Introduction

The agriculture sector remains the mainstay of livelihood for a significant chunk of the country’s population, approximately 40%1. However, it faces numerous challenges, including environmental degradation2, inefficient resource utilization3, and limited adoption of modern technologies4. Studies have identified that a lack of access to information5, limited digital literacy6, and poor technological uptake are key barriers to agricultural modernization7. According to8, lack of integrated systems for disseminating advanced agricultural information to farmers is the main challenge to agricultural growth and transformation in Pakistan. Apart from the availability of modern technologies and inputs, the sector has been underperforming considering that the growth rate has averaged below 3% for the past two decades9. Notably, there is no vertically integrated system that connects information from various agricultural stakeholders including agricultural research institutes, farm advisory agencies, markets, metrological stations, and farmers for services required10,11. As a result, farmers often rely on traditional practices, which require higher energy and labor inputs, hindering agricultural productivity, farmers’ welfare, and long-term sustainability. Although the use of the internet in agriculture has emerged to disseminate technological information and provide access to knowledge about modern farming practices, there is still limited evidence on how digital technologies can specifically promote the adoption of energy-efficient agricultural practices in Pakistan.

The agriculture sector in the country is facing serious challenges including climate change12, biodiversity loss13, desertification and drought14, and rising food prices15. Recently, researchers, and conservationists have acknowledged the benefits of adoption of energy-smart agricultural (ESA) practices in lowering agricultural carbon footprints and enhancing farmer income16,17,18. The ESA practices lie within the realm of climate-smart practices. There are five technologies within the scope of ESA practices: 1) zero tillage (ZT), crop residue management (CRM), legume integration (LI), direct-seeded rice (DSR), and precision water management (PWM)16,19,20. The adoption of underlying practices improves optimal use of key input – like water, energy, and fertilizers—and enhances productivity16,20. Given this, in today’s milieu, the agriculture sector has evolved into a knowledge-intensive field21, yet there remain notable research gaps concerning the interplay between digital training participation, ESA practices adoption, and farmers’ welfare in developing countries. However, this backdrop offers an opportunity for the increasing access to modern ICT systems (e.g., web access and mobile connectivity) to channel valuable information to farmers8,22. Effective internet usage can provides information on key agriculture inputs, sustainable practices, and technologies; however, farmers’ improved skills are fundamental to interpret internet information into relevant formats to support agricultural decision-making23. Given this, there exists the need of digital training for farmers’ capacity building to enhance their ICT skills to enhance productivity, input efficiency, resilience, and farm income.

The agriculture sector holds significant potential for fostering pro-poor economic growth24. Evidence demonstrates that enabling farming communities to access knowledge banks, networks, and institutions through information and communication technologies (ICTs)—hereafter e-agriculture—substantially improves productivity, food security, and employment opportunities20,25. E-agriculture has emerged as a burgeoning field aimed at enhancing agricultural and rural development through improved information and communication processes. Notably, it improves learning, practicing and knowledge-sharing in rural areas through capacity building on basic internet skills26.

The unprecedented growth and affordability of ICTs technologies, particularly broadband access and mobile phones, has largely transformed tools for accessing, sharing, and implementing information in agriculture27. Numerous ICT-based applications have been developed, with varying degrees of success, to support a transformation toward resilient and sustainable agriculture28, and improving farmers’ incomes, and reducing risks29. Existing literature provides cues on the impact of ICT adoption on various outcomes, including the adoption of green technologies and climate change mitigation30, poverty reduction31, productivity and income32, as well as preventive measures, early warning systems, and weather information33. Further, there is evidence regarding the significant impact of ICT in market participation and decision-making29, access to farm credit34, and farm advisory services8. The studies have investigated the connection between integration of ICT and remote learning of rural farmers35. Yet, the link between digital training and the adoption of ESA practices has largely remained unexplored. Moreover, the influence of digital training on smallholder farmers’ welfare has been overlooked so far.

The investigation into the synergies of digital training brings forth several useful findings for policy formulation and practitioners. First, through farm-level empirical evidence, this study addresses the research gap regarding the effectiveness of digital training programs in promoting energy-smart and sustainable agriculture in developing nations. Second, the study findings improve understanding of the role of digital technologies in modernizing farm advisory services by facilitating better outreach of the latest technologies. Likewise, the study sheds light on the potential contribution of internet use to enhance productivity, profit, and technology adoption. Third, it provides valuable insights for policy makers, academics, farming communities, and other stakeholders. It helps to design more effective interventions to promote energy and input use efficiency and the welfare of farming communities while ensuring sustainable development in the agriculture sector. Moreover, the study can aid in developing evidence-based planning mechanisms to support rural development, resilience, and welfare outcomes in developing countries.

Doing so, this study distinguishes itself from previous research in several important ways. Most existing studies, though varied in focus, predominantly center on regions such as China5, sub-Saharan Africa36, and parts of Asia37, but there has been no comprehensive investigation into the adoption of ESA practices, particularly in Pakistan. Second, earlier research tends to highlight mobile phone adoption as the primary ICT tool (e.g. 38,39) often neglecting the significant impact of training interventions. Third, many studies rely on qualitative methods40 or linear regression models like propensity score matching41, and difference-in-difference42. In contrast, this study uses endogenous switching regression to deliver more precise and reliable estimates of the effects of training participation on ESA practices. Additionally, this study prioritizes farmer-centric outcomes, including productivity, ESA adoption, and farm profitability.

The remainder of this paper is structured as follows: Section 2 provides details regarding the digital training program. Section 3 covers research methodology, followed by the empirical results in Section 4. Section 5 presents the discussion. Finally, Section 6 offers conclusions.

Background regarding digital training

According to FAO,Footnote 1 formal training for farmers, such as digital training, is an interactive educational approach that brings together small-scale food producers to address production challenges through sustainable agriculture. This method promotes hands-on, group-based digital learning, practical use of internet technologies, crop management mobile applications, social media, critical thinking, and enhanced decision-making within local communities. In our study, we assess the impact of digital training provided by “Digital Dera,” located in the Pakpattan district of Punjab, Pakistan. This initiative, launched in 2021, is a public-private partnership between Agriculture Republic, Hayat Farm Pakistan, and Pakistan Telecommunication Company Limited (PTCL), the country’s largest telecommunications provider. The program was implemented in Pakpattan, an area with limited internet access and low awareness of digital agricultural technologies. Its goal is to improve the digital literacy of rural and farming communities while testing the feasibility of scaling similar projects across the country. The knowledge resources offered at Digital Dera cover a wide range of topics, including weather forecasts, soil analysis, climate-resilient seeds and fertilizers, smart irrigation techniques, crop insurance, financing options, marketing opportunities, and other aspects of digital agriculture. Serving as a hub, the center provides essential internet skills to help bridge the digital divide and support informed decision-making. According to43, training programs like this are crucial for improving farmers’ decision-making abilities, enabling them to seize market opportunities and enhance income growth.

This intervention was implemented exclusively in Punjab Province, specifically targeting farmers in the Pakpattan and Okara districts. These districts were chosen due to their significant agricultural output and representation of small-scale farming. Additionally, these districts are major suppliers of milk to large cities, making them an ideal setting for promoting ESA practices to transform dairy production into a climate- and energy-smart model. At the village level, the training was carried out by inviting farmers through local market publicity, engagement with community leaders, and collaboration with the agricultural department. We used a stratified random sampling approach to select treatment households, ensuring diversity in terms of crops, farm size, and farming practices. Although the project regions were predetermined, we employed a rigorous methodological approach to capture a representative sample of farming households within these districts.

Figure 1 presents the conceptual framework outlining the potential impact pathways of digital training participation. It highlights factors influencing farmers’ decisions to participate in digital training. Participation enhances farmers’ digital and business skills by connecting them with a broader community consisting of experts, researchers, leading farmers, and innovative practitioners. This connection could lead to increased adoption of ESA practices. The main research objectives are as follows: First, to examine the mixed effects of factors like market access, proximity to Digital Dera, and farmers’ social circles on smallholder farmers’ decisions to participate in digital training. Second, to assess the significant influence of digital training participation on ESA practice adoption. Third, to evaluate the impacts of smallholder farmers’ digital training participation on productivity, and net farm returns (profit). Figure 2 summarizes the overall potential impact of internet-enabled ESA practices on farmers, rural communities, and the environment at the community level.

Fig. 1
figure 1

Conceptual Framework of Digital Training Participation. Source: Authors.

Full size image
Fig. 2
figure 2

Community-Level Potential Impact of ESA Practices. Source: Authors.

Full size image

Research methodology

Data collection

The study utilized cross-sectional data gathered from Okara and Pakpattan districts of Punjab through a well-structured questionnaire. To obtain a robust sample for consistent estimates of training participation, we followed a systematic process based on the training protocol, recruitment strategy, and participation rates. The Digital Dera used a blended learning approach, combining both face-to-face and online learning modules. These covered key topics such as soil analysis, market information access, weather forecasts, smart irrigation techniques, modern agricultural technologies, and certified seeds. Additional online modules were provided for follow-up learning. The recruitment strategy included several methods: advertisements in nearby input-output markets, local community outreach, and word-of-mouth through farmer networks. Local village-level leaders also played a significant role in recruiting farmers. In total, Digital Dera invited approximately 2,000 farmers in two phases, of which 1,357 farmers participated, resulting in a participation rate of 68%. The non-participation rate was mainly due to time constraints and geographical barriers. Of the participants, nearly 91% were from within a 19 km radius of the training center, which includes villages from both the Pakpattan and Okara districts. While the training was conducted in Pakpattan district, farmers from 9 Okara villages within this radius also took part in the program. These districts were chosen due to the presence of a Digital Dera in Pakpattan and their geographical proximity, ensuring equal opportunities for training participation in both districts.

The sample size for this study was calculated based on a population of approximately 25,000 registered landowner farmers across both districts, with a 95% confidence level (Z-value of 1.96), a margin of error of 5% (0.05), and a population proportion of 50% (0.5) to capture the maximum variability among farmers. Using the following formula (Eq. 1), we determined the required sample size to be 379 respondents.

$$n=\frac{{Z}^{2}.p.(1-p)}{({E}^{2}.\left(1-N\right)+({Z}^{2}.p.(1-p))}$$
(1)

However, recognizing that many small farmers are not registered as landowners or many farmers instead work on rented or leased land, we adjusted for this by doubling the sample size to ensure a more representative and robust sample. This resulted in a final sample of 723 completed questionnaires which is almost double of the statistically suggested sample size of 379. To collect data from households, a multi-stage random sampling technique was employed. This method offers significant statistical rigor and practical flexibility, making it ideal for a field survey that includes both participants and non-participants from large populations. First, stratified random sampling was used to select the Okara and Pakpattan districts to ensure diversity in the farmers’ decisions to participate. Next, random assignment was used to select both the control and treatment groups. For the treatment group, 400 farmers based on training participants information obtained from Digital Dera (approximately 33%) were randomly selected. It covers farmers from 18 villages (13 in Pakpattan and 5 in Okara) who had participated in successful training sessions, yielding 314 usable responses. In the control group, 409 non-participants were selected from the same districts, but with a minimum distance of 25 km from the training sites (which had a 19 km radius for participants). This distance was maintained to avoid potential biases and to ensure the representativeness of the sample by capturing socio-economically similar farmers with different access to training resources. The final sample consisted of 57% (409) control group and 43% (314) treatment group households. Table 1 presents a summary of the socio-demographic characteristics of the respondents and key study variables, including five ESA practices and various outcome and control variables used for further analysis.

The final field survey for data collection was conducted between December and February of 2024. Before conducting the field survey, a team of local enumerators, consisting of individuals with a Master’s degree (7) and PhD (5) students, underwent training to conduct a pilot survey to identify weaknesses and loopholes in the questionnaire. The study questionnaire and research methods adhered to relevant guidelines and regulations and were approved by the ethics committee of Institute of Agricultural and Resource Economics, University of Agriculture Faisalabad, Pakistan. For the final field survey, the questionnaire was translated into Urdu, and enumerators were also trained to conduct interviews in Punjabi when necessary. Prior to conducting the survey, we clearly explained its purpose and content to all participants. The study ensured complete anonymity, with no personal data collected beyond survey responses. Following standard research practices in Pakistan for such anonymous studies, we obtained verbal and written informed consent from all participants. To minimize bias, we incorporated various control variables, treatment variables, and reverse response-based questions. Notably, our sampling strategy and the design of the questionnaire – including control and treatment variables – help mitigate bias, as confirmed by the validity measures in the results. The finalized questionnaire was comprehensive and well-organized. It covers variables such as farm characteristics, demographics, farmers’ education, information sources, control variables, and digital training. In addition, information on other socioeconomic variables was included.

Variable selection

Control variables

Previous studies has explored the determinants of climate-smart technology adoption44,45, sustainable agriculture46,47, and technology adoption48,49. Building on previous studies and field surveys, particularly focus group discussions, this study incorporates relevant control variables to examine farmers’ access to digital training. Specifically, it includes demographic variables such as farmers’ age, gender, education, and marital status, as well as socioeconomic factors like farm size, farming experience, livestock holdings, smartphone ownership, and social circle. Additionally, institutional variables such as membership in farmer-based organizations (FBOs), access to farm advisory services, and access to credit are considered.

Demographic variables are used to assess the influence of personal characteristics on farmers’ participation in digital training, while socioeconomic variables account for differences in farm size, wealth, and access to information. Institutional variables measure the impact of organizational and institutional factors on participation and their corresponding effects on farmers. Moreover, farming system variables, including wheat-rice and mixed cropping systems, are integrated into the analysis to provide a comprehensive understanding of the factors influencing digital training participation.

Treatment variables

This study used five treatment variables: fines and inspection, access to subsidized technology, access to skill development training, access to markets for selling crop residues, and simultaneous access to both training and subsidized technology. Farmers without access to either technology or skill development services were designated as the control group. Each treatment was represented by dummy variables, where a value of 1 indicated access and 0 indicated no access. A detailed description of the treatment variables is provided in Table 1.

Instrumental variables

In this study, we employed three instrumental variables (IVs) to address potential endogeneity: (1) access to internet connection, (2) distance to Digital Dera, and (3) farm radius. Each of these IVs is directly related to participation in digital training but is not associated with the adoption of ESA practices and other outcome variables.

The first IV is access to internet connection, which can influence a farmer’s participation in digital training, particularly as some training modules require self-learning at home. Farmers with internet access are more likely to participate in the training. However, access to the internet does not have a direct impact on the adoption of ESA practices or farm productivity. The second IV is distance to Digital Dera. A shorter distance to the training center increases the likelihood of participating in the digital training. Therefore, the distance to the training center is positively related to participation but does not have a direct impact on the dependent variables – adoption of ESA practices, farm productivity, or net farm returns. Similar studies have used distance, such as distance to extension services or advisory offices, as an instrumental variable50,51. The third IV is farm radius, which refers to the number of training participants within a 3 km radius of a farmer’s location. A higher number of nearby training participants can encourage participation through word-of-mouth or the sharing of information. However, farm radius is exogenous to the dependent variables and does not directly influence the adoption of ESA practices or farm productivity. A similar study52 used farm radius as an instrumental variable in a comparable context.

Thus, these three IVs satisfy the exclusion restriction, validity (exogeneity), and relevance criteria, as they are not directly related to the dependent variables but are significantly associated with the endogenous explanatory variables.

Outcome variables

The primary outcome variables in this study are the adoption of ESA practices, farm productivity, and net farm returns. ESA practices are measured by whether a farmer adopts Direct Seeded Rice (DSR), Crop Residue Management (CRM), Zero Tillage (ZT), Legume Intercropping (LI), and Precision Water Management (PWM). Farm productivity is assessed by calculating the average per acre yield of rice and wheat. Similarly, the outcome variable for net farm returns is determined by subtracting variable costs from annual farm revenue per acre. Importantly, net farm returns also include income generated from the sale of crop residues of wheat and rice, as a market for crop residues exists in the study districts.

Table 1 Description of the study Variables.
Full size table

Modelling the impacts of digital training participation

We employed an endogenous switching regression (ESR) model to analyze the factors influencing digital training participation, productivity, and net farm return. ESR was chosen for its robustness in addressing endogeneity, a critical issue when evaluating the impacts of agricultural programs. Endogeneity occurs when unobservable factors affect both participation and outcomes, which can lead to biased results when using ordinary least squares (OLS) regression. Unlike experimental data from randomized control trials, cross-sectional surveys lack counterfactual information, making it more difficult to draw causal conclusions.

In our case, the ESR framework offers several advantages over the standard two-stage least squares (2SLS) approach. First, it has greater power in handling treatment effect heterogeneity, as some farmers participated in the training while others did not. Second, ESR is better at accounting for sample selection bias, a potential issue in our study using the 2SLS approach. Since ESR explicitly addresses sample selection, it provides more robust estimates in our scenario. Third, unlike the standard 2SLS approach, ESR uses a unified mechanism to simultaneously handle both treated and control groups. Additionally, since our study sample consists of only two districts, issues with random assignment of treatment could lead to endogeneity and inconsistent estimations due to observed and unobserved factors influencing farmers’ training participation. In this context, the ESR framework is more effective, as it estimates both outcome equations simultaneously, leading to more reliable results.

The study from53 propose analyzing the direct effect of participation by comparing outcomes among farm households. Since farmers make a choice to participate or abstain, the observed outcome variables can take the given expression:

$${Y}_{i}^{0}={X}_{i}{\beta }_{0}+{{\upepsilon }}_{0} (Non-participants)$$
(2)
$${Y}_{i}^{1}={X}_{i}{\beta }_{1}+{{\upepsilon }}_{1} \left(Participants\right)$$
(3)

Equations 2 and 3 model the conditional potential outcomes of internet-based training participation on farm performance. Here: first \({Y}_{i}^{0}\) compute outcomes variable under non-participation, and \({Y}_{i}^{1}\) denotes the outcome under participation, xi captures observed covariates (e.g., farmer education, age, input access), and\({{\upepsilon }}_{i}\) is the error term accounting for unobserved factors. Given this, we can model farmers’ choice for participation in internet-based training (selection equation) as:

$${D}_{i}^{*}={\text{Z}}_{i}\gamma +{\mu }_{i} with {D}_{i}\left\{\left.\begin{array}{c}1, if {D}_{i}^{*}>0\\ 0, Otherwise\end{array}\right\}\right.$$
(4)

Equation 4 specifies a selection model that governs farmers’ decisions to participate in training programs. This model assumes the error terms follow a multivariate normal distribution with zero mean and covariance. This framework corrects for selection bias arising from unobserved confounders (e.g., intrinsic motivation) that correlate with both participation decisions and outcomes.

Given this, we can model outcome equations:

$$E\left[{Y}_{i}^{0}∣{D}_{i}=0\right]={X}_{i}{\beta }_{0}+{{\upsigma }}_{0}{{\uplambda }}_{0}; E\left[{Y}_{i}^{1}∣{D}_{i}=1\right]={X}_{i}{\beta }_{1}+{{\upsigma }}_{1}{{\uplambda }}_{1}$$
(5)

In Eq. 5, we introduced inverse Mills ratio terms (λ0, λ1) correct selection bias. Average treatment effect on treated (ATT) can be expressed as follows:

$$ATT=\left[{Y}_{i}^{1}∣{D}_{i}=1\right]-E\left[{Y}_{i}^{0}∣{D}_{i}=1\right]={\left({X}_{i}{\beta }_{1}+{{\upsigma }}_{1}{{\uplambda }}_{1}\right)-(X}_{i}{\beta }_{0}+{{\upsigma }}_{0}{{\uplambda }}_{0})$$
(6)

Equation 6 estimates the overall impact of internet-based training participation on outcomes (e.g., productivity, and net farm returns). Using ESR, we can use control function approach to correct endogeneity:

$${W}_{i}={Z}_{i}\pi +{\eta }_{i} \left(\text{f}\text{i}\text{r}\text{s}\text{t} \text{s}\text{t}\text{a}\text{g}\text{e}\right)$$
(7)
$${Y}_{ij}={X}_{i}\beta +{{\theta }_{i}\widehat{\eta }}_{i}+{ \epsilon }_{i} \left(\text{s}\text{e}\text{c}\text{o}\text{n}\text{d} \text{s}\text{t}\text{a}\text{g}\text{e}\right)$$
(8)

Equations 7 and 8 helps correct the potential endogeneity of key covariates (e.g., access to farm advisory, off-farm income, internet use).

Sensitivity analysis

We conducted a sensitivity analysis to further validate the results obtained from the ESR model. In this process, we utilized boxplots, as they are more appropriate for studies involving PSM and robustness checks. Compared to line charts, boxplots are better suited for illustrating data variability and effectively represent how outcome variables differ within and across treatment groups.

Robustness checks

In the first stage, we tested the relationship between the instruments and the exogenous regressor using the multivariate Cragg-Donald Wald F-test to address the weak instrument assumption. Additionally, we conducted the Hansen J statistic test for IV models, which evaluates the null hypothesis that all instruments are valid (i.e., the over-identification test). The Hansen J statistic is especially useful in the ESR approach for assessing the validity of the instruments.

Furthermore, we applied the Durbin-Wu-Hausman chi-square test and the Wu-Hausman test to examine the endogeneity, strength, and consistency of the instruments. We also used the Anderson canonical correlation LM statistic to assess potential under-identification of the instruments (relevance). To check for multicollinearity, we tested the variance inflation factor (VIF), and the Breusch-Pagan test statistic was used to investigate the presence of heteroscedasticity, which is a common issue in cross-sectional data. These tests were performed to address endogeneity and self-selection bias, and we proceeded with our ESR estimations by inflating the asymptotic variance of the estimators used.

Results

Mean differences and farmers’ perception

Table 2 presents the summary statistics of variables, categorized by digital training participants and non-participants. There are notable differences between digital training participants and non-participants concerning farm-level household characteristics, corroborated by statistical tests. Regarding the adoption of ESA practices, significant mean differences are observed between digital training participants and non-participants for DSR, CRM, ZT, and PWM. This suggests that digital training participation facilitates the uptake of ESA practices. Moreover, the data reveals that 93% of training participants are male, indicating a predominance of male-headed households among farmers in the study areas. Regarding outcome variables, training participants exhibit significantly higher productivity and income. Results reveal that instrument variables demonstrate a significantly positive perception of internet-based agricultural information among training participants. Similarly, training participants differ from non-participants in their subscription to social media channels for themselves and their social circles. In addition, significant mean differences between training participants and non-participants are observed in education, off-farm income, livestock ownership, access to credit, participation in skill development programs, and membership in FBOs. These differences suggest potential confounding effects on the actual impact of digital training participation on farm productivity, and net farm returns, of smallholder farmers.

Table 2 Mean difference in variables used in Model.
Full size table

We examined the factors influencing internet usage for agricultural activities, as illustrated in Fig. 3. Digital training participants were questioned using a five-point Likert scale to assess various statements, and their responses were recorded. The findings indicate that, among other factors, access to market price information emerges as the primary driver of internet usage, followed by the utilization of online farm advisory services. Information-seeking related to the adoption of ESA practices ranks third in importance. In addition, farmers are utilizing internet resources for accessing weather forecasts, making crop selections, and seeking information on credit and subsidies. These results suggest that the integration of ICTs in agriculture is fundamentally reshaping the landscape of agricultural practices, facilitating vertically integrated information exchange and knowledge sharing.

Fig. 3
figure 3

Determinants of Internet Use Among Farmers.

Full size image

First-stage results: identification, relevance, and robustness of IVs

We employed first-stage IV approach to evaluate the relevance and statistical robustness of three instrument variables (IVs) and their interaction term regarding training participation for adopting ESA practices (Table 3). Our findings reveal a positive and significant relationship between the IVs and the endogenous regressor, adopting ESA practices. Specifically, the first IV, access to internet connection (IV – 1), demonstrates a significant association with ESA practice adoption. Similarly, both the distance to training center (IV – 2) and farm radius (IV – 3) are significantly linked to ESA practice adoption. Furthermore, the introduction of the interaction term, comprising all three IVs collectively, reaffirms their joint relevance to the endogenous regressor, underscoring the statistical robustness and validity of the employed IVs for further analysis.

The results of all models, along with the selected specification tests and checks, are presented in Table 3. The Cragg-Donald Wald F-test (F-stat > 10) confirms that the instruments, both individually and collectively (including the interaction term), are strong, thus invalidating the weak instrument assumption. Additionally, the Hansen J statistic test value (Stat > 0.15) indicates that we fail to reject the null hypothesis, supporting the validity of the instruments for IV models and confirming that all instruments are valid (over-identification test). Moreover, the significant Wu-Hausman test value for all IVs and the interaction term confirms the exogeneity, strength, and consistency of the instruments. The significance of the Anderson canonical correlation LM statistic further confirms the relevance of the individual and joint IVs (including the interaction term). The VIF test values, which are well below the threshold (VIF < 10), confirm the absence of multicollinearity. Additionally, the Breusch-Pagan test statistic value, which exceeds the threshold (p-value > 0.05), confirms that the individual and joint IVs are free from heteroscedasticity. Therefore, we proceed with the ESR framework, ensuring that our IVs are robust, consistent, and statistically sound.

Table 3 Determinants of digital training participation: instrument Variables.
Full size table

Determinants of digital training participation

As outlined in the research methods section, we employed full information maximum likelihood estimation to jointly estimate both equations: the selection (training participation) and outcome equations (net farm returns). The results of the selection equation are presented in Table 4, representing the determinants of training participation in two distinct sets of findings due to slightly different specifications. Since model identification requires that at least one selection equation variable that does not appear in the outcome equation, we utilized variables representing farmers’ access to internet, distance to training centers and farm radius as identifying instruments. We simultaneously incorporated all the three IVs in selection equation. To interpret the model results, we followed the approach outlined by49, treating the results as a normal probit function, as presented in Table 4.

The results from the selection equation reveal the factors affecting digital training participation (Table 4). The results indicate that access to internet connectivity and the presence of individuals within a 3 km farm radius who have already participated in digital training significantly influence farmers’ decisions to engage in such training. Specifically, farmers with internet access and those surrounded by others who have participated in the training are more likely to take part. This suggests that having an individual within a 3 km radius who has previously attended the training positively impacts a farmer’s likelihood of participation, highlighting the presence of a network effect. Additionally, the distance to the training center negatively affects participation in digital training. Social media use is positively associated with digital training participation. The results also indicate that male farmers are more inclined to participate in digital training. This observation indicates the predominance of men in farming communities, where they typically undertake most agricultural activities and tasks. The results reveal the presence of positive relation between farm size and farmers’ participation in digital training. According to previous studies, larger farms have also been reported as being more resource efficient.

Moreover, the study found off-farm income and livestock holdings as significant determinants of digital participation. This implies that small-scale and impoverished farmers are more likely to face training challenges in accessing diverse sources of information and learning opportunities. Findings show that access to farm advisory services has significant a positive effect on the participation in digital training. It can be concluded that farmers who have advisory services have more chances to participate in digital training programs. It may be the fact that FBOs provide information on digital training. Consequently, farmers should be taught about digital technologies so they may have good perceptions of the digital technologies and participate more.

Table 4 Selection equations for productivity and net farm Returns.
Full size table

ESR estimates: determinants of productivity and net farm returns

The impact of digital training on rice productivity and net farm returns for participants and non-participants was assessed and is shown in Table 5. Further, to strengthen the validity of the ESR framework, we have used alternative model specifications employing the two-stage least squares (2SLS) instrumental variable approach. The results are consistent with the ESR findings and are presented in Appendix Tables A1-A3. The analysis shows that the covariance terms for the participants are statistically significant (/lns1). This indicates that participation is indeed self-selected. It implies that the impact of participation in digital training may not be the same for non-participants, if they choose to participate in digital training. Likewise, the negative sign for /lns1 reveals the presence of a positive selection bias. It means that farm households with higher productivity and net farm returns are more likely to participate in digital training. This advocates the significant role of comparative advantage in determining productivity and net farm returns. These results are consistent with findings from49. The statistically insignificant covariance estimates for non-adopters (lns0) indicate that due to unobservable characteristics, in the absence of digital training, there would be no significant difference in the average behavior of participants and non-participants. The estimates from both specifications indicate that the residuals of both productivity and net farm returns are not significantly different from zero. This confirms that these coefficients have been consistently estimated54. The χ2 statistics presented in Table 5. It rejects the validity tests of the overidentifying restrictions which fail to reject the exclusion restriction. Hence, it confirms that the instruments used herein affect adoption only through productivity and net farm returns.

According to the results, the household head’s gender exhibited a statistically significant positive impact on farm productivity of training participants. It indicates that men who engage in digital training are more likely to experience greater productivity gains compared to women. In addition, the productivity of adopters increases with their level of experience. The results reveal that education positively impacts productivity and net farm returns. This implies that education improves farm management practices and therefore significantly enhances farm outcomes. Our results confirm a significant link between farm size and productivity and net farm returns of participants in digital training.

Access to farm advisory services was also found to have a significant positive influence on net farm returns of farmers participating in digital training. However, non-participants with access to farm advisory services did not exhibit significant changes in productivity. Thus, it resonates with the importance of integrating digital training with extension services to realize the anticipated impact on productivity and net farm returns. Regarding household wealth variables, off-farm income and livestock holdings positively and significantly influence both productivity and net farm returns of participants and non-participants of digital training. This implies that farmers with additional resources beyond farming are likely to achieve higher yields per hectare, regardless of their participation in digital training. Such resources can be utilized to purchase improved inputs and finance farm labor, thereby enhancing productivity. Furthermore, farmers with access to off-farm income realized higher net farm returns compared to their counterparts. Moreover, membership of FBOs was found to have a significant positive impact on the productivity of digital training participants.

Table 5 ESR estimates for productivity and net farm Returns.
Full size table

Welfare impacts of digital training participation

Table 6 presents the results for the average treatment effects (ATT), demonstrating the impact of digital training participation on productivity, adoption of ESA practices, and profit. These ATT estimates address selection bias, which may arise from differences in observable and unobservable characteristics between participants and non-participants, potentially confounding the actual impact of digital training participation on outcome variables.

The findings indicate that participation in digital training significantly enhances productivity, adoption of ESA practices, and profit. Farmers engaged in digital training experience an average increase in productivity of 55.21 kg per acre, an improvement in ESA practices adoption by 25.4%, and an increase in net farm returns by PKR14,365 per acre, respectively. This suggests that participation in digital programs yields substantial environmental and economic benefits for farmers and rural communities. Consequently, farmers who opt out of digital training miss the opportunity to enhance their welfare outcomes.

The sensitivity analysis results further demonstrate that training participation is significantly associated with productivity (Fig. 4), the adoption of ESA practices (Fig. 5), and net farm returns (profit) (Fig. 6). Even with the addition of noise, training participation consistently leads to a significant improvement in productivity. Similarly, the findings indicate that training participation enhances adoption of ESA practices and net farm returns. These results confirm the significant positive impact of training participation, providing additional support to the ESR results, and reinforcing the robustness of this study’s estimates regarding training participation.

Fig. 4
figure 4

Sensitivity analysis of the impact of training participation on productivity.

Full size image
Fig. 5
figure 5

Sensitivity analysis of the impact of training participation on the adoption of ESA practices.

Full size image
Fig. 6
figure 6

Sensitivity analysis of the impact of training participation on net farm returns (profit).

Full size image
Table 6 Welfare impacts of digital training Participation.
Full size table

Discussion and new insights

Discussion of results

In developing countries such as Pakistan, the unprecedented growth of ICTs presents numerous opportunities for information access55. However, traditional information-sharing methods through public extension services have been plagued with challenges related to relevance, scalability, responsiveness, and sustainability. Without adequate skills, the utilization of ICTs in agriculture remains constrained, proving to be unsustainable with limited adoption and welfare outcomes56. Therefore, there is a need for rigorous assessment to evaluate the impact of participation in digital training programs. The widespread use of ICTs for disseminating agricultural information underscores their potential impacts. However, before scaling up ICT-based training for farmers, it is imperative to assess its effects on desired outcomes.

Understanding farmers’ perceptions of current ICT-based digital technologies and their effectiveness in agriculture is essential. Our findings indicate that participants in digital training programs hold positive views regarding the suitability and practical applicability of internet information and social media channels for adopting ESA practices. For example, digital training eases the productive use of smartphone usage, which facilitates the adoption of energy-smart technologies at the farm level. These findings are consistent with previous research on farmer field schools57. These results indicate that farmers often use internet resources to seek weather forecasts33, make crop selection decisions8, and search for credit and crop subsidies34. They highlight that the integration of ICTs is transforming the landscape of agricultural practices into one that is more sustainable and resilient. Also, it is facilitating vertically integrated information exchange and knowledge sharing within the family community. These findings contrast with58 assertion that the new agricultural technology extension model enhances farmers’ technology adoption to some extent, with a partial spillover effect, and that farmers of different ages and farmland sizes benefit differently. Therefore, when promoting this extension model, it is crucial to consider information diffusion among farmers who have already adopted it and to ensure its dissemination to elderly and small-scale farmers.

The selection equation results reveal the determinants of digital training participation where internet connection access and farm radius have significant positive connection to digital training participation. Similar findings have been reported in other studies59,60. These findings indicate that farmers’ social circles were found to be enablers of their involvement in digital training. According to61, reported that the network effect of the social circle proves to be a factor that positively influences farmers’ intentions to take part in such activities. The findings also indicates that male farmers are more likely to engage in digital training. This is the same conclusion as those of the previous research that showed gender differences in agricultural learning62. The findings confirm that the bigger the farm, the more farmers are involved in digital training. Research has shown that the bigger farms have also been described as the ones that acquire more knowledge and skills to overcome risks and losses63,64. Besides, our findings show that the wealth-related variables – off-farm income and livestock holdings – have significant positive links with digital training participation65,66. This notion claims that smaller and poor farmers are less likely to take part in digital training. However, these findings contradict Hu’s et al. (2022) argument that farmers with smaller farms show greater interest in processing technologies than those with larger farms. Moreover, FBO membership has an important impact on the participation in digital training. These findings also support the findings of68.

By examining of the effects of digital training participation on the main variables, which are productivity, the adoption of ESA practices on net farm returns, the major question related to the welfare effects is answered. Our study shows that digital training participation considerably improves productivity, the adoption of the ESA practices, and net farm returns. In a similar vein42, reported the significant influence of ICTs on productivity and net farm returns in the People’s Republic of China (PRC) and Kenya, respectively. To sum up, training participation remarkably enhances household income. Our study provides support for earlier work regarding the impact of ICTs on the adoption of green technologies30 and ICTs use for crop protection and weather information33, which shows that digital training participation affects the adoption of ESA practices. These results indicate that engagement in digital programs bring large economic and environmental returns for farmers and rural communities. Thus, it is necessary to motivate farmers to use digital learning programs to improve agricultural resilience and profitability of farming in the wake of climate crises and changing market situations.

New insights

This study contributes several novel insights for policy development and practice. First, it comprehensively addresses endogeneity and selection bias, and provides three robust, relevant, and valid IVs that can be utilized in future research. Second, the study made an empirically robust assessment to confirm that observed changes in outcomes such as productivity and net farm returns are indeed attributable to participation in digital training. By doing so, the study underscores the pivotal role of training activities in realizing broader prospects for sustainable agriculture and farmer welfare. Also, the study supports the argument that declares use of ICTs without farmers’ training as a “fad.” Third, by recognizing the significant influence of internet technologies, especially mobile phones, on access to relevant information, this work highlights how participation in digital training enhances farmers’ skills in problem-solving, understanding ESA practices, and implementing them effectively at the farm level. Moreover, the study examines the mechanisms underlying ATT, which facilitates growth in productivity and profitability. In addition, it offers estimates of the heterogeneity of treatment effects on digital training participation to inform wider interventions. Notably, the findings suggest that digital training enhances farmers’ skills, and ICT technologies promote information sharing among them, which has spillover effects on rural communities.

Conclusions, policy implications, and way forward

This study examines the impact of digital training participation on the adoption of ESA practices and farmers’ welfare using cross-sectional data from 723 households in Punjab, Pakistan. To address endogeneity and selection bias, we employ endogenous switching regression and introduce three valid, relevant, and robust IVs. Descriptive analysis reveals that participants in digital training programs have a positive perception regarding the suitability and practical applicability of internet information. Farmers are actively utilizing internet resources for various purposes, including accessing weather forecasts, making crop selections, and seeking information on credit and subsidies. Regarding the key determinants of participation in digital training, our results from the selection equation indicate that access to an internet connection, positive perceptions, social networks, male gender, farm size, off-farm income, livestock holdings, and membership in FBOs are significantly associated with digital training participation. The endogenous switching regression estimates demonstrate that participation in digital training significantly influences productivity, adoption of ESA practices, and welfare. Specifically, farmers engaged in digital training experience an average increase in productivity of 55.21 kg per acre, a 25.4% improvement in ESA practices adoption, and an increase in net farm returns by PKR14,365 per acre. By analyzing the impact of digital training participation on productivity, ESA practice adoption and net farm returns, our study addresses the key question of its welfare effects. The findings reveal that digital training significantly enhances these outcomes, demonstrating substantial economic and environmental benefits for farmers and rural communities. Therefore, encouraging farmers to engage in digital learning programs is crucial for strengthening agricultural resilience and profitability amid climate challenges and evolving market conditions.

Our findings suggest several policy options for policymakers and practitioners to scale up the implementation of digital training for farmers to enhance the adoption of ESA practices and improve the welfare of rural communities. First, given the positive relationship between access to an internet connection and participation in digital training, governments of Pakistan should prioritize expanding low-cost broadband connectivity through infrastructure investments in rural and remote areas. This initiative can help bridge digital divides and facilitate vertical integration of rural communities into sustainable supply chains since most of the farmers in rural areas have no access to information related to climate-smart technology and markets for better price. Hence, greater access to information via the internet would strengthen their market linkages, support commercialization, and encourage the adoption of energy-smart agricultural technologies. Likewise, in the wake of climate crises, it can enhance the adoption of ESA practices, resilience, and farmers’ integration into markets. Second, the findings reveal that resource-endowed farmers have higher participation in digital training. Given this, there is a need to provide incentives to encourage broader participation, such as input subsidies or cash prizes to smallholder farmers. This recommendation has significant potential to enhance the participation of smallholder and impoverished farmers in the country, as they often lack the resources to cover the costs of training and time commitments. Third, as FBO membership is significantly related to training participation, it can play a pivotal role in promoting the use of ICTs. For that purpose, capacity building initiatives at the FBO and village level, including farm advisory services and awareness campaigns highlighting the effectiveness of ICTs in transitioning current agricultural practices toward ESA practices, is required. Notably, there is a dire need to establish platforms where farmers can share information, knowledge, and field experiences, and seek guidance from experts can significantly enhance learning and knowledge-sharing among farmers. These measures can help realize a transformative shift toward energy-efficient, sustainable, and climate-friendly agriculture. Further, an integrated implementation of these policy tools will accelerate greater economic benefits for farmers in terms of improved productivity, and income.

Last, this study investigates the impact of digital training participation on ESA adoption and welfare outcomes. Future research studies can make comparisons between productivity and income levels of training participants and other internet users who do not participate in training. Likewise, future research can examine the demand, effectiveness, and availability of digital technologies in rural communities.