Improving Cost-Effectiveness and Relevance of Agricultural Censuses in Africa: Lessons from Country Experiences

1. Lessons learned from country experiences Improving cost-effectiveness and relevance of agricultual censuses in Africa: Linking population and agricultural census Naman Keita Senior Statistician FAO Regional Office for Africa ICAS III- MEXSAI: Measuring Sustainable Agriculture Indicators Cancun, Mexico, 2 - 4 November 2004

2. Content of the presentation • Subsistence farming and features of censuses of agriculture in Africa • Why is the issue of sampling frame important for censuses of agriculture in the African context? • What improvements can be obtained by coordinating Population and Agricultural census programmes? • Country examples • Lessons learned

3. I. Subsistence farming and features of censuses of agriculture in Africa • Agricultural census: “large scale periodic statistical operation for collection of quantitative information on the structure of Agriculture”. • Context of traditional subsistence farming imply use of objective measurement is necessary to obtain data on area, yield & production. FAO methods of objective measurement • Crop area (A) and yield (Y): physical measurement (using tapes and compasses) of all plots and crop cutting (sub-plots) to determine yield. • Production (P) derived from: P= A*Y.

4. I. Subsistence farming and features of censuses of agriculture in Africa FAO methods of objective measurement (cont’d) • Provide accurate measures when well applied. • However: • they are highly time consuming and costly, • they imply that complete enumeration is almost impossible given the large number of small farmers and plots to measure. • Consequence: censuses has to be taken on a sample basis to keep cost under limit.

5. II. Why is the issue of sampling frame important for censuses of agriculture in the African context? Sample design used for censuses • Most frequent sample design: Stratified, two-stage sampling with EAs (or villages) as PSUs and AH as SSUs. • PSUs often selected with pps and SSUs selected (within selected PSUs) with equal probability. • Sometimes, PSUs are randomly selected with equal probability when no information is available on size.

6. II. Why is the issue of sampling frame important for censuses of agriculture in the African context? Sampling frame issues • Most frequent sampling frame used: list of EAs (or villages) with indication of size from latest population census. • When population census too far from agricultural census, data can be completely out of date: • EAs may have changed boundaries, composition and size, • villages may have been created or disappeared or changed name. • Even for recent population census, information on size may not be very relevant to agricultural census (number of HH or populations). • Consequence: available frames are often imperfect leading to serious negative impact on data quality (significant contribution to increased non-sampling errors).

7. III. What improvements can be obtained by coordinating Population and Agricultural census programmes? Two major costly data collection operations • Population census and agricultural census are two most important and costly operations for statistical systems in Africa. • Better coordination will allow economies of scale: use of infrastructure of population census to collect relevant data for proper sampling frame for agricultural census. • Alternative ad-hoc exercise to build a sampling frame will considerably increase the total cost (usual cost 3-10 millions dollars) by 20% or more. • Growing resource constraints create new conditions for coordination and several countries have adopted some kind of linkage between the two censuses in recent years.

8. III. What improvements can be obtained by coordinating Population and Agricultural census programmes? Modalities of linkages and advantages • Two types of linkages: • comprehensive pre-census of agriculture taken in conjunction with cartography; • a short agricultural module included in the questionnaire of the population census. • Benefits include: • availability of up-dated, complete and relevant data for proper sampling frame and more efficient sample design; • increased relevance and value of census in case of comprehensive pre-census, as data on AH and rural communities can be presented for small geographical and administrative units; • scope for cross-tabulation of agriculture data and socio-demographic data from population census. • Need for close collaboration between CSO and MoA and availability of additional resources

9. IV. COUNTRY EXAMPLES Approaches taken by various countries • Approaches taken depend on country’s specific conditions: • resources available, • data requirement and • phase of preparation of population census. • PRE-CENSUS: Togo, Senegal, Benin • AGRICULTURAL MODULE: Cote d’Ivoire, Rwanda, Uganda.

10. IV. COUNTRY EXAMPLES PRE-CENSUS OF TOGO • Pre-census of agriculture in 1995 in conjunction with field work for updating Enumeration Area maps (cartography) for population census. • Main purpose: collect data on rural localities and agricultural households in order to build a proper sampling frame for the up-coming agricultural census (96/97). • Two types of questionnaires used: Village level questionnaire and Household level questionnaire

11. IV. COUNTRY EXAMPLES PRE-CENSUS OF TOGO (cont’d) Content of village questionnaire: • Identification and general information • Localization of the village • Types of roads accessing to the village • Infrastructure used for water and electricity • Schools and training centers • Economic infrastructures (market, agro-industries etc..) • Languages spoken • Main economic activities • Fruit production Content ofHousehold questionnaire: • Identification of Household • Socio-economic characteristics of the head & members of HH • Detailed information on the activities of the household (detailed agricultural activities).

12. IV. COUNTRY EXAMPLES AGRICULTURE MODULE: RWANDA • Short agriculture module included in questionnaire of 2002 population census • Main purpose: • Provide a complete enumeration of all agricultural households to serve as a basis for development of an efficient and up-to-date sampling frame. • Provide a limited number of updated information on all agricultural households of the country, for effective sample design. • Analysis of agricultural data which can be cross-tabulated with population data.

13. IV. COUNTRY EXAMPLES AGRICULTURE MODULE: RWANDA Content of module • Identification of Household: province, district, sector, “cellule” (village), enumeration area, household number and name of head. • Identification of Agricultural Households. • Type of agricultural holding operated by the Household (agriculture, livestock, fishery and all combinations) • Crop grown by the agricultural household (8 main crops identified with all combinations) • Livestock raised and numbers (5 main livestock identified) • Ownership of land

14. V. Lessons learned Coordination of the two census programmes at UN level • Experiences come from country initiatives not part of general guidelines or recommendations at international level. • UNFPA and FAO being main organizations technically sponsoring the Population and Agricultural Censuses, appropriate will be highly recommended Modalities of linkage • Modalities of linking the two censuses will vary from one country to another as many factors are to be considered. • Additional funds will always be required and are critical to cover for the extra costs added to population census.

15. V. Lessons learned Type of questions and variables for an agriculture module • Type of questions and variables most appropriate are simple often qualitative questions (type of activity, type of crop grown, etc...) not requiring precise measurement. • Attempts to get precise numbers (like livestock numbers) have failed to provide accurate results. Timing of the population census field operations • The most appropriate period for collecting adequate data on agriculture is the cropping season.

16. V. Lessons learned Data processing and analysis • The timely processing depends on advance planning and provision of adequate resources before the data is collected. • Data analysis has to be also planned in advance and resources provided for if a report is to be prepared in time. • Cross-tabulation of agriculture data with population data has not been done but in Rwanda this option is still being envisaged.

17. END