4.1. Comparison of the Business Participants and Non-Participants
In Table 1
, we compare participation in the inclusive business initiative in terms of the households’ key socio-economic characteristics. The descriptive statistic output indicates that the participants’ households have access to relatively more land than non-participants, averaging at 12 and 7 acres, respectively. The average age of participating households’ head is higher (58 years) compared to that of non-participating households (48 years). The differences in sizes of land and age of the households’ head are statistically significant. Only 8% of all the households are headed by a youth (<35 years), who owns an average of 3 acres.
The annual crop and livestock incomes are also considerably different between the two categories. The participating households have significantly more annual average crop income (Ksh. 147,542) than non-participants (Ksh. 68,125). Although the difference in the annual average livestock income is not significant, non-participants seemingly fare better (Ksh. 61,100) than participants (Ksh. 42,355). With respect to number of mango trees, participating households had significantly more trees in total (127) than non-participants (88). This indicates that the number of mango trees is proportionate to the size of land owned by the household. The period in which the trees were planted varies, with majority being planted in the 2000s and only a few in the early 2010s. All the households were reported to depend on family labor for mango production. Being a perennial crop, however, means there is not much labor involved in mango farming other than during the harvest season. It was reported by the smallholders during focus group discussions that the majority of the harvesting labor is provided by the buyer (the cooperative) at a cost. Nevertheless, it is apparent that, based on the family size, participants have a slightly bigger labor pool (seven) than non-participants (six), although the difference is not statistically significant.
compares participation in terms of the gender and education level of the household head, and whether the household had acquired a loan in the 12 months leading to the crop harvest period. We find that although only 14% of the sampled households were female headed, comparatively more female-headed households (75%) participate in the business than male-headed households (65%). Education-wise, in all levels, we find that participant households’ heads are more educated than non-participants’: primary (56%), secondary (70%), and higher education (89%). Although both participant and non-participant households had taken loans, the participant farmer group has a significantly higher number of the borrowers (47% against 25%). Besides agriculture, a significant proportion (59%) of the households also engage in off-farm livelihood activities, but involvement differs between participants and non-participants: the number of the households’ heads with formal employment is higher among participants (61%) than non-participants (39%). A similar case applies to business ownership, in which participants represent 71% and non-participants represent 29%. Conversely, a relatively higher proportion of non-participant households’ heads earn their off-farm income from wage labor (53%) compared with participants (47%). Only the participant households receive remittances.
The results of the probit regression model, estimating the likelihood of participation in the mango business, are presented in Table 2
. The model’s output (χ2 = 27.55, p
= 0.001) indicates the presence of adequate information to explain the association between participation and smallholders’ socio-economic attributes. The marginal effects are reported. We find that a household’s land size is positively and significantly associated with participation. The estimated marginal effect indicates that an increase in land size by a single unit (an acre) increases the likelihood of participating in the business by 8.5 percentage points. Furthermore, loan procurement has a positive and significant effect on participation. Having a loan increases the likelihood of a household’s participation in the business by 18.3 percentage points. Family size, in reference to a household’s labor capacity, is also positively associated with participation. The estimated marginal effect of family size reveals that an increase by one unit (one person) is associated with increased likelihood of participation by 3.9 percentage points. Male-headed households have a significantly lower probability of participating, while those with a more educated head have a significantly higher probability of participating. The estimated marginal effect indicates that being a male-headed household reduces the likelihood of participation by 29.2 percentage points. Completion of secondary education and higher education by the household head increases the likelihood of participation by 18.8 percentage points and 35.9 percentage points, respectively.
The age of a household’s head and the number of mango trees the household owns are also positively and significantly associated with participation in the business. Nevertheless, these variables were excluded in the present regression model to avoid multicollinearity [106
]. Both variables were highly correlated with the variable “land” and consequently were causing “adverse effects on the estimated coefficients” of the model [107
]. The variable “total annual income” does not have a significant effect on business participation.
These results corroborate claims made in interviews by smallholder farmers, who were excluded from the business initiative, that the cooperative only targeted farmers capable of supplying a high volume of produce. While we could not obtain specific information about this claim, the descriptive information on the disparity in the land size, number of mango trees, access to loans, and the annual income substantiate these assertions. Nevertheless, according to the cooperative chairman and the board members interviewed, all smallholders in the community are welcome to join the cooperative membership. By default, therefore, everyone has access to the business opportunity irrespective of production volumes. Despite this, the board indicated that they found it difficult to recruit the farmers. Our household surveys and interviews with non-participants pointed to a lack of interest in working with the cooperative. Their reasons for not joining: up to 83% of all the non-participants surveyed expressed either a lack of trust in a cooperative and/or an absence of adequate information about its existence and operations. As identified by a handful others, trust issues that relate primarily to the fear of collapse leading to loss of revenue, which is a common occurrence in Kenya’s cooperative sector, prevent farmers from joining. Indeed, cooperatives in Kenya “suffer common problems associated with weak legislation, poor financial management, leadership, governance and political interference among many others” [108
There is an apparent absence of youth in the business participation. Based on our survey, out of all smallholder households engaged in mango farming, only 20% are participants. This situation is largely attributed to the inaccessibility of land. Much of the land is owned by parents, who have not yet shared it with their children. Furthermore, even if they did receive their share, there is high likelihood that their portion would be relatively small, similar to that of the non-participants in the present analysis. Nevertheless, a handful of the youths in the community have benefited from the business initiative through job opportunities that have emerged from the establishment of the processing plant. Based on interviews with all the key informants, some 450 youths from the community have been employed along the value chains to perform various activities, including fruit picking, weighing, loading, offloading, and even in the processing activities.
It is evident that resource-poor smallholder farmers do not participate in the business. As a result, we expect the contribution of the business initiative to these households’ food and nutrition security to be negligible or nonexistent. For participating households, according to interviews and FGDs, the business has had a positive impact on income, hence better household food and nutrition security is to be expected. Following their integration in the business initiative, participating households maintained that they have seen a considerable increase in overall returns, a change they linked to the establishment of the fruit plant, which, unlike in the past, has allowed them to sell all their produce. According to one of the smallholder farmers (W):
“Previously, brokers and other buyers would only pick fruits that are “best in quality”, often in small quantities and would leave the rest. This would lead to major losses that we are currently able to escape (W).”
Nevertheless, to ascertain the overall contribution of the mango business initiative to local food and nutrition security in Makueni, it is critical to closely review how the participants and non-participants compare in their respective households’ food and nutrition security statuses.
4.2. Contribution to Local Food and Nutrition Security
presents the results of the independent sample t
-test, comparing the Household Food Insecurity Access Score (HFIAS) and Household Dietary Diversity Score (HDDS) between participating and non-participating households. We find that, on average, participants have a significantly lower score (4.2), which implies better food security status than non-participants (6.0). There is no significant difference in the average HDDS between participants (7.1) and non-participants (7.2). Despite participants having a slightly bigger family size, their average weekly spending on food is slightly lower (Kshs. 1094) than for non-participants (Kshs. 1170). This difference, though, is not statistically significant.
The lack of a significant difference in HDDS implies that there is no variation in the type of diets consumed locality. Figure 2
presents the type of food consumed as part of the local diet. Cereals, vegetables, dairy products, oil and fats, sweets, spices, condiments, and beverages predominate the local diet. Based on the data, chai, a hot beverage made up of milk, tea, spices, sugar, and water, is the primary source of dairy products and represents dairy products, sweets, spices, condiments, and beverages in the diet. Up to 15% of households do not consume legumes, nuts, or seeds. The consumption of fruits, white roots, and tubers is notably low, as they are only consumed by 14% and 31% of the households, respectively. None of the households consume fish, and only 1% and 3% consume eggs and meat, respectively.
Remarkably, while improvement in income among participating households likely contributed to better food security as measured by the HFIAS, compared to non-participating households, it is revealed that it has not affected these households’ nutrition security (Table 3
). The analysis on income spending between business participants and non-participants, whether on food or other needs—education, healthcare, clothing, agricultural inputs, labor, paying a loan, transport, assets, and home improvement—does not vary significantly. Furthermore, we find that the annual income spending compared to mango income spending on main items, including food, education, and agricultural inputs, is consistent across the households, whether they are participants or non-participants (Figure 3
). This implies spending trends across all households match closely. In addition, the results of weekly spending on food indicate an insignificant difference between participant and non-participant households. This implies that non-participants, with a lower resource endowment and capacity to produce, spend proportionately more than their counterparts, who have more ability. Overall, one factor that partly explains a lack of improvement in nutritional diet, despite the larger income accrued by participating households, is the fact that smallholders primarily buy staples and “commercial” ingredients rather than food that is essential for improving overall dietary diversity. The following statement summarizes interviews with key informants and focus group discussions on the issue:
“The food bought items from mango income is mainly staples (maize and beans) during ‘hunger periods’ and generally common ingredients for food and hot beverages preparations such as cooking oil, onions, tomatoes, tea leaves and sugar. Only on rare and special occasions do households spend money on ingredients from special meals such as wheat flour for chapati, rice, eggs and meat.”
Notably, an impending impact on local food security of the mango business initiative relates to land-use change associated with local mango plantation sizes. It is anticipated that the market opportunity provided by the plant and its promise of economic growth has prompted an expansion in mango plantations in Makueni. The following quotes from smallholders X and Y represent a growing determination to increase their mango tree plantations:
“I have been planting new mango trees. I have done so because in the future I expect to fill one lorry or more to be taken to the factory (X).”
“I have been grafting the local variety to Apple or Ngowe varieties and increasing the total number of trees to increase their productivity and marketability (Y).”
Indeed, the majority (60%) of the interviewed smallholders indicated that they had increased the number of mango trees on their land. Those that did not either replaced the old ones with new varieties recently or do not have enough land to expand their plantation, as informed by two different respondents.