverse mix of small scale farmers, whilst elsewhere national income is drawn largely from one natural resource for which the price is determined in world markets. In conclusion, one has both validity and reliability issues with official African
data. This might cause serious problems for cross-country and inter-temporal growth comparisons. To gauge exactly the seriousness of this problem is complicated by the fact that there is no direct way of knowing the extent and variation of the ‘unrecorded’ element through time and space. This paper first (Section II) examines how these data problems are manifested in official statistics to get a measure of the timing, size and cause of data inconsistencies. Section III then considers why this matters for the interpretation of African growth. The conclusion offers guidance to scholars using historical data to interpret economic change in Africa.
