19. Some Precautions in the Preparation and Use of Forecasts

As fertilizer supply is a politically sensitive subject, governments tend to play safe by overestimating demand to avoid any possibility of shortage. Similarly, fertilizer marketing companies may underestimate demand to keep down imports and create a seller's market. To overcome this problem provincial departments of agriculture and the leading fertilizer company operating in the province should be asked to forecast demand independently on the basis of their knowledge of local conditions and to arrive at a consensus through discussion.

Regional forecasts recognise local conditions and, therefore, are less susceptible to error than country-wide forecasts which treat diverse agro-climatic zones and farming systems as a single unit and ignore market forces operating in each area.

Growth trends for N, P & K are not similar because of differences in soil and crop needs. Therefore, care should be taken to assess these trends separately. For example, in estimating the demand for DAP (which contains 18% N and 46% P), the projection may show an increase of, say, ten percent per year for the product. However, continued low application of N may retard the response to P leading to a fall in demand. For short-term forecasts, product as opposed to nutrient quantities are acceptable.

Part of a good forecasting procedure is to compare actuals with the predicted demand or sales and identify reasons for variations. This is an important step in improving forecasting accuracy over time. In the light of periodic reviews, factors influencing demand may have to be given different weights and new factors may have to be considered.

Caution should be shown when using averages. The arithmetic average is affected by extreme values in the data. Fertilizer demand during drought and floods can be very low but these are not normal occurences. Neverthless, they affect the average consumption. The average values of two or more series can be identical but the spread can be very different. The spread is referred to as "standard deviation" by statisticians. For these reasons, averages do not explain the data adequately. The following table contains fertilizer consumption details of two dealers:

Although the averages are the same, there is a marked difference in the performance of the two dealers. Dealer 2 is a more consistent performer.

In interpreting available data a clear distinction should be drawn between "consumption", "distribution" and "supply". In several countries, what is distributed to sales depots from the port or transit stores plus opening stock at the depots is treated as availability and is equated with consumption. Obviously this is incorrect as consumption can be arrived at only by deducting the closing inventory from the sum of opening stock and fresh arrivals. On the other hand, sales to wholesalers or retailers in a given period can be equated with consumption as they normally purchase what they can sell. Therefore, sales by primary distributors to wholesalers or retailers can be assumed to be roughly equivalent to "consumption" and such data should be easy to collect. Only sales to wholesalers and direct sales to retailers should be included and not stocks moved to the distributor's own storage points.

The demand forecast should not be treated as the supply plan. The latter will depend on (i) the opening inventory at the begining of the year under forecast (ii) the size of pipeline stock needed, and (iii) desired reserve stock provision. In any given situation some stocks always remain in stores or in transit, unavailable for consumption. These are said to be in the pipeline. The weaker the infrastructure of the country or the larger the country the larger is the pipeline requirement likely to be. The minimum in normal situations is ten percent of the expected demand and as much as thirty percent in countries which are landlocked or which have difficult terrain or poor transport availability. The supply programme is arrived at by adding these provisions to the quantity forecast and deducting the opening inventory.

For countries that depend both on domestic production and imports, additional precautions are necessary in calculating the procurement programme from the forecast. The anticipated domestic production should be deducted from the supply programme as well as the opening inventory. In so doing, only that part of the year's domestic production that is likely to reach the field in time for use by the farmer should be deducted. The standard format for the supply plan suggested in Annex XVI can be adapted to suit specific situations.