4.2.4. Flooding

Climate change-induced flooding is one of the impacts of climate change, which has vast economic and social consequences. It has been occurring in Ethiopia for years inflicting heavy social and economic costs [56]. Flooding during rainy seasons is one of the primary climate-induced e ffects in the study area. Of the 402 households asked about their exposure to flooding, 74% (*n* = 297) said that flooding had a ffected them in the last ten years. Most parts of the study area, especially the western part, have su ffered from devastating flooding due to the seasonal overflow of the Bilat River. On the other hand, since the area is located in lower elevations, there are run-o ffs from the surrounding highlands that destroy terraces built to control soil erosion. Also, several gullies occurred because of heavy rain that has a ffected the landscape as noted by the researcher during field visits. Flooding during rainy seasons causes property and harvest loss incurring a significant economic cost to the farmers. The social and economic e ffects mentioned by households include loss of human lives and livestock, property damages, destruction of crops, water-borne diseases, and soil erosion.

## 4.2.5. Climate Change-Induced Hunger

Famine is a condition where people lack food for a longer period whereas hunger is a seasonal lack of food because of natural or human-made causes. Hunger becomes famine when it persists and affects many people [57]. RF shortage is one of the main reasons for famine and hunger in SSA [4,5]. Of the 401 respondents included in this study, 77% (*n* = 309) did experience hunger in the last ten years. Data collected from the Sidama Zone Agriculture and Natural Resources Management Department show that climate change-induced hunger did happen in the three districts in the previous fifteen years. Many people received emergency food aid in 2000, 2004, 2009, and 2016. These years coincide with the years of lower SRA discussed in part 4.1. According to key informants, signs of hunger include a delayed start of rain during rainy seasons (*belg* or *kiremt*) or its absence during critical sowing seasons. Both a ffect crop and livestock productivity and food security.

#### *4.3. Determinants of Climate Change-Induced Impacts on Smallholder Farmers*

#### 4.3.1. Determinants of Exposure to the E ffects of Drought

The result of the logit model shows that several reasons decide the likelihood of exposure to the e ffects of drought in the study area. The model is statistically significant for the determinants of drought (N = 357, Chi<sup>2</sup> = 59.99, *p* = 0.0000). The pseudo R<sup>2</sup> and adjusted R<sup>2</sup> (after marginal e ffect) values for this category of explanatory variables are 18.56% and 88.83%, respectively (Table 5). Out of the explanatory variables considered in the model, educational status, growing enset, membership in 1 to 5 groups, and being head in 1 to 5 groups a ffect the exposure to the e ffects of drought. Besides, the use of chemical fertilizer and the participation of women in family decision-making are found to be the main determinants of exposure to the e ffects of climate change-induced drought. Specifically, the findings show that a one year increase in schooling decreases the likelihood of being a ffected by drought by 10% (at *p* < 0.05 level).

Growing enset is another determinant of climate change-induced drought. Households who grow enset are 111.3% less likely to experience the e ffects of drought than those without *enset* at *p* < 0.05 level (Table 5). *Enset* (false banana) is a perennial crop known for its drought-resistant nature and is wildly used as a staple food in the most southern part of Ethiopia [58,59]). *Enset* is important for the local economy because its products, locally called *waasa* and *bu'la*, are sold in markets and are a source of income. Its roots, leaves, and stem are also used to feed animals during drought years. The Peasant Association (Kebele Administration) is the lowest administrative unit in Ethiopia that is accountable to the district (district) administration. Locally, households are organized in small groups of five households known as the 1 to 5 group. The governmen<sup>t</sup> designed it without having local farmers say in its formation. The governmen<sup>t</sup> widely uses this group for economic, political, and social purposes. Of the total 402 households involved in this study, 60% (*n* = 242) were members in the 1 to 5 group organization and the remaining 40% (*n* = 160) were not. On the other hand, 11% (*n* = 42) were serving as group leaders. There is widespread debate and controversies on the merits and demerits of such grouping. However, the purpose here is as a social organization, to what extent being a member or a leader in 1 to 5 groups is related to the likelihood of facing climate change-induced impacts?

The model revealed that being a member or having a position in the 1 to 5 groups a ffects the likelihood of exposure to climate change-induced drought. Being a member of the group increases the likelihood of exposure to drought e ffects by 120.2% (at *p* < 0.01 level). On the other hand, having a position of leadership of a group decreases the exposure to climate change-induced drought by 149.7% at *p* < 0.01 level (Table 5). A further investigation on the socio-economic status of the 1 to 5 group heads pointed out that they are were individuals with a better resource base. Furthermore, the group heads have more decision making power on resources owned by the community. On the other hand, the members of these groups were poor farmers who joined the group to ge<sup>t</sup> governmen<sup>t</sup> support. Despite the claims of the governmen<sup>t</sup> to support the poor organized in the groups, the members were most vulnerable compared with others. Also, the model examined agricultural technology use and the likelihood of exposure to climate change-induced impacts. The results of the model show that except for improved animal fodder, other agricultural technologies did not decrease the likelihood of exposure to climate change-induced impacts. Using chemical fertilizers (Urea and NPS) increases the possibility of experiencing the effects of drought and harvest losses. It increases the likelihood of experiencing drought effects by 150.4% (at *p* < 0.01 level). This is contrary to the claims of governmen<sup>t</sup> officials who argue using chemical fertilizers as rewarding for farmers. Interviews with farmers revealed that when RF onset is late, the application of chemical fertilizers is calamitous to farmers because the fertilizers wilt the crops in their early stages. Thus, applying chemical fertilizers during drought seasons increases their likelihood of exposure to the effects of drought. Moreover, those farmers who use chemical fertilizers are more likely to experience bankruptcy during drought years for they have to pay back the fertilizer loan, which is expensive according to key informants. Thus, the debt and risks associated with borrowing in precarious situations are found to be factors for the higher vulnerability of farmers.


**Table 5.** Marginal effects due to independent variables (the coefficients table).

> \* *p* < 0.1; \*\* *p* < 0.05; \*\*\* *p* < 0.01; (SE) = standard error.

#### 4.3.2. Determinants of Facing Harvest Loss

The logit model results show that several factors affect the possibility of exposure to harvest loss in the study area. The model is statistically significant for the determinants of harvest loss (N = 358, Chi<sup>2</sup> = 85.78, *p* = 0.0000). The pseudo R<sup>2</sup> and adjusted R<sup>2</sup> values are 24.44% and 87.75%, respectively (Table 5). Among the explanatory variables, membership and headship in 1 to 5 groups, PSNP, and

animal fodder control the likelihood of experiencing harvest loss. Also, the use of chemical fertilizer controls the likelihood of harvest loss. The PSNP is a social protection program that governmen<sup>t</sup> instruments in drought-prone and food insecure areas of Ethiopia. Because of the widespread food insecurity, the governmen<sup>t</sup> has been carrying out the PSNP in the study area. It aids vulnerable farmers in cash and kind in the study area. Of the sampled households, 29% (*n* = 114) were PSNP beneficiaries, whereas the remaining were not. The results of the model revealed that HHs supported by the program are less likely to experience the e ffects of harvest loss by 91.9% (at *p* < 0.01 level) than the non-users of the program. Like the case of drought exposure, membership in the 1 to 5 group increases the possibility of facing a harvest loss by 159.8% (at *p* < 0.01 level). On the contrary, bearing the leadership position in the 1 to 5 group reduces the likelihood of exposure to harvest loss by 200% (at *p* < 0.01 level). One of the challenges for farmers during the drought season is the lack of pasture and water for their livestock. To solve the problem, the governmen<sup>t</sup> has introduced a plant species that farmers grow to feed their livestock. Farmers call it animal fodder in the study area. The logit model analysis shows that the use of this improved animal fodder reduces the likelihood of facing harvest loss. Farmers who use improved animal fodder are 81.6% less likely to experience harvest loss than the non-users (at *p* < 0.1 level). This is because, without the fodder, farmers use crops to feed livestock. Using chemical fertilizers in crop fields increases the chance of experiencing harvest loss. Farmers who apply chemical fertilizers in their crop field are 37% more likely to experience harvest loss during a drought season because crops wilt under inadequate rain conditions (Table 5).

#### 4.3.3. Determinants of Exposure to the E ffects of Flooding

Flooding is one of the climate change-induced impacts experienced by the HHs in the study area. The model is statistically significant for the determinants of experiencing flooding (N = 357, Chi<sup>2</sup> = 39.62.78, *p* = 0.0958). The pseudo R<sup>2</sup> and adjusted R<sup>2</sup> values of the model are 9.58% and 75.7%, respectively (Table 5). Among the variables considered, land use right certificate, being a beneficiary of PSNP, and membership in the 1 to 5 groups are the determinants of exposure to flooding. Also, taking part in climate change adaptation decisions a ffects exposure to climate change-induced flooding (Table 5). Land use right certificate a ffects the likelihood of experiencing climate change-induced impacts. The likelihood of facing flooding decreases by 64.5% for farmers who have land ownership certificate. This result is statistically significant (at *p* < 0.1 level). This is because farmers with a land use right certificate have a stronger landownership feeling than those without and this encourages farmers to carry out di fferent soil and water conservation practices. Households with larger landholding size are less likely to experience flooding than HHs with a smaller landholding size. A unit increase in landholding size decreases the likelihood of experiencing flooding by 43.9% (at *p* < 0.05). Being a PSNP beneficiary also decreases the likelihood of facing climate change-induced flooding by 92.3% (at *p* < 0.01 level). On the other hand, membership in the 1 to 5 groups controls the likelihood of exposure to flooding impacts. Being a member in 1 to 5 group decreases the chance of experiencing flooding impacts by 85.2% (at *p* < 0.01 level). Participation in climate change adaptation decisions at the local level is also relevant. Farmers who take part in local level decision-making are 52% less likely to experience climate change induced flooding impacts than the non-participants (Table 5). The incident of flooding is common for low-lying areas, but the model did not consider topography and is a limitation of the model.

#### 4.3.4. Determinants of Facing Hunger

Climate change and variability-induced hunger are one of the impacts experienced by the HHs in the study area, especially during drought periods. The model is statistically significant for the determinants of experiencing hunger by the HHs (N = 357, Chi<sup>2</sup> = 82.49, *p* = 0.0000). The pseudo R<sup>2</sup> and adjusted R<sup>2</sup> values are 21.18% % and 83.25%, respectively (Table 5). The logit model result identified sex, educational status, farmland size, membership, and headship in the 1 to 5 groups as determinants of exposure to hunger. Besides, the use of improved animal fodder, chemical fertilizer, and improved seed variety also control the chance of exposure to hunger. Furthermore, female participation in family decision making also decides the likelihood of exposure to change-induced hunger. The sex of the household head is associated with the likelihood of facing climate-induced impacts; being a male household head decreases the chance of exposure to hunger by 55.9% (at *p* < 0.1 level). This means female-headed HHs are more vulnerable to the exposure of hunger than the male counterpart. This is due to issues related to access to and control over resources owned by the family and the community.

On the other hand, a unit increase in education decreases the likelihood of facing hunger by 11.6% (at *p* < 0.01 level). This suggests the role of education in minimizing the impacts of climate change induced hunger. Household's total farm size also lowers the chance of facing hunger. A unit increase in farm size reduces the likelihood of facing hunger by 57.1% (at *p* < 0.01 level). This means as land holding size decreases, farmers become more vulnerable to climate change-induced hunger. Like the case of exposure to drought, membership in the 1 to 5 groups increases the likelihood of exposure to hunger. The likelihood of facing hunger increases by 147.5% for members in 1 to 5 group and decreases by 142.7% for the heads of the group (at *p* < 0.01 levels). Using improved animal fodder decreases the likelihood of experiencing hunger. Users of improved animal fodder are 88.7% less likely to face climate change-induced hunger than non-users (at *p* < 0.1 level). Women's involvement in family decision-making such as selling assets and renting out lands a ffects the likelihood of exposure to hunger. It reduces the chance of experiencing climate change-induced hunger by 101.5% (at *p* < 0.01 level). The use of chemical fertilizers and drought-resistant varieties also a ffects the possibility of experiencing climate change-induced hunger. Using these inputs increases the chance of experiencing hunger by 142.9% and 69.7%, respectively (Table 5). Key informants' interviews to justify the reasons behind these findings suggested the use of chemical fertilizer and improved seed varieties is disadvantageous for farmers. This is because farmers who use agricultural inputs usually ge<sup>t</sup> them in the form of a loan. The farmers have to sell their assets to pay back the debt since there is no surplus produce to sell in the market during drought years. This further degrades their asset base, exposing them to food security.
