International Migration, Remittances and Adaptation Practices to Climate Change of Agricultural Households in Coastal Oases (Southeastern Tunisia)

1. Introduction

Small-scale agriculture in many arid regions of Africa has become an insufficient source of income and food security, partly due to climate-change impacts. Facing these challenges, agricultural households adopt various strategies to compensate for the fragility of their economic systems. Out-migration of at least one household member to high-income countries is among these strategies [1], generally decided collectively to diversify income sources for those remaining in the origin areas [2,3].

Although migration intensifies in contexts of environmental degradation and climate-related threats [4,5,6], it can also increase household income and thereby enhance the likelihood of investing in adaptation practices that support productivity. This suggests potential linkages between migration and climate-change adaptation. The African regions most threatened by climate risks are also those most affected by migration and remittance inflows [7,8,9,10].

Adaptation and migration are conceptually distinct, yet one may influence the other. Climate-risk management is an adjustment process aimed at reducing agricultural vulnerability and increasing resilience [11]. While this process often involves public and private institutions [12,13], the agricultural household remains the primary actor responsible for implementing adaptation practices [14]. Water-saving can be achieved through precision irrigation [15,16,17] ; soil fertilization and reduced tillage help optimize yields and mitigate environmental impacts [18] ; crop diversification has gained increasing interest among farmers in arid zones [19,20]. These practices are generally proportional to household income and financial capacity [21], forming the core of productive investment that can generate economic surplus through improved productivity and strengthened resilience.

Labor migration is an economic strategy intended to diversify income sources and reduce socio-economic risks for household members who remain in origin areas. As a response to low land productivity [22], migrants aim—through remittances—to reduce vulnerability in their home regions [2,23]. They act as financial intermediaries, providing capital that can be invested in the well-being of household members [3], although their contribution to productive investment remains debated.

Economic studies on the relationship between migration and climate-change adaptation remain limited. Much attention has been given to remittances in analyzing migration’s impact on household welfare. The New Economics of Labor Migration (NELM) considers remittances as a form of insurance against socio-economic risks for origin households [23]. Most remittances are spent on consumption and living-standard improvements, with limited allocation to productive investment [24,25,26].

In highly vulnerable arid zones, the impact of migration—particularly remittances—on adaptation practices and agricultural production remains a recurrent debate. The lack of a clear theoretical framework complicates the discussion [27], although existing case studies help fill this gap [27,28]. Results vary: in South Asia, remittances have supported climate-resilient technologies [14] ; in the Sahel, their impact on agro-system resilience and production appears limited or neutral in the absence of public policies promoting productive investment [29].

Tunisia, heavily affected by international migration, is vulnerable to climate extremes in North Africa [9]. Migration is widespread across the country, with migrant numbers increasing from 65,927 (2009–2014) to 156,497 (2019–2024). The diaspora counts 1.4 million people [30], and remittances reached USD 2.97 billion in 2025 (5.3% of GDP), exceeding foreign direct investment inflows [31].

Managing migration has thus become a major policy challenge, especially in arid oases with high added value and strong migration trends [32,33,34,35]. Although recent studies have addressed climate-change adaptation in these oases [20,36], the effects of migration—particularly remittances—remain understudied.

Migration may stimulate productive investment and enhance adaptation capacity through remittances, but this remains a hypothesis requiring empirical validation. This study analyzes the impact of migration on adaptation practices and agricultural production in the coastal oases of southeastern Tunisia using descriptive statistics and a Cobb-Douglas production function with a migration indicator variable, based on survey data from 212 households.

2. Methodological Approach

2.1. Site Description

The study area (Figure 1) covers approximately 2,874 ha, representing 40% of the coastal oasis surfaces in southeastern Tunisia [37]. The climate is arid, influenced by coastal conditions, with annual rainfall ranging between 100 and 220 mm [20]. The three-layered cropping system is dominated by date palms and pomegranates, associated with various fruit species and diverse vegetable crops. Farmers cultivate fodder crops, industrial crops (henna, tobacco), and condiments (parsley, mint, spinach). Agricultural activities take place on small plots, most under 1 ha per farmer [38].

These oases face environmental challenges, notably water scarcity due to climate change [20,39], and demographic pressures linked to migration [32,33,34,40,41]. In the Gabès Governorate, where the oases are located, the departure abroad, which appeared early in history with a stock of 23,680 people in 2008 [42], nevertheless shows a certain continuity over time. In 2024, the number of migrants from Gabès exceeded 6,000 people [43].

2.2. Methodology

2.2.1. Data Sources and Sampling

Data collection was carried out at the end of 2024 as part of the research activities of the Laboratory of Rural Economics and Societies at the Arid Regions Institute (IRA) of Tunisia. The survey was designed to gather information on households and on the agricultural production system. The collected data are both quantitative and qualitative and can be classified into the following two dimensions :

• demographic and socio-economic characteristics of agricultural households at the time of the survey (2024) ;
• agricultural production and operating costs. The information collected refers to the year of the survey (late 2024), except for data related to the construction of water basins and the installation of electric pumping and drip-irrigation systems, which were implemented in previous years. These assets are considered stock variables, as they continue to affect agricultural operations during 2024.

The survey covered 212 households, without prior sorting according to migration status (migrant vs. non-migrant households). The sample was selected using simple random sampling, based on Yamane’s formula (1967). The margin of error is identical to that used in the survey conducted under the “Jeffara of Southeastern Tunisia” project [44].

$$n={\displaystyle \frac{N}{1+\left(N\times e^2\right)}}$$

where:

$n$: sample size

$N$: number of farmers = 11,628 [37]

$e$: margin of error = 6.8%

$$n={\displaystyle \frac{\mathrm{11,628}}{1+\left(\mathrm{11,628}\times {0.068}^2\right)}}=212$$

The data collected during the survey phase were subsequently subjected to statistical and econometric processing using Excel and EViews.

2.2.2. Analytical Method and Econometric Model Specification

Two embedded analytical approaches were employed in this study. The first consists of a descriptive analysis, by migration status, of households adoption of adaptation practices and their agricultural production. The second relies on an econometric regression of an agricultural Cobb–Douglas production function, incorporating an indicator variable to test the impact of migration.

For the purposes of this study, a household is defined as a group of individuals—biologically related, unrelated, or both—who live in the same dwelling, share meals, and sleep under the same roof. A household is classified as a migrant household if at least one of its members has migrated abroad to work outside the region of origin for a period of at least one year [45]. Within such households, some members may have returned from migration ; return migration is considered in this study to examine the effects of income from former migrants [24]. Conversely, a non-migrant household is composed of members who have never left their region of origin to work abroad for at least one year [46].

The econometric approach consists of estimating a Cobb–Douglas production function, commonly used in neoclassical economics to describe the relationship between output (Y) and production factors (capital and labor). The capital employed in production includes practices necessary for climate-change adaptation. As it is sometimes difficult to identify all such practices, we selected those reported by respondents and referenced in the literature (Table 1).

The adoption of adaptation practices generally contributes to optimizing productivity and reducing climate-related risks [51,52]. This process often requires farmers to invest part of their saved income in order to increase their production. Access to credit and government subsidy policies are not considered in this study. Household income alone is treated as the source of self-financing for production factors. Accordingly, the classical Cobb–Douglas production function can be written as follows:

$$Y=A{K}^{\alpha }{L}^{\beta }$$

where:

• Y: observed level of production (output)
• K: capital
• L: labor
• A: parameter of overall production efficiency
• α and β: elasticities

Transforming the function into a log-linear form yields the following model :

$$\mathrm{l}\mathrm{n}\left(Y\right)=\mathrm{l}\mathrm{n}\left(A\right)+\alpha \mathrm{l}\mathrm{n}\left(K\right)+\beta \mathrm{l}\mathrm{n}\left(L\right)+ϵ$$

where:

$ϵ$ : error term

• ln(A) : model constant

For this model, it is important to introduce an indicator variable equal to 1 if the household is a migrant household and 0 otherwise. This variable allows testing the impact of migration on agricultural production. Other sociodemographic variables (Table 2) may also influence production under climate constraints [53]. The selection of these variables follows similar empirical studies [54,55], even if conceptual differences may exist. The final form of the chosen model is :

$$\mathrm{l}\mathrm{n}\left(Y\right)=\mathrm{l}\mathrm{n}\left(A\right)+\alpha \mathrm{l}\mathrm{n}\left(K\right)+\beta \mathrm{l}\mathrm{n}\left(L\right)+\delta ENF15+\lambda EDUCAT+\rho TAIL+\theta MIG+ϵ$$

where:

• α, β, δ, λ, $\rho$, θ: coefficients to be estimated

The model was estimated using the Ordinary Least Squares (OLS) method. Durbin–Watson, Breusch–Pagan, and VIF tests were performed to verify, respectively, the presence of autocorrelation, heteroskedasticity, and multicollinearity in the regression.

3. Results & Discussion

This section begins with an analysis of household profiles, their adaptation practices, and their agricultural production. It then discusses the results of the econometric regression of the Cobb–Douglas production function, and finally provides an overall interpretation of the findings.

3.1. Profile of Surveyed Households

Table 3 summarizes the descriptive statistics related to several characteristics of the surveyed households. Migration, which is embedded in the traditional culture of the Tunisian population [56], concerned 42% of respondents, most of whom were male household heads (44.9%). The migration of other household members (sons, sisters, brothers) represents 23.6% of all migrant households. In addition, return migration is also observed in these oases, with a rate of 31.5%.

Female household heads are rarely observed, representing only 2.83% of the sample. The average age of household heads is 58.92 years, indicating an aging population that remains highly active in the agricultural sector. The average household size is 5.47 persons, exceeding the national average in Tunisia (3.45 persons per household) [43]. On average, each household includes 0.87 children under 15 and 1.90 educated persons beyond primary level. This demographic context reflects the rural nature of the population [57].

Agriculture is among the main occupations of households in the oases. The average cultivated area is 0.55 ha per household, with a predominance of smallholders cultivating less than 0.5 ha. Water management is ensured by a Collective Interest Group (GIC) responsible for water distribution. Only 29.25% of respondents expressed satisfaction with their ability to secure a minimum level of irrigation water access.

Household income sources are relatively diversified and depend on several sectors of activity. For migrant households, the average volume of remittances is estimated at 5,746 Tunisian Dinars (TND) per year. Income derived from the agricultural sector averages 2,664.445 TND per year, with higher values observed among migrant households. Conversely, non-agricultural activities, which generate an average of 2,783.03 TND per year, primarily concern non-migrant households (Figure 2).

Thus, it clearly appears that remittances have contributed significantly to improving household income in these oases. Migrant households receive an average total income almost twice as high as that of non-migrant households. This observation is consistent with the New Economics of Labor Migration (NELM), which posits that remittances constitute one of the most important externalities in terms of income generation and wealth circulation in migrants’ regions of origin [2,58].

3.2. Descriptive Analysis by Migration Status of Adaptation Practices and Agricultural Production

3.2.1. Adoption of Adaptation Practices

According to the investigation results, households used their own income to invest in adaptation practices without resorting to credit or government subsidies. Households receiving higher income were more likely to allocate financial resources to adopt these practices. In the oases, the proportion of migrant households practicing soil fertilization (organic and mineral fertilizers) is high (78.8%) compared with non-migrant households (60.5%). Furthermore, 30.5% of migrant households adopted drip irrigation—a technique with a direct effect on water savings—versus 17.1% among non-migrant households.

The construction of water basins with motor pumps for collecting accumulated irrigation water concerned 22.7% of migrant households, compared with 15.2% of households that have never migrated. Additionally, crop diversification is more frequently observed among migrant households (31.9%) than among non-migrant households (28.6%) (Figure 3). According to the survey, this technique mainly involved new varieties of fruit trees (orange, plum), high-value vegetable crops, fodder crops (barley, sorghum), and condiment crops (parsley, mint, spinach).

3.2.2. Agricultural Products

For all tree crops cultivated in the oases, modest differences in production quantities are observed (Figure 4). This cropping system, composed of an upper layer dominated by date palms and a middle layer consisting of olive trees and various fruit trees, remains widespread in the oases, with a slight predominance among migrant households. Indeed, the average quantity of date palm production is higher among migrant households, reaching 1.8 tons per farmer, compared with 1.1 tons for non-migrant households. Moreover, migrant households produce more pomegranates than households that have never migrated, with respective average quantities of 6.5 tons versus 3.3 tons. For other tree crops—such as olive, almond, fig, orange, plum, and apple—production levels are consistently higher among migrant households (Figure 4).

Forming the lower layer, vegetable and fodder crops are mainly cultivated by farmers producing onion, turnip, carrot, and alfalfa. Although production levels vary across households, the results clearly confirm a growing orientation toward the development of these crops [59]. A simple comparison between the two household categories reveals a slight predominance among migrant households. This lower layer also includes other crop varieties with varying production quantities. Farmers cultivate grapevine, parsley, celery, Swiss chard, spinach, cucumber, pepper, tomato, potato, henna (Lawsonia inermis), tobacco, and others (Figure 5).

Animal production in the oases, although the average livestock size per household is low (6 head per household), remains widespread among the local community. Thanks in particular to the development of alfalfa cultivation, livestock farming has been able to maintain its sustainability. Annual alfalfa production reaches 10.95 tons per migrant household, compared with 9.7 tons for non-migrant households.

3.3. Econometric Analysis of the Cobb–Douglas Production Function

The results of the econometric regression and the statistical tests are presented in Table 4. The linear fit of the model (adjusted $R^2$) is 34.5%, with a generally significant F-statistic. The model also shows low multicollinearity among predictors, as VIF values are within acceptable ranges. The Durbin–Watson test and the Breusch–Pagan test respectively confirm the absence of autocorrelation in residuals and the absence of heteroskedasticity issues. Statistically, the model explains agricultural production in the oases to a certain extent. At the 1% significance level, the statistically significant variables are capital (K) and labor (L). In addition, the indicator variable capturing migration (MIG) shows a significance level of 10%. These variables positively affect agricultural production. Conversely, the variables related to the number of children under 15 (ENF15), the number of educated persons (EDUCAT), and household size (TAIL) do not have significant effects in this application. In what follows, we interpret the production elasticities with respect to their explanatory variables, in order to understand the degree of sensitivity of each variable in the model.

According to the model results, capital shows a positive and highly significant coefficient (1% level). A 1% increase in capital leads to a 0.52% change in production, all else being equal. The inelastic relationship (0.52) indicates that capital influences production, but in a less than proportional manner. An increase in capital generates only a modest gain in output. This finding does not necessarily reflect weak self-financing of adaptation practices ; rather, it may result from constraints such as resource under-utilization, misallocation, soil salinity, and water scarcity. This result is consistent with studies conducted in Latin America [60] and North Africa [36], which show that production remains non-proportional to capital under conditions of natural resource scarcity and ecosystem degradation intensified by climate change.

Furthermore, the coefficient associated with labor hours (L) is positive and highly significant (1% level). A 1% increase in labor hours raises production by 0.45%, all else being equal. The inelastic relationship (0.45) between these variables indicates that labor has a limited impact on production. Increasing labor hours is no longer proportional to output. This result appears to be influenced by land constraints in the oases, characterized by homogeneous and small cultivated areas [61], which limit labor requirements. It may also stem from poor organization and distribution of labor [62,63].

Additionally, the coefficient of the migration indicator (MIG) is statistically significant at the 10% level. The idea that migration can have a positive impact on agricultural production is confirmed. Since the dependent variable is in logarithmic form, the estimated coefficient is interpreted as:

$$\left(\mathrm{e}\mathrm{x}\mathrm{p}\left(0.305\right)-1\right)\times 100=35.67\%.$$

Thus, for identical levels of capital and labor, migrant households produce on average 35.67% more than households that have never migrated. This result supports earlier findings showing a slight predominance of migrant households adopting adaptation practices that enhance productivity. It also aligns with studies demonstrating that migration contributes to the socio-economic well-being of household members who remain in the origin region through remittances [14,32].

The production function exhibits decreasing returns to scale (0.52 + 0.45 = 0.97 < 1), meaning that when inputs increase proportionally, output increases less than proportionally. Thus, despite households’ efforts to invest in production factors—particularly in climate-change adaptation practices—production remains non-proportional due to the complexity of adaptation and the degradation or depletion of water and soil quality [56].

3.4. Discussion

As an economic surplus contributing to the socio-economic well-being of households remaining in the origin region, remittances constitute an important source of self-financing for the agricultural sector in the face of climate change. The impact of remittances on agricultural production is far from negligible. Migrant households, which receive more money than non-migrant households, invest more in capital, including adaptation practices. These practices have direct positive effects on production. For all oasis crops, a simple comparison between the two household categories reveals a slight predominance of migrant households. Thus, increased income and savings have primarily led to greater productive self-financing. Even though capital and labor influence production in a less than proportional manner (decreasing returns to scale), migration remains a determining factor in the self-financing process.

Despite the vulnerability of these arid oases, households have managed to invest part of the income derived from migration. Although their primary concern is to improve the living conditions of family members who remain in place [24,26], part of this income is allocated to productive self-financing [59]. This result is consistent with studies showing that migration strengthens farmers’ financial capacity to improve their resilience and production systems [32,35,64].

Clearly, migration constitutes one of the adaptation strategies that can be particularly effective in the face of progressive environmental degradation and climate-related threats. Nevertheless, non-migrant households also contribute to this process, even if with smaller financial volumes. Benefiting from agricultural and non-agricultural income, non-migrant households also finance adaptation practices. This result is expected, as surplus income from permanent or temporary household activities increases the likelihood of self-financing adaptation practices and boosting production [20,36].

Finally, this study highlights certain limitations whose consideration could improve future results. Expanding the sample size and integrating new explanatory variables into the econometric model could enrich the findings. Variables that are difficult to quantify—such as extension services, local policy, and governance—could be valuable additions to the model.

4. Conclusion & Perspectives

The impact of migration on adaptation practices and agricultural production in the oases was tested using two complementary methods. The descriptive analysis shows that the highest proportion of households adopting practices to adapt and reduce their vulnerability to climate change effects belongs to migrant households. The most widespread practices include water-saving techniques, soil fertilization, and crop diversification. These practices have had positive impacts on agricultural yields, particularly among migrant households. This finding is confirmed by the econometric regression of the Cobb–Douglas production model : a migrant household produces on average 35.67% more than a non-migrant household, all else being equal.

However, international migration—sometimes triggered by declining agricultural yields due to climate change—has contributed to improving the resilience of the oasis agro-system. The role of migration, particularly remittances, remains crucial in self-financing capital, including adaptation practices. Despite these potential advantages, challenges persist regarding the management and efficiency of producers in allocating production factors within an agro-system facing environmental constraints, especially water scarcity under climate change.

Thus, the gains derived from migration should be further leveraged with the support of actors beyond migrants themselves. The role of local actors, including the participation of non-migrant populations, becomes essential in climate-change adaptation strategies. Migrants should be more actively engaged in this process. Alternative pathways must be identified, both in terms of strategy and agricultural policy. Strengthening links between the territory, local actors, and the diaspora abroad is necessary to inform migrants about climate challenges, local development plans, and new agricultural opportunities in the context of climate change. The objective is to establish an adaptation governance strategy based on reforms in agricultural investment, training, awareness-raising, extension services, and the management of local cooperatives.