1 Introduction

In light of the current climate and food crises and the associated alarming reversal of poverty alleviation (World Bank, 2020), there is an urgent need for land use policies to build the resilience of smallholder farmers. Three billion people, roughly 75% of the world’s poor, live in rural areas and derive the major portion of their incomes from agricultural activities (World Bank, 2018, 2020). In India, the agricultural sector provides direct employment to more than 50% of the nation’s workforce. Small and marginal farmers with landholdings of less than 2 ha comprise 82% of all agricultural labor, reflecting high land fragmentation and low economies of scale (FAO, 2022). Mainstream extension and agricultural support services fail to meet farmers’ needs, limiting their capacity to access improved production inputs and technology (Singh et al., 2020). A typical family practices rain-fed agriculture, with few inputs and technologies. Unable to ensure their livelihoods with their meager harvests, most farmers migrate seasonally with their families to work elsewhere, typically under exploitative conditions, for several months every year. They are caught in a poverty cycle that perpetuates the low standard of living across generations (Asadullah & Yalonetzky, 2012; Mehta & Shah, 2003) and persists (Thorat et al., 2017), even though the Government of India follows affirmative action policies and despite Indians in general profiting from increases in wealth.,

Many programs and efforts have been initiated in the past to eradicate rural poverty, but the results have often been disappointing (Bhaumik & Chadha, 2008; Bjornlund et al., 2020; Pittock et al., 2020; Roumasset, 2007; Wiggins, 2014), leading to growing dissatisfaction with the performance of many of the agricultural projects and programs (Roumasset, 2004). Development practitioners widely agree that rural agricultural projects are among the development projects that are the most challenging or even prone to fail. In their survey of the past century’s literature on agricultural development, Barrett et al. (2010) concluded that knowledge about how to foster rural development remained incomplete, limiting the ability to pursue it.

Especially in challenging places where highly vulnerable people and highly degraded lands are interconnected by the ‘natural resource degradation and human vulnerability nexus,’ projects might exacerbate the vulnerability of poor resource users if they fail to involve the people or benefit some families while harming others (Coleman et al., 2021; Ramprasad et al., 2020; Rozelle et al., 2000). In an evaluation of their own projects that apply natural resource management practices, the World Bank concluded that they failed to sufficiently tackle the vulnerability of people, who depend on natural resources (Independent Evaluation Group, 2021). A study by Turner (2021) examined projects as part of the larger goal of the Great Green Wall, a visionary effort aimed at restoring depleted lands and minimizing the vulnerability of the poor smallholders in the Sahel. It revealed that some of the most vulnerable were either neglected or excluded, and overall, little attention was paid to the improvement of their livelihoods. Bhattacharya et al. (2010) examined the insights gained over twenty years of Joint Forest Management (JFM) in India. A major drawback was the lack of people’s participation. Worsdell and Sambhav (2020) explained that afforestation efforts in India resulted in the forced eviction of communities from their traditional lands, leading to homelessness and adverse impacts on half a million tribal and forest-dwelling people in India. Moreover, in global mitigation policy, the unfavorable trade-offs with food security are frequently overlooked (Doelman et al., 2020). Oxfam (2021) states that commonly used offsetting measures such as reforestation and planting new forests can put food security at risk. Instead, it is better to use nature-based solutions like agroforestry, because these sequester carbon and at the same time allow people to use the land for food.

As there are only a limited number of examples of agricultural initiatives that have achieved long-term success, and due to continued doubt regarding financial backing for rural farming programs, evidence of potential success is urgently needed. “Failing to learn the lessons of the history of development policies may have doomed us to repeat them” (Roumasset, 2007, p. 36). Despite the fact that the past decade witnessed a movement toward evidence-based policy and increased emphasis on learning from evaluations (Maredia, 2009), the available literature to date provides little systematic guidance for agricultural policymakers. Standard evaluations of projects managed by development banks are mostly conducted in the form of short field visits and rapid rural appraisals, which cannot be compared to scientific analysis based on quantitative data. Furthermore, most impact assessments are small-scale, short-term or very context-specific types; only a few reports try to generalize findings and provide evidence of what works in agricultural development, such as Millions Fed: Proven Successes in Agricultural Development (Spielman et al., 2009). Despite billion-dollar investments, there is relatively little knowledge about the net impact of most programs (Center for Global Development [CGD], 2006). As the Development Impact Evaluation (DIME, 2021) states, “[t]he Sustainable Development Goals (SDGs) urge the international community to make the investments needed to double agricultural incomes of small-scale food producers. However, current yield trends suggest a need for path-breaking innovations to come to the rain-fed areas of the world to meet this target. Astonishingly little evidence exists to rigorously inform the investments needed to meet this urgent goal.”

We address this knowledge gap by rigorously scrutinizing the Wadi program, which has anecdotal evidence of success and is a potential bright spot in this otherwise bleak picture. Can it transform the livelihoods of chronically poor smallholders? Initiated in the early 1990s, it encourages integrated farming practices by developing water resources and orchards and has since been rolled out in several Indian states, covering geographically diverse regions. There is reason to expect that the Wadi concept may provide guidance for future development efforts, which we describe in the next section. However, no scientific assessment of impacts has been undertaken so far. We use mixed methods, including a survey of 1,860 randomly sampled households (from over 188,000) in Rajasthan, Gujarat, Maharashtra and Karnataka, to assess both the impacts of the Wadi program and the underlying causal relations that likely drive them. To foreshadow our findings, we discover significant improvements across a variety of parameters—higher incomes, higher life satisfaction, perceived positive life changes, as well as better long-term perspectives—when comparing household participants of the Wadi program with non-participants. The objective of our study is to provide the much-needed evidence of program success to inform the science-policy dialog about development programs by rigorously assessing the Wadi program’s potential to help farmers escape chronic poverty and to synthesize key lessons. This way, it can serve as a role model for future agricultural development programs.

2 Background: integrated farming and the Wadi program

In terms of indirect evidence, there is reason to believe that integrated farming systems, such as those promulgated by the Wadi program, generally play an important role in rural agricultural development. Argade and Wadkar (2013) reviewed the literature on integrated farming systems’ contribution to the achievement of sustainable rural livelihood security and noted the inevitability of a paradigm shift in research toward small and marginal holders and the role of integrated farming systems. They pointed out that almost all Indian farmers adopted some sort of integrated farming system; however, without scientifically designed and appropriate farming models, they remained mostly subsistent and rather uneconomical, unable to harness the benefits. Especially for small and marginal farmers, integrated farming is considered the most powerful tool for sustainably enhancing prosperity, employment and nutritional security, as well as fostering regional economic growth and resilience to climate change (Argade & Wadkar, 2013; Singh & Dhayani, 2014). Sharma et al. (2017) assessed the challenges and opportunities of agroforestry systems in India. They did not explicitly mention the Wadi program but included agri-horticulture, defined as “fruit orchards/fruit tree-based cropping systems” as well as “horti-pastoral systems in hilly orchards for soil conservation” (based on Dagar et al., 2014). There is the possibility to increase the agroforestry area, particularly on degraded land and wasteland, according to the Central Agroforestry Research Institute (CAFRI, 2015). But even so, there has not been enough study done on agroforestry concepts. In India, most studies are of short-term nature and have been conducted at research stations on relatively small plots and/or in laboratories, with few ecosystem or landscape-level studies (Sharma et al., 2017).

In this study, we analyze the Wadi program in India. Wadi translates to “fruit orchard” in the Gujarati language, since the Wadi program first took hold in Vansda, Gujarat, in the 1990s. From then on, the concept has been widely replicated in 12 districts of Gujarat, 18 districts of Maharashtra, 15 districts of Rajasthan and 8 districts of Karnataka, with almost 200,000 participating households.

The program's design begins by acknowledging the strong emotional connection smallholders have to their land. Typically, each family receives assistance for four years to establish a Wadi on one acre of land that is considered unsuitable for rain-fed crop cultivation. The main focus of the program is to assist families in growing fruit trees. The families are consulted on which fruit tree varieties, such as mangoes and cashews, to plant, based on the local climate and market potential. Usually, a block of 50 to 60 fruit trees is planted with seasonal intercrops in between. To protect the area from open-grazing cattle, the block is enclosed by 250–400 multipurpose forest tree species and a thorny live hedge, as shown in Fig. 1. Depending on the location, family housing and the water source might be within the Wadi or in its surroundings.

Fig. 1
figure 1

Schematic depiction of a Wadi

Full size image

Because these plantations are set up on degraded land in places prone to soil erosion and water shortages, soil and water conservation, as well as water resource development, are critical. Despite receiving moderate to high rainfall, most precipitation is limited to the monsoon season, and due to the steep and rocky terrain, most of the water is lost, potentially washing away the topsoil in the process. Soil conservation measures have the objectives of stopping further degradation and improving soil fertility. By reducing the velocity of runoff water, the water has more time to trickle into the ground and recharge the aquifer underneath. The intent is to reduce the problem of drinking water scarcity, as well as to ensure a sufficient supply of water to the cultivated plants, especially during the growth stage in the summer. Depending on the site-specific needs, new decentralized water resources are developed, or existing ones rehabilitated. The measures are planned through interactions with the respective farmers and customized to each Wadi plot.

Beyond tree planting and water conservation and development, another important measure is capacity building. This reflects the fact that smallholders have rich traditional knowledge but are not acquainted with the management of horticulture and cash crops. Capacity building is implemented through regular training sessions and field demonstrations. This also involves creating cooperative community organizations to more effectively address difficulties in marketing agricultural goods and to guarantee fair prices. These are set up and guided in aggregation, marketing, value addition and forward integration. This creates local job opportunities, such as mango trading or cashew processing, from which landless people profit as well.

One more critical component is the initial financial assistance. Due to the fairly long gestation period of 3–5 years, during which no considerable revenue is expected, except from intercrops (implying an irrigation system is available), farmers get compensation for working on their own land. This allows them to stay on their farms rather than turning to seasonal migration for subsistence (Table 1).

Table 1 Measures of the Wadi program
Full size table

So far, no rigorous scientific study of the Wadi program’s impact and effectiveness exists. However, anecdotal and indirect evidence suggests that the program has been successful. Trebbin and Hassler (2012) examined the actions of a cooperative established under the Wadi program, which was involved in processing and marketing mangos and cashews in a specific region. They conducted a set of 45 interviews with key stakeholders, including six members of the cooperative. Overall, they found that the stakeholders were positively impressed with the actions of the cooperative, and this reflected well on the Wadi program more generally.

3 Methods

3.1 Mixed methods

We employed a mixed-methods approach, beginning with qualitative evidence and progressing to quantitative evidence to determine important findings before concluding with further qualitative evidence to expand our knowledge of the mechanisms at work.

We began with field visits, participatory rural appraisal, participant observation and focus group discussions with village representatives and women's self-help groups. We also used techniques that accounted for illiteracy, such as ranking and drawing. We conducted interviews with various stakeholders, including participants, non-participants, local and international experts, NGO representatives, ministers, health professionals, teachers, bank professionals and representatives of tribal institutes and development agencies. Based on this qualitative data, we developed a quantitative survey that was conducted in four Indian states, face-to-face. The survey lasted between one and two hours and asked questions about several socioeconomic indicators and aspects of farming, food security, seasonal migration, education and well-being. We also included questions about the respondent's capability to understand of the questions, the geographic context and an open question for remarks and GPS data collection—for additional insights, validity and reliability checks. This questionnaire was developed in an iterative process, including rigorous field-testing and translation. We conducted two training sessions for the enumerators, who were familiar with the local dialect and context. As part of the training, they practiced how to use the Open Data Kit (ODK) software (Hartung et al., 2010) on their smartphones. By utilizing mostly (closed) multiple-choice questions and enabling direct digital input of the data, potential errors and the need for extensive data cleaning were minimized. After the survey was completed, the interviewers discussed their personal experiences, challenges and solutions in a report. During analysis, we conducted further interviews with stakeholders or experts as needed for verification or clarification.

4 Research design and sampling

First, we explain here the implementation procedure and program’s rules for selecting participants because these rules allowed us to pick a robust method and counterfactual group. The Wadi program was gradually implemented in different geographical areas, i.e., in form of a randomized rollout. More precisely, all eligible beneficiaries in a particular village or region were covered, before transitioning to the next one. The program was introduced by organizing meetings in villages and hamlets and facilitating exposure visits, ensuring that all eligible participants were provided with an equal chance to participate. Automatically, this system has produced conditions comparable to a randomized evaluation design that can provide good internal validity and can rely on weaker assumptions compared with other methods (Gertler et al., 2016). All villages situated in the program zone were provided with the option to enroll in the program during a particular year, whereas the villages located outside the area had to wait for the program to expand further. As a result, these households and villages located immediately beyond the program's coverage area, who had not yet been given the chance to participate, were used as control villages. Consequently, we were able to specify three different control groups (control villages, recent participants and non-participants in Wadi villages) to compare with the treatment group, as depicted in Fig. 2 and described.

Fig. 2
figure 2

The research design consists of one treatment group (n = 1132) and three control groups: recent participants (n = 202), non-participants in program villages (n = 380) and households in control villages (n = 146)

Full size image

Recent participants: These randomly sampled households are already enlisted as participants but have not received any benefits yet. There is no self-selection bias as they are technically participants; however, the program has not had any impact yet, which makes them an ideal counterfactual group.

Non-participants in Wadi program villages: These are households that were randomly selected and who chose not to take part in the program when it was introduced in their villages.

Households in control villages: These households were randomly sampled from villages situated 5–20 km away from the program areas. The choice of control villages was made after consulting with local experts and based on specific criteria to ensure that their circumstances were similar to those of the program villages before the program began and to take into account any external factors.

The survey was conducted among 1860 households, which were randomly sampled from a total population of 188,231, made up of all households that participated in the program across four states. Of the 1860 households surveyed, 1132 were program participants, while 728 were non-participants. For the sampling of the participants, the lists from the project implementation were available; for the non-participants, the lists of inhabitants were provided by each village administration. Random sampling was conducted in Excel. In case the residents of a sampled household could not be located during the first and second visits of the interviewer or refused to participate in the survey, a backup shortlist of randomly sampled households was prepared to ensure that the required number of households was interviewed.

4.1 Analysis

We conduct a number of analyses to fully explain the causal relationship between program participation and its effects. First, we analyze income and the number of income sources by comparing treatment and control groups (using the t-test and the Wilcoxon rank-sum test). Then, we analyze the increase in income and income sources over time. We appreciate the fact that the program has been implemented in different villages at different times over a 23-year period because this makes it possible to analyze a trend in possible impacts of the program by (ceteris paribus) comparing farmers who have established their Wadis at different points in time—from 1995 to 2017. In this assessment of development over time, the time course is represented by the explanatory variable years since participation, where control households are considered to have 0 years since participation and each treatment household has its specific number of years since participation. This enables an understanding of the trend as a development pathway and constitutes a valuable addition to the classic statistical analyses of treatment versus control groups. Furthermore, we use multiple regression analysis to control for other factors that are potentially confounding. With this, we analyze income and the number of income sources, as well as positive life changes over the last 5 years, life satisfaction and perceived opportunities for children. These findings are based on questions, answered on a 5-point Likert scale, to examine the farmers’ perceptions on whether (1) life has changed in the last 5 years, (2) they are satisfied with life, and (3) their children and grandchildren will have more opportunities than they have had themselves. Any statistical result is cross-checked with our qualitative findings and verified with stakeholders, while our quantitative analysis confirms the qualitative conclusions.

5 Results

We want to know whether the Wadi program can transform livelihoods of chronically poor smallholders. For these farmers, income is a key factor of well-being, as it can help fulfill many urgent needs, such as food, shelter or education. Our statistical analysis and field verification showed that the program participants had increased their incomes. On average, they generated an annual net income of INR 73,000 (EUR 920) compared with INR 40,000 (EUR 500) without intervention (t-test, t (1774.2) = − 12.96, p < 0.001). Since outliers in income were found in both groups, the test results were also confirmed by nonparametric testing (Mann–Whitney (W) = 254,480 and p < 0.001). With d = 0.59, a moderately strong effect presented itself (Ellis, 2010). This significant increase in income as a positive result also applied to families with small landholdings of 1–2 acres. Households with 3–5 acres, fertile soils and good management profited even more and reported net incomes of INR 200,000. In exceptional cases, for example, where farmers started an additional business of producing mango seedlings, incomes of up to INR 400,000 were earned.

As illustrated in Fig. 3, this significant difference can also be observed when scrutinizing the control groups separately (Fig. 3, left graph) (F (3, 1787) = 55.04, p < 0.001). On average, the participants’ annual income was almost twice those of the control groups (i.e., households in control villages, non-participants in program villages and recent participants). The number of income sources, on average, was almost 25% higher.

Fig. 3
figure 3

Average income in rupees and number of income sources of control groups and treatment. Whiskers represent standard errors

Full size image

A small difference among the three control groups can be noted and explained. The recent participants (i.e., the control group that just signed up for the program but did not receive any benefits yet, thus ruling out any selection bias) earned a similar low income on average, as the farmers did in the control villages, located just outside the program area, confirming the high similarity between the control villages and the program villages before the program started. The non-participants in the program villages (i.e., people who theoretically had the opportunity to participate but did not join) had an even lower income, on average. This raises the question of barriers to participation, which is beyond the scope of this paper. Overall, some people who did not participate were among the most marginalized and disadvantaged, in line with their being the control group with the lowest income, as depicted in Fig. 3.

Regarding the number of income sources (Fig. 3, right graph), similar differences between the control groups and the treatment groups were found. The household participants of the program had, on average, 8 different income sources, while the control groups had around 6. It should be noted that the information on income sources was collected per category; for example, tomatoes, onions and cucumbers were grouped under the category vegetables and were therefore counted as one income source. Thus, counting the different vegetable species grown would result in even higher numbers of different income sources.

Figure 4 shows the Wadi program’s effects on income and income sources over time. The left-hand graph illustrates a typical participating household’s increase in annual income from INR 30,000 to INR 100,000 or even higher. The right-hand graph shows the increase in the number of income sources, that is, the typical participating household’s diversification of its income sources over time.

Fig. 4
figure 4

Increase in income and income sources over time. The x-axis categories indicate number of years a participating family has been active in the Wadi program

Full size image

We found that the farmers’ incomes continued to increase as a result of the intervention, even though the support ended after four years. This finding is consistent with the increase in harvest over time, due to the trees’ growth. The farmers also continued to apply their newly acquired knowledge on improved farming techniques, as well as their newly adopted technologies, such as drip irrigation, enabling them to generate up to three harvests per year. Some farmers currently perceive themselves as agricultural entrepreneurs, whereas before their participation, they were impoverished smallholders.

We observed that many farmers developed an entrepreneurial spirit and diversified their income sources over time. The right-hand graph in Fig. 4 illustrates that diversification especially takes off after a few years, when the establishment of the Wadi is completed. The farmers then rely on up to 10 income sources. In the questionnaire and the quantitative analysis, different types of vegetables were counted as one income source. Nonetheless, our qualitative data showed that most farmers cultivated several different types of vegetables and therefore diversified their income sources even more than the quantitative results indicated.

The results depicted in Figs. 3 and 4 are in line with the multiple regression analysis results shown in Table 2; the regression analysis allows us to control for other factors that are potentially confounding.

Table 2 Regression models of (1) income in INR depending on treatment, (2) income in INR depending on years since participation (3) income in INR depending on treatment and years since participation, (4) number of income sources depending on treatment and (5) number of income sources depending on years since participation
Full size table

As shown in Model 1 (Table 2) of this regression analysis, the variable treatment indicates that the average annual income of the participants is INR 29,423 higher than that of the non-participants. As presented in Model 2, each additional year in the program (years since participation) is associated with INR 1,679 additional income, significant at p < 0.001. With R2 = 0.191 and F (8, 1734) = 51.142, p < 0.001, a significant and moderate variance explanation for the model can be noted. Model 3 reveals that when treatment and years since participation are both fed into the model as independent variables, treatment still shows a significant effect. Model 4 shows that the participants (treatment) have, on average, more than one additional income source compared with the control group members. Finally, Model 5 shows a significant increase in the number of income sources for every year of participation.

Besides treatment and years since participation, further factors are significant for annual income, as well as the number of income sources. Regarding the number of income sources, it should be noted that it is not the case that an entire additional income source is added every year, which is why fractions of 1 appear in Models 4 and 5 of the regression analysis (Table 2). On average, each additional acre of a farmer’s own landholdings means an extra income of more than INR 5000 per year. Each additional year of education completed by the adults in a household means an extra income of over INR 900 annually. Electricity also plays a role, as each additional hour of electricity leads to a yearly income increase amounting to INR 1,000. Distance to secondary school, a proxy for remoteness, shows, on average, a slightly higher income for more remote households. The variable distance to nearest water source indicates the importance of water; for every 100 m less, the annual income is, on average, more than INR 8000 higher. The analysis shows that more people in a household, i.e., household size (total people,) means generally less income but more income sources. The time that someone in a household spends on seasonal migration (sum of total months of migration per year) increases the household’s annual income, on average, to more than INR 4000.

The increase in income is further reflected in an improved standard of living. Qualitatively, many Wadi households have acquired electric appliances or other assets for the first time. Examples include pump set/irrigation pump, radio, television, mobile phone, bicycle or scooter and livestock, such as oxen and/or dairy cattle, apart from home renovation or construction of new solid brick houses. Participants were able to generate cash income through fruit trees, while also meeting most of their food needs from intercropping and kitchen gardens, both of which were encouraged as part of the program. All this led to substantially improved food security and dietary diversity. Subsequent to participation, livestock has become an integral part of the household assets to meet food, meat, egg, animal power and fertilizing needs.

While an increase in income can be used to finance many urgent needs of smallholders, money is not everything. Eager to find out more about the important but non-tangible aspects of life as well, we asked questions, answered on a 5-point Likert scale, to examine the farmers’ perceptions on whether (1) life had changed in the last 5 years, (2) they were satisfied with life, and (3) their children and grandchildren would have more opportunities than they had themselves.

As shown in Table 3, treatment has a positive impact on all three dependent variables. Across the three models, income is also highly significant, but due to the nominal value of the Indian rupee, the coefficients themselves are so small that only zeros are shown. Finally, the survey respondents’ comment that they would rather farm than seasonally migrate for the same income is also reflected in Model 2, where life satisfaction is negatively correlated with migration. This is not surprising, given the often exploitative nature of seasonal migration. (For our analysis of the program’s impact on seasonal migration, see Nienkerke et al., 2023).

Table 3 Regression models for positive life changes over the last 5 years, life satisfaction and perceived better future/perspectives/opportunities for children
Full size table

We always analyzed all households together, as well as each state separately. Comparing the state-wise analyses, we only found small differences, similar to differences between villages and regions within states, which are not worth reporting here. In line with our qualitative data and field visits, the conditions and challenges faced by smallholders in these marginalized areas are similar, independent of state boundaries. Factors playing a role are part of the regression analysis, such as water availability. We found the same causal relationships in all areas and therefore took all states as one sample.

Taken together, this evidence shows that while control groups continue to live in poverty, the participants convert wastelands into productive assets, generate sustainable incomes, diversify their income sources, lead more satisfied lives and report positive life changes and perspectives.

6 Discussion

Can a development program initiate a pathway that enables marginalized smallholders to break out of a vicious cycle of chronic poverty? The answer is yes. Our findings show that through the Wadi program, farmers from the poorest groups of society convert wastelands into profitable assets, thus breaking the cycle of resource degradation and chronic poverty.

Regarding our statistical analysis, it should be pointed out that if the control groups benefit from external development benefits, this means that the program’s impact is even underestimated.

In the regression analysis, we discover that in addition to treatment and years since participation, various factors—own landholdings, education of adults, electricity, remoteness, water availability, the number of people in the household and the number of months of seasonal migration—have a significant impact on both annual income and the number of sources of income. Even though farmers with 1-acre landholdings can substantially increase their incomes through program participation, farmers with more lands can cultivate more crops and, on average, earn even higher incomes. Education is neither a prerequisite nor a guarantee for success; however, educated participants are more likely to acquire new knowledge, adopt improved farming techniques and manage their farming business successfully. Electricity and a nearby water source both facilitate irrigation, but this does not mean that constant electricity and water supplies automatically lead to an endless increase in irrigation, harvest and income. Interestingly, remoteness (we use distance to secondary school as a proxy) has a small but significant positive impact, which means that remote farms perform better on average. This is in line with our qualitative findings that remoteness is not necessarily a drawback, despite more difficult market access. Remotely located farmers often focus more on their Wadis and take especially good care of them. Perhaps because they are less distracted or tempted by alternatives, which might provide quick but low earnings in the long term, compared with orchards. The impact of family size (i.e., the number of people in a household) can be diverse. On the one hand, a household with more members means more hands to help; on the other hand, there are more mouths to feed. Whether the members help or need help generally depends on the family’s age composition; therefore, this factor should not be generalized, and we cannot discuss it in depth here. What is clearer is that on average, the more time spent on seasonal migration by a household member who is not needed in farming, the higher the amount of money that can be contributed to the household income from non-farming activities (for more details on the program’s impact on seasonal migration, refer to Nienkerke et al., 2023).

Besides the direct effect of increased income, the program has many positive, indirect, somewhat intangible impacts as well. It stimulates the local economy by providing jobs—also for landless families. Twice the amount of employment per hectare of production is required in this integrated farming system compared with cereal crop production. Off-farm job opportunities are also created, for instance processing and marketing of cashew or managing the cooperative activities. As a result, a general growth of agri-businesses and markets can be observed in project areas. Social benefits include a rise in self-confidence, gained through newly acquired abilities and knowledge, women empowerment and improvements in dietary diversity, health and quality of life. Stakeholders speak of an atmosphere of confident hope, whereas before the program implementation, people were desperate. Owning a Wadi is a status symbol, representing prosperity and security, as ancient kings and landlords had the capacity to invest in this type of agricultural activity, which has a longer gestation period yet appreciates in value year after year. Working hands-on to convert their own wastelands into valuable orchards helps the participants feel empowered and competent. This explains the happiness and strong emotional connection to the Wadi, as expressed by some women: “We can consider the Wadi our earning son who will stay with us forever.” Many farmers view an established Wadi as a type of pension. Even when cultivated with a low level of intensity, it provides a steady source of income. According to a widow, she still cherishes the Wadi very much; even though she moved away to live with her son’s family, she returns every year to harvest the fruits, which grow even with zero input. Furthermore, the trees exhibit good resistance to fluctuating rainfall; some farmers report a loss of yield from cereal crops over the past few years as a result of unusually dry periods during the monsoon season; in contrast, the fruit trees have sustained no damage.

Through the diversified nature of the Wadi program, multiple income streams are created. This acts as a hedge against the concentration of too much risk on a single survival option. The qualitative interviews clearly indicate that diversification of income sources is not only about generating more money but also an insurance against natural calamities and unfavorable market developments. This becomes especially relevant when taking the tragedy of the so-called suicide belt into account, where many farmers who incurred debts to invest in monocropping took their own lives after failed harvests. In contrast, Wadi farmers speak of diversification on various levels: seasonality, gestation periods, shelf life and labor requirements. As a result, they can allocate resources more dynamically, optimize land use and extend income generation to the whole year—even in rain-fed conditions. Before the program support, there was no income source within the village for several months per year, and the local farmers were forced to seasonally migrate for survival. Because families now have stable and secure lives in their local village, they can pay more attention to their children’s education. The program has transformed the lives of poor farmers, and even youths have become enthusiastic about farming again.

Our results are consistent with the benefits of integrated farming systems highlighted in Argade and Wadkar’s (2013) review. This is especially true in terms of productivity (Jayanthi et al., 2002), profitability (Behera et al., 2004; Jayanthi et al., 2009), sustainability (Lightfoot & Minnick, 1991), nutritional quality (Jayanthi et al., 2009), ecological protection (Jayanthi et al., 2009; Lightfoot & Minnick, 1991), employment and income throughout the year (Dwivedi et al., 2007; Mahapatra, 1992; Radhamani et al., 2003; Ugwumba et al., 2010) and resilience (Fraser et al., 2005). Our results offer something new by suggesting that a development program can stimulate integrated farming practices to achieve these benefits. One does not need to wait for farmers to adopt such practices on their own.

7 Conclusion and recommendations

While many agricultural projects fail, from scientific and social standpoints, the Wadi program appears to be a success. Despite small landholdings, initially low soil productivity, the lack of irrigation and other resource constraints, small and even the poorest farmers are able to increase their incomes, improve their livelihoods, enhance their life satisfaction and expand their perspectives substantially. However, these do not occur overnight but need strong support. These require financing, good-quality inputs, technical knowhow to adopt improved practices and an organization, such as a cooperative, to collect, grade, process and market the produce. It has become evident that complex problems require holistic solutions (see Nienkerke and Patt (2022) for details on the program’s holistic approach and its broad range of positive transformations). It is equally essential to involve the beneficiaries themselves at every stage of the development process, as well as provide the initial financial support. The investment is well worth it, as the return on investment is high, and the economic life is up to 100 years.

Our rigorous assessment has enabled us to not only generate knowledge that can help the development of future programs but also identify central success factors, as follows:

  • Building good relationships with local people, mutual trust and transparency

  • Involvement of participants in management since the beginning

  • Extension service through local field guides instead of external technicians

  • Potential participants’ exposure visits to established Wadis

  • Disbursement of cash component directly to each participant’s bank account

  • Protective irrigation during gestation of fruit trees

  • Live hedge against free-grazing cattle

  • Simple, affordable technologies

  • Well-defined central theme but with considerable built-in flexibility

  • Options for rural communities within their own local resource bases rather than looking for answers outside their villages

  • Replicability across various socioeconomic conditions and agro-climatic zones

  • Better market access through cooperatives

The Wadi model can be scaled up to efficiently use limited water resources, small landholdings and high labor–land ratios in rain-fed, arid agro-ecosystems elsewhere, too. In India, 74% of the farmers hold less than 2 ha of land. It is also estimated that India has roughly 10 million ha of under-utilized wastelands, which can be brought under the Wadi program without curtailing the areas dedicated to food crops. Globally, rain-fed agriculture also accounts for sizeable portions of the lands and of total production in the continents of Africa, Asia and the Pacific.

Our results suggest the need for further investigation into the particular Wadi composition that provides the most successful output and is the most climate-resilient in other geographic areas. Further areas for valuable research include the program’s impact on seasonal distress migration. Many participants said that they were able to stop migrating; however, this is beyond the scope of this paper. We found anecdotal evidence of the Wadi program’s positive impacts on groundwater levels and soil quality; unfortunately, without robust measurements, this cannot be proven as of now.

Gandhi believed in human dignity. Every person has the right to live in dignity, but in the case of the poor, this gets lost in their struggle for survival. The Wadi model can provide a reliable source of livelihood for the rural poor throughout the year in order for them to escape the vicious cycle of chronic poverty. This has become particularly crucial in the face of the present food and climate crises.