Nsanje is the southmost district of Malawi and lies in the Lower Shire River Valley, which is the most disaster-prone area of the country. The Valley, almost entirely, lies below 100 m above sea level on the Great East African Rift Valley floor. The district is adjacent to Chikwawa district and Thyolo district in the east, while it is surrounded by Mozambique state in the west, south, and east. The district is spread over a total area of about 1942 km², and the population is estimated at over 299,168 inhabitants [31]. Nsanje district is home to two Traditional Authorities (TA). The TA of Nyachikadza, a lowland community (wetland), and TA Ndamera, an upland community (dry-land) [32]. The majority of the population in the district is rural, with over 90% involved in agricultural activities, growing food crops (maize, millet, sorghum, rice, groundnuts, chili pepper, and cash crop (cotton) production [33]. Local communities are also engaged in other livelihood activities such as capture fishing from the Shire River, animal husbandry, petty trading, charcoal production, firewood, and grass (thatch) harvesting for sale. The cropping season runs from November to April. The district exhibits a subtropical climate with three distinct seasons: a cool and dry winter, which runs from May to August, a hot and dry season from September to October, and a warm and wet season from November to April. Mean annual temperatures range between −3˚C and −27˚C. Annual precipitation ranges between 600 mm and 1200 mm.

Source: Sundqvist (2023).

Figure 1.Overview map of the study area.

Using a participatory research approach, data were collected between October and December 2023. The approach involved administering individual semi-structured questionnaires. The household was the basic unit for research within each of the villages. This was through a door-to-door survey; no criteria were set for the selection of households to be assessed in each village. As such, the interviews were conducted individually; participation was random, regardless of the participants’ level of education or occupation. The individuals were interviewed by research assistants, who had prior knowledge of the study area. In instances where potential participants were unable to read and understand English, the Chichewa and Sena dialects were used by the research assistant to ensure proper understanding. Chichewa and Sena are the widely spoken languages in the study area. Traditional leaders, government, NGOs staff, and other opinion leaders working in the study area were purposively selected and interviewed. The information from this group (key informants) was primarily used to clarify areas that needed some explanation. For this reason, the questionnaires for farmers and key informants were similar. Before interviewing the farmers, the research team had to consult government officials in the area for official permission. A total of 385 interviews were conducted in the TA of Ndamera and Nyachikadza villages (a total of ten villages were visited). The survey collects information on household characteristics, value of agricultural output, agricultural assets, non-farm income-generating activities, access to credit, livestock owned, and use of irrigation. The dataset also contains information on climate shocks experienced by households and the coping strategies.

Descriptive statistics and multiple regression models were used to analyze the data. The Statastatistical package was used for the multiple regression models to estimate the impact of extreme weather events on farmers’ income. The multiple regression models were employed because they consider the linear relationship between one dependent and two or more independent variables, Equation (1), [34].

where, Y i is the dependent variable and X i is the set of independent variables. The parameter β 1 gives the log odds of the dependent variable and β 0 is a constant. ε i is the error term. Here, the dependent variable is the farm income. As defined by [24], it is the sum of all incomes from agricultural activities, encompassing revenue from selling crops, livestock, and rent from agricultural land, farm machinery, and agricultural wages. In Equation (1), endogeneity may be a problem due to the endogenous nature of some of the explanatory variables. ε i is the error term for which a strict exogeneity condition is assumed to hold. Errors are independently and normally distributed with zero mean and constant variance. The value of one variable is not affected by any other variable in this model at any time. Therefore, for this study, the multiple regression model used for estimation is given in Equation (2).

The description of the variables used in the regression model, along with their expected signs, is given in Table 1. The explanatory variables have been selected on the basis of insights from the relevant literature and the ground realities of the study area. The use of irrigation was included in the regression model because it enhances agricultural productivity, thus crop income. Furthermore, it can serve as a coping strategy against drought impacts by increasing crop yields and resilience in the face of changing conditions. However, in this article, we considered the variable as one of the determinants of the value of agricultural output, not as a coping strategy. Extreme weather events reduce the income of farming households; as such, we do not include the variable as a coping strategy. This is because most farming households lack the financial resources and knowledge to apply irrigation systems [35][36]. For the variable of weather shock, the study relied on subjective reports of measures of extreme weather events experienced by households. The study considered whether a household experienced flood, drought, and heavy rains during the 2022/2023 cropping season as indicators of climate risks. Therefore, households that reported having been affected by floods, droughts, and heavy rains during that period were considered for analysis, as described by [37] and [38].

Table 1.Description of variables and expected sign with the response variable.

Table 2 presents descriptive statistics for socioeconomic variables, with an average age of 38.22 years and an average household size of 4.28 members. Meanwhile, farmers had on average 3.72 years of education and a farm size of 0.89 hectares. Also, the average farm income was 11576.16 Malawi Kwacha (MWK), and 13% of the farmers used irrigation. Further analysis found 37% engaged in non-farm employment and 18% had access to agricultural extension services.

Table 2.Descriptive statistics of socioeconomic and shock variables.

Note: 2019 exchange rate was 1 USD = 789 MWK.

About 88.3%, 8.1% and 3.6% of the farmers had perceived an increase, a decrease, and no changes in flood frequency. Similarly, 76.1%, 19.1% and 9.8% had perceived an increase, decrease, and no change in drought frequency, respectively. Also, 20.5%, 68.3%, and 11.2% thought the frequency of heavy rains had decreased, increased, or remained unchanged (Table 3).

Here, the household adaptation strategies included measures adopted before (ex-ante) and after (ex-post) the occurrence of the different extreme weather events. As seen in Figure 2Figure 2, farmers adopted a number of measures before the occurrence of extreme weather events. However, about 30% of the farmers did not take any actions in advance to cope with climatic risks. The majority of the farmers did more ganyulabour.

Table 3.Perceptions of the farmers on weather shocks in the past ten years.

Figure 2.Ex-ante coping strategies.

The adoption of coping strategies to climatic risks after the occurrence of extreme weather events (ex-post adaptation) is presented in Figure 3Figure 3. The main farming household ex-post measures included ganyu labour, borrowing money, crop diversification, changing eating patterns, receiving remittances from friends/family, and planting early.

Figure 3.Ex-post adaptation strategies.

Ganyulabour is a kind of off -farm work done by individuals on a casual basis [39]. The key ex-ante adaptation mechanisms adopted by farming households in the study area wereganyu labour, crop diversification, planting crops early, planting drought-tolerant varieties, and planting crops adapted to waterlogging.

In Table 4, the findings indicate that the occurrence of droughts and floods has a significant and negative impact on farm income. This study also found a significant and negative relationship between household size and farm income. Furthermore, results show that there is a positive and statistically significant relationship between farm income and non-farm employment. Non-farm employment is economic activities not related to agriculture. Additionally, applying irrigation, years of schooling, access to credit, and farm size have a significant positive impact on the farm income.

Table 4.Regression results of the determinants of farm income.

Note: ***, **, *indicate significance of estimated coefficient at 1%, 5%, and 10% levels of probability, respectively.

Majority of the farmers did not adopt any strategy before the occurrence of the climate risks. The reason behind this was that they had limited information about the occurrence of specific types of extreme weather events. In addition, farmers lack financial resources that restrict them from adopting adaptation measures in advance. These findings are in line with the findings by [8], who observed that a lack of finance and limited warning information on the occurrence of specific extreme weather events restricted farmers from taking action in advance. The major ex-ante adaptation strategies included doing ganyulabour, crop diversification, planting crops early, and planting drought-tolerant varieties. However, farmers adapted to crop diversification, doing ganyu labour, and planting crops early as both ex-ante and ex-post adaptation measures. Crop diversification provides the farming households with a broader choice of growing a variety of crops within a given farmland. [40] also found that crop diversification expands production-related activities on various crops, reduces the possible risk of climate variations, and permits farmers to change crop types that are better suited to the prevailing climate regimes. Farming households did more ganyu labour in order to protect and prepare their crops from the extreme weather events. Ganyu labour is an important household buffering mechanism in response to low farming household income, especially in the case of low crop yields. According to [40], this practice ensures a constant flow in a household’s off-farm income, which buffers the impacts of climate risks on income. Furthermore, farming households that forecasted drought based on their previous experiences selected drought-tolerant crops. Many farmers who used drought-tolerant varieties posited that these varieties had a positive impact on their crop yield. This finding is in line with that of [41], who showed that farm households who were exposed to drought adopted drought-tolerant varieties. Farmers who have access to weather forecasting information often adjust to early planting of crops according to the occurrence of extreme events. [42] also reported similar findings. Another important ex-post measure was borrowing money, changing eating patterns, and receiving remittances from friends/family. Receiving remittances was able to minimize the negative effects on farming household income after the occurrence of weather events. Remittances are usually made by family members and other acquaintances who have temporarily or permanently migrated abroad or to a bigger town, or who live in the same location. According to [43], farming households that received remittances were able to offset the impacts of extreme weather events on agricultural income in the rural areas of Ghana. Farming households explained changing eating habits through either reducing the frequency of mealtimes a day. Some attest to reducing the quantity of each meal and skipping meals. Similar findings were reported by [44], who revealed that farm households changed their eating patterns as a coping strategy against climate risks in Southwestern Ethiopia. Also, borrowing money was found to be another strategy adopted by farmers to cope with extreme weather events. The source of borrowing money is mostly informal and entails relying on family members and friends during an emergency. The use of informal credit is common due to limited access to formal credit facilities at the local level. [37] also found that borrowing money was a household adaptation measure to buffer the effects of climate risks.

Being exposed to droughts has a negative and significant impact on agricultural income. This indicates that rising temperatures and changing rainfall patterns are detrimental to agricultural production, hence income. It may be due to the fact that higher temperatures and deficiencies in precipitation hamper crop growth. This usually occurs when rainfall does not meet crop moisture and soil water requirements, thereby reducing crop yields, thus the value of crop output. This result is in line with that of [45], who revealed that drought reduced crop yields due to inadequate water supply available for crop growth in the United States. With the application of irrigation systems, agricultural production losses could be reduced during a drought. Furthermore, floods had a negative influence on farm income. This is due to the fact that flood destroys and wash away crops and cropland. This finding is consistent with the work of [46], who also found that floods destroyed crop production of rural farmers in Northern Benin. The negative coefficient of heavy rains suggests that as heavy rains increase, farm income tends to decrease (an inverse correlation), which aligns with theoretical expectations that excessive rainfall can cause crop damage, soil erosion, rot of crops in the fields, extreme moisture content levels, and other problems. The positive and significant coefficient of access to credit signifies that an increase in credit improves a household’s financial capacity to expand agricultural production. For instance, credit facilitates households to purchase and apply crop production inputs as well as to timely implement farm management decisions. This observation is in line with that of the study carried out by [47]. They revealed that credit availability plays an important role in agricultural production in rural Ethiopia. Still, from a financial viewpoint, farmers involved in non-farm employment had an increase in farm income per hectare. This is due to the fact that income from non-farm employment may be used to purchase farm inputs such as fertilizer and seedlings so as to augment agricultural production. This observation is related to that by [48], who postulated that income from off-farm employment assisted in the purchase of farm inputs such as fertilizer and seeds in Ghana. On the other hand, off-farm income can be used to finance irrigation systems during droughts. Implementing irrigation may boost crop yields, thus bringing higher returns to the farmer [49]. The number of years of schooling has a positive and significant correlation with farm income. This implies that an increase in the educational level enables the farmer to be more skilful in the search for information and the application of new techniques of farming [50]. An increase in the age of the farmer has a positive and significant impact on agricultural income. It may be due to the fact that older household heads have the required experience to cultivate more crops and adapt to climate change. Experienced farmers have equipped most farmers with various farm management practices and techniques that can be used to increase crop yields and, hence, farm income [51]. An increase in farm size has a positive and significant impact on farm income. This means that with an increase in farm size, revenues from crop production increase. This is due to the fact that an increase in farm size enables farmers to diversify crops. This finding is in line with [24], who found that an increase in farm size permits farmers to diversify crop systems. The increase in the household size has a negative impact on the farm income. When a household is largely composed of dependents, the production input on participating in most farming activities may be outweighed by other household expenses. This is due to the fact that the dependents do not assist in any agricultural activities. This observation is similar to that by [52], who found that household size affects its agricultural activities.