Next, we explore the incidence of various shocks across zones. As discussed in section 4.2, the study sites reflect diversity in agro-ecological potential, poverty incidence, population density, and culture (ethnicity). We find significant differences in prevalence of shocks across zones.
Certain shocks are more prevalent in some zones than others (Table 4.1). The differences in distribution of the most commonly reported shocks (those reported by four per cent or more of all households) were statistically significant. Across divisions within the zones, the incidence of shocks do not differ significantly, except for livestock losses in Eastern lowland (Table C1 in appendix). 31
Health expenses, ill-health and livestock losses were the most commonly reported shocks in Eastern lowland (EL). The proportion of households reporting ill-health (37%) and livestock losses (22%) was higher in this zone than other zones. The high proportion of households reporting livestock losses is consistent with the fact that Eastern lowland is prone to droughts
31 Within each zone, we sub-divided the sample further into divisions (relatively homogenous units). Overall, the sample cuts across eight divisions, each with a sample size ranging from 30 to 56 households.
that result in livestock death. The recent droughts were experienced in 2005 and 2009 (as reported from the community focus group discussions). Funeral expenses and land sub-division shocks were reported by less than 10% of the households each.
Table 4.1: Distribution of shocks across regions (per cent of households) Eastern
1The Chi-square statistic is calculated for the most commonly reported shocks only.
* p < 0.1, ** p < 0.05, *** p < 0.01
Health expenses, ill-health, funeral expenses and death of major income earner were the most commonly reported shocks in Western lowland (WL). This zone had the highest proportion of households reporting death of a major income earner (19%) compared with other regions.
These results are consistent with the high disease burden (HIV and malaria) in this zone. The HIV prevalence rate in Western lowland was estimated to be 15.3% in 2007 — the highest in Kenya (NASCOP, 2009). This region is also one of the areas with relatively high risk of malaria in Kenya as some areas within the zone are located around the shores of Lake Victoria (Noor et al., 2009). Surprisingly, this zone also had the lowest proportion of households reporting ill-health shocks. In situations where ill-ill-health is widespread among a population, it might not be considered a shock to individual households. Also, in the context of health shocks and health status, people‘s perceptions of their own health are likely to be related to their education, occupation, and household income (see Gertler et al., 2000). Funeral expenses were reported by a substantial proportion of households in the region (the second highest after the Western
transitional zone). Among the Luos, adult deaths are often associated with heavy funeral expenses that include customary slaughter of a household‘s livestock assets coupled with extended periods of mourning. Therefore, the death of an adult not only leads to loss of human capital, but is often associated with expenditure shocks. Other common shocks reported in the region were livestock losses (11%) and accidental losses (9%). The WL is occasionally affected by drought, explaining the substantial proportion of the households reporting livestock losses.
Health expenses, ill-health, funeral expenses and land sub-division were the most commonly reported shocks in the Western transitional zone (WT). This zone had the highest proportion of households reporting health expenses (50%) and funeral expenses (26%) compared with other zones. The findings for funeral expenses reflect the underlying cultural differences across the zones. Similar to the Western lowland zone, the death of an adult among the Luhyas is often associated with heavy funeral expenses, and is thus often associated with expenditure shocks. Similarly, land sub-division shocks reflect the high population pressure in this area leading to land fragmentation. Livestock losses were reported by less than 10% of the households. Surprisingly, the WT is the only zone where dowry payment shocks were reported.
This may be explained by the underlying customs and traditions surrounding marriage among the Luhyas. Regardless of welfare status, it is customary for men to pay bride price (dowry) — mostly in the form of livestock (particularly cattle). Failure to pay dowry can lead to lack of respect for the man (husband) by his relatives, community members, and in-laws. In addition, there are cultural penalties in the event that a wife dies before dowry is paid that include compulsory payment of dowry before the man can be allowed to bury the wife.32 In the worst case, the in-laws may decide to bury their daughter.33
In Central highland (CH), the most commonly reported shocks are similar to those of the Western transitional zone. The Central highland had the highest proportion of households (16%) reporting land sub-division. Among the Kikuyu, property is often sub-divided among sons and unmarried daughters. The prevalence of land sub-division may also be a reflection of the high population pressure in this area. The average land holding per household of 3.4 acres is lower than in other zones. Land sub-division may also be highly driven by the cultivation of cash crops
32 The man is also expected to give separate cattle specifically to the wife‘s family for slaughter during the funeral or to take it away after the funeral.All these are done under tough negotiations and threats by in-laws who normally would have an upper hand in such situations.
33 If this happens, it is traditionally believed that the spirits of the dead wife would remain to haunt the man and the community forever.
(tea and coffee). In some sites within the zone, for example, smaller plots may not be sub-divided further, yet the tea bushes on these plots are sub-sub-divided among sons. Surprisingly, this is the only region where households reported conflicts (6%). These conflicts may be attributed to alcoholism and other underlying cultural factors specific to our sites within the Central highland.34
Other studies also report an association between the incidence of shocks and geographical location. Tesliuc and Lindert (2004) found that the largest relative differences in the incidence of shocks in Guatemala occurred across geographical locations, with regional characteristics explaining most of the variation in incidence of shocks. Other household characteristics such as poverty status, gender, and ethnicity of the household head were also associated with the relative incidence of shocks. However, these associations were partly due to the association between poverty, geographical location and ethnicity. Geographical clustering of certain shocks such as those related to human diseases — malaria, tuberculosis, typhoid and cholera — have been reported in Kenya and Madagascar (Mills et al., 2004; Noor et al., 2009).