Introduction
Critical task analysis
Extensive recall methods
The time allocation technique
References
Collecte de données sur la main- d'euvre
Summary of discussion session 6.
Résumé des débats de la sixième séance
Barbara E. Grandin
Anthropologist, Arid Zones (Eastern and Southern Africa) Programme, ILCA,
Kenya
Labour plays a crucial role in any agricultural production system. In a pastoral system where grazing and water are theoretically of free access within a resource-sharing group, capital (in the form of animals) and labour power are the essential means of production which must be acquired and managed for a successful operation. This paper addresses issues regarding the collection of data on labour inputs for pastoral production. First it briefly considers the utility of data on labour as part of pastoral systems research, and then discusses research methods for data collection.
The need for labour data in pastoral systems research
We have already noted that unlike FSR, PSR on the whole suffers from a paucity of baseline information and understanding of system dynamics. This is perhaps more true of labour than of any other input to production. Anthropological studies, which comprise the bulk of the research done on pastoral societies, have tended to focus on the social organisation of labour while paying little attention to the actual labour demands of the system. However, a proper understanding of labour as a system component must pay attention to both its social and technical aspects.
The energy demands of many tasks necessary for pastoral production are much less than the energy demands for crop production. A myth has arisen about the idle pastoralist, because inputs in energy were confused with inputs in time. It has been assumed on the basis of little evidence that labour rarely if ever constitutes a constraint to pastoral production. Hence, traditionally, little attention has been paid to the labour demands of pastoral production. Labour data gathered by several ILCA teams, as well as other researchers, however, are beginning to demonstrate that in fact there can be large demands for labour time in pastoral production.
As labour is a critical input into pastoral production, it is necessary to understand both the labour demands of the households and the labour supply available, In the descriptive and diagnostic phase of PSR informal surveys can be used to gather information on the following aspects of labour:
1. The critical production tasks, their timing and seasonality.
2. The culturally accepted age and sex division of labour.
3. The possible recruitment pathways for labour.
4. Labour demand/supply variations between households and social mechanisms for adjusting these.
5. The constraints due to insufficient labour including its effect on productivity.
Ideally there should be sufficient information to make some estimation of the economic returns to labour.
Through the use of a labour profile which details the time and amount of labour inputs into production activities by category of worker and season, it may be possible to determine whether there are labour constraints to production. It is also possible to determine what the opportunity cost of proposed interventions might be in terms of the shifting of time allocation from one activity to another.
These factors must be considered for designing interventions. If proposed interventions require significant changes in labour inputs or patterns, it is essential to know: Is labour available and/or can it be obtained at the required time? What would be the opportunity costs forgone with the desired changes of labour input? Will the age/sex group normally responsible for a task or group of tasks be willing or able to undertake a proposed intervention?
Occasionally, proposed interventions cross the accepted age/ sex division of labour. For example among Maasai, women have the responsibility for fetching water: although they occasionally bring water to the dwelling site for young or ill animals, their primary responsibility is to fetch water for domestic use. It is the responsibility of the men to see to the watering of animals. Thus an intervention which suggests or recommends the carrying of water to the dwelling site on a regular basis for young animals would most likely meet with resistance on the part of women because they do not see it as their responsibility to water animals except in extreme instances, and on the part of men who do not see it as their responsibility to bring water to the dwelling site.
In determining labour availability (particularly with a view to introducing a new technology) it is important to collect data on the household level. Householdsvary significantly in labour needs, labour availability as well as in their labour need/supply ratio.
In one Maasai group ranch, poor and rich families have a total work force of 6.3 and 9.8 people respectively. (Workers are defined as all adults and children over the age of six years). However, although richer households are much larger, their livestock holdings are also much larger, and consequently the ratio of livestock units per worker is much higher. For example in Olkarkar, whereas rich households average 26 livestock units per worker, poor households average only 6 livestock units per worker. These different ratios of livestock units per worker affect the management strategies and time allocation between rich and poor households, although they may be offset by social and market mechanisms for acquiring labour.
With the ideas of building up a repertoire of methods for gathering labour data, this section will briefly review some of the more traditional methods available, and provide more in-depth information on a relatively new technique. The better-known methods include: critical task analysis (which may include some direct observation) and recall; the emerging technique is called the random visiting or time allocation technique. The critical task technique involves specifying the most important production tasks and trying to quantify the time necessary for them. Recall techniques involve interviewing various workers and asking them how much time they have spent on a certain activity for a certain specified period. Recall techniques range in intensity from seasonal to daily interviews. (In the case of the seasonal interviews they are more like critical task specification as it is impossible for the worker to remember actual hours worked). Lastly, the time allocation technique is based on random visits at which time the current activities of all workers in the household are recorded. These methods are not mutually exclusive; as appropriate, they can enhance one another.
There is a particular dearth of labour input data for pastoral systems. Swift (1979), Torry (1977) and Dahl (1979) have applied the critical task technique to pastoral systems of the Tuareg of Mali and the Gabra and the Borana of Kenya respectively. In critical task analysis, through observations and interviews, the researcher develops a list of the most important production tasks and the approximate amount of time that it takes to complete these tasks. The quantification may be done by direct observation (parallelling the work study approach used in farm management studies). Dahl describes a fairly large number of both the minimum livestock and domestic tasks necessary to sustain a Borana household, but the does not attempt to quantify the time involved. Torry, on the other hand, estimates only labour time required for herding, milking, and watering the major species kept by the Gabra, as well as labour needed to collect water for domestic consumption. Using these species specific task estimates, he then estimates the labour that would be required for an independent Gabra household. He estimates that a total of 2,494 hours are required each month for these critical tasks, and suggests that this labour intensive system necessitates the pooling of the labour of several households and thus joint herding. Swift also pays primary attention to labour input. to livestock management but he refines the technique by estimating necessary inputs by season as well as by management unit size.
These critical task approaches to livestock production provide useful and interesting data in a cost-effective way. They are similar in many respects to Collinson (1972)'s limited-visit survey. As Collinson notes, "The same type of pattern is formed for components of the labour profile by repetition from season to season... limited visit techniques deliberately seek to exploit this pattern in eliciting answers based on experience rather than on historical recall of labour use for that particular season." They suffer, however, from limitations inherent in the method. They deal only with general descriptions of major tasks and provide no information on the actual inputs of labour by various workers and variations in allocation of labour which might be related to season, skill or wealth.
Cossins (personal communication) uses a variation of the critical task technique in his on-going study among the Borana of Ethiopia. Preliminary survey results indicated that labour necessary for watering animals in traditional Borana wells might be a key constraint to productivity. All animals are watered from wells for four to six weeks each year; thus the amount of water that can be extracted from wells at this time is a limiting factor of production.
He is dealing with this particular labour input on several levels. At the first level' for the whole production system, he sought to determine in gross teens whether there is sufficient labour available to work the wells, i.e. whether labour is a constraint to water production in the dry season.
In the Borana study site, of the 30 or so major well groups, 24 were studied in detail and sufficient information was gathered on the remaining wells to allow extrapolation. Data on total numbers of workers required for each task (by age/sex category) were obtained. This was compared with the size of the population for these age/sex categories which had been obtained from a demographic survey. This showed that in overall terms labour was not likely to be a limiting factor. The second level, now underway, is to look at labour demand/ supply for each well group and then each well.
Multi-visit recall techniques for collecting labor data are currently being used by Swift in Mali and White in Niger. They use resident enumerators to collect labour input data from all workers in sample camps at twice-weekly intervals. As Swift (1982) notes, "The notion of elapsed time created some difficulties." The use of watches was not successful, so that the local concepts of time (broken into seven categories) were used to estimate time spent on various activities. Both Swift and White collect the labour data with a battery of other data as part of a much broader study, so that the enumerator costs were not high.
In designing a multi-visit recall study for labour data, perhaps the two most important decisions are the frequency of the visits, and whom to interview.
1. Frequency
Collison (1972) stated that "where work organisation is complex and highly irregular even from day to day,... information for two or three days previous may be too much to ask and liable to heavy transfers." This is likely to be less of a problem in pastoral societies where the bulk of the labour inputs are regular. However, if detailed information is required, especially on tasks other than herding and watering, twice-weekly recall is probably as necessary with pastoralists as it is with farmers. This is the frequency being used by Swift and White. As noted earlier, however, high frequencies can lead to problems with respondent and enumerator fatigue.
2. Respondents
The decision on whom to interview depends both on the breadth of data that is required and the cultural practices in the area under study. The herd owner at least should be interviewed. He would probably be able to say which people looked after each group of animals for the previous few days, and in what major management tasks they engaged. The researcher would have to be particularly wary of reporting biases, e.g. if the herder really doesn't know, but won't admit the fact, or if due to labour shortages, the household has non-conventional patterns of labour input to which the herder would not want to admit. For more precise information, each worker would need to be interviewed separately as Swift and White have done. This greatly increases the costs of the survey (both in data collection and analysis), but would significantly increase accuracy.
Either limited-visit or extensive recall data may be able to yield sufficient data for determining labour bottlenecks. However, these methods have several drawbacks related to the fact that they tap only major inputs into productive, and possibly domestic, activities. First, as many smaller inputs are missed, it is not possible to calculate returns on labour in pastoralism (about which we know very little as Eicher and Baker have noted). Secondly, the researcher is not likely to know what other activities the workers are engaged in. He will thus not be in a position to understand what activities would have to be forgone should labour intensive innovations be suggested.
Because of these felt difficulties with critical task and recall methods of labour data collection, in the Kenya Maasai study, the ILCA human science team used a data collection technique called "time allocation" recently developed by American anthropologists. The method relies on randomly timed visits to a household during which descriptions of the activities of all its members are recorded. These records produce a series of verbal snap-shots sufficient in number to provide a thorough description of activities by such parameters as age, sex and season. The time devoted to a particular task is extrapolated from the percentage of all activities devoted to that task (See Grandin, 1982 for more details).
The requirements of the time allocation technique are the following :
1. Sample households which are included in the study must be censured so that the name, age and sex of all the workers in the household can be determined2. Enumerators must stay near the homesteads so that all the important hours of the day might be covered. Thus the technique is difficult with a highly nomadic population in which encampments or households split seasonally.
3. A computer is needed for analysis as the number of snapshots or records is quite large. Although the Kenya team has used a main-frame computer, a micro computer would be sufficient.
The time allocation technique has one disadvantage in addition to its basic requirements. It is not adequate for specifying time spent on certain activities away from the home. For this category of activity it is useful if it can be implemented by some specific observations or at least interviews as is done in critical task analysis. For example in the watering of livestock, if a producer leaves the household in the morning with animals to water, it would be useful to know how much time he spent walking to the water, how much time he spent waiting for the animals to be watered, how much time the animals spent watering and how much time the man spent socialising or gathering information. The time allocation technique in and of itself cannot supply these data.
The advantages of the time allocation technique, however, are numerous.
1. It easily covers all potential workers.2. It is not plagued by the problems of recall since activities are recorded on the spot.
3. There is little respondent fatigue; in fact almost any person in the household can be interviewed. There is no requirement for uniformity of respondents as long as one can be sure, within the context of the specific society, that the person interviewed is likely to know the whereabouts and activities of the various members of the household. Recording of the information does not take very long. Five to ten minutes is normally sufficient for a household of 10 people.
4. Apart from a household census, little pre-survey labour data is required. It is not necessary to specify critical tasks beforehand, and little needs to be known on the age and sex division of labour.
5. The time allocation technique covers the complete range of tasks whereas critical task techniques depend on the specification of just the few most important tasks. Time allocation will record minor but important tasks, the cumulative effect of which can be very important for the production system. For example where critical task techniques tend to concentrate on time to herding and watering, in our Maasai study 70% of the time devoted to livestock management (excluding milking) was devoted to herding and watering. This critical task specification would have missed 30% of livestock management tasks.
6. Although we have not done so in the Kenya study, it is possible with time allocation to record multiple simultaneous activities. Thus a woman who was sitting outside, watching a child while making butter and keeping an eye on smallstock can be recorded as doing all those tasks. Were she to be interviewed on what she had done at that period of time, it is likely she would have said "making butter" as that is the most discrete, least regular activity.
7. Lastly, and perhaps most important, the time allocation technique provides critical information on production unit variation. If the sample size is sufficient, households can be compared according to their labour supply, wealth, neighbourhood or any other parameter which is thought to affect labour input.
Although the time allocation technique used by the Kenya team was continous for a period of just over a year, we see no a priori reason why the technique could not be adopted to more limited periods of time. With the help of the biometrician at ILCA's headquarters, the team is doing retrospective analysis to determine the minimum number of visits deemed to be necessary in a season in order to provide a labour profile for that season. Preliminary results suggest that little precision is gained by having more than 10 visits per household per season. On the other hand, whereas a minimum of 10 people per person category desired for analysis would be required, precision would be increased by increasing the number of people studied. We suspect that frequent visits to a number of households (for example two weeks of visits to each household four times a day) should provide the necessary quantity of data to estimate production inputs and other activities. Of course, in order to use seasonal sub-sampling more background information on labour inputs and especially constraints would be required to determine the optimal timing.
Sample results of the time allocation technique
This section presents a few preliminary results from the time allocation study by the ILCA team in Kenya. Visits were made to each sample household twice a month at randomly chosen times between the hours of 6 a.m. and 8 p.m. during which period the activities of every member of the household were hosed. Table 1 shows how the observations were used to calculate the percentage of livestock management tasks done by each age/sex category of worker. For example, there were 1,121 observations of livestock management activities out of a total of 3,530 observations. Thus 41% of all the observations were of livestock management (representing approximately 4.3 hours per worker per day). For watering, there were 123 observations, of which 15% were male children, 5% were female children, 75% were adult males and 6% were adult females. While this table presents data on the percentage of specific tasks done by various age/sex categories, the time allocation data can also be used to calculate hours per day spent on these tasks, either in total or by age, sex, season etc...
Table 1. Age/sex roles in livestock management at Olkarkar.
|
Activity |
Percentage of tasks done by each group |
Livestock management observation |
|||||
|
Children¹ |
Adults |
Total |
N³ |
% |
|||
|
M |
F |
M |
F |
% |
|
|
|
|
watering |
15 |
5 |
75 |
6 |
101 |
123 |
11 |
|
herding |
47 |
45 |
3 |
5 |
100 |
641 |
57 |
|
dipping |
5 |
5 |
68 |
21 |
99 |
19 |
2 |
|
boma livestock work |
18 |
24 |
33 |
26 |
101 |
274 |
24 |
|
other livestock work |
30 |
6 |
48 |
16 |
100 |
64 |
6 |
|
No. of observations |
389 |
363 |
245 |
124 |
|
1121 |
|
|
% total observations per age/sex |
42 |
47 |
48 |
10 |
|
31 |
|
N.B. There were no differences in task percentages between wet and dry months.
¹Children 6 years and above.
2Percentage errors due to rounding.
³N = number of observations of livestock management activities.
Table 2 shows the time devoted to livestock management by age/sex of worker, for all workers and finally by livestock unit. The data demonstrate that whereas each worker in a rich household spends more time on livestock management than a worker in a poorer household, richer households devote far much less time per livestock unit held. This is because rich households herd their animals in much larger groups, thus benefitting from economies of scale; they also are able to marshal! labour from poorer households with whom they frequently herd.
Table 2. Labour devoted to livestock management at Olkarkar.
|
Stratum |
Mean number of hours of labour per day |
||||
|
Children¹ |
Adults |
All workers per livestock unit |
|||
|
M |
F |
M |
F |
||
|
1 |
5.0² |
4.62 |
6.3 |
0.8 |
1.29 |
|
2 |
5.72 |
7.1 |
6.0 |
1.5 |
0.60 |
|
3 |
6.42 |
7.6 |
7.9 |
1.5 |
0.25 |
¹Children 6 years and above.
²Children spent a mean of approximately 2 hours a day at school.
These examples indicate just a few of the analyses possible with the data gathered by the time allocation technique. Although the time allocation technique is being increasingly used, to date there are not, to my knowledge, any studies available comparing results gained from this technique to those gained from recall or critical task analysis. The Kenya team is in the process of designing a brief study which will involve concurrent collection of recall and time allocation data from a sample of households in the Maasailand. It is hoped that the results produced will aid in assessing the validity and reliability of the two methods under pastoral conditions.
Collinson, M.P. 1972. Farm management in peasant agriculture: A handbook for rural development planning in Africa. Praeger, London. P. 226, 229.
Dahl, G. 1979. Suffering grass: Subsistence and society of Waso Borana. Dept. of Anthropology, Univ. of Stockholm, Stockholm.
Grandin, B.E. 1582. The allocation and labour inputs on a Maasai group ranch: Preliminary findings from Olkarkar. Manuscript ILCA, Nairobi.
Swift, J. 1979. The economies of traditional nomadic pastoralism. The Tuareg of the Adras N Ifras (Mali). Unpublished Ph. D. dissetation, University of Sussex.
Swift, J. 1982. Les systèmes de production pastoraux au Mali. Internal document ? ILCA, Addis Ababa. p. 308.
Torry, W. I. 1977. Labour requirements among the Gabra. Paper presented at the Conference on East African Pastoralism, Nairobi, 1977.
Résumé
La main-d'oeuvre joue un rôle très important dans tout système de production agricole. Dans un système pastoral à caractère collectiviste où les pâturages et l'eau sont théoriquement d'accès libre, le capital (sous forme d'animaux) et la maind'oeuvre constituent les moyens essentiels de production à acquérir et à gérer pour une exploitation productive.
Pour bien comprendre la main-d'oeuvre conçue comme une composante des systèmes, il faudrait se pencher sur ses aspects sociaux et techniques. Etant donné l'importance de la main-d'oeuvre dans la production pastorale, il est nécessaire de bien connaître la demande et l'offre de main-d'oeuvre des ménages. Dans la phase de la description et du diagnostic de la recherche sur les systèmes pastoraux, des enquêtes informelles peuvent être utilisées pour la collecte de données sur les tâches de production les plus importantes, sur le moment de leur exécution et leur caractère saisonnier, sur la division culturellement acceptée du travail par âge et par sexe, sur les divers modes de recrutement des travailleurs, sur les variations offre/demande de main-d'oeuvre entre les ménages et sur les mécanismes sociaux d'ajustement de celles-ci et les contraintes dues à l'insuffisance de la maind'oeuvre, y compris ses effets sur la productivité.
Il y a une indigence particulière de données sur la main-d'oeuvre des systèmes pastoraux. Par le biais de l'analyse, d'observations et d'interrogations, le chercheur dresse une liste des tâches de production les plus importantes et du temps approximatif nécessaire pour les mener à bien. La quantification peut s'effectuer par l'observation directe tout comme pour l'approche sur l'étude de la main-d'oeuvre utilisée dans les études sur la gestion de l'exploitation agricole.
L'analyse des tâches les plus importantes de la production pastorale fournit des données utiles et intéressantes à un prix abordable.
Cette approche souffre cependant de limites inhérentes à la méthode. Elle se contente de descriptions générales des tâches les plus importantes et ne fournit pas d'informations sur les travaux effectués par divers travailleurs et sur les variations de la répartition du travail qui peuvent être lices à la saison, à la compétence ou à la richesse. Cossins utilise une variante de la technique de l'analyse des tâches les plus importantes dans ses études sur les Borana d'Ethiopie.
La technique de rappel fondée sur les visites multiples, telle que celle utilisée par Swift au Mali et White au Niger se fonde sur l'utilisation d'observateurs résidents pour collecter des données sur le travail effectué par tous les travailleurs dans des campements échantillons deux fois par semaine. Dans la conception d'une étude de rappel à visites multiples pour l'obtention de données sur la main-d'oeuvre, les deux aspects les plus importants ont certainement trait à la fréquence des visites et au choix des personnes à interroger Le rappel extensif ou les visites limitées pourraient fournir suffisamment de données pour déterminer les goulets d'étranglement en matière de main-d'oeuvre. Toutefois, ces méthodes comportent plusieurs inconvénients étant donné qu'elles n'exploitent que les facteurs les plus importants dans les activités de production et éventuellement dans les activités domestiques.
La méthode de la répartition du temps se fonde sur des visites programmées de manière aléatoire dans un ménage, au cours desquelles, les activités de tous ses membres sont enregistrées. Ces enregistrements produisent une série de tableaux verbaux suffisants en nombre pour fournir une description complète des activités par des paramètres tels que l'âge, le sexe et la saison. Le temps consacré à une tache particulière est extrapolé sur la base du pourcentage de l'ensemble des activités consacrées à cette tâche.
Les ménages-échantillons inclus dans l'étude doivent être recensés de sorte que le nom, l'âge et le sexe de tous les travailleurs du ménage puissent être déterminés. Les enquêteurs doivent rester à proximité des ménages de sorte que les heures importantes du Jour puissent être couvertes. Une ordinateur est nécessaire pour l'analyse, étant donné que le nombre d'entrées est assez important.
La technique de la répartition du temps comporte un inconvénient en ce sens qu'elle ne permet pas de déterminer avec précision le temps consacré à certaines activités effectuées loin de la maison. De telles activités peuvent être couvertes par des observations ou par des interrogations spécifiques comme dans l'analyse des tâches les plus importantes. Les avantages de la technique de la répartition du temps sont multiples. Elle couvre tous les travailleurs potentiels et n'est pas Limitée par les problèmes que pose le rappel; les enquêtés ne sont pas agacés et seul un nombre limité de données de pré-enquête sur la maind'oeuvre s'avère nécessaire; elle couvre toute la gamme des taches; elle permet d'enregistrer des activités multiples et simultanées et fournit une information vitale sur les variations au niveau des unités de production.
Certains résultats préliminaires de l'étude sur la répartition du temps effectuée par le Groupe du CIPEA au Kenya ont été présentés. Ils mettent l'accent sur quelques unes seulement des analyses possibles avec les données collectées sur la base de cette technique. L'équipe du Kenya est sur le point d'élaborer une courte étude qui portera à la fois sur la collecte de données basées sur le rappel et sur la répartition du temps dans un échantillonnage de ménages vivant en pays Maasai. On espère que les résultats contribueront à évaluer la validité et la fiabilité des deux méthodes dans les conditions pastorales.
Chairman: Prof. Gunnar Sorbo (Norway)
Discussion led by Dr Befekadu Degefe (Ethiopia)
In referring to Dr Grandin's paper on Livestock Transactions, Dr Wilson pointed out that off-take data are taken by animal scientists as a matter of course. Indications of off-take and deaths can be got from initial surveys, as was done in Niger.
Mr Sandford pointed out that labour intensity was different from labour energy intensity' and both were different from the question of whether labour was a constraint on pastoral production. Even if people were totally idle for 11½ months of the year, if labour was in short supply for just 2 month, e.g. for watering livestock from wells, then labour shortage may be the critical constraint on the system. Dr Grandin agreed and said that people might not work hard but in feet were not idle as they may have to work long hours. If a labour constraint was only for a short period of time, understanding the dynamics of this would come from a critical task/season specification and an analysis of the detailed labour input information. Mr Sandford said that he suspected that the technique of the 'time allocation study'-might overestimate the labour requirements of a system by failing to distinguish between those tasks which were truly essential and those which were just undertaken to fill in time. In the 'recall' technique, respondents were more likely to indicate priorities by not mentioning the time spent on tasks which were inessential.
Dr Zulberti warned that it was very easy for scientists to become data addicts and to try to collect everything from everyone. Anyone could look at his or her own work and recognise many occasions on which data was collected and never used. Such data collection could postpone real action in the field and could be an expensive endeavour providing little or no effect on the wellbeing of the pastoralist. He said that the final objective was not to know the pastoralist but to know how one could be useful to him. Mr Sandford agreed that it was important to be cost effective in research, but he pointed out that there was a great danger in restricting one's collection-of data exclusively to those items which before the research started were thought to be the most important. He said that sometimes breakthroughs in knowledge occurred through the analysis of data which was collected in a much more open-minded way. In project preparation, for example, one may need to prioritise much more strictly. But in research one needed to be much more open minded. This was the essence of the systems approach - to find out what was important, not to predetermine it.
Dr Zakary referred to Dr Solomon's observations that the expenditure of pastoralists was mainly devoted to buying food and very little was spent on systems inputs, and that perhaps parameters of development should increase the demand so that pastoralists could sell their cattle. Dr Zakary felt that although the latter might improve the pastoral sector's contribution to the national economy, at the same time pastoralists might then become slaves to demand this could be very dangerous during a drought. Dr Zakary felt that the present system of using pastoral areas was good - developers should above all allow pastoralists to organise themselves. Dr Solomon agreed.
Dr Tilahun stated that the usual economic contention was that low income households spent a large proportion of their income on food and that this ratio declined as household income increased. But he didn't think one could see such a relationship in pastoral societies. Dr Solomon said that he had evidence from the Kenya study that the wealth status of pastoral households greatly influenced the per caput expenditure on food. Dr Solomon agreed with Dr Tilahun when he said that information on decision making responsibilities of household members on the use and disposal of livestock was important for household income and expenditure studies. The analysis of such information gave an insight into the extent to which management and marketing practices varied due to household size and composition.
Dr Thompson said that the controversy regarding the need to understand the complete system as opposed to part of the system in the descriptive/diagnostic phase was resolved if one remembered that the systems approach was iterative. This allowed one to return to the initial phase to study either another well-defined aspect of the system or to gather further information needed for designing and testing solutions to a constraint previously identified. But the time-span of an iteration must be short; thus a narrowly focused, short duration diagnostic/descriptive survey was preferable to a complete systems survey.
Président: Prof. Gunnar Sorbo (Norvège)
Débats dirigé par le Dr Befekadu Degefe (Ethiopie)
Faisant allusion au document de Mlle Grandin sur les transactions de bétail, M. Wilson a mis l'accent sur le fait que les relevés des données sur l'écoulement sont effectuées par les zootechniciens de manière systématique. Des données relatives à l'écoulement et aux décès peuvent être abtenues à partir d'enquêtes initiales comme ce fut le cas au Niger.
M. Sandford a souligné que l'intensité de main-d'oeuvre était différente de l'intensité d'énergie de la main-d'oeuvre et que ces deux éléments n'intervenaient pas dans la question de savoir si oui ou non le travail constituait une contrainte à la production pastorale. Même si les travailleurs étaient totalement oisifs pendant 11 mois et demi et qu'au cours de la dernière quinzaine de l'année il y ait carence de main d'oeuvre, par exemple, pour l'abreuvement du bétail au puits, la carence en main-d'oeuvre constituerait la contrainte la plus grave sur le système. Mlle Grandin a accepté cette idée et a déclaré que la main-d'oeuvre ne travaillait peut-être pas dur mais en fait qu'elle n'était pas oisive étant donné qu'elle pouvait avoir des journées de travail très longues. Si une contrainte en matière de travail était pour une courte durée seulement, on pourrait en saisir la dynamique par une spécification saison/tâche vitale et par l'analyse des données détaillées du facteur travail. M. Sandford a déclaré qu'il avait l'impression que la technique de la répartition du temps pourrait surestimer les besoins en travail d'un système, notamment parce qu'elle ne permet pas de distinguer les tâches réellement essentielles de celles qui sont effectuées pour meubler le temps. Dans la technique du rappel, il était probable que les enquêtés mettent l'accent sur leurs priorités et qu'ils passent sous silence le temps consacré à des tâches qui n'étaient pas essentielles.
Le Dr Zulberti a fait savoir qu'il était très facile pour qu'un chercheur devienne un maniaque des données et qu'il essaie de collecter tout ce qui est possible et imaginable. Nous pouvons tous trouver dans nos travaux plusieurs exemples où nous avons collecté des données que nous n'avons jamais utilisées. Une telle collecte de données pourrait retarder l'action sur le terrain et pourrait constituer une entreprise coûteuse ayant un effet limité ou nul sur la qualité de la vie de l'éleveur. Le Dr Zulberti a déclaré que l'objectif final n'était pas de connaître l'éleveur mais de savoir comment on pouvait lui être utile, M. Sandford a reconnu que le rapport coût-efficacité était important en matière de recherche mais il a souligné qu'il était très dangeureux de limiter la collecte de données aux informations que l'on pensait être les plus importantes avant le démarrage des activités de recherche. Il a déclaré que quelquefois les progrès scientifiques intervenaient à la suite de l'analyse de données qui avaient été collectées de manière beaucoup plus générale. Dans l'élaboration des projets, il est possible de fixer les priorités de manière beaucoup plus stricte. Mais dans la recherche, il fallait être beaucoup plus ouvert. C'était là l'essence de l'approche par système: trouver ce qui était important et non pas le prédéterminer.
Le Dr Zakary a fait allusion aux observations de M. Solomon selon lesquelles les dépenses des éleveurs étaient essentiellement consacrées à l'achat de produits alimentaires, que très peu de ressources étaient affectées à l'achat de facteurs de production et que, peut-être, les paramètres de développement devraient accroître la demande afin que les éleveurs puissent vendre leurs bovins. Le Dr Zakary estimait que quoiqu'une telle démarche puisse augmenter la part du secteur pastoral dans l'économie nationale, elle pouvait en retour transformer les éleveurs en esclaves de la demande et cela pouvait être très dangereux en période de sécheresse. Le Dr Zakary a déclaré que le système actuel d'utilisation des zones pastorales était bon, que les responsables du développement devraient par dessus tout permettre aux éleveurs de s'organiser. M. Solomon a exprimé son accord sur cette idée.
Le Dr Tilahun a déclaré que l'hypothèse économique habituelle était que les ménages à faible revenu consacraient une proportion importante de leurs revenus à l'achat de produits alimentaires et que ce rapport diminuait avec l'accroissement du revenu du ménage. Mais il ne pensait pas qu'on puisse observer une telle relation dans les sociétés pastorales M. Solomon a déclaré qu'il avait des preuves émanant de l'étude, selon lesquelles la richesse des ménages pastoraux influençait considérablement les dépenses alimentaires par habitant. M. Solomon a accepté le point de vue du Dr Tilahun lorsque celui-ci a déclaré que les informations sur les membres du ménage investis du pouvoir de décision sur l'utilisation et la disposition du bétail étaient importantes pour les études sur les revenus et les dépenses des ménages, L'analyse de telles informations donnait une idée de la manière dont la gestion et les pratiques de commercialisation variaient en fonction de la taille et de la composition du ménage.
Le Dr Thomson a déclaré que la controverse concernant la nécessité de connaître l'ensemble et non pas une partie seulement du système comme c'est le cas dans la phase de description/diagnostic était dépassée dés lors que l'on se rappelait que l'approche par système était itérative. Cela permettait de retourner à la phase initiale soit pour étudier un autre aspect bien défini du système, soit pour collecter d'autres informations nécessaires pour la conception et l'essai de l'élimination d'une contrainte entièrement identifiée. Mais la durée d'une itération doit être courte; ainsi, une enquête de diagnostic/description bien délimitée et de courte durée était préférable à une enquête complète sur les systèmes.
