1 INTRODUCTION

The rapid spread of information and communication technology (ICT) in Asia offers the promise of a “digital revolution” for agriculture in the region (Asian Development Bank [ADB], 2021). The dissemination of ICT supports farmers by improving access to markets through real-time data on prices; weather forecasts; and information on pests, seed varieties, and planting techniques (Aker, Ghosh, & Burrell, 2016; Fabregas, Kremer, & Schilbach, 2019; Trendov, Varas, & Zeng, 2019). This can in turn accelerate the transformation of agriculture and food system by increasing farm productivity and efficiency; improving farmers' access to output, input, and financial markets; and facilitating the design and delivery of agriculture policies (Schroeder, Lampietti, & Elabed, 2021). Especially encouraging is the potential of mobile technologies, particularly smartphones, in disseminating agricultural extension services and advice to smallholder farmers in emerging markets and developing economies (Aker et al., 2016; Baumüller, 2018; Fabregas et al., 2019; Trendov et al., 2019). Evidence to date suggests that digitally delivered advice to farmers in these economies increases yields by around 4% and the probability of adopting recommended inputs by 22%, which may be critical to reducing poverty, food insecurity, and environmental risks (Fabregas et al., 2019).

However, the potential of ICT for reducing poverty and transforming agriculture in Asian emerging markets and developing countries could be thwarted by the “digital divide” within rural areas (Deichmann, Goyal, & Mishra, 2016; Fabregas et al., 2019; International Telecommunications Union [ITU], 2021; Mehrabi et al., 2021; Quimba, Rosellon, & Calizo, 2020; Schroeder et al., 2021; Trendov et al., 2019). That is, substantial gaps exist between different farming populations and regions in terms of both the availability of ICT and data services and their ability to use them. In rural Asia, the digital divide between those who do and those who do not have access to ICT is defined by two interrelated factors: access to technology and adequate skills to use it (ITU, 2021; Quimba et al., 2020; Schroeder et al., 2021; Trendov et al., 2019).

Both factors are linked to poverty. First, poorer and more remote communities in rural Asia have limited access to ICT because of high costs and a general lack of reliable electricity and other infrastructure. Second, the smallholders within these communities have insufficient human capital skills and knowledge to reap the benefits of digital technologies for farming, and they do not have the income and wealth to invest in the necessary education and training. In short, isolated rural communities are most in need of improved digital connectivity to compensate for their remoteness, but they are often the least connected to allow access to ICT. Equally, poorer smallholders in such communities could benefit significantly from the additional productivity and income from adopting ICT for farming but lack the skills and means to do so.

This interplay between poverty, remoteness, and low ICT skills in rural areas can lead to digital poverty traps. Smallholder households inhabiting regions with poor market access and infrastructure often have insufficient wealth to accumulate the human capital necessary to adopt ICT for agriculture. Consequently, such households are prone to persistent poverty, from which they are unable to escape. If sufficiently large numbers of smallholders and communities in rural Asia suffer from digital poverty traps, then this outcome may have economy-wide implications. First, it will slow down the ability of ICT to transform agriculture and food systems in Asia by boosting farm productivity and efficiency (ADB, 2021; Schroeder et al., 2021; Trendov et al., 2019). Second, if digital poverty traps are widespread, they will perpetuate the problem of “poverty hotspots” in rural Asia, which is becoming a hallmark of current and projected global poverty trends (Ahmadzai, Tutundjian, & Elouafi, 2021; Barbier & Hochard, 2018; Cohen, Desia, & Kharas, 2019; Graw & Husmann, 2014; Kharas & Dooley, 2022).

An important contribution of this paper is to explore why digital poverty traps may be occurring in rural Asia and to suggest policies to help alleviate the problem. Drawing on novel data and trends, the paper first examines the digital poverty traps in Asia and focuses on its two driving factors, that is, low accessibility to ICT due to a lack of infrastructure and inadequate skills to use digital technologies. The paper then develops a model to show how these two conditions can lead to digital poverty traps for rural smallholders in remote regions with poor market access. Since such a digital poverty trap model has not been previously developed, it can be considered a novel contribution of this paper. The insights gained also help inform policies to overcome such traps. Increased public investments to extend “last-mile infrastructure” in digital services for smallholders in remote regions of rural Asia are necessary but not sufficient for overcoming such traps. Other policies and investments are also needed, such as expanding off-grid solar and other forms of reliable and affordable electricity supplies, improving rural infrastructure, developing appropriate ICT business models and services, and advancing the digital literacy of smallholders. Such a comprehensive strategy for overcoming digital poverty traps in rural Asia is essential if the promise of a “digital revolution” for agriculture in the region is to be realized.

The organization of this paper is as follows. Section 2 provides background on the emergence of a possible digital divide within rural Asia. There is growing evidence that large numbers of poorer smallholders in more remote communities may be prevented from adopting ICT for farming because they lack the skills and means to adopt these technologies. Such conditions can lead to digital poverty traps, whereby significant numbers of poor smallholders inhabiting remote regions are unable to take advantage of the benefits of ICT for agriculture and thus remain mired in poverty. Section 3 then develops a model to illustrate this digital poverty trap phenomenon. The results indicate that, if a smallholder household located in a remote region is too poor and cannot accumulate sufficient ICT skills, it cannot escape a long-run poverty trap equilibrium characterized by low levels of per capita income and capital stock. Section 4 discusses the policy implications of such digital policy traps for disseminating ICT for farming in rural Asia. Section 5 summarizes the main findings and the key areas for future research.

2 THE DIGITAL DIVIDE AND POVERTY

Over the past two decades, ICT infrastructure, access, and use have expanded considerably in 37 Asian emerging markets and developing economies (see Figure 1). They include 8 countries from Central Asia, 8 from South Asia, and 21 from East Asia and the Pacific (see Table A1 in Appendix A). This group of countries also consists of 24 low- and lower-middle-income economies and 13 upper-middle-income countries. Despite this diversity in geography and income, ICT has spread quickly across developing Asia since 2000.

Whereas in 2000 less than 2% of the population in Asian emerging markets and developing economies had access to the internet, by 2020 over 54% of the population were internet users (see Figure 1). Mobile cellular use also grew rapidly over this period. In 2000, there were around two subscriptions per 100 people, but by 2020 they numbered 102 subscriptions per 100 people. Both the increase in internet and mobile cellular use has been spurred by the expansion in electricity access, especially in rural areas, which increased from 61% of the rural population in developing Asia in 2000 to 92.5% in 2020 (see Figure 1). Mobile cellular coverage, which is the percentage of the population that lives within reach of a mobile cellular signal, is also close to 100% across Asia (ITU, 2021). Just over 96.1% of the population is now within reach of a third-generation (3G) signal, and 94.2% is within reach of a long-term evolution mobile broadband signal (ITU, 2021).

Yet, despite this progress in ICT coverage and use over the past 20 years, there remains a significant rural–urban digital divide in Asia. In 2019, only 37% of rural households had access to the internet in Sri Lanka, compared to 70.4% of urban households (ITU, 2021). About 46% of farming households across Asia have internet access, but only 31% have access in India, 21% in Pakistan, and 12% in Tajikistan (Mehrabi et al., 2021). Figure 2 further indicates the extent of this digital divide across low- and middle-income economies in Asia. Across these countries, the share of the population using the internet is negatively correlated (r = −0.66) with the share of the population that is rural.

Computer and digital literacy are also significantly lower for those living in rural as opposed to urban areas. For example, 40.4% of urban households in Sri Lanka are considered computer literate, whereas in rural areas the proportion is only 27.5% (Quimba et al., 2020). Levels of education, household income, and the degree of urbanization appear to be important factors behind the digital divide in Asia (Maji & Laha, 2020, 2022).

Poverty, especially in rural areas, appears to be a key underlying cause. Over the past 20 years in Asian emerging markets and developing countries, populations have become less rural and extreme poverty has fallen (Figure 3). Access to key basic services in rural areas, such as drinking water and sanitation, has also improved. But poverty still remains a pervasive problem in the region. Around one in five people on average across Asian emerging markets and developing economies lives in poverty, and this share rises to 23.6% in the low and lower-middle-income countries of the region (Barbier, 2022). Moreover, as with all developing regions in the world, the poor in Asia are increasingly rural, dependent on agriculture, and are predominantly young (Castañeda et al., 2018).

There is also growing evidence that the digital divide in Asia and other developing regions is globally linked to the problem of persistent poverty in rural areas characterized by large numbers of smallholders farming disadvantaged and remote agricultural areas, who lack the skills to access ICT or the wealth to invest in acquiring digital literacy (Aker et al., 2016; Birner, Daum, & Pray, 2021; Deichmann et al., 2016; Fabregas et al., 2019; ITU, 2021; Mehrabi et al., 2021; Quimba et al., 2020; Schroeder et al., 2021; Trendov et al., 2019). This phenomenon is reflected in the positive correlation (r = 0.60) between internet access and a key indicator of the extent of rural poverty, which is the share of the rural population that has access to at least basic sanitation (see Figure 4). Overall, countries with greater rural poverty also appear to have a smaller share of their population with access to the internet.

Within rural areas of Asia and elsewhere, there is also a substantial gap between different farming populations and regions in terms of both the availability of ICT and data services and their ability to use them. Only 24%–37% of farms less than 1 ha in size are served by 3G or 4G mobile cellular services, compared to 74%–80% of farms greater than 200 ha in size, and croplands with severe yield gaps, climate-stressed locations, and food-insecure populations have poor service coverage (Mehrabi et al., 2021). The use of basic mobile phone technology as a form of digital extension and dissemination of agricultural advice appears to favor richer, younger, and more educated farmers with better digital access and skills (Fabregas et al., 2019). Such a digital divide within rural areas and between different farming populations might perpetuate the growing problem of rural poverty traps.

Currently, 80% of the extreme poor and 75% of the moderate poor live in rural areas (Castañeda et al., 2018), and within these areas, the poor are increasingly concentrated in poverty “hotspots,” which are often characterized by poor agricultural potential, susceptibility to drought and climate change, and insufficient market access (Ahmadzai et al., 2021; Barbier & Hochard, 2018; Cohen et al., 2019; Graw & Husmann, 2014; Kharas & Dooley, 2022). Of particular concern is how many poor people in these areas are smallholders farming 1–2 ha of land or less. Smallholders dominate agriculture in much of rural Asia. For example, in East Asia and Pacific (excluding China) and South Asia, about 70%–80% of farms are smaller than 2 ha and they operate on 30%–40% of the agricultural land (Lowder, Skoet, & Raney, 2016).

Many smallholder farmers lack the skills to access ICT or the means to acquire them (Aker et al., 2016; Birner et al., 2021; Deichmann et al., 2016; Fabregas et al., 2019; ITU, 2021; Mehrabi et al., 2021; Quimba et al., 2020; Schroeder et al., 2021; Trendov et al., 2019). In rural Asia, they are often found in remote communities with limited access to digital technologies because of high costs and a general lack of reliable electricity and other infrastructure (Rapsomanikis, 2015). In addition, the poor rural households and farmers within these communities are unable to reap the benefits of ICT for farming, as they do not have the income and wealth to invest in acquiring the education and training necessary to attain such skills and knowledge (ADB, 2021; Schroeder et al., 2021; Trendov et al., 2019). Such conditions could produce large digital poverty traps in parts of rural Asia, whereby significant numbers of poor smallholders inhabiting remote regions are unable to take advantage of the benefits of ICT for agriculture and thus remain mired in poverty. The next section explores further how a digital poverty trap might arise for a representative poor smallholder living in a remote and agriculturally disadvantaged region.

3 WHY DO DIGITAL POVERTY TRAPS OCCUR?

This section develops a single-equilibrium poverty trap model to illustrate why smallholder households inhabiting remote regions, who have insufficient wealth and thus difficulty in accumulating the human skills to adopt ICT for agriculture, are prone to persistent poverty. The normal assumption for such models is to compare outcomes for two representative households that may differ in a crucial way, such as the type of geographical location, but otherwise are identical in initial household characteristics, such as education, literacy, and land holding size (Ghatak, 2015; Kraay & McKenzie, 2014). This is the approach adopted here.

Consequently, the following model distinguishes two otherwise identical smallholder households farming agricultural land in different geographical regions. The first household lives in a region that has good market access. This means that the household is near to not only input and output markets for farming but also related services, such as agricultural extension, credit and banking facilities, and education and training opportunities. The smallholder household in such a favorable location is therefore able to accumulate relatively easily the human capital needed for incorporating new ICT into agricultural production. In comparison, the second household is located in a remote region with poor market access, which means it has much more difficulty in accumulating ICT human skills. Because of its remote location, the second household also uses more traditional farming techniques than the household with good market access.

This digital divide is critical to generating a poverty trap for disadvantaged smallholder household. Compared to the household farming in the area with favorable market access and ICT human capital accumulation, if the household located in the remote region is initially too poor, it cannot escape a digital poverty trap. This outcome is characterized by a stable steady state with low levels of per capita output and capital stock and insufficient accumulation of human skills to adopt ICT for farming. If the household falls below a threshold level of initial capital or income, it will remain caught in this digital poverty trap and converge to the low-productivity steady state. This outcome can be illustrated as follows.

Now suppose a second smallholder household is located in a remote region with poor market access. It has the same initial human capital H₀ as the first household (e.g., same level of education, and literacy), but because of its disadvantageous location, the remote household has more difficulty in obtaining ICT human capital skills. The remote location also means that the household uses only traditional production methods, which implies a fixed level of technology that yields lower total factor productivity than on land with more favorable market access; that is, $B<A$. Thus, for the more disadvantaged smallholder household, per capita output 1 can be specified as $q=B{k}_1^{\mathrm{\alpha }}{H}^{\mathrm{\beta }}$, and it is able to accumulate human capital at the rate $H=H_0{\left(k-k_1\right)}^{\mathrm{\epsilon }},0\le \mathrm{\epsilon }<\mathrm{\gamma }$. Note that the latter condition implies that the household in the remote region has more difficulty in accumulating human capital for ICT than the smallholder household located in the region with more favorable market access.

For either household, let per capita income be $y=\mathit{rk}+w$, where r is the rate of return on capital and w is the opportunity cost of using its own household labor in production, which is equivalent to the local market wage rate. Assume that output prices are normalized to 1, and household capital depreciates at a constant rate δ. If all households save the same fraction s of per capita output and all savings are devoted to capital accumulation, the net change in the capital–output ratio over time $\dot{k}$ is governed by $\dot{k}=\mathit{sy}-\left(\mathrm{\delta }+n\right)k$, where n is the rate of growth in household labor.

Figure 5 indicates that, if a household located in a remote region is too poor, it cannot escape a poverty trap outcome, which is the stable steady state with low levels of per capita output and capital stock k**. This outcome occurs for any household with initial capital per person below a critical threshold value k_T. It is called a digital poverty trap because a smallholder household in this remote region that is initially too poor (i.e., $k_0<k_T$) cannot accumulate enough human capital skills to adopt ICT for farming and thus converges to a low-level equilibrium of output and capital per person as represented by k**.

Consequently, if a smallholder household located in a remote region is initially too poor $k_0<k_T$, it will not be able to accumulate sufficient skills to adopt ICT for farming and will fall into a digital poverty trap. If the smallholder is sufficiently wealthy $k\ge k_T$, it can afford to accumulate human capital for digital farming. Like the household living in a more favorable region with good market access, it can avoid the digital poverty trap outcome k^** depicted in Figure 1.

4 POLICY IMPLICATIONS

Policies to overcome digital poverty traps in rural Asia must tackle both the lack of access by smallholders to ICT for farming and their insufficient skills to use the new technologies (Deichmann et al., 2016; Fabregas et al., 2019; ITU, 2021; Mehrabi et al., 2021; Quimba et al., 2020; Schroeder et al., 2021; Trendov et al., 2019).

As Mehrabi et al. (2021) point out, substantial gaps still exist in the availability of and access to ICT for many smallholders in rural Asia and other developing regions, which must be addressed if the potential gains of these technologies for farming are to be realized. The authors suggest that an important and necessary complement to extending “last-mile infrastructure” in digital services for smallholders in remote regions is to improve the supply of cheap and reliable sources of energy, such as replacing diesel generators with cost-efficient fuel cell generators or solar energy. Although the recent growth of electricity services in rural areas of Asian emerging markets and developing countries is encouraging (see Figure 1), many rural households in remote regions still do not have access to a reliable supply of electricity with shorter hours of power supply, frequent power cuts, and voltage fluctuations (Rockefeller Foundation, 2021). This can be a significant constraint that can limit the adoption and impact of ICT in agriculture for many poor smallholders in such regions (Deichmann et al., 2016; Mehrabi et al., 2021; Schroeder et al., 2021; Trendov et al., 2019).

One possible solution is the expansion of solar energy “safety nets,” which have been proposed for Bangladesh, India, and other Asian emerging markets and developing countries (Barbier, 2022; Zaman, van Vliet, & Posch, 2021). Solar safety nets are targeted social assistance programs to provide solar power as an off-grid solution to solving the lack of access to energy for poor rural households in remote locations. Both Bangladesh and India have piloted such schemes, which provide clean energy access to remote rural households through the free distribution of solar home systems and panel installations (Zaman et al., 2021). Similar solar safety net programs could be funded through a fossil fuel “subsidy swap,” whereby the savings from subsidy reform for coal, oil, and natural gas consumption are recycled to expand such programs throughout rural Asia (Barbier, 2022).

Off-grid solar and other solutions, such as inexpensive fuel cell generators and small-scale renewable energy systems (10 MW or less), are essential to providing reliable and affordable energy for the adoption of ICTs for farming by poor smallholders in rural Asia (Mehrabi et al., 2021; Rockefeller Foundation, 2021; Trendov et al., 2019). This is also important for reducing the high upfront costs for smallholders in adopting ICT for more efficient application of irrigation, seeds, fertilizers, and pesticides (Trendov et al., 2019). Off-grid solar may also facilitate irrigation, which is often an essential input for smallholders realizing many of the benefits of ICT for farming (Lefore, Closas, & Schmitter, 2021). Finally, investments in solar and small-scale renewable energy in remote rural regions of Asia could yield wider economic benefits, such as boosting agricultural processing and generating or improving millions of jobs (Rockefeller Foundation, 2021).

In addition to inadequate energy supply, the lack of other complementary rural infrastructure–such as roads, credit facilities, postharvest storage, and logistics–can limit the adoption and impact of ICT for agriculture in remote regions (Deichmann et al., 2016; Schroeder et al., 2021). Public investment in overcoming such bottlenecks to agricultural development in remote rural areas may also be essential to the adoption of ICT for farming by poor smallholders. As pointed out by Deichmann et al. (2016, p. 31), ICT may improve the dissemination advice to farmers but “they may not necessarily be able to act on that information because of inaccessibility of alternative markets and the complex interlinked relationships between buyers and sellers in poor developing economies.” Studies of poverty hotspots in rural areas also indicate that agricultural research, rural roads, and education are the most effective investments in improving market access and reducing poverty (Ahmadzai et al., 2021; Barbier & Hochard, 2018; Cohen et al., 2019; Graw & Husmann, 2014; Kharas & Dooley, 2022).

Digital payments are an important component of the availability and spread of ICT for agriculture, yet they may also be a significant obstacle to the adoption of the technologies by poor smallholders in remote regions (Aker et al., 2016; Baumüller, 2018; Birner et al., 2021; Deichmann et al., 2016; Fabregas et al., 2019; Schroeder et al., 2021). One key challenge for improving the supply of ICT to underserved rural populations is for service providers to target a large enough customer base while keeping rates affordable and investment costs low (Birner et al., 2021). Promising applications that smallholders appear willing to pay for include largely automated information services, insurance pay-outs triggered by sensory data, and mobile banking or supply chain management using mobile technologies to collect and transmit delivery data (Baumüller, 2018). In contrast, much more general information obtained digitally, such as general agroeconomic advice or market prices, may be less suitable because they do not provide the localized and timely information required by smallholder farmers, which can undermine their incentives to pay (Aker et al., 2016; Baumüller, 2018; Birner et al., 2021; Fabregas et al., 2019).

Public policies can play a key role in encouraging service providers to develop ICT systems that disseminate information of use to smallholder farmers and communities through determining what agriculture-related data are most valuable to poor smallholders, designing appropriate legal and regulatory frameworks, and putting in place appropriate and financially sustainable business models for such services (Baumüller, 2018; Birner et al., 2021; Deichmann et al., 2016; Fabregas et al., 2019; Schroeder et al., 2021). For example, the most common ICT platform for disseminating agricultural information and collecting payment is mobile phone services (Baumüller, 2018; Birner et al., 2021; Fabregas et al., 2019). But studies suggest that such services are more suitable for smallholder farmers if the service provider or its intermediaries can connect farmers to a range of services, such as collection points, processors, financial institutions, extension agents, farmers' organizations, agricultural input suppliers, or dedicated m-service agents (Baumüller, 2018). In addition, a range of business models in developing countries are possible for such services, including providing them for free to farmers, traders, and consumers enabled by funding from network operators, donors, and input companies or providing them against a fee, which can be paid as part of a subscription or on a pay-per-use basis (Birner et al., 2021).

Finally, as this paper has demonstrated, a key factor underlying digital poverty traps is that smallholder households inhabiting remote regions have insufficient wealth to accumulate the human capital necessary to adopt ICT for agriculture. Addressing this challenge will require specific policies and investments to develop affordable skills training and education programs targeted at improving smallholder adoption of digital technologies for farming (Aker et al., 2016; Schroeder et al., 2021; Trendov et al., 2019). Such programs can stimulate demand for ICT from farms but only if they significantly improve the digital literacy of smallholders, which are the basic skills that enable farmers to use digital tools, such as cell phone messaging, apps, and platforms developed for farmers. This also suggests an important role for extension and advisory services to deliver such support to farmers in improving digital literacy and familiarity with the agricultural information and advice available through ICT. It also requires tailoring different forms of digital learning to the diverse stakeholders in remote rural regions, such as women who dominate smallholder farming and different racial and ethnic groups (Aker et al., 2016; Rapsomanikis, 2015). Extension services can also work with existing farmers' organizations and associations to facilitate farmers' education in the use of ICT by using such groups to promote digital literacy and community-wide learning. In addition, government partnerships with private service providers of mobile networks and platforms can be used to design technical support and digital learning programs designed for smallholders in remote regions with limited digital skills and means to acquire them.

5 CONCLUSION

In rural Asia, the digital divide between those who do and those who do not have access to ICT is defined by two interrelated factors: access to technology and adequate skills to use it (ITU, 2021; Quimba et al., 2020; Schroeder et al., 2021; Trendov et al., 2019). Both factors are linked to poverty. Poorer and more remote communities in rural Asia have limited access to digital technologies because of high costs and a general lack of reliable electricity and other infrastructure constraints. Second, the smallholders within these communities may lack the human capital skills and knowledge to reap the benefits of ICT for farming, and they do not have the income and wealth to invest in acquiring the education and training necessary to attain such skills and knowledge. As this paper has demonstrated, the result could be a significant digital poverty trap for many poor smallholders living in remote and agriculturally disadvantaged regions.

Such digital poverty traps could be considerable in Asian emerging markets and developing countries. Smallholders dominate agriculture in much of rural Asia, and they are often found in remote regions without adequate market access or infrastructure (Lowder et al., 2016; Rapsomanikis, 2015). Digital poverty traps could further exacerbate the growing problem of poverty hotspots, which are often characterized by poor agricultural potential, susceptibility to drought and climate change, and insufficient market access (Ahmadzai et al., 2021; Barbier & Hochard, 2018; Cohen et al., 2019; Graw & Husmann, 2014; Kharas & Dooley, 2022). Currently, 80% of the extreme poor and 75% of the moderate poor live in rural areas, and they largely depend on agriculture and are predominantly young (Castañeda et al., 2018). Digital poverty traps will not only perpetuate poverty in rural Asia but also thwart the widespread potential of ICT for agriculture to foster rural development and economy-wide growth.

As this paper has emphasized, increased public investments to extend “last-mile infrastructure” in digital services for smallholders in remote regions of rural Asia are important but not sufficient for overcoming digital poverty traps. Other policies and investments are also needed. These include the expansion of off-grid solar power and other ways of providing reliable and affordable electricity supplies; improving rural infrastructures such as roads, credit facilities, postharvest storage, and logistics; encouraging service providers to develop ICT systems and business models that disseminate information of use to smallholder farmers and communities; and policies and investments to develop affordable skills training and education programs targeted at improving smallholder adoption of digital technologies for farming. Such a comprehensive strategy for overcoming digital poverty traps in rural Asia is essential if the promise of a “digital revolution” for agriculture in the region is to be realized.

If such a strategy is successful, it could also lead to positive externalities from ICT adoption at the community level in disadvantaged rural areas of Asia. As individual smallholders gain more skills to adopt ICT and more “last-mile” infrastructures for these technologies are provided, there would be positive spillovers from household to household and, eventually, across entire communities. Although the model of this paper does not include such community-level externalities, doing so would lead to multiple equilibria outcomes, including the possibility that positive spillovers would boost ICT human capital accumulation by poor smallholders and allow them to escape the long-run poverty trap.

Future research should identify and develop the key elements required for an effective strategy to overcome digital poverty traps in Asia. One important priority is to identify the smallholder groups and remote regions that are highly susceptible to digital poverty traps and their specific infrastructure and skill requirements. As a starting point, this paper has highlighted the key characteristics of such traps and the corresponding infrastructure and skills gaps. But clearly answering the question of how we can identify digital poverty traps in rural Asia requires additional attention.

This paper has also identified how the digital poverty trap may be a constraint on boosting farm productivity and incomes among smallholders in disadvantaged regions and communities in rural Asia. Others have also noted how this constraint could be a major deterrent to the promise of digital technologies in agriculture (Aker et al., 2016; Fabregas et al., 2019; Mehrabi et al., 2021; Schroeder et al., 2021; Trendov et al., 2019). There is clearly a need for additional case studies and empirical evidence of the extent to which digital poverty traps impact farm productivity and incomes for different regions and communities in rural Asia. Such studies and evidence will be essential in designing a comprehensive policy strategy for addressing this problem.

Lastly, more research is required into appropriate and financially sustainable ICT systems and business models for disseminating information of use to poor farmers and communities, determining what agriculture-related digital data are most valuable to smallholders, and designing the legal and regulatory frameworks needed for such services to be effective.