Victims of old-fashioned statistics

At least half of the women in the world are directly dependent upon agriculture for their livelihood. Most of these are engaged in subsistence agriculture, either entirely responsible for feeding their families or working together with men on the family or communal land. In many countries, especially in Africa, women are the mainstay of the agricultural economy as in other countries they are the core of the domestic economy, obtaining food, clothing and shelter for their families. Yet, even in Africa, the number of men employed in agriculture is estimated at almost twice that of women. For the whole world, the estimate of the number of men working in agriculture is 1.68 times that of women. In 1970, the International Labour Office (ILO) estimated that 287 million women in the world were employed in agriculture, while the corresponding figure for men was 481 million.

With the exception of a few countries, women who support themselves or their families as farmers or agricultural workers are statistically invisible. It is difficult to imagine how women who labour from dawn until dusk on weeding, hoeing, drying grain, tending chickens, carrying water, feeding their families and often also selling some fruits and vegetables in the market could not be counted as working and not contributing to the national economy. Yet, this is what national statistics often show. This statistical neglect is not without cost. Since they do not figure in national statistics, women are too easily excluded from national development planning, with the result that development programmes directed to women are few. Programmes directed to men are at times directly contrary to the interests of women and consequently not well received, sometimes even sabotaged by women. The result may be a badly limping development effort that accentuates sex differences and fosters progress for one sex at the expense of the other, and often no development at all.

The reasons for ignoring women's work statistically are many and varied. In a large part of the world, it is the result of old-fashioned statistical systems, as ill-adapted to developing countries as they were to the colonies of the past. In other parts, it reflects the dichotomy between industrial and household production, between social and domestic labour, and the resultant tendency to split workers into two groups - the labour force and those outside the labour force - rather than measuring degrees of work. The reasons touch closely upon the definition of economic and non-economic activities, between activities to be measured and included in GNP and statistics outside its sphere. And, no doubt, they also reflect, at times, a tendency to consider the activities of men more worthy of measuring and documenting than those of women.

To be meaningful, development planning must be based on adequate data for the entire population, not just the male half. Even estimates of agricultural output or gross domestic product per worker - rather straightforward procedures - cannot be made without accurate estimates of the agricultural labour force, i.e., both male and female workers. If large numbers of women active in agriculture are omitted from estimates of the labour force, this underestimate of the agricultural labour force will result in an overestimation of agricultural output and productivity per worker.

To get better data on women working in agriculture, particularly women active in small-scale and subsistence agriculture, we need a new statistical outlook using concepts, measures and methods of collecting and analysing data better suited to the work of women.

For a new statistical outlook to be useful, however, it is also important to bear in mind that it should produce data more useful for planning purposes; show greater sensitivity to the policy issues of development and the integration of women in develop meet, in particular; be adaptable to a wide variety of local circumstances, yet be internationally comparable; and be compatible with existing measures of economic activity, both to provide comparability with the past and to be useful in the overall statistical system.

The need for compatibility with existing statistical measures suggests, paradoxically, that a system of statistics on women's work in agriculture must be built on standard measures of economic activity, however deficient these may have shown themselves to be in past practice.

To fill out a bare picture

Although women's work has special characteristics and measurement requirements that must be taken into account, special measures and questions on women can never be substituted for standard measures. Rather, they must be seen as an addition to basic labour force measures, to fill out an otherwise bare picture.

Standard labour force concepts have shown themselves deficient with respect to the very definition of economic activity and with respect to the whole range of agricultural work, seasonal work, multiple activities, "women's work," part-time work, work for payment in kind, work as an unpaid family member, work done at home, work in the informal sector of the economy - agricultural or not. Since it is also in these types of work that so many women are found, expanding and refining traditional employment statistics on them will improve the coverage of women's work in agriculture.

In addition, the dichotomization of economic activity into two categories, the economically active and the inactive, too often tends to put women into the inactive category, even when they do some work. If, instead, labour force participation is viewed as a continuum, it becomes possible to collect and tabulate information on different amounts of work, and the less intensive involvement of women can more easily be documented.

Statistics on women may also be improved by adopting the "gainful worker" approach in combination with the "labour force" approach, particularly when it comes to seasonal work, which characterizes in large part women's work in agriculture. In the gainful worker approach, a person is considered economically active if normally engaged in an occupation for direct or indirect remuneration. The reference period is usually one year. In the labour force approach, on the other hand, a person is economically active if a) at work for pay or profit during a specified brief period, either one week or one day, b) with a job but not at work or c) unemployed.

The gainful worker approach is less apt to omit a woman's work because she was not working in the week preceding the survey, but in stressing usual activity, it sometimes tends to classify women as housewives and overlook their seasonal and intermittent work. Care should be exercised that seasonal work be recorded even when it is not a full-time activity and even when non-economic activities such as housework are performed more regularly.

The labour force approach, by recording the economic activity of the preceding week, can provide better coverage of work not done on a regular basis, but will miss all work, including seasonal work, not taking place in the reference week.

The two approaches should therefore be used to complement each other: the gainful worker approach to "catch" regular, including seasonal activities, and the labour force approach to record the entire span of activities, regardless of their frequency, of the preceding week. When the labour force approach cannot be repeated several times a year to cover the work of the entire agricultural season, it is particularly important that it be combined with the gainful worker approach.

In rural areas of developing countries, many people piece together a living by engaging in several kinds of work. A woman may help with the harvest, make baskets and grow chillies, which she sells in the market for cash. A man may be a blacksmith but also own and cultivate land. Thus, unless a survey questionnaire probes for multiple work activities, working time, income and production may be underreported for men as well as women. Since domestic duties occupy much of women's time, small-scale and short-duration activities are particularly easy to overlook, unless special care is taken to record them. Similarly, an investigation of activity patterns and time use will yield the necessary information on multiple activities.

A lower time limit

Traditionally, those working less than full time are classified as part-time workers if working one third or more of normal working hours, usually not less than 15 hours. Those working less than 15 hours per week are usually classified as not economically active. Again, the large majority of these workers, classified as not working and not being in the labour force, are women. In order to include the work of women who are not even counted as part-time workers often because they were busy with housework and worked fewer hours than the usual minimum requirement for part-time work, it is in many cases useful to adopt a lower time limit for recording less than full-time work. In the measurement of underemployment, ILO has proposed the following categories of hours per week: less than 15, 15-34, 35-39, 40-47, 48 and over. Although these categories may be somewhat too precise for work in agriculture, adopting a categorization, including the less-than-15 category, would be especially helpful in "catching" more of women's invisible work.

It is also important to pay more attention to what constitutes work in agriculture - both in the monetized and in the non-monetized sector. Even when they are not responsible for a certain crop, women often help out or engage in a number of subsidiary activities in the fields and gardens. They also participate in a whole range of agricultural work within the confines of the home or farmstead that cannot be labelled "housework" and ignored statistically. Agricultural production, we must remember, includes, in addition to preparing the soil, sowing and harvesting, also weeding, tending, processing, transporting, storing and marketing the product, as well as harvesting and processing by-products of the main crop. In many cultures a large part of this work is done by women, along with the care of small animals, poultry or dairying. Not infrequently, women also provide services to others working in agriculture, e.g., cooking and transporting food to those working in the fields.

Such activities should be included in measuring work in agriculture and farming - not only because they are frequently performed by women, but also because they constitute an essential part of agricultural production. If emphasis is placed on statistics of work rather than of workers, these activities will naturally be included without undue strain on standard statistical concepts. A survey of activities or a time-use study is especially suited to collecting such data. In a regular survey, it is the work that women do as unpaid family work - often on a part-time basis when more labour is needed - work for payment in kind rather than cash, and work done at home that should receive the additional attention required to define and record it accurately, for it is an essential input to agricultural production.

Other easily omitted activities are those that are on the borderline between housework and economic work, e.g., raising chickens, cultivating vegetables near the house, processing food which is at least partly for sale, taking in laundry, knitting, weaving and the like. Since these activities contribute to the gross domestic product, they constitute market work. Omitting them lowers the employment estimates for women compared to those for men, for these borderline activities, by their very nature, tend to be women's work. To reduce this type of underreporting, it is necessary to probe extensively by checking through a list of local agricultural activities or by a chronological recording of the activities of the preceding day.

No longer as rigid

Whether to include "housework," i.e., cooking, cleaning and child care, as economic activity or not remains a major point of debate. Without settling this complex issue, it is nevertheless possible to suggest a solution as part of our overall framework. If economic activity is measured and tabulated as a continuum rather than as a dichotomy of those inside or outside the labour force, the distinction between housework and economic activity need no longer remain as rigid as before. Several different measures of time spent on work, housework or market work, may be devised, e.g., market production, home production and housework activities, or production intended for sale or for pay and for household or family use. The different types of work should be recorded separately and then tabulated separately and together. The dividing line between household production and market production has never been easy to draw, but when information on several different types of activity is collected and tabulated, the classification becomes less restrictive.

In getting better coverage of women in agriculture, efforts should be directed to a wide variety of statistical sources. Population or agriculture censuses must be complemented by sample surveys. It must be remembered that the different vehicles for data collection, whether censuses surveys or even administrative records, lend themselves to collecting fundamentally different types of data. We cannot expect a census to produce a great deal more than a general overview and a sampling frame for later, more detailed analysis for separate areas or for the whole country. For instance, a 1-percent sample in a census or a national survey based on the census sampling frame can be used to obtain the further detail needed. However, even such surveys cannot be expected to provide all details for all data users. Although the national statistical systems with considerably expanded coverage and detail on women should always be the mainstay in data collection, small-scale and special surveys provide flexibility, independence and openness to new issues that are of value.

With repeat visits

The timing of the survey or census is also of importance in measuring the seasonal work of women. In order not to interfere with the busiest times of the agricultural year, censuses or surveys are often taken during the slack periods. Women who are very busy during peak-labour seasons but much less active in the slack season are then easily left out of the labour force. Ideally, surveys should be designed with repeat visits spread over the year. When this is not possible, the longer reference period of the gainful worker approach and the shorter reference period of the labour force approach should be used together.

In measuring women's work, the role of the interviewer is of paramount importance. Without a thorough understanding on the part of the interviewer of the special problems in measuring women's work, conceptual innovations come to naught. It falls upon the interviewer to ascertain all the kinds of work women do, i.e., economic as well as non-economic activities. This distinction (which in practice has become a rather loose statistical construction around the basic concept of direct contribution of labour to the gross domestic product) is far too complex to be made by the respondent herself, or even by the interviewer, and should instead be made when data are coded and classified. This, of course, will also permit a less rigid set of classifications as already discussed.

But the interviewer must also be made aware that statistical tradition, his own expectations, a woman's eagerness to provide the expected and socially most acceptable answer, her husband's - and thereby also her own - social standing as the competent provider of the family, and the immediate expectations of those present conspire toward the underestimation and statistical invisibility of women's work.

To overcome these difficulties, the interviewer must inquire about all of women's activities in detail. He or she must ascertain the type of crops grown and the kind of animals raised in the household, and specifically ask each woman about work with each of these. Also, he or she must not be content with one answer but must probe by asking repeatedly about all the different types of work performed to ensure that the marginal or part-time activities that women so often do are included.

How the results are tabulated, finally, will make a significant difference in the information provided on the labour force participation of women. Making all tabulations by sex is a basic step in studying the work of women. Particularly when the work of women is investigated with the detail recommended above, tabulations by sex provide rather a complete picture of the agricultural labour force. Gross-tabulations for the various categorizations of pay, no-pay, cash-kind, family work, work outside the family, at home, away from home, etc., fill out the picture further.

By leaving the distinction of economic activity to the coding phase of the survey, a more precise and consistent definition of economic activity becomes possible. It is no longer necessary to use the rather artificial dichotomy between economic activity and non-economic activity. Rather, work may be measured in several degrees of intensity, also useful in the study of the utilization of labour, to complement as well as to expand upon the two-way classification of of those inside and outside the labour force. Borderline activities may or may not be included in the count of economic activity and the results given together. Also, estimates of the labour force may be made cumulatively, so that the first activity only, then first and second activity together, and finally first, second and third activity together may be presented. Classifications of this sort give a considerably fuller and more correct account of women's many activities without losing comparability with the simple two-way classifications of the past.

The sexual division of labour

In addition, it is frequently useful to compare the two sexes on different aspects of work in agriculture through indices of percent female or the sex ratio, i.e., the ratio of males to females, on each item tabulated. For instance, the sex ratio of labour force participation, the sex ratio of unpaid family workers illustrate the sexual division of labour in agriculture.

In reshaping our thinking about development to combat poverty and improve the standard of living rather than bring about a mechanical increase in the GNP, it becomes essential to have data on the daily life and work of the population, including that of women. Because of the key roles that women play both in the sustenance of their families and the socialization of the young, in community affairs and in agricultural production, it is clear that statistics on the economic roles of women are an essential part of the data needed for both economic and social development planning.