4.2.1 Household economy in average rainfall years
4.2.2 Labour
4.2.3 Livestock marketing
4.2.4 Dairy processing and marketing
4.2.5 Dairy marketing and calf management
4.2.6 Grain cultivation
A profile of Borana households during average rainfall years in the 1980s was primarily developed using results of three independent surveys that involved a total of 247 households (i.e. Negussie Tilahun, 1984; Holden, 1988; and Mulugeta Assefa, 1990) and one preliminary synthesis (Cossins and Upton, 1987). In sum, the objectives of these studies were to: (1) define production units and describe households in terms of people and livestock holdings; (2) characterise contributions of livestock production, cultivation and trade to food procurement; (3) derive indicators of household wealth and depict wealth stratification; and (4) enumerate important sources of cash income and expenditure.
The first survey entailed a baseline survey of 49 Borana households within nine encampments (olla). Throughout nine madda were surveyed from June 1981 to July 1983. This was a period of average rainfall and a high population density of cattle (see Section 2.4.1.4: Climate, primary production and carrying capacity and Section 7.2: A theory of local system dynamics). Madda included Marmaro, Orbati, Hobok, Dilo Goraye, Dokole, Melbana, Web and Did Hara (Negussie Tilahun, 1984: p 4; see Figure 2.10). Enumerators lived in encampments and interviewed men and women in each household at monthly intervals to record dynamics of family composition, cash income and expenditure as well as sales and exchange of animals and other possessions. Production units were largely defined on the basis of households managed by married couples in association with live-in relatives. Livestock units were scaled as: Calf = 0.25; other cattle = 0.75; sheep and goats = 0.15; and camels = 1.4. African Male Equivalents (AAME; see Cossins and Upton, 1987) were scaled as: Males aged 16 years and over = 1; females 16 and over = 0.8; and children less than 16 = 0.6. One-way ANOVAs were employed to assess effects of region on response variables.
In a second survey, Holden (1988) selected 108 households on a stratified-random basis for interview from encampments located within 30 km of the towns of Dubluk and Mega during 1987. This was a year of average rainfall and moderate population density of cattle (Section 7.2: A theory of local system dynamics). Her study primarily dealt with dairy marketing and the methods are reported in full later in this section. Selected data pertaining to household structure and cash income from Holden (1988) are compared and contrasted to those in Negussie Tilahun (1984).
In a third survey, Mulugeta Assefa (1990) profiled 633 households in terms of per capita cattle holdings and conducted detailed interviews with 90 households concerning cattle production and management during 1988 in the Did Hara and Dubluk madda. This was a year of average rainfall and a higher population density of cattle compared to 1987. Selected data pertaining to wealth stratification in Mulugeta Assefa (1990) will be presented here. His main body of results are presented in full elsewhere (see Section 5.3.3: Cattle production and pastoral wealth and Section 7.3.3.5: Calf mortality mitigation).
Conceptual models for resource use and food production and demand for an average Borana household (i.e. one that managed eight breeding cows plus followers) were analysed by Donaldson (1986) and Cossins and Upton (1987, 1988b). They calculated average numbers of stock/household based on aerial survey data (Milligan, 1983; Assefa Eshete et al, 1987; for methods see Section 4.2.6: Grain cultivation). Estimates of numbers of people and households/encampment were based on hut counts from the air (Milligan, 1983; Assefa Eshete et al, 1987) supplemented with ground surveys (Negussie Tilahun, 1984; Donaldson, 1986; Coppock and Mulugeta Mamo, 1985). Human diets were collated from survey data in 1982 (Negussie Tilahun, 1984; Donaldson, 1986) in which 20 families reported daily intakes of common foods that were standardised on a gross energy (GE) basis.
Profiles of labour supply and demand at the encampment level of resolution were compiled from data presented in Cossins and Upton (1987) and Negussie Tilahun (1984). This mainly focused on labour required to raise water from the deep wells and that needed for herding. Descriptions of how labour is organised at the deep wells is described elsewhere (see Section 2.4.1.7: Water resources).
A preliminary study of seasonal time budgets for married women was carried out by Mulugeta Assefa (1990: pp 62-64) to clarify implications of labour constraints for interventions related to improved calf management (see Section 7.3.3.5: Calf mortality mitigation). A total of 30 Borana women were interviewed from Dubluk and Did Hara madda (15 in each) and asked to estimate the frequency and average duration/event in which some 20 independent activities were carried out in each of four seasons (i.e. long rains, cool dry, short rains and warm dry; see Section 2.4.1.4: Climate, primary production and carrying capacity). Composited labour profiles were calculated based on mean values. Each activity was scored as more or less important against others using Friedman's nonparametric ranking test (Steel and Torrie, 1980). This provided a means to value seasonal priorities independent of time allocation. Details are in Mulugeta Assefa (1990: pp 62-64).
Monitoring of livestock markets was conducted from 1981 to 1984 in the following locations: Moyale (for 27 months); Hidi Lola (34 months); Mega (26 months); Yabelo (27 months); Teltele (23 months); Negele (24 months); Agere Mariam (28 months); Kera (9 months); and Finchewha (5 months). The last two sites varied with regards to year of observation, but otherwise data roughly covered the same period elsewhere. More details can be found in Dyce (1987: p 33) and Negussie Tilahun (nd).
The objectives of this work were to characterise (1) who supplied animals to markets and why; (2) who purchased animals and why; (3) volume and composition of throughput in terms of species, age and sex; and (4) sources of variation in supply of livestock over time. On market days the following information was collected (Dyce, 1987: pp 33-34): (1) number of animals brought to market, reasons for sale, type of supplier and time taken to reach market; (2) livestock species, age, sex, colour, body condition and live weight or head girth; (3) number of animals purchased, prices paid, reasons for purchase, and type and origin of buyers; and (4) opinions of buyers and suppliers as to market prices, price expectations and reasons for unsold animals. Data analysis involved collation of nearly 70000 records (Negussie Tilahun, nd). Simple linear regressions were used to correlate relationships between: (1) monthly rainfall and number of cattle marketed for two locations; and (2) prices for mature cattle/unit of live weight at five markets over 21 months from August 1981 to April 1983. Prices were determined by assuming the average live weight for mature males and females to be 318 and 225 kg, respectively (Alberro, 1986), and summing recorded prices on a monthly basis (Dyce, 1987: p 73). Selected results from these studies are presented here.
Coppock (1992b) conducted extensive interviews with 30 leaders of Borana society to examine hypotheses concerning the traditional rationale for animal production. Topics investigated included cultural constraints on livestock sales, motivations for retaining or selling livestock, perceived changes in Borana society in terms of livestock marketing behaviour and to what degree perverse supply factors could be expected to operate (Doran et al, 1979; Sandford, 1983a). Perverse supply response is the concept that throughput of marketed animals could decline over time in response to higher prices. This considers that pastoralists only need a certain amount of cash income per year and manage herd assets in a manner which minimises sales. This is because herd assets have other traditional social and economic functions besides income generation. Sandford (1983a) contends that the evidence to support or refute the perverse supply hypotheses for pastoral systems is equivocal. Some 100 logic questions were formulated. Respondents were selected nonrandomly and ranged from 35 to 60 years of age. Interviews were open-ended to adequately solicit unstructured responses and were carried out for several hours over two consecutive days per person. Respondents were widely distributed throughout a 10000 km2 region on the central plateau.
Studies of household milk allocation and processing were conducted during 1986-88 and are summarised in Coppock et al (1992) and Coppock et al (in press). Seasonal patterns of milk production and use were quantified for two households at each of four encampments in the Melbana, Medecho and Did Hara madda. One enumerator lived in each encampment and collected data for the two households for seven consecutive days during each of four main seasons (long rains, cool dry, short rains and warm dry) during 1987-88. Daily data collection consisted of interviews and measurements to establish major pathways of milk allocation. This primarily included use of fresh milk for consumption, sale or storage and fermentation. Milk was typically stored to produce: (1) milk fermented for a short term (stored for <5 days) for family consumption or butter-making; or (2) milk fermented for a longer term (typically stored for up to 30 days, but reportedly as long as 60 days), as ititu (a special food commonly reserved for guests). Data were also collected on use of stored products including production and use of butter and buttermilk. Seasonal allocation patterns were standardised on a gross energy (GE) basis (i.e. where 1 kg of fresh whole milk = 3.3 MJ GE; 1 kg of fermented milk = 3.9 MJ GE and 1 kg of butter = 29.8 MJ GE (ENI, 1980; Nicholson, 1983a)).
The role of dairy marketing in the household economy and its possible consequences for animal production was the focus of several studies (Holder, 1988; Holden et al, 1991; Coppock et al, in press; Holden and Coppock, 1992). Wealth stratification of households evident in Negussie Tilahun (1984) served as a basis for work designed to analyse effects of per capita livestock wealth, distance to market and season on quantity of dairy products sold/person/day and the role of dairy income in the household economy. A sample of households was drawn within a 30-km radius of the market towns of Dubluk (population 500) and Mega (population 3000), located within 40 km of each other on the main tarmac road that runs south from Addis Ababa to Moyale. Demand for dairy products in these towns is a combination of local demand plus a demand for butter for export to the southern highlands via traders on the public transport system (Holder, 1988). The 2800-km2 study area surrounding each market town was divided up into three concentric sub-areas (i.e. 0 to 10,11 to 20 and 21 to 30 km from market). Six encampments were randomly selected within each sub-area using compass coordinates. One family from each of the wealthy, intermediate and poor wealth classes was selected by the senior male leader (aba olla) in each encampment and interviewed for a total of 108, or 54 from each market area.
Although similar scalars were used to quantify livestock units and people as used in Negussie Tilahun (1984), they were amended in the dairy study to be lactating livestock units (LLU) while people were scored as African adult male equivalents (AAME) as in ILCA (1981). This is reviewed in Holden (1988: p 20).
Data for different seasons that influenced milk production were collected from all families using a one-time questionnaire that recorded reported dairy sales for the current dry period of August/September 1987 and for the rainy and transition (or post-rainy) seasons in the previous 12 months. The senior woman (i.e. the key person for dairy marketing) in each household was asked to recall: (1) maximum and minimum sale volumes/market trip/season (calibrated using local containers); (2) frequency of market trips; (3) number and type of livestock milked/day in each season; (4) daily milk offtake (also calibrated using local containers; and (5) income from dairy sales and uses of the money.
Income from livestock sales and other domains controlled by men was reported by husbands. Dependent variables were calculated on a seasonal or time-weighted annual basis and included daily quantity of dairy products sold per AAME and per LLU, and dairy income as a per cent of reported total income. Dairy products were expressed in litres of fresh milk equivalents derived from GE content (above). Independent variables were distance class to market (km), wealth class, season and market site. A four-way ANOVA calculating least-squares means with repeated measures was used to analyse the data (SAS, 1987). The ANOVA of dairy income as a per cent of seasonal cash income had to employ two, rather than three, distance categories because of empty cells due to some reports of zero income. Simple linear regressions were used to analyse marketing behaviour by correlating: (1) frequency of marketing or (2) quantity of dairy products sold/market trip with: (1) annual dairy sales/AAME, (2) wealth or (3) distance to market. Details are available in Holden (1988: pp 17-21) and Holden and Coppock (1992).
Data from 15 encampments in the Dubluk marketing area in Holden (1988) were combined with information on calf morbidity and mortality (Mulugeta Assefa, 1990) to analyse risks of milk marketing to calf management in households of varying wealth (Holder et al, 1991). The hypothesis was that poorer families living closer to a market would be affected by the opportunity to sell dairy products and that this would intensify competition between people and calves for milk, with negative implications for the vigour and health of calves. In addition to the data collection and scaling reported in Section 4.2.4, one woman/household was asked to report amounts of milk and grain consumed by themselves and one child under the age of four during the previous 24 hours and these were converted to metabolisable energy equivalents (see appendix 3 in Holden, 1988). Quantities were estimated by calibrating local containers. As part of another questionnaire on livestock production that was similarly set up to measure variability due to distance from market and wealth, the perceived daily milk yield, daily milk offtake and calf performance (births, deaths and morbidity) were recorded for offsprings of up to six randomly selected lactating cows in each of the 45 households (or five per wealth and distance class). The 233 cows were evenly divided into "good", "average" and "poor" milking classes (as defined by the respondents) with milk offtakes ranging from 2.3 litres/day over nine months for the highest producing cows to 1.5 litres/day over seven months for the lowest-producing cows (Mulugeta Assefa, 1990: p 19).
Calf morbidity and mortality were reported only for animals prior to weaning and rates were calculated based on two samples of calves born in three years previous to the survey (i.e. 1985-87). These years had average rainfall (see Section 2.4.1.4: Climate, primary production and carrying capacity). A calf was considered to have experienced morbidity if it had recovered at least once from a life-threatening ailment related to nutrition management. Respondents apparently had no difficulty reporting such information and data were cross-checked with other family members.
Effects of wealth class and distance to market on human food intake were analysed using a two-way ANOVA. Milk offtake/lactating cow was analysed using a split-plot three-factor design with families as the main plot with the two factors of wealth and distance to market. Cows were sub-plots within the factor cow class. Binomial data for calf mortality and morbidity were analysed using a "maximum likelihood" logistic analysis for the same three factors using PROC CATMOD procedures in SAS (1987). Details are available in Holden et al (1991).
Cereal cultivation has important implications for change in the Borana System. Incidence of cultivation and pastoral opinions concerning farming among the Boran and Gabra were assessed from a series of household surveys (Coppock and Mulugeta Mamo, 1985; EWWCA, 1987; D. L. Coppock, ILCA, unpublished data; R. J. Hodgson, CARE-Ethiopia, unpublished data; Coppock, 1988; Webb et al, 1992). The extent of cultivation before and after the 1983-84 drought was determined by Assefa Eshete et al (1987) using aerial survey methods.
The 15475-km2 study area was divided into a 10x10-km UTM (Universal Transverse Mercator) grid which was further sub-divided into 619 (5x5 km) sampling cells. Forty parallel flight lines, 5 km apart and running north to south, were sampled along the midline of each cell. The sampled area was limited to a fixed width along each side of the flight line using a sighting device composed of parallel rods. At a flying height of 180 m the projected sighting strip at ground level was 300 m wide, which gave a sampling intensity of 12% (Assefa Eshete et al, 1987: p 4). Counts of huts, farming plots and livestock in the sampling strips were supplemented by photography if numbers were high (i.e. above 10) and validated in the office.
