(Item 7 of the agenda)78. In response to previous recommendations by the commission for a documentation of methodologies in crop forecasting, document APCAS/00/9 Food Crop Production Forecasting: Statistical Considerations was presented during the session.
79. The commission was informed that in most developing countries, in addition to several agronomic and economic factors, crop production depended heavily on the vagaries of weather. Policy decisions relating to procurement, stock management, marketing and distribution would be enhanced if supported by a strong system for food crop production forecasting. Crop production forecasting and early warning systems (EWSs) were sometimes construed to be serving the same purpose. It was stressed however, that EWSs were more action oriented providing timely information to policy/decision makers on impending food shortages and/or surpluses. Timely availability of reports was, therefore, considered a crucial and necessary characteristic of EWSs; often, the reliability and accuracy characteristics being considered as only secondary. Crop forecasts, on the other hand, to be useful, should be timely, reliable and with known level of precision. Useful crop forecasts should be made soon after planting (early forecasts), during the growing season (mid-season forecasts) and some weeks before harvesting (final forecasts). Since pre-harvest forecasting of crop production was being focused on the estimation in advance of actual harvest of production of crops, which were already in the field, methodologies for forecasting yield should be strengthened to enhance utility of production forecasts. The use of a statistically valid crop forecast sample survey was recommended.
80. The commission noted that the system of crop forecasting had not been institutionalized in many countries. The status of crop forecasting activities, including institutional set-up, commodities covered, data availability, methodologies used for preparing forecasts, their level of disaggregation and methods of dissemination varied widely from country to country. Available forecasts of food crop production suffered from a number of shortcomings: (a) forecasts were largely a product of subjective reporting of crop area and yield; (b) forecasts were available only for very few commodities, generally for main cereal crops like paddy rice, wheat or maize; (c) forecasts were generally available only at national level while sub-national level forecasts were rarely available; and (d) conflicting figures in instances where more than one national agency were issuing crop production forecasts.
81. The commission was informed about experiences including methods used in preparing crop production forecasts and advance estimates in selected countries of APCAS, e.g. India generating advance estimates through the Timely Reporting Scheme, Philippines using farmers intentions schedule, and Thailand utilizing econometric modeling in addition to other forecasting methods. It was reported that in the Philippines, results from farmers intentions gave 3 - 5% margin of error as compared to actual production data. The commission emphasized that advance estimates were an important aspect of information on crop production.
82. The commission heard the recent attempt in Sri Lanka to introduce food crop forecasting. It was reported that this was a laborious undertaking and created difficulties in the collaboration sought from Government officials concerned as they required additional payment. The commission was of the opinion that if the exercise had been introduced as an integral part of the existing annual survey programme, many of the problems encountered, including selection of the sample and staff collaboration, might have been avoided.
83. The commission was informed that in order to improve the reliability of crop production forecasts, data used should be objective, quantitative and verifiable. The use of probability sampling techniques to collect data on food crop area and yield was stressed. The statistical accuracy of forecasts should be sufficiently high: national level forecasts should not have more than 1 - 2% sampling errors; sub-national level forecasts or estimates of minor crops might have lower levels of accuracy. Timeliness, on the other hand, could be improved if countries laid down stringent timetables for all agencies involved, at various administrative levels, to observe prescribed deadlines in collection and submission of data for food crop production forecasts. It would also be necessary to identify data sources and ensure that the data flows were maintained. Guidelines on how to deal with exigencies like non-response, missing or incomplete data would provide useful support mechanisms for ensuring smooth functioning of the system. The use of a rapid appraisal enquiry could also be considered as a useful and economical option for a more timely collection of data.
84. Considering that one of the uses of food crop production forecasts was to support food security programmes, the commodity coverage should be expanded; in particular, forecasts for roots and tuber crops, horticulture crops and even livestock should be made regularly. These commodities were contributing substantial shares to total national food supply and were both complementing and supplementing cereals as sources of food. The number of commodities to be included in the food crop production forecasting programme should be decided on the basis of their total contribution to the required calorie supply as well as importance in certain pockets of the country. In addition, the food crop production forecasting system should ensure that the entire geographic areas (administrative as well as agro-ecological) of a country were being covered. Various farming sub-sectors such as small or traditional farmers, large-scale farmers, commercial farmers and government farms producing food crops and commodities should also be brought under the umbrella of the forecasting system. The system should also have a mechanism for assessing the losses in crop production due to natural calamities (floods, drought, etc.), a common phenomenon in many countries in the region.
85. The commission noted that while there were existing crop yield forecasting models, the use of data based on statistically accepted sample survey design would still be preferred. For early forecasts, estimates of area and yield could be made using time series data for these variables. A survey of planting intentions could also be considered. It was also stressed that crop forecasting should always be treated as a multi-sectoral and multi-disciplinary effort. In this respect, strong coordination of activities, specifically between the Department of Agricultural Extension of the Ministry of Agriculture and the Agricultural Statistical Organization of the Department of Statistics should be maintained.
86. The commission confirmed the importance of, and need for, production forecasts in respect of main as well as minor crops. Major funding agencies, such as the IMF and the World Bank required such forecasts for planning and decision making regarding loans and support programmes to Governments.
87. The potential of remote sensing techniques for forecasting crop production was reported by some delegates. It was noted that the application of these techniques was still in experimental stage. The commission was informed of the use of remotely sensed data by FAO as inputs for the Global Information and Early Warning System (GIEWS). FAO monitored the cloud cover over the Sahel countries using satellite imagery, from which rainfall was regularly estimated. The pattern of this estimated rainfall were reported to be useful for monitoring crop growing conditions in the Sahel countries; thus providing information on impending major food shortage areas or countries. Such information, if any, would be disseminated to all parties concerned.
88. The commission recognized the difficulties in estimating the adverse effect of major hazards and natural calamities on crop yields. It recommended, in the case of cereals, the inclusion of methods involving counting of crop ears prior to the harvest as a possible way for forecasting food crop production. The commission also noted that in countries where crop yields had been observed to be stable over the years, forecasting of production was simply reduced to obtaining reliable data on harvested area.
89. The commission observed that if forecasts and actual production figures were under the responsibilities of one and the same agency, there might be an integrity problem in connection with these data. It was felt that an agency other than the Statistics Agency might be a more appropriate organization to be in charge of food crop forecasting. It was reported that ABARE (Australian Bureau of Agricultural and Resource Economics) carried out forecasting activities in Australia while the Australian Bureau of Statistics was not involved in this exercise. In this regard the commission stressed the important role of a working group on crop forecasting in countries where many agencies prepared forecasts for food crop production. The commission recommended that an inter-agency Working Group on Crop Forecasting be established at country level to coordinate activities in this field with the aim of producing forecast figures acceptable to all parties concerned.
90. The commission appreciated FAOs initiatives to implement its earlier recommendation for the preparation of a users guide on food crop forecasting. These included a series of meetings of experts in crop forecasting organized by the Statistical Development Service (ESSS) in Asia and Africa in 1999 and 2000. An important output of these meetings was the suggestions made on the contents of a proposed Manual on Guidelines for Food Crop Production Forecasting. The commission reviewed the draft outline of the proposed manual and made useful suggestions. It recommended that FAO give priority attention to this activity and expressed the wish that the manual be finalized and disseminated soon for use of member countries. It further recommended that member countries should endeavor to include statistical activities such as regular crop forecasting surveys, crop monitoring systems, objective area and yield surveys, and the like, in their respective national agricultural statistical programmes and that food crop production forecasts be integrated with the Food and Agricultural Information System of the countries.
91. In view of the increasing importance of agribusiness in member countries, the commission recommended that the planned FAO manual on crop forecasting include a section covering high value crops. The commission was pleased to note that FAO was planning to issue a series of manuals on crop forecasting, each covering a particular aspect of this specific topic, e.g. cereal crops, root crops, livestock, and possibly fisheries and forestry. Funds permitting, such manuals would be prepared over the next several years.