Assessing Crop and Food Supplies

Michael Daw, Department of Agriculture & Forestry,

University of Aberdeen

(Presentation to TAA Meeting, Edinburgh, 18/12/00)

Introduction

The main focus of this presentation is on forecasting food situations at national (macro) or regional level, as practised by FAO and other international organisations. The assessment of local (micro) situations involves many different approaches – used by WFP, NGOs and relief agencies. These involve “Risk Mapping”, “Vulnerability Analysis and Mapping”, “Household Food Economy Analysis” and a whole jargon of methods from “Proportional Piling” to “Pair-wise Ranking Matrices”. Although these have useful applications at the local level, I am not going to concern us with them – my experience has been much more with the forecasting of national and regional food situations, usually with FAO, and in this area there are fewer objective techniques available. In most situations, it is a matter of combining and interpreting a mix of information in a logical way.

Since the 1984/5 drought and food deficits in the Horn of Africa, considerable growth has occurred in “food supply forecasting” and “early warning systems” because: donors recognised that lead times are long for procuring, mobilising and distributing food aid assistance, and governments are much more aware the need for good information on impending food situations (especially in LIFD countries) – for planning state procurement, pricing, storage and prudent levels of exports/imports, and for providing services and incentives to farmers in the following season.

Institutional Background

Individual governments normally have their own Agriculture Departments or Statistical Authorities – which have collected historical data on crop production, and produced long-standing time series and, more recently, made pre-harvest forecasts. Some countries have invested in specialist National Early Warning (NEW) units often with external assistance (for staff, training, hardware, image receiving facilities etc.) located in an appropriate department (sometimes Agriculture).

NEWUs may also link up regionally eg. SADC, CILS, IGADD – covering 24 African countries – mostly LIFD. They produce regular reports (national and regional) on the developing crop situation, run-down of stocks and price movements. They may use primary data and other information, for example, from USAID’s FEWS, FAO’s various remote sensing products or food/crop/nutritional data from NGOs. These are important inputs to Government tactical planning, to the private sector and to the international community. The emphasis is on cereals and pulses, and the systems are most developed in grain-eating countries (N, E, S. Africa and Asia).

{Examples of NEWU reports}

In most cases, NEWUs maintain close liaison with UN (mostly with FAO) which collates national information; as well as providing training, international staff, hardware and various satelitte-based products to NEWUs.

Also available data is received at Rome and incorporated into 3 separate regular planning publications: Food Outlook (monthly); Food Crops and Shortages (6 per year), the “Africa Report” (6 per year). All are available on-line or by regular mail. They include commentary on the world grain situation plus identification of problems, import needs, price movements etc. The work is carried out by GIEWS within ESC, Rome which monitors 30 African countries, 12 Asian countries, some Latin American countries and 9 FSU countries (more recently). Desk officers in GIEWS are responsible for monitoring a number of countries.

GIEWS also produces “Cereal Balance Sheets” - historical time series and early estimates for the current season {Example of wheat in Ethiopia} – continually revised with new information so as to identify deficits and import requirements in good time. Many countries now have 10 years’ Balance Sheets, by cereal and for all cereals. They are tied back to countries’ official crop statistics after finalisation, so historical figures are official, whereas the forecast is based on GIEWS’ best estimate. All are held on a large spreadsheet managed in Rome.

Crop and Food Supply Assessment Missions (CFSAMs)

In addition to monitoring and contributing to the above reports and balance sheets, GIEWS also engages in primary data collection by mounting CFSAMs in-country where a crisis is expected, or quantitative forecasts are needed, and countries request assistance, and when there is finance available (regular programme/external sources).

CFSAMs are pre-harvest. They collect unique information but also act as catalyst for local staff to engage in fieldwork and pull together all relevant data. They also bear costs of the travel which might otherwise be difficult. They are expensive, because of logistics, and are usually conducted jointly by FAO and WFP HQ staff, with local officials and local consultants.

FAO mounts about 40 CFSAMs annually, usually following a critical situation e.g. 1997 Indonesian economic crisis, 1998 Bangladesh flood damage, 1998-99 North Korea, 2000 Tajikistan. Some LIFD countries have regular CFSAMs e.g. Sudan, Eritrea, Ethiopia. Mostly there is one mission per year but there may be two where the situation changes (eg. rainfall) or where there are two distinct crop seasons (eg. Somalia, Sudan, Kenya).

Objectives:

Ø finalise Production and BS for previous year

Ø forecast food crop production and availability for current year

Ø estimate B.S. for following year (imports as residual)

Ø present draft findings to Gov., donors, NGOs

Ø write “Special Report” for distribution to donor community

{examples- Special Reports and Alerts}

Staffing

Typically 2 FAO staff + 8 local staff (agronomists and economists).

1 week in capital, 2-4 weeks in field, 1 week in Rome. From start of mission to distribution of final report, 6 weeks.

Sources of Information

Forecasts are made using information in the following 5 categories:

Ø historical time series (crop stats)

Ø rainfall monitoring (records, agromet models, water satisfaction indices, yield estimations)

Ø remote sensing products (CCD, NDVI)

Ø market price analysis (seasonal and spatial time series)

Ø field work (farmers, traders, crop inspection).

Information needed at start of mission (HQ or capital):

Ø historical time series of crop areas and production (by crop, region, sector)

Ø previous national balance sheets

Ø information on growing season (weather, insect damage, disease etc)

Ø price data for season

Ø official stocks

Ø trade in food crops

Ø Macro-econ information

Ø Remote sensing images {show examples}

Ø Outputs from NEW units, donors (FEWS), NGOs etc.

Ø Preliminary crop planting data (if available)

During Mission:

Field visits with small interdisciplinary teams to surplus and deficit areas. Attempt to cover whole country. Common checklist. Aim to get areas planted/harvestable plus factors affecting areas and yields:

Ø pre-planting prices

Ø stock levels at start of season

Ø seed, improved seed, fertiliser availability and prices

Ø availability of other inputs (fuel, labour etc.)

Ø relevant weather data

Ø pests, diseases, weed problems and crop protection measures

Ø prices (grain, livestock, wages)

Ø farmers’/traders’ expectations.

Visit MOA local offices, farmers, traders, NGOs and inspect crops first hand.

The main steps in such missions are:

Ø Develop areas, yields and production forecasts in the field against a background of satellite images and agricultural statistics. Neither images nor time series are predictive (quantitatively) but satellite data are very useful inputs to estimating rainfall, crop conditions and yields. Statistical series may give “the bounds” of area and yield estimates.

Ø Develop forecasts at localised level and aggregate to region and national level (initially very disaggregated, then build up).

Ø Teams return from field with draft forecasts and explanations/justifications for each “district” or zone. Presented, discussed and criticised in the larger forum of the whole mission (8-10 people). Continually checked against time series, satellite data and rainfall records. Forecasts agreed at dis-aggregated level.

· Finalised – single point estimates

· Aggregated to regional & national level Produce tables of areas, production etc by crop/sector/season {examples}

· Build up next year’s national BS

· Estimate import requirement (commercial and food aid).

· Calculate regional production, surplus/deficits {examples}

· Present preliminary findings (national and regional to Government, donors, UN, NGOs in-country)

· Write “Special Report” and debriefing FAO and WFP Rome. {examples}

· Special report approved and distributed, possibly with a formal Appeal to donors.

Example Forecasts (Ethiopia 1999 Crop Year)

Table 1: Area, production, yield, 1999/00 by individual cereal/pulses,

by region

Table 2: Area, production and yield (cereals, pulses) – last 3 years

actual and 1999/00 forecast, by zone.

Table 3: National Cereal Balance Sheets, 1999 and 2000.

Table 4: Regional Cereal Balances.

Figure 1: Seasonal Price Movements.

Figure 2: Forecast Net Surplus per caput, by zone, 2000.

Observations

Ø Despite large volume of information, in the end, the forecasts are judgmental; no fully objective method. Depends on skills and experience of C.A. teams – consistent teams desirable. Objectivity of Agromet models limited by good rainfall data (beginning to use CCD in some situations) but, even then, must still obtain good data on areas planted and on insects, disease, weeds etc.

Ø Satellite images and statistical time series very useful but do not lead to point forecasts – need careful interpretation. But very useful information against which to check local opinion and observations.

Ø Views of farmers and traders and local MOA staff, and field inspections are still a vital input to whole exercise.

Ø Care needed with overly optimistic opinions of agronomists and extension staff, and pessimistic views of traders. Farmers more reliable. Checks and balances provided by historical time series, satellite images and price trends.

Ø Still partly an art rather than a mathematical science but accuracy is still improving..