Economic Diversification and Risk Management

GL-CRSP Pastoral Risk Management Project (PRMP)
Technical Report 02/99
July, 1999

by
Peter D. Little, University of Kentucky
Barbara Cellarius, University of Kentucky
Christopher Barrett, Cornell University
D. Layne Coppock, Utah State University

Proper citation: Little, P.D., B. Cellarius, C. Barrett, and D.L. Coppock. 1999. Economic Diversification and Risk Management among East African Herders: A Preliminary Assessment and Literature Review. GL-CRSP Pastoral Risk Management Project Technical Report 02/99. Utah State University, Logan. 40 pp.

This publication was made possible through support provided by the Office of Agriculture and Food Security, Global Bureau, United States Agency for International Development, under Grant No. PCE-G-98-00036-00. The opinions expressed herein are those of the authors and do not necessarily reflect the views of the U.S. Agency for International Development.

Project on 'Improving Pastoral Risk Management on East African Rangelands', Global Livestock Collaborative Research Support Program (GL-CRSP)

This preliminary assessment addresses one aspect of the risk management focus of the GL-CRSP project on 'Improving Pastoral Risk Management on East African Rangelands': the role that economic diversification⁽¹⁾ assumes in risk management. It analyzes existing studies and secondary data from the region as they pertain to income diversification, and indicates a set of variables to guide field research on diversification. This document, including the associated bibliographies (annexes 1 and 2) are meant to serve as a 'working' resource for the project that can be 'updated' as additional literature and data become available.

Eleven general variables are used here to assess pastoral diversification. These include: (1) average annual rainfall; (2) ethnicity; (3) ratio of pastoral to non-pastoral income sources; (4) type of non-pastoral activity; (5) income source by herder wealth category; (6) gender; (7) 'other' indicators of income flows (e.g., food aid transfers); (8) human population density; (9) per capita livestock holdings; (10) distance to urban centers; and (11) date of data collection.

The review found that there are several gaps in our understandings of pastoral diversification. First, there is little understanding of the gendered nature of diversification and its role in risk management, particularly how income from remittances and livestock sales are distributed among household members. We know very little about how a range of non-pastoral activities affect women, except to caution that the higher-revenue pursuits (livestock trade and retail businesses) seem to be dominated by males. Second, analyses of non-pastoral income diversification needs to be coupled with better understandings of the returns to investments in education (human capital formation) among herders. This is particularly the case on the Kenyan side, where herder investment in education is substantial in certain parts of the study region. Field studies need to look specifically at how pastoral investment in education affect herder responses to drought and other disasters. Finally, the literature raises a series of spatial questions related to diversification and risk management, but provides little analytical treatment. Proximity to urban centers (differentiated by scale and economic activity) and infrastructure (e.g., roads) does have an affect on non-pastoral strategies for managing risk. How important it is needs to be addressed in the field studies, and incorporated as a criterion when selecting communities for subsequent field research.

Pastoralists of East Africa increasingly pursue non-pastoral strategies to meet consumption needs and to buttress against shocks caused by climatic fluctuations, animal disease, market failures, and insecurity. Recent literature and databases for the region verify this tendency and show a marked increase in this phenomenon during the 1980s and 1990s, even in areas considered to be remote (Holtzman 1996; Little 1992; Straight 1997; Kituyi 1992) (see also the bibliography in Annex 1). This recent trend contrasts considerably with the focus of earlier studies (1960 to 1985) and challenges many of the theoretical and methodological paradigms that were based on assumptions of pastoral livestock specialization. What are the social, economic, and ecological variables that explain economic diversification among East African pastoralists; and has diversification assisted herders to better withstand the social and economic shocks that have especially characterized the region during the past 20 years?

This preliminary assessment addresses one aspect of the risk management focus of the GL-CRSP project on 'Improving Pastoral Risk Management on East African Rangelands': the role that economic diversification⁽²⁾ assumes in risk management. The paper is not meant to definitively verify or invalidate this relationship; existing literature and data would not allow this. Rather, the objective is to review and analyze existing studies and secondary data from the region as they pertain to income diversification, and to begin to assess the critical variables that may help guide later field-based research. Important gaps in existing literature and data also are highlighted. This document, including the associated bibliographies (annexes 1 and 2) are meant to serve as a 'working' resource for the project that can be 'updated' as additional literature and data become available. Further refinement of the data and variables discussed below could permit more formal statistical analyses than is attempted here.

The exercise started with the literature (especially unpublished reports) and data on diversification that had been accumulated by Dr. Little during the past 20 years, supplemented with standard library collections at universities in the US and elsewhere. Little and Barbara Cellarius, an advanced graduate student in development anthropology at the University of Kentucky (UK), did extensive literature searches at UK, and at other university collections accessible through inter-library loan and the Internet. They also acquired leads on local data sources from colleagues working in the region. Additional literature and secondary data were collected by Little during visits to Kenya and Ethiopia and to the national development studies library at the Institute for Development Studies, University of Sussex during June and July, 1998. The assessment included as much 'gray' literature and publications from the region as possible. However, it is likely that some project reports and other data related to pastoral diversification from the region were missed With a few notable exceptions, the searches were limited to the project study area--a north-south transect from Hagre Mariam, Ethiopia to Isiolo and west to Samburu and Baringo, Kenya , including the Isiolo and Baringo areas. The study region encompasses a range of social, economic, and ecological variation which makes it especially appropriate for examining different strategies of diversification.

The exceptions are important to point out. Since considerable theoretical and empirical work on rural diversification has been conducted outside the study region, some of this literature was reviewed, especially the work of Reardon on rural labor markets (1996a; 1997a; 1997b; see other citations in Annex 1). Numerous, excellent comparative papers have been written by Reardon, but virtually all relate to cultivating or agropastoral populations outside of East Africa. These works provide helpful insights into the ways in which income diversification relates to asset accumulation and risk management behavior.

Other exceptions entail the inclusion of the Maasai of southern Kenya, the Mukogodo Maasai (see Herrens 1991), and the Borana-related Orma of Kenya (see Ensminger 1992), all groups which fall outside the study region. This was justified on the basis of the excellent data sets that exist for these groups and because these groups are very similar to those in the project's study region; in the case of the Mukogodo Maasai, they reside immediately south of Isiolo along the region's border.

The first step was to devise a list of variables and indicators to guide the literature review. Critiques of diversification often fail to appreciate the differences between study areas and the ways in which the simplest variables--for example, annual rainfall amounts--can explain particular patterns. Consequently, the literature is unnecessarily polarized. For example, confusion exists about the importance of herder cultivation as a risk-mitigating strategy without sufficient attention to climatic and ecological constraints in different sites. Thus, on the one hand, cultivation is perceived as an important mechanism to sustain consumption and assets during disasters; while, in other areas, cropping is seen as a strain on labor for herding, a cause of increased vulnerability to drought, and a reason for localized pockets (around settlements) of ecological degradation. Both sides assume they are correct, but rarely appreciate the fact that conditions vary so much across different locations that blanket generalizations are difficult. Within our study region, agroclimatic variability is considerable (from 300 mm to 900 mm annually), allowing the project to address hypotheses about the role of farming in decreasing/increasing risk.

Eleven general variables are used here to assess pastoral diversification. These include:

These variables require explanation. It should be noted that concepts and units often were defined differently across study sites. For example, in some cases livestock were converted into Tropical Livestock Units (TLU), a weight/production measurement, while other times they were equated to a Standard Stock Unit (SSU) based on market value. Other inconsistencies were found that make comparisons between data sets difficult.

The first variable, annual rainfall, provides some indication of aridity and the feasibility of rainfed agriculture. Although not always a good indicator of agricultural and drought risk, this figure was available for most parts of the study region. A more accurate assessment of cultivation risk would require analyses of variance and standard deviations--rather than statistical means--and data on daily and monthly rainfall. For the purposes of this exercise a measurement of average rainfall is adequate, although its limitations should be acknowledged. Lowland areas of northern Kenya--parts of Marsabit, Moyale, and Samburu Districts--where average annual rainfall is below 400 mm and agriculture is least viable in the study region.

Ethnicity is a second variable that is highlighted. It generally did not show up as an important explanatory variable in and of itself, but in association with other variables it can be informative. Historically many ethnic groups within the study region received very little formal education and thus their participation in the labor market is limited to unskilled jobs. With the exception of those groups near the southern border of the study region, like the Chamus and Tugen of Baringo, Kenya, the other groups in the study received very little education and are relatively isolated from important markets and centers of employment. Where diversification into agriculture is climatically feasible, communities with a strong pastoral legacy often pursued some investment in crop production. These include highland Samburu, highland Rendille, and Borana, where important pockets of cultivation are found in medium-to-high potential areas. Different income diversification strategies are found regardless of ethnicity.

The ratio (percentage) of pastoral to non-pastoral sources of income is another important variable, but often is not uniformly defined or applied. In some studies, non-livestock sources of income include income from herders who engaged in livestock trading. In others any activity related to livestock (including income from livestock or milk trading) is included as livestock-based income, even if it takes place in towns or is done as a form of self-employed work. Other studies categorize non-pastoral income, as any revenue earned away from the unit of production or herd, which means that a women selling milk in town is engaged in a non-pastoral activity. Some of the most detailed data sets define all forms of self-employed trading (including income from sales of animals and animal products) as non-pastoral income (see Ensminger 1992). We would suggest that the project think carefully about how to define non-pastoral activities. For anthropologists, petty trading--even if it involves sale of livestock products--usually is treated as non-pastoral. Economists, on the other hand, tend to take a more sectoral approach and treat any income earned from livestock as 'on-farm' pastoral income (see Reardon's work, Annex 1).

A related variable that was used in the assessment is the source or type of non-pastoral income (waged employment, cultivation, etc.). As will be discussed later, this indicator varied across studies vis-a-vis proximity to urban centers, climate, and other variables. There also are significant differences between Kenyan and Ethiopian communities within the region, because of the more complex labor markets, commercial activities, and training (educational) options in Kenya than in Ethiopia.

Numerous studies show how diversification strategies often vary by differences in household wealth. Thus, a fifth variable, herder wealth category, was used to capture these differences, although in many cases data were lacking on it. However, certain data from the region illustrate why poor livestock owners often diversify into particular income-earning strategies out of necessity to survive, while wealthy pastoralists pursue non-pastoral activities as an accumulation or investment strategy and often take on more lucrative options (Little 1985). The poor are 'pushed' or forced into diversification through poverty, while the more fortunate choose to diversify to maximize economic returns and to manage risk. Unskilled waged employment and petty trading often are options for the poor, while wealthy herders can pursue more capital-intensive activities, such as irrigated agriculture or business ownership.

While few studies differentiate income-earning strategies by wealth level, even fewer disaggregate income activities by gender, another variable that is included in the assessment. Gender-specific diversity strategies are included in some databases and an effort was made to address this variable. For a field study to do so, data collection at the intra-household level, atime-consuming process, is required. Some good data on the frequency with which certain non-pastoral activities are dominated by one gender group over another is available. Where data are available, they depict non-pastoral strategies (petty trading and vegetable production) where women are the key income earners.

The variable, 'other indicators of non-livestock income,' mainly addresses food-aid transfers. Food aid is found throughout the study region but rarely is it addressed in detail in the literature. In most cases only the absence/presence of food-aid transfers were indicated and only anecdotal data on amounts per recipient or market value are provided. There is some evidence from one part of the study region (Baringo) that food-aid transfers have allowed agropastoralists to direct consumption savings into farm and non-farm investments (Bezuneh et al. 1988, 1989). Elsewhere in the study region (e.g., Samburu District, Kenya) anecdotal data point to the negative effects of food aid that has resulted in dependency and has allowed herders to maintain subsistence (or below subsistence) levels of consumption without productive investments in non-pastoral activities and assets (GFA 1993; Sperling 1987b ).

Both human population density and per capita livestock holdings are variables that can explain patterns diversification. Population density is a crude measurement which varies considerably across the study region, while the importance of per capita livestock holdings differs for different populations. In terms of the former, it can fluctuate over a distance of less than 40 kms; population densities can decline fivefold as one moves from highland to lowland zones. In the study region, population densities varied from a high of 41 to a low of 1 per square kilometer; at the higher end of densities some form of non-pastoral activities must take place. While per capita livestock holdings generally declined in the more densely populated areas, this was not always the case (see Table 1, discussed later in the paper).

Measurement of per capita livestock holdings is not always consistent in the studies. As indicated earlier, TLUs were used in some cases and SSUs in others. In other instances livestock holdings were given on a per household or family basis rather than a per capita measurement. Where it was possible to convert into a standard unit, we opted to use an estimated market unit (Standard Stock Unit: 1cattle=five small stock; 2 cattle=1 camel) instead of the more complex TLU. As will be discussed later in the paper, the per capita indicator proved to be a generally important measurement, and the tendency to diversify generally increases as per capita livestock holdings decline.

Distance to urban centers is a critical variable in explaining patterns of economic diversification, but unfortunately studies provided little data on geographic locations vis-à-vis centers of different scale. An important activity to pursue in the next three months is to reconstruct the distances to urban centers of the different field studies. Opportunities for trade, informal sector activities, and waged employment clearly are affected by distance to urban centers. Not only the distance but also the size of the urban center is likely to be an important factor; small rural market centers open up petty trading opportunities, but do not offer the types of formal and informal waged employment opportunities that a regional or primate city do. A rough four-fold scale could be utilized: centers of (1) less than 10,000; (2) 10,000 to 100,000; (3) 100,000 to 1,000,000; and (4) 1,000,000+ population. Access to smaller market centers are important--especially for marketing opportunities--but centers of less than 10,000 do not afford many opportunities for diversification. Throughout the study region low levels of urbanization exist. Most important centers lie outside of the study region: Nanyuki, Nakuru, and Nairobi on the Kenya side; and Nazareth, Awaso, and Addis Ababa on the Ethiopia side. Consequently, with few exceptions most waged employment among pastoralists in the study region requires migration to an outside city.

It also should be noted that Kenyan centers contain more commercial activities than a similarly-sized town in Ethiopia. This could make trans-border comparisons difficult. For example, a town of 10,000 in Kenya will have far greater commercial activity and opportunity for diversification than its counterpart. 'Geo-referenced' settlement data should be available when the current rapid appraisal field studies are completed. When these data are available and market distances carefully calculated, it will be possible to say more about the effects of market center/town distance on diversification strategies. Distance-to-town will be a very important variable for determining site selection for the project's community and household-level studies to begin in 1999. At present we have not included this variable in our summary tables, because we do not have sufficient data or adequate analyses. In general, however, sites at the southern end of the study region, where access to major urban centers is relatively easy, show higher levels of non-pastoral activities than other sites in the area. In Table 1 these include the Chamus and Tugen areas of Baringo, Kenya. There are exceptions. For example, even with relatively good access to Nairobi, some communties of Maasai maintain (as of 1981-1983) relatively high levels of pastoral specialization, with in some cases more than 85 percent of cash income deriving from livestock-based activities (see Bekure et al. 1991). Relative to most of the groups in our study region the Maasai maintain relatively large herd sizes and thus have less need to diversify into non-pastoral activities

The date (year) in which the data were collected is the final variable to be considered. Data collected in the 1970s are less likely to show high levels of diversity, than more recent studies. The bulk of good data on pastoral diversification were collected in the 1980s. What time-series data on diversification exist in the study region are very limited, covering no more than 7 to 10 years in most cases. Yet, the pace of change has been such that major shifts in diversification can show up in less than 10 years.

Very few existing studies contain data on all of the above variables (see Table 1). In terms of income diversification, much of the data is presented in terms of percentages rather than actual income figures. For example, the percentage of households with members engaged in non-pastoral activities might be presented, rather than numbers of individuals or actual income amounts of activities. The most comprehensive data on income diversification cover the Ariaal, Chamus, Gabbra, and Rendille in the study region; and the Maasai and Orma (a Borana-related group) outside of the study area. Important study groups, like the Borana and Samburu, have information but it is limited and not easily comparable. Indeed, most of the Samburu data are presented in terms of percentages (rather than actual amounts) and disaggregated by generation or age set (see Holtzman 1997; Sperling 1984; Straight 1997), which constrain comparative analyses. These data may indicate whether an individual participates in waged labor or another non-pastoral activity, but not its relative value to household welfare, earnings, or risk reduction.

With the exception of Coppock's (1994) work, little is known about diversification in southern Ethiopia,. A recent publication by Diriba (1995) has some data on diversification among the Borana, but detailed income information are lacking. We know from qualitative data that the Borana of southern Ethiopia are focused more on pastoralism than their Kenyan counterparts, but--like in Kenya-- agriculture and sedentarization are increasing in importance. Petty trading is limited to settlements near the few important towns of the Borana Plateau, while waged labor is relatively infrequent--a major contrast with the Kenyan data. Until recently labor markets were heavily regulated by the state, a marked contrast with Kenya. Diriba notes that on the Ethiopian side there was virtually no rural labor market in the 1980s: "This means that, due to restrictive state policies affecting the rural labour market, rural households are not permitted to employ or to be employed even when it is desirable (1995:117)." Diriba goes on to say that off-farm employment in southern Ethiopia even limited, even with less restrictive policies. In principle labor market policies have changed since 1994, but the impact of this transition is not known.

As noted earlier, most data on diversification were collected in the 1980s, which makes them approximately 10 or more years old. For instance, the IPAL data on the Gabbra and Rendille are from 1981-1984; the Chamus data are from 1981 and 1984; the Tugen data from 1987 (although there is more recent data that has not yet been written up); the Kenya Borana data from 1979; and the Ethiopian Borana from 1980-1989. An exception to this is the recent study of diversification among the Samburu by Holtzman (1992-1994); this shows that about 35 percent of Samburu households currently are involved in waged labor. The O'Leary/IPAL (Gabbra, Rendille), Fratkin (Ariaal), Sperling (Samburu), Little (Chamus),and Coppock data could be updated through revisits and this should be encouraged. Outside the study region, the Ensminger (Orma), Herren (Mukogodo), and Bekure et al. data sets provide sound baselines to measure changes in diversification.

The demographic data from the region generally point to some important implications for diversification, but their gross aggregation reduce their usefulness (Table 1). With the rapid growth of settlements in the study region and the very uneven nature of population density, district-level (Kenya) or woreda-level (Ethiopia) population data are problematic. Nonetheless, there are important relationships that can be implied. Population densities (per square Km) are highest for the Tugen (23,41,and 14), the Chamus (14), and Mukogodo (31), while each of these groups also depend heavily on non-pastoral sources for household cash income (from 58 to 65.2 percent of total cash income). Those groups with the lowest population densities, in turn, show a different dependence on non-pastoral income sources. Both the Rendille and Gabbra have very low population densities (about 1-2 per square km), as well as little reliance on non-pastoral sources of income (about 19-20 percent of total income). Population density is generally a good measure of the land available for livestock production-- although this will vary greatly according to ecological and climatic factors--and the extent to which populations can live off pastoral-based income and products.

Before any statistical analysis of the relationship between per capita livestock holdings and diversification is carried out, additional work is needed to standardize the measurement of livestock units. This was been pointed out earlier in the paper. With standardization we are likely to find important statistical relationships. Even with existing data we know that pastoral populations with high levels of per capita livestock holdings are generally less likely to diverse than poorer ones, although this will vary by the wealth category of owners. Rich herders also diversify, but in activities generally different than poor and middle herders (see Table ). Wealthy livestock owners are less likely to migrate out of the region to pursue low waged jobs, nor are they as likely to engage in petty trade as poor and middle herders.

With few exceptions existing databases reflect single stints of fieldwork, with only Ensminger (Orma) and Fratkin Ariaal) carrying out detailed followup studies on some of the variables discussed . The majority of studies capture patterns of diversification over one or more years but from a single period of field research (for example, 1981-1984).

As noted earlier, the goal of this review exercise was not to carry out analyses to establish statistically significant relationships between sets of variables and patterns of diversification. Nonetheless, we have shown that there are important trends about diversification in the region that can be observed from existing studies and the eleven variables indicated (see Table 1). Another goal of this exercise was to highlight trends that are important to consider in the design of the project's subsequent field studies. First, there is little question that across the study region (and generally throughout pastoral areas of East Africa) per capita livestock holdings have declined considerably since 1980. In these areas there is now a substantial population of stockless or near-stockless herders, representing as much as 10 to 12 percent of the pastoral population in areas like Baringo and Samburu Districts. We would venture to say that very few groups in the region would currently have per capita holdings of more than 6 Standard Stock Units/person, and most groups would on average have less than 2-3 Standard Stock Units/person. For example, data for the Ariaal of Marsabit, Kenya show a decline in per capita holdings of 8.2 TLU in 1976 to 4.0 in 1995 (Fratkin et al. 1996); while among the Orma per capita holdings declined from 5.0 in 1980 to 3.5 TLU in 1987 (Ensminger 1992). The Chamus also have seen a reduction in per capita livestock stock holdings of about 40 percent during 1978-1990 (Little 1992).

Second, with declining per capita stock holdings, there is little question that many herders (male and female) have had to diversify their income-earning activities. What is surprising is how rapid and how much of this has occurred since 1980. For areas where agriculture is feasible, there continues to be a continued expansion of agriculture into formerly rangeland areas. These include areas on the Borana plateau, southern Ethiopia; around irrigated perimeters in Baringo; the Marsabit mountain in Kenya; and the Leroghi Plateau, Samburu District. Much of this expansion has been carried out by herders themselves, but a large percentage reflects agricultural encroachment by non-pastoralists. For areas where agriculture is not feasible diversification mainly has entailed waged labor and trading/business activities. The waged labor usually entails migration out of the study region (e.g., to Nairobi), but this varies by location and proximity to centers of employment (e.g., towns or employment-intensive activities like irrigation schemes).

Third, there are data that show not only has non-pastoral diversification increased since 1980 but that different categories of herders--rich/poor and male/female-- respond differently. Little shows that among the poorest category of herders in Baringo more than 50 percent of household income came from non-pastoral activities, while among the richest category it was less than 15 percent. For the Orma Ensminger (1992) shows similar variations. In 1987 the richest Orma herders received about 37 percent of income from non-livestock activities, while those in the poorest category received about 58 percent. In the Orma case, Ensminger documents the widespread income diversification that occurred in only about a 7 year period (1980 to 1987). For the poorest herders in all groups unskilled waged labor seems to be the most common non-pastoral option; while for the wealthiest it tends to be trading, business, and skilled (higher income) waged labor (see Table 1).

Sedentarization/settlement often provides some increased income-earning opportunities for women, especially for members of low-income households. This is especially true in petty trade (milk and vegetable trading), handicrafts, informal brewing of beer, and local waged employment, where women assume prominent roles (see Coppock 1994; Little 1992; and Fratkin and Smith 1995). Wealthier women herders are likely to rely more on income from livestock and milk/ghee sales, than on other revenue sources. Shop ownership, retail business, and labor migration are activities that remain predominantly male.

Fourth, increased income from trading activities also seems to be an important pattern of diversification since the 1980s. The rapid growth in market towns in the study region--in some cases (Marsabit, Isiolo, and Maralal towns) at rates of more than 10 percent annually since the 1980s--reflects the increased importance of trading activities in the area.

Fifth, there is little doubt that major non-pastoral income transfers in the form of food aid (either grants or food/cash-for-work schemes) increasingly characterize the region, as well as impact strategies of diversification. We were well of aware this before writing this paper. What we are surprised at is the extent to which food aid is a critical component of local economies, and how many families depend on it on a regular basis. In parts of the study area, more than 40 percent of the population receive some food aid during the year. For example, in lowland Samburu more than 45 percent of women were regularly receiving food aid during the late 1980s (GFA 1993; Kielmann et al. 1994). As Table 1 demonstrates, more than 80 percent of the studies reviewed indicated the presence of food aid. The data on food aid is much more readily available for Kenya than it is for Ethiopia.

Finally, for the more favorable pastoral areas where rainfed and/or irrigated agriculture is possible, there is some data to show that cultivation allows herders to better manage risk. They seem to respond better to drought-induced shocks, than do other pastoralists; and in these higher rainfall areas pastoralism requires less mobility and thus generally requires less labor than in drier rangelands. Campbell's work (1978) shows that Maasai agropastoralists were less affected by the 1977 drought than were specialized herders, and less dependent on famine relief than other pastoralists. Little (1992), in turn, describes how herders who had partially diversified into irrigated agriculture more quickly rebuilt their herds after the 1979-1980 and 1984 droughts than did others; while Hogg (1980) shows how the Borana of Isiolo who have diversified into agriculture and trading better withstand 'bust' years, than do others. The extent to which cultivation allows Borana of southern Ethiopia to better manage risk is currently being addressed under a joint ILRI/CRSP project (the work of Brent Swallow and Nancy McCarthy) and data should soon be available.

What about the links between diversification and improved risk management in drier pastoral zones? In these areas the issues are more complex and some diversification strategies may directly compete with labor for herding and reduce herder mobility, an occurrence that can have negative ecological impacts. The studies from Marsabit, Kenya (Fratkin, O'Leary and others) show the negative ecological affects of pastoral sedentarization and diversification there. We do not have sufficient data to demonstrate the relationship between diversification and risk management in these areas. Most studies have not paid sufficient attention to the differences in non-pastoral income activities and what they mean for the herder, the ecology, and production system generally. Nor have they acknowledged that certain diversification strategies do not always lead to sedentarization. Indeed, a herder family with a member (s) engaged in a lucrative non-farm job can help the family maintain a pastoral livelihood through remittances, as well as provide capital to rebuild herds after a disaster.

This paper raises a number of issues that will need to be addressed in the design of field studies in 1999, as well as points to additional data needs. Perhaps the biggest gap in our understanding of the role that diversification plays in risk management is the gendered nature of these responses, particularly to how income from remittances and livestock sales are distributed among household members. We know very little about how a range of non-pastoral activities affect women, except to caution that the higher-revenue pursuits (livestock trade and retail businesses) seem to be dominated by males. At the intrahousehold level we know that women have important rights to livestock, but we know little about how rights to sell livestock are negotiated and determined between household members; how income from sales are distributed within the household; and how different (and at time conflictive) rights to sell animals may affect investments in non-pastoral activities. Primary data collection on these questions will be required and will pose important methodological questions.

Analyses of non-pastoral income diversification needs to be coupled with better understandings of the returns to investments in education (human capital formation) among herders. This is particularly the case on the Kenyan side, where herder investment in education is substantial in certain parts of the study region. Although waged employment is a major source of income diversification, little data exist on how pastoral investment in education relates to success in the labor market. Field studies need to look specifically at how pastoral investment in education affect herder responses to drought and other disasters.

This paper raises a series of spatial questions related to diversification and risk management, but has left their treatment to subsequent analyses and fieldwork. Proximity to urban centers (differentiated by scale and economic activity) and infrastructure (e.g., roads) does have an affect on non-pastoral strategies for managing risk. How important it is needs to be addressed in the field studies, and incorporated as a criterion when selecting communities for data collection. Spatial access is likely to have a major effect on the options available for income diversification.

This paper has shown that there are certain areas of the study region where relatively good baseline data are available (Baringo and Marsabit). By updating them we could learn much about pastoral diversification and risk management. While areas where little field data are available (for example, parts of southern Ethiopia) should be covered in field studies, we are likely to learn more by emphasizing sites where time-series/baseline are already available. If the project uses field sites (e.g., Baringo and highland Samburu) where diversification is widespread and others where it is not (parts of Marsabit), it will be possible to do comparative work around an important project hypothesis: that diversification allows herders to better manage risk.

Finally, the border (Ethiopia/Kenya) that traverses our study region has a major impact on economic diversification. Levels of commercial activity, market development, and urbanization are considerably higher on the Kenya than on the Ethiopia side. This difference will also allow important comparative work to be done, but while caution against overgeneralization between the two countries. Ironically, road networks--especially of the primary routes--are better in southern Ethiopia than in northern Kenya. The extent to which urban centers of southern and central Ethiopia influence marketing and employment patterns in the study region are not well known.

In sum, this paper is a first step in summarizing and conceptualizing some of the key issues of pastoral income diversification as it relates to risk management. While statistical analyses were not carried out as a part of this exercise, additional effort could be made in quantifying and standardizing existing data sets and in conducting regression and other formal analyses. This would point to important causal relationships to be explored further in the project's field studies.

TABLE 1. SUMMARY OF DATA ON PASTORAL DIVERSIFICATION

Ethnic Group	Pastoral versus non-livestock income sources	Sources of non-livestock income	Data by wealth category	Data by Gender	Other indicators of non-livestock income	Avg annual rainfall	Pop. Dens (/km²)**	Livestock per capita	Date	Source
Ariaal					Food aid programs in 1970s and 1980s. % of HHs with wage workers: 1976 pastoral 10%, 1976 town 36%; 1985 pastoral 19%, 1985 town 43%. 1976 0.13 HH members employed, 1985 0.25 HH members employed	500mm	1.2	1976: 8.2 TLU/ person 1995: 4.0 TLU/ person	1974-75, 1985, 1990	Fratkin, various years
Borana/ Orma	90% of income cattle related, remainder from other animal sources				Recent increase in farming, studies cited ranged from 33% to 88%.	700mm	7.3	2.3 TLU/ person	1980-1989	Coppock 1994
				17% men involved in wage labor, 17% had 1 or more sons in wage labor.	Food aid programs present. Some received cash from farming.	305mm	1		1979	Hogg 1981
	All stock related (SR): 1980: 74.8% 1987: 51.3% (Data exclude subsistence production and gifts)	1980: wage labor (WL) 18.6%, trade (T) 6.5% (includes 3.4% stock trade) 1987: wage labor 33.2%, trade 15.4% (includes 3.4% stock trade)	Poor 1980: SR 61.9, WL 33.2, T 4.8; 1987: SR 42.1, WL 52.3, T 5.5. Middle 1980: SR 77.0, WL 16.7, T 6.3; 1987: SR 49.0, WL 32.7, T 18.2 Rich: 1980: SR 84.9, WL 6.8, T 8.4: 1987: SR 62.9, WL 14.8, T 22.4		Significant employment opportunities beginning in 1980s. Wide-scale river farming beginning in 1970s. Government famine relief.		**2	1980: 5.0 TLU/ person 1987: 3.5 TLU/person	1978-81, 1987	Ensminger 1992
Chamus	63.5% of total income is livestock related (LS) 65.2% of cash income	Agriculture (Ag) 14.3%/0.8%, Perkerra (Pk) 1.5%/2.6%, fish (Fi) 3.4%/2.8%, nonfarm (NF) 17.2%/28.6% (1^st figure total income, 2^nd cash, data in next column is cash income)	Cat I (most): LS 85.5%, Ag 4.0%, NF 10.5% Cat II: LS 30.9%, Ag 2.8%, NF 66.3% Cat III-H: LS 64.2%, Ag 10.8%, Fi 1.9%, NF 23.2% Cat III-L: LS 44.2%, Ag 3.3%, Pk 10.9%, Fi 7.1%, NF 34.4% Cat IV: LS 59.5%, Ag -1.9%, Fi 1.8%, NF 40.7% Cat V (least): LS 47.3%, Ag 10.5%, Fi 11.7%, NF 30.5%		Food aid program present. 41% of household have 1 or more members employed for wages. Less than 10% of working age population is employed.	640mm	**14	19 cattle, 29 goats, 73 sheep per homestead (=6.72 people); 5.85 LSU/capita	1980-81	Little 1992
Gabbra	80.2% of income from sale of livestock and livestock products	Wage remittances 7.9%, gifts 4.9%, other (esp. trade) 5.0%, handicrafts 2.0%			Food aid program present. Settlement schemes in mountains 1976 Increasing reliance on non-pastoral sources of income, but livestock remains a mainstay May 1971: 16% of district population was receiving famine relief			12 camels, 15.1 cattle, 76.1 sheep, 68.7 goats average per HH (= 6.3 persons); 10 LSU/capita	1981-1984	O'Leary 1985, 1990
Maasai	"Pastoralists": 93.4% livestock "M. Farmers": 35.4% livestock (cash income)	"Pastoralists": 1.3% relatives, 2.4% business, 1.4% labor, 1.5% other "M. Farmers": crop sales 9.9%, egg sales 1.3%, duka 6.5%, labor 8.0%, land rent 4.1%, business 34.9%			Famine relief received by 41% M. farmers, 67% M. pastoralists	794mm	**6		1977	Campbell 1978
	58% of household cash income from livestock (LS)	Remittances from migrants (RM) 31%, other activities (O) (beer and liquor brewing, charcoal making, beekeeping) 11%	V Poor: LS 63%, RM 22%, O 15% Poor: LS 31%, RM 60%, O 9% Medium: LS 59%, RM 31%, O 10% Rich: LS 96%, RM 1%, O 3%		Increased labor migration in 1982-83 (down country)		**31	15 LSU per household; est. 2 LSU/capita	1987-88	Herren 1991
	Olkarkar: 79.8% of cash income from livestock products (LSP) Merueshi: 87.2% Mbirkani: 67.7%	Olkarkar: wages (WL) 12.4%, money transactions (MT) 7.4%, beer brewing (BB) 0.3%, other (O) 0.0% Merueshi: WL 0.8%, MT 10.0%, BB 1.9%, O 0.1% Mbirkani: WL 11.5%, MT 19.4%, BB 0.0%, O 1.4%	Poor: LSP 71.9%, WL 8.9%, MT 13.9%, BB 2.3%, O 3.0% Medium: LSP 80.6%, WL 11.8%, MT 7.0%, BB 0.5%, O 0.0% Rich: LSP 88.6%, WL 8.4%, MT 2.6%, BB 0.2%, O 0.2%			463- 584mm	6	About 3 cattle and 8 small stock per person; est. 4.4 LSU/capita	1981-83	Bekure et. al 1991
				11-12% of men currently employed, less than 1% of women currently employed. Few employed work outside district			6-8	Lemek: 127 cattle & 131 shoats per household	1989-90	Holland 1992
Rendille	77.8% of income from sale of livestock and livestock products		Poor: livestock 77.8%, wage remittances 13.4%, gifts 8.9% Better-Off: livestock 80.5%, wage remittances 11.9%, gifts 7.6%		Food aid program present. Settlement schemes in mountains 1976 Increasing reliance on non-pastoral sources of income, but livestock remains a mainstay Wage labor during 1938 drought May 1971: about 16% of district population receiving famine relief	167-800mm	**1	157.1 sm. stock, 14.4 camels, 7.9 cattle, 0.4 donkeys per HH (=5.7 persons); 11.9 LSU/ capita	1981-1984	O'Leary 1985, 1990
					1976 0.0 HH members employed, 1989 0.37 HH members employed. Food aid programs. Increasing settlement in towns with opportunities for wage labor and selling milk or vegetables.	500mm	1.2	1.9 TLU/ person (1995)		Fratkin, various years
Samburu				Expansion of wage labor opportunities for men post WWII.	Farming began in late 1970s, some crops sold: 13% at Ilkilorili farm (bush/ grasslands); 91% at Baawa farm (more moist/fertile); 100% at Lorian farm (forested/good for farming) HH may include people in city who send money home.	500-700mm	1.4 to 2.1 house-holds per km²	3.25 cattle per person	1981-82	Perlov 1987
					Food aid program present. Increasing involvement in wage labor, esp since droughts of 1980 and 1984. E.g., members of age sets from 1948-1976: 54% had wage labor experience; members of age sets from 1893 to 1936: 14% had wage labor experience	250-500mm	**4	1.7 cattle per person	1983-84	Sperling 1987
				Samburu women: 51% in one location and 45% at another had participated in food/cash for work activities						GFA 1993
			Data presented on participation in wage labor by wealth category (based on livestock): Richest 25.0% in wage labor, well off 28.6%, average 35.3%, stock poor 35.5%, stockless 46.9%	Widespread wage labor: 36.6% of males currently; 64.8% of males at some time. Women work locally (no data).	Spread of cultivation during study period. High involvement in non-traditional activities: 42.6% of households in brewing; 15.8% in agriculture, and 42.8% in wage labor	Lowlnd 400-500mm Highlnd 500-900mm			Jan 1992-Aug 1994	Holtzman 1996
				Over 40% of younger men involved in wage employment at some time. Women and married men work locally, only moran travel significantly for jobs.	Increasing involvement in wage labor and settling in towns or near missions. From time allocation study -- town location: earning cash accounts for 15% of dry season and 18% of wet season activities; typical lowland: cash takes up 8% dry season, 13% wet season activities; remote site: cash à 4% dry, 4% wet season activities. Table 6.8 stratifies this data by age and sex (see Straight 1997)	Range= 400mm lowlandto 1250 mm highlnd	(adult popula-tion given for each site, but not density)	Figure 6.1 presented with %age of HHs w/ different amounts of stock	July 1992 to Aug 1994	Straight 1997
Tugen	60.0% of Gross Output Value from livestock	Crop production 7.4% of GOV; Salary employment 5.3% of GOV; Non-farm activities 27.3% of GOV	Poor: LS 55%, Ag 24%, NF 21% Medium: LS 59%, Ag 18%, NF 24% Richer: LS 29%, Ag 7%, NF 64% (units unclear, labeled "source of income (net margin)")		Increasing involvement in crop production and off-farm activities in recent years. 93% of household have livestock and 54% of households sell livestock for cash. 99% of HHs cultivate grains and 41% of households sell grains for cash. 67% of HHs have off-farm activities	940mm	23, 41 (data shown for two locals)	12.5 LU per household (= 8.8 members è 1.4 LU/person)	1987-88	Vendeld 1990, Vendeld and Lusenaka 1991
			Rich: 66% more income from business than from livestock. All have non-livestock income, more from business than wages. Average: 26% LS and O income of = importance; 37% LS more important; 37% O more important. Poor: 46% LS more important; 31% LS and O of = importance, [leaves 23% with O more important]		57% of households rely on salaries to some degree. 18% no business involvement and rely totally on livestock. Farming is a subsistence activity. This study only included people with livestock.	300-700mm	**14	44 goats, 18 sheep, 7 cattle per herd owner (i.e., excludes folks w/o livestock)	1987	Meyerhoff 1991

** If population density figures were not provided in the cited text, figures listed are for the district as a whole as reproduced in Fratkin 1993 (data for 1979).

Annex 1: Pastoral Diversification Study: List of all sources reviewed (** indicates article data have been summarized)

Baxter, PTW, Jan Hultin, and Alessandro Triulzi (eds.). 1996. Being and Becoming Oromo: Historical and Anthropological Enquiries. Uppsala: Nordiska Afrikaninstitutet.

Beaman, Anne Windsor. 1981. The Rendille Age-Set System in Ethnographic Context: Adaptation and Integration in a Nomadic Society. Ph.D. Dissertation, Boston University: Boston. (One chapter discusses pastoral economy, no mention of non-pastoral activities.)

Bekure, Solomon PN de Leeuw, BE Grandin and PJH Neate (eds.). 1991. Maasai Herding: An Analysis of the Livestock Production System of Maasai Pastoralists in Eastern Kajiado District, Kenya. ILCA Systems Study 4. Addis Ababa: ILCA (International Livestock Centre for Africa). **

Bezuneh, Mesfin, Brady J. Deaton, and George W. Norton. 1988. Food Aid Impacts in Rural Kenya. American Journal of Agricultural Economics 70(1): 181-191.

Bezuneh, Mesfin, Brady J. Deaton, and George W. Norton. 1989. Farm Level Impacts of Food-for-Work in a Semi-Arid Region of Kenya. Eastern Africa Economic Review 5(1): 1-8.

Bezuneh, Mesfin, and Brady Deaton. 1997. Food Aid Impacts on Safety Nets: Theory and Evidence - A Conceptual Perspective on Safety Nets. American Journal of Agricultural Economics 79: 672-677.

Campbell, David J. 1978. Coping with Drought in Kenya Maasailand: Pastoralists and Farmers of the Loitokitok Area, Kajiado District. Institute of Development Studies Working Paper 337. Nairobi: University of Nairobi. **

Campbell, David J. 1984. Responses to Drought Among Farmers and Herders in Southern Kajiado District, Kenya. Human Ecology 12(1): 35-63. **

Charnley, Susan. 1997. Environmentally-Displaced Peoples and the Cascade Effect: Lessons from Tanzania. Human Ecology 25(4): 593-618.

Coppock, D Layne. 1994. The Borana Plateau of Southern Ethiopia: Synthesis of Pastoral Research, Development, and Change, 1980-1991. Addis Ababa: International Livestock Centre for Africa. **

Diriba, Getachew. 1995. Economy at the Crossroads: Famine and Food Security in Rural Ethiopia. Addis Ababa: Care International.

Ensminger, Jean. 1992. Making a Market: The Institutional Transformation of an African Society. Cambridge: Cambridge University Press. **

Ensminger, Jean, and Jack Knight. 1997. Changing Social Norms: Common Property, Bridewealth, and Clan Exogamy. Current Anthropology 38(1): 1-24.

Evans, Hugh Emrys, and Peter Ngau. 1991. Rural-Urban Relations, Household Income Diversification and Agricultural Productivity. Development and Change 22: 519-545.

Fratkin, Elliot, 1986. Stability and Resilience in East African Pastoralism: The Rendille and the Ariaal of Northern Kenya. Human Ecology 14(3): 269-286.

Fratkin, Elliot. 1987. Age Sets, Households, and the Organization of Pastoral Production: The Ariaal, Samburu, and Rendille of Northern Kenya. Research in Economic Anthropology 8: 295-314.

Fratkin, Elliot. 1989. Household Variation and Gender Inequality in Ariaal Pastoral Production: Results of a Stratified Time-Allocation Survey. American Anthropologist 91(2): 430-440.

Fratkin, Elliot. 1991. Surviving Drought and Development: Ariaal Pastoralists of Northern Kenya. Boulder: Westview Press. **

Fratkin, Elliot. 1992. Drought and Development in Marsabit District, Kenya. Disasters 16(2): 119-130. **

Fratkin, Elliot. 1992. Maa-Speakers of the Northern Desert: Recent Developments in Ariaal and Rendille Identity. In T Spear and R Waller, Being Maasai: Aspects of Identity. London: James Curry Publishers.

Fratkin, Elliot. 1993. Problems of Pastoral Land Tenure in Kenya: Demographic, Economic and Political Processes Among Maasai, Samburu, Boran, and Rendille, 1950-1990. Paper presented at the African Studies Association Annual Meetings. December 2-6, 1993. (Copied table with population density by district. Looks like it was later published in Nomadic Peoples 34/35 (1994): 55-68, with title "Pastoral Land Tenure in Kenya: Maasai, Samburu, Boran, and Rendille Experiences, 1950-1990.")

Fratkin, Elliot, and Eric Abella Roth. 1990. Drought and Economic Differentiation Among Ariaal Pastoralists of Kenya. Human Ecology 18(4): 385-402.

Fratkin, Elliot, Kathleen A. Galvin, and Eric Abella Roth (eds.). 1994. African Pastoralist Systems: An Integrated Approach. Boulder and London: Lynne Rienner.

Fratkin, Elliot, Eric Abella Roth, and Martha A Nathan. 1996. Sedentarization, Commoditization and Development Among Rendille Pastoralists of Kenya. Paper presented at the American Anthropological Association Annual Meetings. San Francisco, CA. Nov. 20, 1996. **

Fratkin, Elliot, and Kevin Smith. 1995. Women's Changing Economic Roles and Pastoral Sedentarization: Varying Strategies in Alternative Rendille Communities. Human Ecology. 23(4): 433-454. **

Galaty, John G., and Pierre Bonte (eds.). 1991. Herders, Warriors, and Traders: Pastoralism in Africa. Boulder: Westview.

Gebre Mariam, Ayele. 1991. Livestock and Economic Differentiation in North East Ethiopia: The Afar Case. Nomadic Peoples 29:10-20. **

Gesellshaft für Agraprojekte, MBH. 1993. Baseline Survey for Samburu District Development Programme. Hamburg, Germany: GFA.

Grandin, B.E. 1981. Group Ranches in Kaputiei: The Impact of the Kenya Livestock Development Project, Phase I. Draft report for IBRD. Nairobi: ILCA.

Grandin, Barbara E. 1988. Wealth and Pastoral Dairy Production: A Case Study from Maasailand. Human Ecology 16(1): 1-21.

Grandin, Barbara E., Peter N. De Leeuw, and P. Lembuya. 1989. Drought, Resource Distribution and Mobility in Two Maasai Group Ranches, Southeastern Kajiado District. In Coping with Drought in Kenya: National and Local Strategies, edited by Thomas E. Downing, Kangethe W. Gitu, and Crispin M. Kamau. Pp. 245-263. Boulder and London: Lynne Rienner Publishers.

Grandin, Barbara E. and Solomon Bekure. 1982. Livestock Offtake and Acquisition: A Preliminary Analysis of Livestock Transactions in Olkarkar and Mbirikani Group Ranches. Background paper for the sub-programme committee review of EARLS Research in Kenya.

Hart, Keith, and Louise Sperling. 1987. Cattle as Capital. Ethnos 3-4:324-338.

Herren, Urs J. 1991. "Droughts Have Different Tails": Responses to Crises in Mukogodo Division, North Central Kenya, 1950s-1980s. Disasters 15(2): 93-107. **

Hogg, Richard. 1980. Pastoralism and Impoverishment: The Case of the Isiolo Boran of Northern Kenya. Disasters 4(3): 299-310.

Hogg, RS. 1981. The Social and Economic Organization of the Boran of Isiolo District of Northern Kenya. Ph.D. Dissertation. Manchester University. **

Hogg, Richard. 1986. The New Pastoralism: Poverty and Dependency in Northern Kenya. Africa 56 (3): 319-333.

Hogg, Richard. 1987. Development in Kenya: Drought, Desertification and Food Scarcity. African Affairs 86: 47-58.

Hogg, Richard. 1988. Water Harvesting and Agricultural Production in Semi-arid Kenya. Development and Change 19: 69-87. **

Holland, Killian. 1992. The Diversification of a Pastoral Society: Education and Employment among the Maasai of Narok District, Kenya. Ph.D. Dissertation, Department of Anthropology, McGill University, Montreal.

Holsteen, Melbourne Edward. 1982. Continuity and Change in Samburu Education. Ph.D. Dissertation, University of Florida. (Brief mention of increasing involvement of men in wage labor and trade, no quantitative data on this.)

Holtzman, Jon David. 1996. Transformations in Samburu Domestic Economy: The Reconstitution of Age and Gender-based Processes of Production and Resource Allocation among a Kenya "Pastoral" People. Ph.D. Dissertation, Department of Anthropology, University of Michigan, Ann Arbor. **

Kielmann, A.A., N.S. Kielmann, A.A.J. Jansen, D.N. Njama, G.K. Maritim, R. Mwadime, and K. Saidi. 1994. Nutritional profile of the propulation in a food-for-work project area: A case study from Samburu District, Kenya. Food and Nutrition Bulletin 15(3): 215-226.

Kituyi, Mukhisa. 1990. Becoming Kenyans: Socio-economic Transformation of the Pastoral Maasai. Nairobi: ACTS Press (African Centre for Technology Studies). **

Lesorogol, Carolyn K. 1990. A Little Tea, A Little Sugar: The Role of Women's Networks in Pastoral Production and Reproduction. M.s.

Little, Peter D., 1992. The Elusive Granary: Herder, Farmer, and State in Northern Kenya. Cambridge: Cambridge University Press. **

Livingstone, Ian. 1986. Rural Development, Employment and Incomes in Kenya. Aldershot: Gower. (Also earlier version (1981) of paper with same name published by International Labour Office, Jobs and Skills Programme for Africa.) (Very aggregated figures. Has information on income sources at household level for different regions in Kenya on p. 81-82.)

Meyerhoff, E. 1988. Socio-Economic Changes in the Kampi Ya Samaki Area, Baringo: Effects on Baringo Fuel and Fodder Project Development Activities. M.s.

Meyerhoff, Elizabeth, 1991. Taking Stock: Changing Livelihoods in an Agropastoral Community. Nairobi: ACTS Press (African Centre for Technology Studies). **

Molenaar, Maarten. 1986. "I Cannot Beat": A Study Into Irrigation Management in Eldume Irrigation Scheme. Provincial Irrigation Unit, Nakuru, Kenya. (Copied tables only.)

Mwangi, Lucy Ngendo. 1994. Optimization of Pastoral Subsistence Strategies Under Drought and Marketing Risks: Maasai Model Ranch Study. Ph.D. Dissertation, Colorado State University.

Njiru, George K. 1982. The Rendille Economy: Some Preliminary Findings on Trading Activities. Institute for Development Studies Working Paper No. 389. Nairobi: University of Nairobi.

Oba, Gufu. 1994. Kenya, Boran: Sharing and Surviving: Responses of Impoverished Pastoralists to Food Insecurity. The Rural Extension Bulletin 4: 17-22. University of Reading Agricultural Extension and Rural Development Department. (Examines role of sharing and begging in food security. Data are stratified by wealth.)

O'Leary, Michael F. 1985. The Economics of Pastoralism in Northern Kenya: The Rendille and the Gabra. Integrated Project in Arid Lands Technical Report Number F-3. Nairobi: UNESCO. **

O'Leary, Michael F. 1990. "Drought and Change Among Northern Kenya Nomadic Pastoralists: The Case of the Rendille and Gabra" in Gísli Pálsson (ed.), From Water to World-making: African Models and Arid Lands, pp. 151-174. Uppsala: Scandinavian Institute of African Studies. **

O'Leary, Michael F. 1990. Changing Responses to Drought in Northern Kenya: The Rendille and Gabra Livestock Producers. In WTW Baxter with Richard Hogg (eds.), Property, Poverty and People: Changing Rights in Property and Problems of Pastoral Development, pp. 55-79. Manchester: Department of Social Anthropology and International Development Centre, University of Manchester. **

O'Leary, Michael, and Gísli Pálsson. 1992. Pastoral Resources and Strategies: Similarities and Differences between the Rendille and Gabra of Northern Kenya. Security in African Drylands: Research, Development and Policy. Uppsala: Research Programme on Environment and International Security, Departments of Human and Physical Geography, Uppsala University. Pp. 105-122.

Ole Kuney, Reuben. 1994. Pluralism and Ethnic Conflict in Tanzania's Arid Lands: The Case of the Maasai and the WaArusha. Nomadic Peoples 34/35: 95-107.

Perlov, Diane Catherine. 1987. Trading for Influence: The Social and Cultural Economics of Livestock Marketing Among the Highland Samburu of Northern Kenya. Ph.D. Dissertation. University of California at Los Angeles. **

Reardon, Thomas. 1997. Using Evidence of Household Income Diversification to Inform Study of the Rural Nonfarm Labor Market in Africa. World Development 25(5): 735-748.

Reardon, Thomas, Eric Crawford, and Valerie Kelly. 1994. Links Between Nonfarm Income and Farm Investment in African Households: Adding the Capital Market Perspective. American Journal of Agricultural Economics 76: 1172-1176.

Reardon, Thomas, and J. Edward Taylor. 1996. Agroclimatic Shock, Income Inequality, and Poverty: Evidence from Burkina Faso. World Development 24(5): 901-914.

Reardon, Thomas, J Edward Taylor, Kostas Stamoulis, Peter Lanjouw, and Arsenio Balisacan. N.d. Effects of Nonfarm Employment on Rural Income Inequality in Developing Countries: An Investment Perspective. (M.s. submitted to Journal of Agricultural Economics in April 1996; also an earlier version of paper was invited for the symposium "Rural Diversification in the Developing World," Agricultural Economics Society Annual Conference, 25-28 March 1998, at the University of Reading.)

Roth, Eric Abella, and Elliot Fratkin. 1991. Composition of Household Herds and Rendille Settlement Patterns. Nomadic Peoples 28: 83-92.

Sieff, Daniela F. 1997. Herding Strategies of the Datoga Pastoralists of Tanzania: Is Household Labor a Limiting Factor. Human Ecology 25(4): 519-544. **

Spear, Thomas and Richard Waller (eds.). 1993. Being Maasai: Ethnicity and Identity in East Africa. London: John Curry.

Sperling, Louise. 1984a. Labor Organization in a Nomadic Pastoral Society, the Samburu of Kenya: A Theoretical and Methodological Framework for Research. Institute for Development Studies Working Paper No. 400. Nairobi: University of Nairobi.

Sperling, Louise. 1984b. The Recruitment of Labor Among Samburu Herders. Working Paper No. 414. Institute for Development Studies. Nairobi: University of Nairobi.

Sperling, Louis. 1984c. The Adoption of Camels by Samburu Cattle Herders. M.s. IDS.

Sperling, Louise. 1987a. Wage Employment Among Samburu Pastoralists of Northcentral Kenya. Research in Economic Anthropology 9: 167-190. **

Sperling, Louise. 1987b. Food Acquisition During the African Drought of 1983-84: A Study of Kenya Herders. m.s. May 1987. (Also published as "Food Acquisition by the Samburu Herders During the Drought of 1983/84" in Coping with Drought in Kenya: National and Local Strategies, edited by Thomas E. Downing, Kangethe W. Gitu, and Crispin M. Kamau. Pp. 264-279. Boulder and London: Lynne Rienner Publishers, 1989.)**

Staal, Steven, Christopher Delgado, and Charles Nicholson. 1997. Smallholder Dairying Under Transaction Costs in East Africa. World Development 25(5): 779-794. **

Straight, Bilinda Sean. 1997. Altered Landscapes, Shifting Strategies: The Politics of Location in the Constitution of Gender, Belief, and Identity Among Samburu Pastoralists in Northern Kenya. Ph.D. Dissertation (Anthropology), University of Michigan.

Vedeld, Pål. 1990. Household Viability and Change among the Tugens: A Case Study of Household Resource Allocation in the Semi-Arid Baringo District. Nomadic Peoples 25/27: 133-152. **

Vedeld, Paul, and Frank W. Lusenaka. 1991. Agriculture and Social Change in Kenya: A Case Study of Household Resource Allocation in Semi-arid Baringo District. Pp. 1-40. In Management of Natural Resources Articles/Summaries from M.Sc. Theses, 1988: Case Studies from China, Kenya, Nepal, and Tanzania Development and Environment No. 6. Noragric occasional papers series C/Norwegian Centre for International Agricultural Development at the Agricultural University of Norway. Norway. **

Webb, Partrick, and Thomas Reardon. 1992. Drought Impact and Household Response in East and West Africa. Quarterly Journal of International Agriculture 3: 230-246.

Annex 2: Bibliographic Annotations of Key Studies (complied by Barbara Cellarius)

Source: Fratkin, Elliot, and Kevin Smith. 1995. Women's Changing Economic Roles and Pastoral Sedentarization: Varying Strategies in Alternative Rendille Communities. Human Ecology. 23(4): 433-454.

Location: Marsabit District, Northern Kenya. In 1992, 5 communities were surveyed. Some of these were villages, others were towns. Populations ranged 250 to 2,000.

Estimated distance from nearest large town, name of the town and its estimated population: Town included in study

Absence/presence of food aid or food for work programs: Food aid in response to droughts in 1970s and 1980s.

Absence/presence of donor-funded or government funded agricultural schemes: Missions sponsored agricultural projects beginning in 1973.

Summary: Text generally discusses women's economic roles in the process of sedentarization. Compares four types of communities: nomadic cattle keeping, sedentary cattle keeping, settled agro-pastoralists, and urban town dwellers. Sedentarization began in 1960s with civil war and bandits. Droughts in the early 1970s and mid 1980s led to food aid and the growth of towns around food distribution centers. Living in town is described as providing greater opportunities for selling milk and/or vegetables. There is a narrative discussion of how women earn money, their reasons for preferring town life. Tables on labor time allocation, total income for women, and anthropometric measures.

Source: Fratkin, Elliot. 1992. Drought and Development in Marsabit District, Kenya. Disasters 16(2): 119-130.

Estimated distance from nearest large town, name of the town and its estimated population:

Population density: District density of 1.2 persons per km²in 1979 (Fratkin 1993)

Absence/presence of donor-funded or government funded agricultural schemes: projects to improve water resources?

Summary: By 1985 about half of the pastoral Rendille (total: 21,000) had settled in or around towns and were living on grain and wages. Towns are said to provide markets, services like education and health care, wage opportunities. Wages, often from the project that encourages the decrease in the size of the herds, are often converted into livestock. People in towns are often poor with few or no livestock or are elderly.

There is increasing economic stratification. There were aid projects to settle pastoralists and reduce herd size. Also food aid and projects to improve water resources, roads, and livestock marketing. This article seemed to be more of a commentary/description than a recitation of quantitative data. Much of the data that isn't filled in on this sheet can be found on the sheets for the other articles by Fratkin.

Source: Fratkin, Elliot. 1991. Surviving Drought and Development: Ariaal Pastoralists of Northern Kenya. Boulder: Westview Press.

Location: Marsabit District, Northern Kenya. He lived in Lewogso Lukumai, a village of about 250 people in the 1970s. It was in the desert but near a mountain (e.g. highland pasture access)

Estimated distance from nearest large town, name of the town and its estimated population: The town of Ngurunit was located about 15 km away

Annual rainfall: topographically diverse region. District average annual rainfall is 500 mm with a range of 200 mm to 1,000 mm across space. It is erratic and irregular in timing and quantity. There are usually two rainy seasons when it rains. Temperature is more consistent.

Drought frequency: 1934, 1939, 1942-45, 1959, 1954-55, 1965, 1970, 1973, 1976, 1980, 1982-84, 1987-88

Altitude: ranges from lowland desert (200-800m) to highland forests (>2,000m)

Livestock per capita: 8.2 TLU/person (1976), for 1985 it looks to be between 5.3 and 6.0 TLU/person (i.e. average TLU/household X no. of households / no. of people). In plain numbers: Ariaal 1976 (pop = 269) 1.6 camels, 4.6 cattle, 5.2 goats, 2.7 sheep; Ariaal 1985 (pop = 239) 2.7 camels, 3.6 cattle, 8.8 goats, 4.4 sheep; Rendille (aerial survey 1976-81; pop = 12,900) 1.4 camels, 1.9 cattle. 5.0 goats, 3.3 sheep.

Absence/presence of food aid or food for work programs: food aid programs in 1970s and 1980s.

Absence/presence of donor-funded or government funded agricultural schemes: Even after the drought there were programs to introduce maize cultivation to some groups.

Summary: The book deals with adaptation to drought, missionaries, and international development projects. The Ariaal were pressured by the missionaries and the government to settle. UNESCO-IPAL project: intent was to reduce herd size and encourage pastoralists to go into alternative ways of gaining income (goal was to reduce desertification by reducing pastoralism, he describes the project as unsuccessful).

The Ariaal raise cattle, camels, and small stock (there are said to be useful in terms of survival during drought). Town life is described as providing new opportunities, especially for people without enough animals.

Poor Ariaal with few or no animals look for wage labor in towns or take to hunting and collecting honey in the mountains.

Permanent settlements have been formed around missions (often associated with food aid programs). Move to towns not so much for famine relief as for greater security for their families, but most continue pastoral existence close to but not within urban areas.

Figures on p. 110 about wage labor: opportunities related to development projects.

Towards the end of the book he makes a comparison with other pastoral groups in Kenya. Constraints/factors pushing groups out of pastoralism include land restrictions on herding ranges, civil unrest, encouragement to give up herding and settle near population centers, expansion of agriculture or commercial growers. He says that these factors have not affected the Ariaal.

Source: Fratkin, Elliot, Eric Abella Roth, and Martha A Nathan. 1996. Sedentarization, Commoditization and Development Among Rendille Pastoralists of Kenya. Paper presented at the American Anthropological Association Annual Meetings. San Francisco, CA. Nov. 20, 1996.

Estimated distance from nearest large town, name of the town and its estimated population:

Livestock per capita: 1.9TLUs/cap at Korr (lowland Rendille camel community), 4.0 TLUs/cap at Karare (highland Ariaal cattle community) - looks like these data are for 1995.

Summary: Following recent droughts of 1968-73, 1983-84, about one half of the Rendille have settled in or around 3-4 towns in the region in response to environmental decline and active intervention by relief agencies. The people most likely to settle are poor without enough animals to survive or rich families who relocate part of the household to take advantage offered by living in town. However, their findings indicate that town life is not necessary beneficial to the health and nutrition of children. They are better off in nomadic communities with sufficient livestock herds.

Source: Campbell, David J. 1984. Responses to Drought Among Farmers and Herders in Southern Kajiado District, Kenya. Human Ecology 12(1): 35-63.

Ethnic group(s): Maasai herders (42%); Maasai herder-farmers (32%) and non-Maasai farmers (34%) (percentages out of 391).

Date data collected: February/March 1977 and March/April 1978 regarding drought between 1972 and 1976. Seems to have been a regional survey, rather than from a particular location.

Estimated distance from nearest large town, name of the town and its estimated population:

Absence/presence of food aid or food for work programs: Famine relief present

Absence/presence of donor- or government-funded agricultural schemes: not discussed

Summary: Conclusion that a combination of livestock and crop production was more successful way to overcome drought related shortages than either activity alone. Specific data on coping strategies was rare. Maasai farmers: 36% cash income from petty trading, 31% livestock sales, 10% sale of crops. Also notes that this varies from one family member to another. Famine relief has been given to 41% of Maasai farmers, 53% of non-Maasai farmers, and 67% of Maasai pastoralists. This is seen as evidence that Maasai farmers are the best off of the three groups. About 45% of non-Maasai farmers have off-farm income. Also asked questions and thus presents data about precautions that folks surveyed would take against future drought.

Source: Campbell, David J. 1978. Coping with Drought in Kenya Maasailand: Pastoralists and Farmers of the Loitokitok Area, Kajiado District. Working Paper 337. Nairobi: Institute of Development Studies, University of Nairobi.

Ethnic group(s): Maasai pastoralists (42%), Maasai farmers (23%), Kikuyu farmers (20%), and other farmers (14%)

Date data collected: Survey at end of 1972-76 drought, prior to rains of March/April 1977.

Estimated distance from nearest large town, name of the town and its estimated population:

Annual rainfall: study (from report cited below, they are 13 to 14 year averages) from Loitokitok (elevation 6050 feet) - 913 mm. Rambo Mission (elevation 3700 feet) -- 794 mm.

Summary: Land-use issues in region: increase in agriculture, especially following independence. Also creation of national parks. Some farmers are recent arrivals to the area and hadn't adapted their cropping patterns to conditions of low rainfall.

91% of pastoralists said that this drought was the worst that they had experienced.

96% of non-Maasai farmers said that this was their first drought that they had experienced.

Animals raised are sheep, goats and cattle. Lots of data on livestock numbers and losses. Author says that many Maasai have taken up agriculture but with relatively little success. 29% of Maasai respondents see wildlife as a source of food during bad years. Farmers don't eat wildlife.

67% of respondents from Loitokitok had received famine relief. Fewer Maasai farmers received famine relief (41%), which is attributed to their mixed economy allowing them to better cope.

Maasai farmers: 36% of average cash income from 'biashara', 31% from sale of livestock, 10% crop sales. Activities vary according to status in household: head of household most likely to be involved in trade, sons to work in town, and wife to sell food. Principle source of money to buy food for Kikuyu farmers was business activities such as shop-keeping and remittances from wage earners in town. About 25% of cash was from sale of crops. Kamba respondents (second largest non-Maasai farmer group in study) earned large share of their cash from crop sales but also earned cash through wage labor.

Source: Ensminger, Jean. 1992. Making a Market: The Institutional Transformation of an African Society. Cambridge: Cambridge University Press.

Location: Tana River District and Lamu District, SE Kenya. She specifically worked with the Galole subgroup of the Orma, who live in the middle part of the district. Her survey data of what is described as a scattered and nomadic population is from a 20 by 20 mile block near the market center of Wayu. The xeroxed table is for a sub-sample of households living in two villages in the study area (Wayu and Koticha, about 8 km from Wayu).

Estimated distance from nearest large town, name of the town and its estimated population: By late 1970s, the combined population of Wayu and its sister village of Wayu Boro (3 km away) was 74 households. At the time about 39 percent of the entire Golole population was living in such market towns. In 1987, the combined population of the same two settlements was 119 households and 63% of population was living in market towns. District headquarters of Hola was about 50 km away from Wayu (based on a rough estimate off of her map).

Drought frequency: Notable droughts and stock losses in 1933 (rinderpest), 1943, 1974-75, 1984-85.

Absence/presence of food aid or food for work programs: Government famine relief (present intermittently since colonial times) during 1970s and also during 1985 drought.

Absence/presence of donor-funded or government funded agricultural schemes: Irrigation schemes are mentioned as source of territorial encroachment.

Summary: 1980s: Significant employment opportunities. Wide scale farming along river beginning in 1970s.

Source: Sieff, Daniela F. 1997. Herding Strategies of the Datoga Pastoralists of Tanzania: Is Household Labor a Limiting Factor. Human Ecology 25(4): 519-544.

Estimated distance from nearest large town, name of the town and its estimated population: Not specified. Agricultural centers (e.g., to buy grain) are located 1-3 days walk away.

Annual rainfall: Average annual precipitation 363mm/year, mostly between November and April. Noted that there is year to year variation in rainfall.

Livestock per capita: average household has 12.1 residents. Households own average of 58 cattle, 22 goats, 27 sheep, and 6 donkeys.

Summary: Cattle represent 81% of TLU, goats 10%, and sheep 10%. Agriculture is negligible although they tried to grow maize and millet since 1987. These efforts have generally been unsuccessful. No other discussion of diversification or time series data is provided.

Source: Little, Peter D., 1992. The Elusive Granary: Herder, Farmer, and State in Northern Kenya. Cambridge: Cambridge University Press.

Location: Baringo District, Kenya (most intensely in neighborhoods of Loropili, Kailerr, and Salabani, but also some more regional investigations)

Date data collected: Main fieldwork 1980-81, with shorter visits in 1984, 1985, and 1986

Estimated distance from nearest large town, name of the town and its estimated population: This was not clear.

Annual rainfall: Rainfall varies with elevation, from averages of 500 mm to 1500 mm a year. There is a potential for rainfed agriculture mostly between 900 and 1400 meters, with 400 to 750 mm of rainfall a year. Another place says that average rainfall is 640mm/year. Rainfall distribution is described as erratic during the year. In addition the area sees high temperatures and evaporation rates.

Drought frequency: Droughts occurred 1921-22, 1924-25, 1927-28, 1931-33, 1979-81 and 1984.

Altitude: There are several ecological zones in the region, ranging in elevation from 1,000 to 2,500 meters above sea level.

Livestock per capita: average homestead = 6.72 people. Average homestead herd has 19 cattle, 29 goats, 73 sheep.

Absence/presence of food aid or food for work programs: This is not a new phenomenon for the area. The area saw famine relief between 1926 and 1933, mostly food-for-work or subsidized barter programs. More recently, there was food aid between 1971 to 1981, both from both governmental and nongovernmental donors. In 1982 there was a food-for-work project.

Absence/presence of donor-funded or government funded agricultural schemes: There have been small-scale irrigation projects and a project to support dryland agriculture and conservation.

Summary: The Il Chamus have a mixed agro-pastoral economy, although there is an emphasis on pastoralism. They raise cattle, sheep and goats. According to the author, the reasons for diversification depend on the economic status of the household. For example, for some cultivation for purposes of securing land is a reason to invest in agriculture even if returns are low. About 41 percent of homesteads have one or more members employed for wages. Wage employment is not new, but the magnitude is, the result of an expansion in the opportunity for wage labor. It is estimated that less than 10 percent of the working age population is employed.

Source: Kituyi, Mukhisa. 1990. Becoming Kenyans: Socio-economic Transformation of the Pastoral Maasai. Nairobi: ACTS Press (African Centre for Technology Studies).

Estimated distance from nearest large town, name of the town and its estimated population:

Annual rainfall: Rainfall is low and erratic, averaging less that 750 mm a year. On average there is a bimodal distribution during the year, but with great spatial and periodic variation - from less than 200 mm to more than 1000 mm.

Summary: The author explains that he is taking a regional approach and there is no specific discussion of fieldwork. Almost no quantitative data, and what there are are very regionalized.

He does discuss increasing involvement in no-herding activities, specifically, agriculture, wage labor, and business. Also some discussion of who tends to be involved in these activities.

Source: O'Leary, Michael F. 1990. "Drought and Change Among Northern Kenya Nomadic Pastoralists: The case of the Rendille and Gabra" in Gísli Pálsson (ed.), From Water to World-making: African Models and Arid Lands, pp. 151-174. Uppsala: Scandinavian Institute of African Studies.

Estimated distance from nearest large town, name of the town and its estimated population:

Drought frequency: He has detailed information on drought occurrences (as defined by the local people) on pages 171-74. Also says that there were droughts in 1965, 1970-71, 1975-76 and 1980.

Livestock per capita: household size of 5.7 persons. 1984 livestock per household: small stock 157.1 (27.6/person), camels 14.4 (2.5/person), cattle 7.9 (1.4/person), donkeys 0.4 (0.07/person)

Absence/presence of food aid or food for work programs: Famine relief/food aid beginning with droughts in the 1970s on.

Summary: The study population is described as primarily nomadic pastoralists, who raise camels, sheep, goats, cattle, and donkeys. The Rendille began to settle in large numbers following the droughts of the 1970s. Paper discusses drought responses in 3 time periods: 1918 to 1939, 1940 to 1963, and 1963 to 1980. It is mostly a discussion of the different drought response strategies, such as temporary migration, resettlement, and wage labor. The importance of non-pastoral sources of income is increasing, but livestock remains the mainstay. No quantitative data on diversification.

Source: Vedeld, Paul, and Frank W. Lusenaka. 1991. Agriculture and Social Change in Kenya: A Case Study of Household Resource Allocation in Semi-arid Baringo District. Pp. 1-40. In Management of Natural Resources Articles/Summaries from M.Sc. Theses, 1988: Case Studies from China, Kenya, Nepal, and Tanzania Development and Environment No. 6. Noragric occasional papers series C/Norwegian Centre for International Agricultural Development at the Agricultural University of Norway. Norway.

Location: Southern Baringo, Baringo District, Kenya. Pokor/Keben and Kakimor locations, on main road between Nakuru and Marigat. For the survey, a 15 percent sample was selected of households in each of 16 villages in the study area.

Estimated distance from nearest large town, name of the town and its estimated population: Emining is a marketing and administrative center. It is 60 km to Nakuru (from where is not clear)

Annual rainfall: The district is described as semi-arid with low and erratic rainfall. The average annual rainfall at Emining in the study area is 940 mm and erratic. There is a unimodal distribution with 5-6 months of dry weather.

Population density: : in 1979: 23 persons per km² at Pokor/Keben and 41 at Karimor

Absence/presence of donor-funded or government funded agricultural schemes: Authors note that the study area was chosen in particular for a lack of active project implementation.

Summary: The study population cultivates millet, sorghum, maize, and beans, and they keep cattle, sheep and goats. 93 percent of households own livestock and livestock are described as the main source of livelihood. 99 percent of the households in the sample cultivate grain. 67 percent of households are involved in off farm activities: salary work, kiosk keeping, matatu business, tailoring, shoe repairing, running grain mills, butcheries, casual work, trading, and charcoal production. Household heads running small business may employ poor landless folks.

Labor allocation figures presented are based on estimates and figures (co-efficients) from other studies), rather than from any formal data collection at this site.

In the average household, 60 percent of gross output value is estimated to come from livestock, 8 percent from crop production, and 32 percent from off-farm activities. This varies according to the economic status of the household. Low-income households tend to emphasize crop production and have low off-farm income. Well-to-do households get about 64 percent of their income from off-farm sources.

Source: Vedeld, Pål. 1990. Household Viability and Change among the Tugens: A Case Study of Household Resource Allocation in the Semi-Arid Baringo District. Nomadic Peoples 25/27: 133-152.

Location: Southern part of Baringo District, Kenya. Pokor/Keben and Kakimor locations

Estimated distance from nearest large town, name of the town and its estimated population: main market and administrative center is Emining

Annual rainfall: Rainfall is low and erratic. Unimodal distribution with 5-6 months dry. Average annual rainfall is 940 mm.

Population density: in 1979: 23 persons per km² at Pokor/Keben and 41 at Karimor

Livestock per capita: average household in survey (8,8 members) controlled 12.5 L.U, 65% cattle, 35% goats/sheep

Summary: In recent years the study population has become involved in crop production and off-farm activities in addition to livestock production. They were settled through land reform. The author says that because of the small amount of land there has been off-farm diversification of household resources. There is a skewed distribution of land between households, and also in the number of livestock.

The author attributes resource allocation to physical constraints, resource endowments, market constraints, household aspirations and motivations, and socially related values with respect to production and output distribution. Livestock are the main source of livelihood. 93% of households have livestock and 54% of households sell livestock for cash. 99% of households surveyed cultivate grains and 41% of households sell grains for cash. 67% of households are involved in off-farm activities. Labor time estimates are based on figures from other sources.

Source: Hogg, Richard. 1980. Pastoralism and Impoverishment: The Case of the Isiolo Boran of Northern Kenya. Disasters 4(3): 299-310.

Estimated distance from nearest large town, name of the town and its estimated population:

Annual rainfall: Garba Tulla (455 meters) 312 mm/yr; Mado Gashe (245 meters) 294 mm/yr; Habaswein (195 meters) 262 mm/yr. But great yearly variability

Drought frequency: drought years listed on p. 303. 8 major droughts in the last 60 years.

Summary: Evidence of some irrigated agriculture in the past (there is the possibility for this along rivers), perhaps as early as the 1940s. Since end of Shifta several irrigation schemes have been established (from 1969 onward). Cultivation of maize is an important insurance against hunger. (p. 307) his estimate of involvement in agriculture: says that less than 13% of Borana/Sakuye are directly involved in cultivation Involvement in wage labor is usually outside the district. In areas of high population density, there is involvement in petty trade. This is especially the case in female- headed households. He associated trade with market opportunities, not only in the context of the disasters.

Source: Hogg, RS. 1981. The Social and Economic Organization of the Boran of Isiolo District of Northern Kenya. PhD Dissertation. Manchester University.

Estimated distance from nearest large town, name of the town and its estimated population: Largest local town is Mado Gashe (4,000) about 35 miles from the river (study villages seem to be along the Ewaso river); the administrative center for this part of the district (Merti - 2,000) is located on the river; there is a market in Garba Tula (3,500) about 15 miles from the river.

Annual rainfall: semi-arid, bimodal rainfall but highly variable in distribution and amount. Figures presented for three sites: average annual rainfall at Gabba Tula is 408 mm, at Mado Gashe 305 mm and at Habaswein 265 mm

Drought frequency: 1918-19, 1928-29, 1933-34, 1943-44, 1959-60, 1970-73, 1975-76, 1979-80

Absence/presence of food aid or food for work programs: in 1979 food aid, food for work

Absence/presence of donor-funded or government funded agricultural schemes: Irrigation schemes - beginning in early 1970s.

Summary: In the early 1950s, the Boran relied primarily on their cattle, camels, and small stock with only marginal involvement in market economy. Now they are part time pastoralists (after drought and war). There is increasing reliance on the goods and services available in towns. Some are involved in government/donor sponsored irrigation schemes. Others are petty traders or beggars in town or involved in labor migration. Trade is described as an occasional source of income. Rapid growth in outmigration since shifta. Of 160 married men surveyed, 17% were involved in wage labor themselves and 17% had one or more sons in wage labor. Most often this was as watchmen. With all this outmigration, a significant feature of village life was women with husbands absent.

Farming was also a way to get cash. Irrigation agriculture along the rivers. In the late 1970s "only" 900 families were involved in irrigation schemes/directly farming.

Incentives for involvement in wage labor depended on economic status of the household.

He said that he had trouble collecting detailed family budgets. Concentration of villages along the Ewaso river (irrigation schemes starting in the early 1970s).

Author sees economic diversification as the key to successful adaptation to permanent settlement and farming.

Source: Hogg, Richard. 1988. Water Harvesting and Agricultural Production in Semi-arid Kenya. Development and Change 19: 69-87.

Estimated distance from nearest large town, name of the town and its estimated population:

Annual rainfall: Turkana: 200-700mm depending on location; semi arid zone at Baringo 400-750 mm; Kitiui district 400 mm in lowlands to 1,000 mm in highlands

Absence/presence of food aid or food for work programs: Turkana: food for work projects; not so massive at Baringo

Summary: Article primarily looks at development projects, specifically water harvesting/water conservation. Examines such projects in three different sites with periodic references to off farm employment.

Turkana: Increasing interest in water harvesting/spreading following 1979/80 famine. Also related to failure of irrigation projects and lots of people on food for work projects. Water harvesting is expensive in time and labor (traditional preference of Turkana for growing crops in unimproved sites is attributed to high labor costs and marginal rates of return). Also, as Turkana households become economically diversified, labor resources are stretched and thus it is more difficult to allocate resources to farming which has marginal rates of return.

Kitiui: labor scarcity also an issue here , part of which is from young men away for wage labor.

Source: Staal, Steven, Christopher Delgado, and Charles Nicholson. 1997. Smallholder Dairying Under Transaction Costs in East Africa. World Development 25(5): 779-794.

Estimated distance from nearest large town, name of the town and its estimated population: data grouped into within city limits and within 60 km of city

Summary: Only a discussion of milk production. Doesn't discuss other economic activities.

Source: Gebre Mariam, Ayele. 1991. Livestock and Economic Differentiation in North East Ethiopia: The Afar Case. Nomadic Peoples 29:10-20.

Estimated distance from nearest large town, name of the town and its estimated population:

Annual rainfall: For arid regions: average annual rainfall between 1962 and 1982 was 225 mm with a range of 152 mm to 322 mm. For semi-arid regions: average annual rainfall between 1966 and 1982 was 561 mm with a range of 444 mm to 681 mm.

Livestock per capita: ration of cattle, camel, sheep and goats to people was 1.1:1; 0.5:1; 3.8:1; and 3.1:1

Summary: The Afar raise camels, cattle, sheep, goats, donkeys, and horses. When poor households are "sloughed off" from the pastoral activities due to epidemics, drought, and diseases, they move into poaching, the sale of firewood, charcoal, mat fibre, aromatic trees and ropes along with labor migration, irrigated farming, and keeping animals for Oromo lowlanders. This is about the only discussion of income diversification other than to note that the Afar are unlike the Maasai and Somali who invest wealth gained through livestock production in other sectors of the local economy. There is diversification in types of li

ECONOMIC DIVERSIFICATION AND RISK MANAGEMENT AMONG EAST AFRICAN HERDERS: A PRELIMINARY ASSESSMENT AND LITERATURE REVIEW