Posted December 1999

Using agricultural census data for demographic purposes: some pros and cons

Background

This paper focuses on the census of agriculture, which so far has been insufficiently explored as a possible data source for the examination of interactions between population and agricultural factors. Although agricultural censuses are primarily designed to describe agricultural holdings and thus can cover only a part of the agricultural population - i.e. holders, members of their households, and agricultural workers employed on holdings -, they could be a powerful tool for analysis because they provide a unique opportunity to link the information on the techno-economic structure of holdings with the demographic and employment characteristics of persons attached to them. The author gratefully acknowledges the helpful comments of J. du Guerny, A. Marcoux, S. Ogbuagu, M. Villarreal (Population Programme Service, Women and Population Division, FAO), L. Kabat, L. Naiken and A. Gonzalez-Villalobos (Statistics Division, FAO), and J. Vercueil (Agriculture and Economic Development Analysis Division, FAO), on a previous draft.

1. Introduction

Agricultural census results can be used in many different ways, ranging from very general to very specific technical applications. From a strictly statistical viewpoint, the census data represents one of the most important components of the information system in a country and can serve as the basis for many other statistical activities related to food and agriculture, such as conducting various agricultural sample surveys. While the raison d’être of the agricultural census is no doubt the use of census data for agricultural development planning and formulation of national agricultural policies, a much wider application of census results is possible.

This paper looks at agricultural censuses from the point of view of their applicability for the analysis of linkages between certain aspects of rural population change, natural resources and agricultural factors. Its objective is threefold: (a) to provide an overview of demographically relevant items available from censuses of agriculture, (b) to discuss selected subject areas that can be studied using population data from agricultural censuses, and (c) to identify some of the problems of using agricultural censuses to examine interactions between agricultural factors and demographic phenomena. Because the agricultural census is of particular importance to countries in which significant segments of the population depend on agriculture for their livelihood, the focus of the paper is on developing countries.

2. Demographic information obtainable from a census of agriculture

Box 1 An agricultural holding is an economic unit of agricultural production under single management comprising all livestock kept and all land used wholly or partly for agricultural production purposes, without regard to title, legal form, or size. Single management may be exercised by an individual or household, jointly by two or more individual or households, by a clan or tribe, or by a juridical person such as a corporation, cooperative or government agency. The holding's land may consist of one or more parcels, located in one or more separate areas or in one or more territorial or administrative division, providing the parcels share the same production means utilized by the holding, such as labour, farm buildings, machinery or draught animals. The requirement of sharing the same production means utilized by the holding is necessary if the various parcels are to be considered as components of one economic unit. Economic units engaged solely in the following economic activities are not considered agricultural holdings and are therefore excluded from the census hunting, trapping and game propagation forestry and logging fishing agricultural services (FAO, 1995, p. 25).

The collection of demographic information in the agricultural census is guided by two basic principles, each having important implications for data utilization:

• Although agricultural holding is the recommended unit of analysis, demographic information is in fact obtained through investigating another unit: the holders’ household (see definition in Box 2). In reality, one agricultural holding can include several agricultural households, and one agricultural household can operate on several agricultural holdings. In developing countries, a one-to-one correspondence between a household and a holding is quite usual, but it is certainly not universal. When using agricultural census data, one has to remember that in some contexts it is common that members of one household operate separate holdings, or that holdings are operated by two or more persons belonging to different households.

  Box 2 The concept of household is based on the arrangements made by persons, individually or in groups, for providing themselves with food or other essentials for living. According to UN (1998, p. 65), a household is classified as either (a) a one-person household, defined as an arrangement in which one person makes provision for his or her own food or other essentials for living without combining with any other person to form part of a multiperson household or (b) a multiperson household, defined as a group of two or more persons living together who make common provision for food or other essentials for living. The persons in the group may pool their incomes and have a common budget to a greater or lesser extent; they may be related or unrelated persons or a combination of persons both related and unrelated.

• The demographic data collected in a census of agriculture refers only to persons attached to agricultural holdings. Demographically speaking, a holding may consist of the holder (defined as the person who makes major decisions regarding resource use and operation management), other persons belonging to the holders’ household, and hired workers who either permanently or occasionally work on the holding. One essential implication of this ‘holding-based’ approach is that other categories of the agricultural population - members of landless hired workers’ households, persons engaged in hunting, forestry and fishery activities or agricultural services - are, by definition, excluded. Thus, agricultural censuses do not cover all persons associated with agriculture.

Each FAO Programme for the World Census of Agriculture attempted, in one way or another, to cover some of the basic demographic and economic characteristics of persons belonging to the population of the holders’ households. FAO recommendations have typically been decided on the basis of extensive consultations with statistical offices in individual countries. Their evolution thus mirrors the collective experience of national and international organizations with the collection of agricultural information. To assist countries in deciding which types of data to monitor in an agricultural census, FAO’s practice has been to distinguish in its recommendations two categories of census items: proposed and essential. While the first category represents the sum of ‘desirable’ characteristics, the latter group comprises a subset of priority items that the FAO considers indispensable for comparison between countries and over time.

The 1970 FAO agricultural census programme (FAO, 1965), spanning the decade from 1966 to 1975, suggested that each person belonging to the population of the holders’ households be classified by age, sex, occupation (agricultural, non-agricultural), and intensity of agricultural activity (permanent, temporary, occasional). For the 1976-1985 decade, FAO recommended that in addition to the above items, information should also be collected about the relationship of household members to the holder. Furthermore, the recommendations stipulated that censuses should monitor the number and sex composition of hired permanent agricultural workers and the utilization of hired occasional agricultural workers on the holding. Apart from the sex of hired workers, all census items related to the population of the holders’ households were recommended as essential ones (FAO, 1976). The 1990 Programme contained similar recommendations, except that ‘relationship to holder’ was removed from the list of items proposed for household members (FAO, 1986). The most recent FAO guidelines on demographic characteristics to be monitored in agricultural censuses during the period 1996-2005 (FAO, 1995) are summarized in Box 3.

In each of FAO’s Programmes for the World Census of Agriculture, countries were cautioned against overloading the census with too broad a substantive coverage as this could complicate collecting and processing of data. Recognizing that countries differ in their capacity to carry out a census of agriculture, FAO Programmes have always included a recommendation that countries should tailor the agricultural census to their unique situation. Countries with poor statistical systems have been advised to restrict the scope to essential items, whereas statistically more developed countries have been invited to broaden their census objectives. Ultimately, however, it is up to the national authorities to choose the statistical topics to be monitored, and the classifications to be used, in the agricultural census in their country.

3. Applicability of agricultural census data for demographic purposes

There are many reasons to be interested in agricultural censuses as sources of information about persons associated with agricultural holdings. To keep the discussion within reasonable bounds, two subject areas will be considered in this paper: (a) demographic structure of the holders’ households, and (b) patterns of economic activity within households. The following paragraphs are largely theoretical and not intended to give concrete examples of analyses undertaken on the basis of available data. However, it needs to be emphasized that for the purposes of demographic analysis, the routinely tabulated information in the various census reports (such as those listed in FAO, 1997) might not be sufficient or suitable, and in most instances one would have to use the original micro-level data sets.

3.1 Demographic structure of agricultural households

Available evidence indicates that at least two basic dimensions of the agricultural system are directly relevant to household structure: (1) the size of holding which a household has at its disposal for production purposes, and (2) the ownership conditions which influence the use of land and the distribution of products from it. These two dimensions have quite different implications for household size and composition:

• Size of agricultural holding is believed to influence household structure primarily through affecting individual households’ assessment of the benefits associated with having (additional) children and the costs connected with childbearing and childrearing. In traditional societies, farm households with larger holdings are able to use productive work of most household members, and therefore they may prefer more children in order to generate a larger income. On the other end of the spectrum, farm households with smaller holdings can be expected to have lower sizes because they have to cope with a less predictable economic climate, especially regarding domestic work for their members. In addition, size of the holding can have some bearing on migration of household members to and from the holding; hiring of agricultural workers; and transfer of funds between those living on the holding and their distant relatives. Of course, in most settings the demographic implications will depend not only on the acreage of the holding as such, but also on the quality of land and its utility within a given farming system.

• Land ownership, as opposed to landlessness, is a key to economic well-being in agrarian societies because it can guarantee an income independent from the owner’s direct involvement in agricultural production, for instance, in the form of a rent or labour return. It has been argued that land ownership reduces the importance of household members as a source of economic security and therefore leads to lower household size. Moreover, farmers who own the land have usually more to lose from subdividing the land for their children. Compared to those who rent land from others or operate communally held land, owners of land may have a greater desire (as well as more resources) to invest in children quality rather than quantity. In contrast, agricultural producers who do not own land have usually little or nothing to lose by having a large number of children.

In reality, of course, the links between land ownership, size of holding and household structure can be much less direct and straightforward than the above generalizations would imply. For instance, wealthy farmers who have enough land to divide up may not necessarily worry about excessive land fragmentation, and therefore may see little benefit in reducing the size of their households. On the other hand, family-formation decisions of small landholders who have little land to divide may be governed by factors beyond land ownership, some of which (such as overall increases in costs of childraising) may depress rather than boost their family size preferences. Moreover, even when an association between characteristics of holdings and household composition is evident, the interpretation of the underlying causal link may be difficult. Is household structure adapting to size of the holding, or vice versa? Does limited access to land lead to larger household sizes, or the reverse? Clearly, the mechanisms through which landownership or size of a holding interact with household composition will largely depend on to the specific geographic, cultural, socio-economic and institutional environments and production patterns within which they occur.

An agricultural census is capable of providing information on the interactions between demographic and economic attributes of the holdings enumerated in the census. In this respect, a particularly important property of the census is that it typically refers to very large numbers of observations and therefore covers a range of socio-economic and environmental contexts. This means that agricultural census data may be disaggregated in a number of ways - by demographic characteristics of members of households, patterns of land use and land tenure, types of agricultural production, agro-ecological zones, etc. - without running into the problem of small sample size, which is so often the case with sample surveys. A proper disaggregation of census data could thus be a very effective strategy to uncover, in an indirect way, the causal links between the demographic composition of the holders’ households and techno-economic features of agricultural holdings.

On the other hand, it must be recognized that certain components of the interplay of demographic and agricultural factors are very difficult to capture in agricultural censuses. For instance, unless the information about the relationship of household members to the holder is monitored, it is virtually impossible to assess the role of intra-household gender relations in influencing demographic and agricultural outcomes. In addition, the quality of demographic information collected in an agricultural census may not be high enough to permit detailed sub-classifications, thus posing serious obstacles to analysis. This may be especially the case in countries where resources for training of enumerators and verification of data are inadequate. In such circumstances, the sample survey approach may be a more efficient way to obtain more detailed and more reliable data on the population engaged in or dependent on agriculture.

While land distribution and tenure are probably crucial to most aspects of rural economic and demographic structure, there are many other themes that could be analysed using data from agricultural censuses. Examples of particularly promising topics include:

• differentials in household size and structure according to types of agricultural production, characteristics of farming systems, etc.;
• associations between characteristics of agricultural structure and various socio-demographic features of households (proportion of the elderly, number of generations present, educational structure etc.)
• migration from and to holdings in relation to agricultural characteristics of the holdings.

3.2 Patterns of economic activity in agriculture

The amount of agricultural work to be carried out on a holding is irregular and may change from day to day, depending on the weather and other agro-ecological conditions. It is also seasonal: there are times of year in which the work is intensive and others when it is negligible. For these reasons, analyses of agricultural census data need to focus on a longer, rather than shorter, time-reference period. Of interest is not only the total volume of work carried out on the holding but also its distribution according to socio-demographic characteristics of workers and the inter-relationships between agricultural work and the size, type and other characteristics of the holding.

The phenomenon of part-time farming is particularly important. Although often not measured, part-time farming has always existed in many developing countries and, recently, it has been gaining importance. Therefore, one of the objectives of the analysis of data on the agricultural population and employment should be the search for a coherent interpretation of this phenomenon in relation to the various economic activities of the members of the holder’s household and the characteristics of the holding.

From a demographic point of view, characteristics of economic activity of members of agricultural households are important primarily through their connection with childbearing and migration. For instance, while traditional patterns of agricultural production usually boost the need for child labour and, at times, foster immigration from poorer rural areas, the transition to more modern forms of agricultural production may reduce the need for labour, resulting in lower levels of fertility and immigration. Moreover, the more advanced forms of farming are often more capital-intensive and the money required for capital inputs is often obtained by temporary migration of some household members to urban areas or through their involvement in the rural off-farm labour market. In some contexts, regulation of household size through selective migration and fertility control may in fact be one of the chief strategies of coping with fluctuations on the local agricultural labour market. Evidently, fertility change and rural-urban migration are complex phenomena that are shaped by many factors, not just agricultural realities. However, ignoring the links between agricultural economic activity and demographic behaviour could lead to erroneous conclusions about the nature of rural population change and rural-urban migration, and, subsequently, to bad population and agricultural policies.

The more specific applications for the agricultural census data would be to bring to light the following topics:

• the intensity and nature of agricultural activities in relation to demographic characteristics of members of agricultural households;
• the linkages between movements of workers from (or to) the agricultural sector to (or from) other branches of economic activities and demographic features of the respective agricultural households;
• the linkages between economic activity and demographic behaviour of persons who belong to agricultural holdings but are not employed in agriculture.

4. Main problems of using agricultural census data

4.1 Census scope

Furthermore, in spite of the widely recognized importance of the agricultural census for the management and study of rural issues, a number of countries do not organize their national censuses according to FAO recommendations. A full-scale agricultural census is an expensive operation which some developing countries may find difficult to support. Besides, many countries have alternative sources for some of the information they need on persons working in or dependent on agriculture (for instance, population census or sample survey evidence) and therefore may be inclined to collect only selected data instead of conducting a full agricultural census. Given that the main aim of the agricultural census is to provide data on physical and technical characteristics of agricultural holdings, it is perhaps not very surprising that many countries tend to reduce the number of census items related to the population of the holders’ households, while some do not collect such information at all.

As an illustration, Table 1 shows to what extent demographic information had been collected in individual developing countries in the 1986-1995 round of agricultural censuses. Of the 56 developing countries which carried out a census of agriculture during that decade, 35 monitored the size of population belonging to agricultural holdings, 32 collected information on the sex composition of this population, and 26 also collected information on its age structure. However, this simple overview conceals substantial inter-country differences in the detail of demographically relevant information obtained. For instance, in monitoring ages of members of households, some countries asked respondents to present the information in single-year age groups, while others instructed them to use five-year age groups or even broader age categories. In some countries, classification by sex is available only for adult household members, etc. The practice of using cruder classifications or omitting certain demographic variables can seriously limit the scope for a demographic analysis of census results, and in some instances it may preclude such an analysis altogether.

4.2 Comparability and reliability of population and employment data

Comparability may also be hampered by different definitions of the minimum size of holding. FAO recommends that the minimum be set as low as practicable because of the importance of small holdings in contributing to agricultural production, but does not set a recommended minimum. Countries thus establish their own minimum, resulting in considerable variation from country to country so that data may not be easily comparable internationally. Yet the definition of the minimum size of holdings can also have important implications for analysis within individual countries. In most countries, people attached to very small holdings are likely to differ from the rest of the farm population in a number of demographic aspects: they may have a distinctive age and sex structure, display very specific patterns of economic activity, childbearing, migration, etc. Setting the minimum size of holding ‘too high’ may mean that the demographic picture obtainable from the agricultural census will be grossly incomplete; setting it ‘too low’ may result in some very atypical holdings being included. Countries should ideally define the minimum size of holding with respect to the existing distribution of agricultural holdings by population characteristics, to ensure that the census produces a demographically fair description.

Finally, a major obstacle to comparability stems from the fact that agricultural information is notoriously difficult to collect and therefore the census enumeration may not necessarily produce reliable data. Quality of census data depends on a host of factors, ranging from the respondents’ willingness and/or ability to supply the correct information to technical aspects of data collection, verification and processing. This is not to say that the agricultural census data are, by definition, of questionable value to population research; it does mean, however, that one should exercise greatest caution in using them.

4.3 Frames of reference

4.4 Cross-sectional nature of census data

A partial solution would be to use data from two or more consecutive censuses, but even such an analysis may not convey a comprehensive picture of developments over time. Besides, not all countries conduct their agricultural censuses in a way to enable inter-censal comparison. Take for example the information on sex of holder. As shown in Table 2, 52 less developed countries monitored this information in either the 1980 or the 1990 census round; however, only ten countries included this census item on both occasions. This illustration indicates that a detailed comparison of census results over time would be possible for only a minority of developing countries.

5. Conclusion

In many developing countries, censuses of agriculture could be one of the best sources of information on persons belonging to agricultural holdings. Evidence from an agricultural census is always important in its own right, and in many countries it may be of interest simply because of the unavailability of better data on rural people. But the agricultural census may also serve as a powerful tool for the study of the interrelationships between the characteristics of the holding and those of the persons associated with it. From the point of view of demographic analysis, the agricultural census provides a unique opportunity to link information on the techno-economic features of the agricultural holding on the one hand and the demographic structure and employment characteristics of the holder’s household on the other. Such a use of agricultural census data is clearly not free of technical problems and hence is unlikely to become a routine undertaking. Nevertheless, agricultural censuses do not deserve to be overlooked by those dealing with rural and agricultural population issues. Agricultural census data is readily available in a number of developing countries, often for several consecutive years, and provides a rich reservoir of information that can be analysed with little additional cost. Although the scope for demographic examination is inevitably limited by existing practices of monitoring population characteristics of agricultural holdings, agricultural censuses should permit the kind of disaggregated analysis which is usually impossible with in-depth surveys. In addition, analyses of the linkages between the organization of agricultural production and demographic behaviour of farm populations have the potential to expand the already existing understanding of rural demographic change and to address significant policy issues. Insights derived from agricultural censuses can then be used to guide policies and programmes to bring about a socially desired path of rural development.

The discussion of potentially important interrelationships between agricultural and socio-demographic factors presented in this paper by no means represent a complete catalogue of opportunities for using the data from agricultural censuses. It is clear, however, that several theoretical frameworks exist for demographic analysis based on this sort of evidence. Some will perhaps argue that an optimum basis for analysis would be obtained by matching the information collected through the census of agriculture and that obtained from the population census and relevant sample surveys. However, such matching is likely to be beset with a plethora of technical problems which are in general very difficult to solve, particularly in countries with weaker statistical systems and/or resource constraints.

References

FAO (1965) Program for the 1970 World Census of Agriculture. FAO: Rome.

FAO (1976) Programme for the 1980 World Census of Agriculture. FAO Statistics Series No. 1. FAO: Rome.

FAO (1978) Collecting Statistics on Agricultural Population and Employment. FAO Economic and Social Development Paper No. 7. FAO: Rome.

FAO (1981) 1970 World Census of Agriculture: Analysis and International Comparison of the Results. FAO: Rome.

FAO (1986) Programme for the 1990 World Census of Agriculture. FAO Statistical Development Series No. 2. FAO: Rome.

FAO (1992) 1980 World Census of Agriculture: Methodological Review. FAO Statistics Series No. 107. FAO: Rome.

FAO (1995) Programme for the World Census of Agriculture 2000. FAO Statistical Development Series No. 5. FAO: Rome.

FAO (1997) Report on the 1990 World Census of Agriculture. FAO Statistical Development Series No. 9. FAO: Rome.

UN (1997) World Urbanization Prospects: The 1996 Revision (Annex tables). UN Population Division: New York.

UN (1998) Principles and Recommendations for Population and Housing Censuses: Revision 1. Statistical Papers, Series M, No. 67/Rev.1. UN Statistics Division: New York.