Can agriculture and food security policies be effective when statistical data is unreliable?


Can agriculture and food security policies be effective when statistical data is unreliable?

In conducting research presented at the AfrEA Conference, I was asking myself if the persistence of hunger and poverty in Africa was not related to the unreliability of statistics.

For this purpose, I analysed the process of primary data collection and the food security indicators produced in my country, Benin. As we all know, the quality of statistics depends on the accuracy of primary data, as primary data ultimately condition all subsequent analyses and policies. My research clearly showed some weaknesses in the official statistics. Based on a literature review of the periodic statistics published by the INSAE, the public structure in charge of national statistics in Benin, and on interviews with some data collection agents used for surveys carried out from 2011 to 2018, I found two main aspects currently affecting the quality of data used in official food security statistics.

1)      Quality of field data collection agents

Collection agents are usually students and recent graduates who are often paid very little (on average 10,000 FCFA/15 Euros for 10 questionnaires of 15 pages per day) and late in the process, at the end of the operation. Furthermore, investigations take place during the academic year, forcing students to be in a hurry to complete the work and not miss classes. I also found that the collection agents typically did not have any means of transportation, even for reaching remote enumeration areas.

2)      Raw data collection process

The collection period proved to be too short, not timely, and ultimately not adequate to provide data reflecting the average household situation throughout the year. In particular, the operations take place either during the lean season, when food accessibility is low, or during periods of abundance when food accessibility is good. Since the lean season and the period of abundance, obviously, exist every year, this should be taken into account when organizing food security surveys. Furthermore, the lack of panel data, collected repeatedly on a certain sample of households, does not allow analysts to make good inferences on their food status throughout the year. Therefore, a household can be food insufficient in one period and food abundant in the other.

3)      The incidence of the reliability of statistics

The study revealed that some established statistics were weak and paradoxical. For example, food insecurity was prevalent in wealthier and richer households, with respectively 20.4% and 16.8% of food insecurity rates in 2011. The rate of households experiencing food insecurity has almost tripled within 2 years, moving from 12% in 2008 to 33.6% in 2010.

The little importance given to statistics in Africa, where funding is not sufficient for the collection, analysis and management of statistical data, results in statistics budgets that are consistently inadequate and generally underfunded by governments and funded by donors.

There is therefore a need to reconsider the statistics in the definition of public policies in Africa, by a consequent endowment of financial resources coming through national budgets for their production and management. This will guarantee a certain level of sovereignty in the production and exploitation of statistics. For, statistics reliability is even more important in the context of the Agenda 2030 and the commitment that governments have made to the SDGs. We have to bear in mind that  progress towards the achievement of the SDG2 by eliminating hunger in all its forms by 2030 is closely linked to our capacity to collect raw data for reliable statistics calculation. That’s the only way to have information which effectively reflects the reality on the ground