Diversity of Sources of Income for Smallholder Farming Communities in Malawi: Importance for Improved Livelihood

: Agriculture is vital to global food production. Around 550 million smallholding households produce most of the world’s food, and many rely on livestock rearing for a living. Smallholder farms must survive and thrive to maintain and increase food production. Baseline information is vital for further extension service interventions. The goal of this Malawian study was to collect quantitative baseline data on crop and livestock production, agriproduct sales, and other indicators through a household survey, and to compare the efﬁcacy (in terms of income) of using the concept of “Lead and Follow” farmer training programs. The baseline study survey was carried out in 44 sections of 11 extension planning areas from Malawi’s ﬁve districts (Dowa, Kasungu, Mchinji, Mzimba, and Rumphi). In total, 1131 smallholder households were interviewed. Crop production, livestock farming, and providing casual labor for others were all identiﬁed as signiﬁcant sources of income for smallholders, implying that all agriproducts (the whole-farm approach) is equally important for improving smallholder livelihoods. On the one hand, the whole-farm approach should improve smallholders’ resilience regarding climate change and poverty. Lower agriproduct sales, on the other hand, indicated that links to the market were frequently poor but an increased market focus should help smallholders sell their produce at a fair margin. In terms of best practices adoption, both Lead and Follow farmers adopted similar farm practices (crops and livestock) to increase income. In general, no signiﬁcant difference in income was calculated from many farm enterprises for both Lead and Follow farmers. However, the income from pigs and ﬁrewood was signiﬁcantly higher for Follow farmers than for Lead farmers. Lead farmers reported signiﬁcantly higher off-farm income sources. Signiﬁcant changes are proposed to the “Lead farmer extension approach”.


Introduction
Agriculture provides the foundational basis for food supply [1,2], and this is particularly true in developing regions. Sub-Saharan Africa must significantly improve productivity in the agricultural sector to combat issues of rapid population growth and climate change. According to the United Nations, small-scale family farms account for more than half of the world's food production [3]. About 1.5 billion people live on smallholder farms, change and improvements in income among agriculture-dependent rural households. The TRANSFORM program is being implemented under the Malawian context, based on low incomes and high poverty levels. Over 80% of Malawi's rapidly growing population relies on subsistence rain-fed agriculture, and limited economic diversification is needed. While some development programs have had a tangible and measurable impact on household resilience, to date, none have reached a scale at which they make a significant impact on poverty levels. If the cycle of hunger and crisis is to be permanently disrupted, a truly transformational program is required.
The objective of this study was to collect quantitative baseline data on crop and livestock production, market access, and other indicators through a household survey. In addition, the effectiveness of the strategy of training Lead Farmers and Follow Farmers was checked. A Lead Farmer (LF) is defined as someone who motivates other farmers to try new technologies [20]; a Follow Farmer (FF) is one who observes the farming practices used by others and then selectively adopts those practices that, in their own opinion, will improve the productivity of their farm (further defined in the next section). The TRANSFORM program Consortium will use this baseline information to compare the project status before and after the implementation of TRANSFORM program activities. Identifying the areas of intervention in crops, livestock production and marketing that can improve the livelihoods of smallholder farmers was our main goal. An important aim of this study was to investigate the income gains based on the LF-led extension approach and compare the earnings of LF and FF groups. This will help to provide further directions for the TRANSFORM program.

Material and Methods
The baseline study survey was conducted in the 44 sections of 11 EPAs from 5 districts (of a total of 28 districts) of Malawi ( Table 1). The districts were selected to cover the TRANSFORM project. The five districts were: Dowa, Kasungu and Mchinji from the central region; and Mzimba and Rumphi from the northern region ( Figure 1). Table 1. List of Extension Planning Areas (EPAs) 1 .

District * EPA List of Sections in EPAs
Methodology: Quantitative data were collected by administering a semi-structured questionnaire to the households in the five study districts of Malawi. After pre-testing, the final questionnaire was structured to acquire information on basic demographics, household assets and income sources, food production and availability, livestock production, marketing, agro-processing and value addition, the adoption of climate-smart agriculture technologies, and access to credit/loans. In this study, data related to the demographics, crop and livestock production, sources of livelihoods, and income comparison for LFs and FFs were analyzed and presented. The TRANSFORM program's partners, and consortium members, had previously worked in Malawi on LF extension programs (such as the Sustainable Agriculture Lead Farmer Program). In a Malawian study, Fisher, Holden [21] used the term "Lead farmers" to refer to farmer trainers. In a report, Regine Andersen [22] has described the detailed procedure for the selection of LFs in Malawi. The income disparity between existing LFs and FFs was used to verify the efficiency of the strategy. The survey questionnaire asked if the respondent was an LF or not. Methodology: Quantitative data were collected by administering a semi-structured questionnaire to the households in the five study districts of Malawi. After pre-testing, the final questionnaire was structured to acquire information on basic demographics, household assets and income sources, food production and availability, livestock production, marketing, agro-processing and value addition, the adoption of climate-smart agriculture technologies, and access to credit/loans. In this study, data related to the demographics, crop and livestock production, sources of livelihoods, and income comparison for LFs and FFs were analyzed and presented. The TRANSFORM program's partners, and consortium members, had previously worked in Malawi on LF extension programs (such as the Sustainable Agriculture Lead Farmer Program). In a Malawian study, Fisher, Holden [21] used the term "Lead farmers" to refer to farmer trainers. In a Sampling techniques: The sampling process for the baseline survey was designed based on the selected EPAs in the program target districts. The study used a multi-stage sampling process to determine the sample of farming families that participated in the household survey. This included purposive, stratified, and simple random samplings. The rationale for adopting multi-stage sampling was to target the potential project beneficiaries across a wide geographic area over the five study districts (Table 1).
Firstly, purposive sampling was adapted for all the project districts where the project has been implemented. Secondly, a stratified sampling technique was used, with extension planning areas (EPAs) as strata. Villages within the targeted EPAs were selected based on a probability proportional to size (PPS), using the size and number of villages in each EPA. Two EPAs were selected in each district except Mzimba, where 3 EPAs were selected, based on the subdivision of the district, into southern and northern zones. Thirdly, a simple random sampling technique was used to select farming households from each participating village to participate in the baseline survey. The survey participants were from a list of farmfamily households obtained from the Agricultural Extension Development Coordinator (AEDC) of the respective EPA.
Sample size (SS): To capture the current demographic, socio-economic, agronomic and nutritional status of the participants in the project, the study unit for the survey was the individual farming family household. The number of households included in the study was determined using the following statistical formula (from Creative Research Systems, Sebastopol, CA 95472 [23]): where Z is the z-score used for creating a 95% confidence interval, p is the proportion picking a choice, expressed as a decimal (0.5 was used for the sample size needed), and C is the confidence interval expressed as a decimal.
Electronic household data collection and capturing: The KoBo Software (KoBo Toolbox at the Harvard Humanitarian Initiative, Cambridge, MA 01138, USA) was used to upload, collect and manage the household data collection. The pre-programmed questionnaire was loaded onto tablets for data collection, and the enumerators were instructed to upload the data onto a server. Most of the 15 research assistants collecting the household data were graduates who were able to speak the local dialects, Chichewa and Tumbuka, so that they were able to communicate well with the interviewees. They underwent a 3-day training course to understand the survey questionnaire and other operational tasks, such as data collection using KOBO and the ethics required during data collection. Before starting the actual data collection, the questionnaire was pre-tested in Waliranji Village (district Mchinji), which was not part of the selected EPAs used for this baseline study. In all, 1131 questionnaires were completed and considered for further analysis.
Data analysis: Welch's two-sample t-tests were used to compare numerical variables between two groups. For the comparison of numerical variables from more than two groups, an analysis of variance followed by Tukey's honest significant difference (HSD) tests were used. The level of significance and the family-wise error rate was set to 5%.
Income sources were grouped based on the number of people who relied on each source, and the average income generated by the corresponding income source was calculated. This means that the income sources shown in the tables below involved the majority of people and generated the most income.
Participation in the study was voluntary. Before each interview, either at the household or community level, a brief introduction about the purpose of the study was given. Verbal consent for participation was sought before starting the survey. Similarly, participants were assured of confidentiality and that the information collected would only be used for the purposes of the survey. Interviewers were also instructed to be neutral and were trained to respect the respondents' dignity and culture.

Results and Discussion
The results of this baseline study were limited to the following: basic demographic characteristics, smallholder agriculture produce, livestock ownership and animal products, ranking of important income sources for smallholder farmers, mean income from all kinds of sources for smallholders across all five districts, and mean income comparison between LFs and FFs (all kinds of income) were presented.
In this survey, 663 women (58.6%) and 468 men (41.4%) were interviewed (Table 2), including 56 LFs, implying that one LF is available for every twenty FFs in the study districts. All respondents were smallholder farmers, so the average land size across all districts was 2.7 hectares. In all districts, only 18.7% (n = 211) of farmers hired labor for farm activities. Rumphi district had the highest percentage (26.5%, n = 56) of farmers involved in hiring labor among the five study districts. The baseline data (Table 1) revealed the variety of different sources of income for smallholder farmers.

Crops and Vegetables Growing
The surveyed smallholder farmers were growing crops of many kinds. Maize: Most (99%) of all surveyed smallholder households who were growing any crop or vegetable reported growing maize. Maize flour is used as a staple food in Malawi.
Soya: With 52% of the surveyed households reporting growing soya, this was the second most popular crop. Households from Mzimba South had most (25%) of the soya growers in the survey, while smallholder farmers from the Kasungu (24%) and Dowa (18%) districts ranked second and third, respectively.
Groundnuts: Groundnuts were the third most popular crop, with around 44% of smallholder farmers growing them. Dowa district had the most (25%), while smallholder farmers from Mchinji (24%) and Rumphi (19%) were the second and third most important groundnut growers in the survey, respectively.
Tobacco: About 20% of all surveyed farmers reported growing tobacco. The Rumphi district contributed 40% of these, while Dowa and Mzimba North provided 25% and 15% of these, respectively.

Livestock Ownership, Sales of Livestock and Livestock Products
Livestock provides an important component of the livelihoods of smallholder households ( Figure 2). Different classes of livestock were found, including dairy cattle, crossbred goats (local goats and Boer goats), guinea fowls, rabbits, and crossbred chickens (local chicken with Black Australorp). The prevalence of each of these classes in each district was assessed.

Livestock Sale
Livestock sale was widespread across the sampled districts, with slightly over 60% of the respondents owned livestock and/ or had livestock sold in the past 12 months. Among smallholder farmers that owned livestock in the respective districts, livestock sales were the most common in Dowa (67%), Mchinji (62%), Rumphi (61%) and Kasungu (61%), and the least common in Mzimba (54%). Overall, 67% of the households owned some type of livestock (Table 3). In Rumphi, this figure was 81%, in Mzimba North, 78%, and in Mzimba South, 71%. In contrast, only 92 (45%) of the 205 respondents in Mchinji owned livestock. The communities reported that the variation in livestock numbers is largely due to drought and the incidence of diseases, such as Newcastle disease in chickens. For example, households sold a high proportion of their livestock in the second half of 2019 because of major feed shortages and a lack of market infrastructure, poor veterinary services, and access to improved breeds. An overview of livestock ownership is shown in Figure 2. The majority of livestock holders (85.2%) owned only one or two kinds of livestock, while 14.1% of farmers were able to generate income from 3-4 different classes of livestock. A very small proportion (0.7%) owned five or more livestock types ( Figure 2). Among all districts, farmers from Rumphi, Mzimba South and Dowa districts owned the highest livestock numbers.

Livestock Sale
Livestock sale was widespread across the sampled districts, with slightly over 60% of the respondents owned livestock and/ or had livestock sold in the past 12 months. Among smallholder farmers that owned livestock in the respective districts, livestock sales were the most common in Dowa (67%), Mchinji (62%), Rumphi (61%) and Kasungu (61%), and the least common in Mzimba (54%).
A reason for livestock sales being somewhat more common in Dowa and Rumphi (close to the central and northern capitals, respectively) could be that a number of fastgrowing trading centers (such as Mponela) are located there, offering an active and competitive livestock market for farmers. Dowa is also closer to the capital city of Lilongwe, while Rumphi is closer to the northern capital of Mzuzu, where restaurants and hotels have a high demand for local chicken and cattle meat. Farmers in other districts sold chickens and other livestock during the lean dry season (February-March) to be able to buy staple foods to feed their families.

Livestock Product Sale
Livestock products were milk, meat and eggs, with Rumphi district being a leading source of these to service the growing market in Mzuzu (the northern capital city) ( Table 2). However, very few households (29 of 1131) reported selling livestock products.
The livestock ownership and sale of livestock in each district are presented in Table 3.

Important Sources of Income for Malawian Smallholder Farmers
The smallholder farmers were asked to prioritize their sources of income to support their livelihoods, such as crop production and sale, livestock rearing and sale, off-farm employment, petty trading/business, artisanal skills, entrepreneurship, and social cash transfer. The majority of the respondents ranked "crop production/sale" as the most important source of livelihood, followed by "off-farm employment" and "livestock production/sale" (Figure 3). Within the agri-production system, livestock was ranked as their second most important source of income.

Important Sources of Income for Malawian Smallholder Farmers
The smallholder farmers were asked to prioritize their sources of income to support their livelihoods, such as crop production and sale, livestock rearing and sale, off-farm employment, petty trading/business, artisanal skills, entrepreneurship, and social cash transfer. The majority of the respondents ranked "crop production/sale" as the most important source of livelihood, followed by "off-farm employment" and "livestock production/sale" (Figure 3). Within the agri-production system, livestock was ranked as their second most important source of income.
Means of incomes per household from livestock (animals and products), forestry products, fishery and small business operations were not significantly different across districts (Table 4 and Figure 4). The farmers of Rumphi had the highest average income (MWK 257,138) from crops, which comprised 66% of their total income (MWK 387,934). People from Mzimba North and Rumphi districts derived more revenue from crops than farmers from Mchinji and Kasungu. In general, crops, fishery, and small businesses of any kind at the local level provided a higher income than other sources in all districts.
(a)   Means of incomes per household from livestock (animals and products), forestry products, fishery and small business operations were not significantly different across districts (Table 4 and Figure 4). The farmers of Rumphi had the highest average income (MWK 257,138) from crops, which comprised 66% of their total income (MWK 387,934). People from Mzimba North and Rumphi districts derived more revenue from crops than farmers from Mchinji and Kasungu. In general, crops, fishery, and small businesses of any kind at the local level provided a higher income than other sources in all districts.

Mean Income Comparison between LFs and FFs Producing Crops and Vegetables
The incomes of Lead and Follow farmers were compared on the basis of average quantity produced (kg), average quantity sold (kg), unit sale price (MWK) and an average value of all the commodities ( Figure 5). LFs were on average producing more of all crops

Mean Income Comparison between LFs and FFs Producing Crops and Vegetables
The incomes of Lead and Follow farmers were compared on the basis of average quantity produced (kg), average quantity sold (kg), unit sale price (MWK) and an average value of all the commodities ( Figure 5). LFs were on average producing more of all crops and vegetables except sugarcane and banana. FFs were selling a larger quantity of cassava, banana, Irish potato, and sugarcane compared to LFs. There was a difference in the quantities of crops and vegetables produced and sold, which could be based on the fact that some agriproduct was consumed at home. FFs gained higher sale prices for vegetables, bananas, Irish potato, cassava, soya, groundnuts and sugarcane. The cumulative income from bananas, cassava, and sugarcane was higher for FFs, while for vegetables, sweet potatoes, Irish potato, maize, beans, groundnut, soya, and tobacco income was higher for LFs ( Figure 5).

Mean Income Comparison between LFs and FFs Producing Livestock
For livestock numbers and income, among all the surveyed households, the proportion of FFs keeping livestock was higher than for LFs. LFs, in contrast, on average maintained larger herds, with the exception of rabbits. Both classes of farmers were selling more chickens. FFs were selling more rabbits, pigs and cattle, while LFs sold more goats. The cumulative income gained from animal sales was higher for FFs in the cases of rabbits, pigs and chickens, while LFs made more from goats ( Figure 6). In particular, the average income gained from selling pigs for FFs was significantly higher than that of LFs (p = 0.0009; Table 5). An overview of the average income (in percent) comparison (between LF and FF) from various sources is presented in Figure 7.

Mean Income Comparison between LFs and FFs Producing Livestock
For livestock numbers and income, among all the surveyed households, the proportion of FFs keeping livestock was higher than for LFs. LFs, in contrast, on average maintained larger herds, with the exception of rabbits. Both classes of farmers were selling more chickens. FFs were selling more rabbits, pigs and cattle, while LFs sold more goats. The cumulative income gained from animal sales was higher for FFs in the cases of rabbits, pigs and chickens, while LFs made more from goats ( Figure 6). In particular, the average income gained from selling pigs for FFs was significantly higher than that of LFs (p = 0.0009; Table 5). An overview of the average income (in percent) comparison (between LF and FF) from various sources is presented in Figure 7.   Table 5 shows that income from most other sources and businesses was not significantly different between LFs and FFs. Significant differences were found for firewood (p = 0.0276) and beer-brewing (p = 0.0038).

Mean Income Comparison between LFs and FFs Having Income from Other Sources
Overall, LFs often had a wider range of income sources (6 vs. 4, p < 0.0001) and the cumulative income from all sources was significantly higher (469343 vs. 239837 MWK, p = 0.0006) than that of FFs (Table 5).   Table 5 shows that income from most other sources and businesses was not significantly different between LFs and FFs. Significant differences were found for firewood (p = 0.0276) and beer-brewing (p = 0.0038).

Mean Income Comparison between LFs and FFs Having Income from other Sources
Overall, LFs often had a wider range of income sources (6 vs. 4, p < 0.0001) and the cumulative income from all sources was significantly higher (469343 vs. 239837 MWK, p = 0.0006) than that of FFs (Table 5).  The utility of LFs as a mechanism for agricultural extension should facilitate knowledge transfer to a higher number of FFs and, thus, increase the adoption of best practices in any specific area. However, this study mostly showed that existing LFs were similar to the FFs and had not improved significantly in regard to adopting best practices. Ragasa [24] reported that the majority of smallholder farmers gave a good rating to all questions about production performance, including questions about the performance of LFs in terms of conveying messages, but a major discrepancy was found when LFs were double-checked for their adoption rate.
Due to their low adoption rate for best agricultural practices, there was not much difference between Lead and Follow farmers' income from farm activities. However, the current study found statistically significant, higher total income (average) for LFs due to off-farm activities such as beer brewing (p = 0.003). This is consistent with the findings of Ragasa [24], and Holden, Fisher [25]. Ragasa [26] checked the effectiveness of the LF approach for technology awareness and adoption by modeling the data of 531 randomly selected LFs. To gain a higher adoption rate, LFs should adapt and implement the best practices on their farms (hence, receiving better economic gains) before conveying their messages to the FFs [26]. In the current study, higher income from off-farm activities revealed that LFs were not implementing best agricultural practices, resulting in lower profitability and less interest in farm activities. Rather than simply delivering "extension messages" to FFs via LFs or extension workers, the focus of the "Lead farmer extension approach" should be on boosting the adoption of best agriculture practices to have a longterm impact on smallholder livelihoods. As a result, when FFs will witness best practices being implemented in the field, they are more inclined to adopt them. Following that process, documenting the adoption of those best practices in the form of "success stories" will have a ripple effect, resulting in increased adoption by the FFs.
Additionally, it is understood that since (in the case of Malawi) the majority of the LFs and FFs are illiterate, concise and innovative, visualized messages, such as diagrams and storytelling, should be used consistently and as often as possible. The purpose of all research-based extension messages and other interventions is to improve production, profitability, and livelihoods. Research institutes are vital to the development of extension materials and protocols, keeping in mind the priority areas of smallholders for extension. Therefore, based on the results (Figures 5 and 6), a strong focus should be placed on linking smallholders with the private sector to increase the profitability of agriproducts. This is the first study of its kind from Malawi that highlights the importance of farmer activities in maintaining the sustainability of their farming enterprises, particularly for small-scale farmers and especially in terms of coping with climate change and increasing production and profitability. The model could be replicated in developing countries where the majority of the population lives in rural areas and where agriculture contributes a significant portion of the country's GDP.
This baseline study serves as a foundation for potential future interventions across Malawi's five districts as part of the TRANSFORM program initiative (2020-2024). A follow-up study should be conducted with a greater emphasis on the linkages established between smallholders and operators in the private sector to improve their livelihoods.

Conclusions
The surveyed smallholder households were subsistence farmers with limited direct market orientation and access. Crops, vegetables, and livestock production are all important for smallholders' livelihoods, and some agriproducts are consumed in the smallholders' household. Adopting best practices to increase agriproduct production will increase sales and, thus, profitability. A significant portion of respondents did not report any sales whatsoever. The sales of both raw and value-added products are often lacking. Among the farm activities, crops were the most important, followed by livestock and off-farm work. Groundnuts, soya beans and maize were essential cash crops when considering both the quantities produced and the market price value. For livestock, cattle production gave the highest return, followed by pigs, goats, and poultry. Farmers mostly sold live animals, but not their produce (meat, milk, processed commodities). Thus, there is an absence of product value-addition and contemporary production skills. The focus should be on commodities that bring the highest income for smallholders, coupled with a needs assessment, systems approaches and farmer-participatory decision making. Farmers rely on a diverse range of products for their livelihood; rather than promoting specific interventiaons, the emphasis should be on the whole-farm approach. The focus of the "Lead farmer extension approach" should be on the adoption of best practices and increasing the long-term impact on smallholders' livelihoods. Furthermore, there is a need to tailor research-extension messages to accommodate the limited education of the farmers.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. TRANSFORM program has all the rights and ownership of all data collected in this study.