Existing product coding systems, principally the UN Central Product Classification (CPC), do not provide enough codes or coding flexibility to meet FEWS NET’s objectives for its work. These include, for example, maintaining a direct link between local crop names and their key characteristics and a more common globally used set of crop names. Both are important for communication at different levels.

In 2014, FEWS NET joined an effort with FAO and WFP to define and meet each institution’s product coding needs within the general bounds of a common product coding structure that would guarantee a high level of consistency and comparability among the codes used by these organizations to classify crop production and monitor food prices.

In a collaborative report published in 2014 by the UN-based Food Security Information Network¹, a common framework and agreement was reached about product coding. Consistent with this agreement, FEWS NET uses a modified product and services coding system based on UN CPC V2 codes to meet its own specific needs, allowing FEWS NET to:

1. Preserve a link between local product names and descriptions and a more global classification of products, allowing appropriate terminology to be used at each level of discussion/investigation.

2. Recognize, group, and distinguish between specific products that play a role in the food security status of a population group or geographic area.

3. Differentiate specific crop plants from others on the basis not only of genetic features but also non-genetic features, often locally identified (taste, color, smell, mouth-feel, health benefits, preparation practices, market preferences, etc.).

FEWS NET product and service code structure

A FEWS NET product code is derived from an existing CPC V2 code and is built in a consistent manner consisting of three parts:

1. Each code begins with a one-character alpha-code prefix that FEWS NET uses to split its product and service codes into 4 distinct sub-groups²:

   1. R: “raw” un-processed³ agricultural product

   2. L: animal/livestock product

   3. P: manufactured or “processed” product

   4. S: service product

2. This is followed by a 2- to 5-character numeric code taken directly from the CPC.

3. FEWS NET appends a two-character “alpha” code (e.g. “AA”) to the end of each to allow for easy differentiation between sub-types of products.

Although the FEWS NET approach allows great flexibility in coding a large number of local product types with different names and codes, this would be a problem if a user could not work at a more global level (e.g. how much wheat was produced in the world in 2016?).  To make such comparisons using FEWS NET codes, a selected portion of the FEWS NET crop code, like “R01112,” must be used.

Crop codes

FEWS NET’s coding system, like the CPC system it has been adapted from, classifies agricultural crops by the uses to which they are put, and the overall structure of each is similar.

Looking at the table of crop code categories below, in a broad sense, cereals are differentiated from vegetables⁴ and fruits because they fulfill different purposes for human consumption and health.  Products consumed by humans are distinguished from fodder used for animals.  Similarly, at a more narrow level, “leafy” vegetables are distinguished from “stalk” vegetables and others that are consumed for their root or for the “flower” they produce.  Spices are differentiated from root crops and from other crops used to produce alcoholic drinks.  Some crops produce medicinal products and scents, and others produce fabric and matter for consumption in a smokeable form.    

The Expected Precision of A Crop Coding System

There are several fundamental reasons why a crop classification system is unlikely to ever produce an objectively perfect way of classifying crops into mutually exclusive categories.

• Similar crop plants may vary significantly in their exact genetic make-up, just like apparently different crop plants (e.g. broccoli, cauliflower, kale, mustard) may share so many genetic features between them that it is difficult to identify a firm genetic boundary between one or the other. Defining that genetic boundary between one plant and another similar one has always been difficult because at some level, most crops are hybrids of others. Genetic differences have become even more difficult to use as a base for classification with the rise of human-engineered gene transfer and hybridization. Note that both the CPC and FEWS NET crop coding systems are based more on how a plant is used, rather than what its genes are.  

• Human perceptions play an important role in defining a crop. Food preferences and market prices often perceive slight but extremely important differences in non-genetic crop features of genetically identical crops like the “common bean”, which FEWS NET codes as R01701AA.  The common bean’s genetically based scientific name is Phaseolus vulgaris, but at the dinner table or in the market, it may be called “black”, “red”, “yellow”, “white”, “kidney”, “navy”, “cannellini”, “alubia”, “turtle”, “pinto”, or hundreds of other names.  And because they are so differently preferred, priced, and consumed, the FEWS NET coding often gives them different two-character suffixes to distinguish them as slightly different but closely related forms of the R01701AA “common bean”.

• Many crops have different uses and are therefore given a different crop code for each use.  For example, the utility of a grain of corn is very different if it is used for preparing a morning breakfast food versus its use as a fodder crop for animals, and each important use has a different code for FEWS NET’s food security-related purpose.  Notably, many vegetables may have different FEWS NET codes depending upon whether they are consumed by their root, stem, leaves, flower, fruit, etc.

• There is not one single system of crop classification that everyone subscribes to for identifying crop plants. There are many.  Even the definitive scientific binomial names we value to identify plants (e.g., Triticum aestivum or common wheat) were originally proprietary terms used by individual researchers, and many, perhaps most, crops have more than one binomial name.  

In summary, each crop classification system is a representation, in some manner, of the needs to which it is being put, and each system almost certainly must allow a certain amount of imprecision and “useful” subjective bias in the categories it creates.  The slight imprecision usually is, however, less important than the value that is provided by being able to broadly group crop products by features that are considered important by users – the reason why “imperfect” crop coding systems persist over time.

Future expansion

Based upon crop data gathered from agricultural production statistics for more than 152 countries and other reporting entities, more than 14,500 crop names⁵ have been coded and associated with one or more of 2,500 scientific binomials (genus + species name) that are used by FEWS NET to identify them in a global environment. More than 40,000 other crop names have been associated with their scientific binomial and could be provided a FEWS NET crop code if needed. 

[1] “Review of Global Food Prices Databases: Overlaps, Gaps and Opportunities to Improve Harmonization”, Food Security Information Network, December 2014

[2] The alpha-numeric code at the beginning of the FEWS NET-derived CPC code serves also to avoid losing the initial “0” which begins many CPC V2 crop codes, when they are used in Excel. 

[3] “Un-processed crops are meant to refer to the crop plant, as it comes from the field, with no other processing applied to it.  Nevertheless, “drying” a crop is not considered a processing operation. The “de-husking” of rice is another notable processing practice which is not considered a “processing” operation. 

[4] The terms “vegetables” and “fruits” are used liberally here to indicate a commonly-perceived grouping of crop plants.

[5] English, French and Spanish names predominate, but many come from other languages.