Structural Declines and Stochastic Escapes 20
3.5 Exploring transition patterns from event-histories
3.5.1 Livelihood strategies and idiosyncratic shocks
The ability to pursue different livelihood strategies depends on livelihood assets. Livelihood strategies in turn shape the livelihood outcomes for households, for example, improved welfare and increased income. Idiosyncratic shocks impact on welfare directly and indirectly (through their influence on assets). In case of a severe shock such as death, households are often forced to dispose of their assets as a coping mechanism. We used cluster analysis techniques to characterize the range of livelihood strategies, and shocks experienced by the households using data from the event-histories. We performed separate cluster analysis for livelihood strategies and shocks.
For livelihood strategies, the cluster analysis aimed at identifying relatively homogenous clusters of households engaged in similar economic activities. With regard to shocks, the cluster analysis aimed at characterizing the different shocks that households experienced over the period 2000–2009. In each case, we used hierarchical agglomerative
clustering to determine the number of clusters. We confine our analysis of the event-histories to poverty transitions over the period 2000–2009 (i.e. over the entire nine years) as it was impossible to separate certain event-histories into the distinct periods. Livelihood strategies, for example, can spread over several years. We use the livelihood strategies and shock clusters, and other factors to explore the characteristics of households in different mobility groups.
Since households pursue a wide range of livelihood activities, our cluster analysis was preceded by an initial factor analysis. The factor analysis identified a smaller number of factors explaining the majority of the variation observed among the groups. Out of 15 livelihood activities, five ‗factors‘ were obtained in this initial step. Next, we performed a cluster analysis of the five factors. The cluster analysis identified eight principal livelihood clusters with relatively homogeneous households pursuing a similar mix of livelihood activities. All households were involved in food crop farming. A detailed description of each livelihood cluster is presented in Table B1 (Appendix B). A brief description of the livelihood clusters is as follows:
Diversified sugarcane farming. This livelihood cluster accounts for 11% of the total sample households (39 households). All households in this group are sugarcane farmers, with almost 70% having increased land under sugarcane cultivation over time by renting in land. In addition, these households have intensified livestock production through adoption of improved breeds of cattle — mainly cross-bred dairy cows. One-third earns their livelihood from formal employment, while slightly over half are involved in trade of non-food products.28 The majority of these households (80%) are from the Western transitional zone.
Indigenous livestock and trade. This is the largest livelihood cluster, made up of 22% of the total households in the sample (78 households). Households in this cluster are mainly from the Western transitional zone. Over 90% of these households earn their livelihood from indigenous livestock farming. Trade in non-food products and formal employment in private or public sectors are also livelihood activities for this cluster.
Diversified livestock farming. This cluster includes households earning their livelihoods primarily from indigenous livestock production. Over half of the households are involved in a range of other livelihood activities: petty trade in food products (77%), sugarcane farming (61%), intensified production of food crops (55%) — vegetables and tomatoes for sale, and
28 The different types of trade in non-food products included livestock trade, retail shops, posho mills, hiring out oxen and plough, weaving (baskets, pots and ropes) and second hand clothes.
trade in non-food products (55%). Over 80% of these households are from the Western lowland and Western transition zones.
Sugarcane farming, trade and farm labour. Virtually all households in this cluster are sugarcane farmers. Over 40% are also involved in trade in non-food products and farm labour, while another one-third are involved in petty trade of food-products. One-third are also farm laborers. Almost 70% of these households are from the Western transitional zone, while 30% are from the Western lowland.
Farm labour and trade. This is the second largest livelihood cluster, making up 20% of total households in the sample (70 households). In addition to subsistence food crop farming, these households pursue relatively low income livelihood strategies: farm labor, trade in non-food products and food products. Over half of these households reside in the Western lowland.
Dairy farming and trade. This is the smallest livelihood cluster, accounting for 6% of the total sample (21 households). About 90% of households in this group are dairy farmers. Over half are also involved in trade in non-food products, while less than 40% are sugarcane farmers. Another 29% earn their livelihood from skilled wage labor.
Tea and dairy farming. This cluster includes households involved in tea and dairy farming as their main livelihood strategy. Other livelihood activities include business in small towns (rental properties, retail shops and transport business), farm labor, and skilled wage labor.
Coffee and dairy farming. Households in this cluster are coffee and dairy farmers, in combination with intensified food crop production, formal employment, and skilled wage labor. One-quarter of households in this cluster have intensified their livestock production strategies by investing in new or different types of animals — dairy goats, pure exotic dairy cows and improved chicken, and the majority of households in this cluster are located in the Central highland.
We identified various idiosyncratic shocks experienced by the households over the period 2000–2009: health shocks, death of a wage earner supporting the household, expenses related to funerals, livestock losses, crop losses, loss of regular employment, and accidental losses (fire, theft of livestock and break-ins for business assets). Since the different types of shocks were few, we did not perform an initial factor analysis. The cluster analysis revealed five groups of households. We distinguish two critical types of health shocks: those related to ill-health that affect the productivity of the household (inability to work), and those related to expenses incurred to access treatment (health expenses). Limited access to health insurance,
poor services in public health facilities, and also cost-sharing in many of the public health facilities, implies that the majority of rural households in Kenya incur expenses for treatment.
Often, the two types of health shocks occur simultaneously. Shocks related to livestock losses were experienced across all groups, but to a limited extent, compared to other types of shocks. Most of these shocks often lead to asset losses. Death, for example, leads to loss of human capital and in some cases households have to dispose of other assets such as livestock as a coping mechanism. However, some are also related to cultural values — for example heavy funeral expenses that includes the slaughter of a household‘s livestock that is prevalent in the western zones. Detailed description of each cluster of shocks is presented in Table B2 (Appendix B). A brief description of the five shock clusters is as follows:
Few shocks. Households in this cluster experienced very few shocks, mainly related to livestock losses and accidental losses due to burglary or fire.
Death and funeral expenses. This cluster includes households who experienced shocks related to death of a major income earner and funeral related expenses. This the smallest cluster of shocks with 8% of sample households (29 households).
Health shocks. This is the largest cluster of shocks, comprising 33% of the sample households (115 households). This cluster represents households that experienced two types of health shocks — ill-health and health related expenses. One-quarter of these households also lost some livestock.
Health and funeral expenses. Virtually all households in this cluster experienced a combination of health shocks (medical expenses) and funeral expenses. This cluster accounted for 14% of the total sample.
Multiple shocks. Households in this cluster experienced multiple shocks compared to the other groups. The shocks included death of primary wage earner, health shocks (ill-health and health related expenses), funeral expenses and also some (26%) lost livestock.
Apart from livelihood strategies and shocks, we identified a number of other factors (positive and negative) that were likely to impact on household welfare over the period 2000–
2009. These factors included help from relatives, large number of dependants (including orphans), land subdivision, dowry payments (bride price), old age, alcohol dependency, and in some cases paying school fees among others. Help from relatives can take various forms:
direct remittances (most common), providing education assistance or school fees, assistance with food in extreme cases, and providing capital for starting up a business.