Digital connectivity at the upstream end of value chains: A dynamic perspective on smartphone adoption amongst horticultural smallholders in Kenya

Introduction

‘Can the Internet reboot Africa?’ This headline in The Guardian (2016) hints to hopes that the Internet could be a stepping stone to leapfrog some of the most pressing challenges in economic development in Africa. It enqueues to a line of thought that has gained widespread momentum in the media, politics, and research community and has led to a debate on the implications of digital connectivity. Since, (digital) connectivity ‘[…] is seen as something that “wants”, “makes” and “does”, rather than as grounded relationships between people, places and processes’ (Graham et al., 2015), the proliferation of information and communication technology (ICT) is discussed controversially with respect to its promises and impact. Under the terminology ‘ICT for development’ (ICT4D), policy papers and development agendas, in particular, tend to envision and promote a transformative potential for the development of economies that lag far behind in terms of connectivity (Fuchs and Horak, 2008). The most common argument for such aspirations is that the broad dissemination of ICTs and especially the Internet can foster inclusive growth by reconfiguring economic processes towards increasing domestic and international competitiveness and, consequently, enable greater value appropriation for industries in the global South (Friederici et al., 2017).

This narrative of ICT being a panacea for economic development is however, widely challenged as being inconclusive and characterized by ‘technological determinism’ (Friederici et al., 2017). More sceptical positions highlight the uneven impact of digital connectivity (Carmody, 2012) and question a far-reaching transformation of global production regimes, power asymmetries and socio-economic disparities in favour of producers from the global South (Etzo and Collender, 2010; Heeks, 2017). Although producers are increasingly digitally connected, they remain thinly integrated. This implies that despite using ICTs for face-to-face, real-time activities, more sophisticated, digitally automated tasks are not pursued (Murphy and Carmody, 2015). Accordingly, thin integration usually cannot significantly reconfigure the relationship between producers and the global economy (Murphy and Carmody, 2015).

Positions that are most opposed to the optimistic narrative go even further by theorizing the Dark Sides of ICT4D (Unwin, 2017). Referring mainly to Internet-based connectivity, it is argued that increased ICT diffusion merely enhances inequalities in an unequal world, which – in value chain terminology – implies that lead firms and buyers from the global North can extend their power, whilst dismantling the agency of workers and producers in the global South.

The fierce debate on digital connectivity is also reflected by the large strand of literature specifically addressing the impact on (smallholder) agriculture in African countries. Remarkably however – and this is the departing point for this work – most scholars base their argumentation on evidence on simple phone-based ICT use, whereas Internet connectivity is often excluded from the analysis or adumbrated when researching agricultural value chains at the producer level (e.g. Baumüller, 2013; Krone et al., 2015; Nakasone et al., 2014; Wyche and Steinfield, 2016). Due to the relative novelty of widespread Internet adoption amongst micro-sized firms in the global South, this is not surprising. However, with increasing Internet availability and decreasing connection costs, the spectrum of ICT artefacts and, thus, digital practices has widened substantially. With emergent digital technologies and practices amongst African smallholders, producers become, however, accessible for digital modes of monitoring, controlling and conclusively Internet-based governance by lead firms to a so far unprecedented extent. For both dynamics, the emergence of digital practices amongst producers as well as the extent of digital coordination and control by lead firms, current empirical evidence is scarce. We, therefore, contribute to ‘updating’ the ICT4D debate based on the example of Kenyan smallholders participating in export-oriented value chains for horticultural produce. By comparing two consecutive surveys covering a period from 2013 to 2017, we establish a dynamic perspective on the evolution and the impact of Internet adoption to discuss the following three research questions: to what extent did Kenyan smallholders adopt the Internet? Which digital practices in relation to agricultural value chains do smallholders apply? And lastly, how does increased Internet use amongst smallholders affect inter-firm coordination at the upstream level of agricultural value chains?

This work is divided into five sections. After the introduction, we connect the literature on horticultural value chains and digital connectivity both conceptually and empirically by introducing the general structure of export-oriented chain arrangements in Kenya and then summarizing the changes discussed that accompany the adoption of smartphones. We continue by describing our case study and our methodological approach before we present the results that follow the structure of the three research questions. Finally, the discussion and conclusion section serve to discuss and summarize the results and link them to the general ICT debate.

Horticultural value chains and digital connectivity in Kenya

Global value chain (GVC) integration of horticultural smallholders is usually highly dependent on international lead firms (e.g. EU retailers) and their regional turnkey suppliers (exporters) (Dolan and Humphrey, 2000; Humphrey, 2008; Ouma, 2010). Using their international networks, their market proximity and their expertise on standardization and certification measures, they are pivotal gatekeepers for smallholders intending to produce for consumer-driven GVCs (Pelupessy and Van Kempen, 2005). Whilst GVC integration comes with the benefit of extended, potential high-value markets, it is, however, shaped by strong power asymmetries. As visually described in the seminal theorization on GVC governance by Gereffi et al. (2005), GVCs can articulate in captive governance forms with high degrees of explicit coordination and power asymmetries when three key conditions apply: high complexities of knowledge transfers, high abilities to codify these and low capabilities in the supply base. In Kenyan export-oriented horticulture, this is especially the case at the interface between producers and turnkey suppliers: here, stringent product and process standards (particularly GlobalGAP) and their certification lead to high complexities of information and knowledge transfers (Asfaw et al., 2010). Lead firms and their subsidiaries can, however, use their economies of scale and a set of transaction cost-reducing measures such as farmer group-oriented trainings and production-pooling as well as standardized input toolkits to simplify and reduce the costs of the transactions (Humphrey, 2008). Further, related to low capabilities and capacities in the supply base, many smallholders cannot access export markets independently and are in parallel, easy to replace (Kariuki, 2014). Thus, GVC integration usually closely resembles the typology of captive governance for smallholders: lead firms and their turnkey suppliers can widely govern under which circumstances (requirements, production model), by whom (which smallholders) and for what (price) production takes place. Smallholders are, consequently, confronted with a non-negotiable take it or leave it deal.

Whilst such highly standardized and captive modes of GVCs for horticultural production have received particular attention, Dannenberg and Nduru (2013) further identified trading practices used to circumvent formal standardization processes. Here, intermediaries (brokers) facilitate commodity flows around standardized certification schemes. Smallholders who are able to independently produce high-grade products that can easily be mixed with certified products can, therefore, indirectly access GVCs through intermediaries via arm’s length market-based exchanges. Hence, the GVC integration of smallholders must not necessarily be highly standardized by contract-based governance structures but can also be shaped by more flexible and informal arrangements with a market-based character. A prerequisite for such intermediated GVC integration is the capacity and capability of smallholders to cultivate high-grade produce and also navigate – or coordinate – along the marketing pathways necessary to informally access GVCs.

In contrast to export-oriented and captive chain arrangements, regional value chains (RVCs) in Kenyan horticulture are predominantly characterized by market-based governance structures. Oriented towards domestic markets, most horticultural RVCs in Kenya are not in line with the GVC concept outlined by Gereffi et al. (2005) per se as they lack the territorial decoupling or international vertical disintegration (Krishnan, 2018). Nevertheless, they play a significant role for domestic producers and consumers and are analysed using common value chain concepts (cf. Barrientos et al., 2016). Contrary to captive GVCs but similarly to intermediated GVCs, farmers participating in RVCs do not have to comply with standardized, captive structures. Here, the interaction with buyers usually occurs on an arm’s length basis (Krishnan, 2018). Accordingly, most RVCs are defined by constantly changing competitive market conditions between farmers, intermediaries (brokers, traders) and direct buyers.

Summarizing these three different value chain forms in Kenyan horticulture, it is important to note the multiplicity of chain options for smallholders. Farmers rarely integrate in one chain, but alternate for strategic or opportunistic reasons between different chain arrangements, thus often producing for domestic and export-oriented markets in parallel (Barrientos et al., 2016; Krishnan, 2018). For smallholders, this complex and highly competitive setting implies a wide and dynamic set of production and marketing practices. Success – or even potential success – in one chain alternative can, consequently, imply increased opportunity costs for the remaining alternatives. This observation is not only based on Krishnan’s (2018) detailed work on strategic diversification processes but also emerged also as a recurring discussion point during the case study presented in this paper: standardized, captive GVC integration is certainly not the ultimate goal for every smallholder. Rather, in the case of given conditions of knowing how to produce for which market and being able to coordinate such market activities, smallholders establish their niche(s) along the value chain options they assess to be best suited to them.

The multiplicity of chain logics as well as the opportunity to alternate between different chain forms emanates into a necessary consideration of producer agency and bargaining power to finalize this section. Bargaining power, as a crucial resource to negotiate how different chain actors interact, can increase when chain actors have reasonable alternatives (e.g. other chain options) and the capacities and capabilities to use these (see Dallas et al., 2019). Viable alternatives to one chain option can consequently create leverage to negotiate better terms for producers or indirectly drive less viable alternatives to improve conditions for producers. Both logics, direct bargaining for better terms as well as the rather indirect adaptation to opportunity costs created by alternative chain options can thus force lead firms to allow for higher value capture and/or better labour conditions. In sum, where there are viable alternatives for producers, they can better negotiate the terms of their global integration.

Smallholder practices in value chains and increasing connectivity

With the widespread use of mobile phones ¹ (texting and calling) amongst African smallholders, a comprehensive spectrum of academic work has contributed to the ICT4D debate (see Aker et al., 2016). These contributions can be roughly summarized based on two major domains with relevance for this article. First, changing connectivity affects what and how knowledge is transferred. Second, increasing connectivity is discussed to reconfigure how coordination at the interface of different chain segments takes place.

The shift towards smartphones amongst smallholders

The shift from phone-based towards Internet-based practices has spurred new hopes and big visions for technology-induced leapfrogging processes (Friederici et al., 2017). Particularly with regard to smallholder farmers, knowledge flows – vertically and horizontally – are discussed as being enriched and intensified in terms of their reach, the complexity of the knowledge and the mode of their transfer. Drawing from her insights on phone-based extension and marketing services in Kenya, Baumüller (2015) suggests that digitally integrated farmers will access a wide variety of media, such as video, images and long texts, and draw from external, international knowledge sources online. In line with such expectations, various online services have been developed offering digital extension to farmers in recent years (e.g. Gandhi, 2016; Karubanga et al., 2016). Furthermore, ‘large virtual networks of farmers that span across countries and borders […]’ (Baumüller, 2015: 50) and exchange knowledge between farmers on a multilateral basis are discussed. These bright expectations of digitally enhanced knowledge flows nurture the hope for emancipation from established vertical knowledge sources (Duncombe, 2018).

The Internet’s technological advances are further associated with a reformulation of the logics under which conditions value chains are coordinated and monitored. For instance, GSMA, the most important representative of mobile network operators worldwide, promotes the digitization of ‘the last mile in value chains’ (GSMA, 2018). Particularly for GVC arrangements, they recommend not only the digital implementation of information services but also of financial transactions as well as product and farm tracking. This set of solutions is supposed to ‘[…] allow lead firms to address a wide array of business challenges and increase farmer loyalty, operational efficiency, and real-time visibility in the last mile’, whilst increasing transparency and decreasing transaction costs for farmers. Moreover, the emergence of digital markets is expected to transform marketing logics. Farmers could use digital marketing platforms as digital middlemen, where trust is built via images of produce, the tracking of produce and user ratings (cf. Macharia, 2013).

These various imaginations of digital connectivity – we have only illustrated a selection of an emerging activity field – feed into the larger debate on the idea of a digital ‘technological fix’ in agriculture (Heeks and Krishna, 2016; Rosner, 2004). Here, the idea that the ‘transfer of established technologies from the West to developing countries’ (Fejerskov, 2017: 947) serves as a ‘leveler in the global economy’ (Ojo, 2018: 2) have rendered the global South as a ‘laboratory of technological experimentation’ in recent years (Ojo, 2018). Dubbed as ‘smart agriculture’, ‘AgTech’ or ‘digital farming’ development practitioners, philanthropists and multinational (agri-) businesses increasingly roll out their technological solutions in this laboratory (cf. Conway et al., 2016).

More critical perspectives draw from other connectivity debates in alluding to the pitfalls of Internet connectivity (cf. the reference to African port infrastructure by Ouma et al. (2019) or the comparison of railway and fibre optic cable construction in East Africa by Graham et al. (2015)). A major conjuncture in this debate lies in the understanding that technology and connectivity is not neutral per se (Ouma et al., 2019). This becomes particularly relevant with the pivotal question about ‘who controls the digital’ (Foster et al., 2018). As elaborated by Mann (2018: 3) in her work on the ‘Big Data Revolution in Development’, Internet adoption opens a so far unprecedented conduit for data extraction under the regime of the ‘data custodians of Africa’s emerging economies’. She argues that data sourced from initially altruistic development programmes are increasingly becoming commercialized and used as a resource to exert power. This power can be exerted directly and indirectly. Most directly, tracking tools that connect and interfere with farmer practices can enable lead firms to trace products as well as the performance of individual producers. Digital supply chain management tools are increasingly used in logistics (Lee and Whang, 2000) as well as some segments of food value chains (Gardner et al., 2019; Pigini and Conti, 2017). It is, therefore, argued that such modes of direct digital control might reconfigure GVCs to be more flexible and, thus, strengthen lead firms’ capacities in managing value chains to their benefit. For instance, Foster et al. (2018) show that particularly the mid- and down-stream segments of GVCs benefitted from increasing data flows in tea value chains in East Africa. For processing, exporting and importing companies, digital data became pivotal to trace logistics as well as process and product standards. Consequently, Internet connectivity is generally adding value to the tea value chain. Looking however at tea producers in such chains, the authors could not find any remarkable impact of Internet access. Smallholders could not use digital data strategically since they did not know ‘where to look’ and further claimed that subsequent chain actors ‘were not sharing access to key information systems’ (Foster et al., 2018).

These results indicate that the added value is captured elsewhere (mid- and down-stream) but not at the producer level. They further hint to the rather indirect – from the producer’s perspective opaque or even invisible – modes of digital control exerted on data streams in other chain segments or meta-data (cf. Aitken, 2017, who illustrates that today all data are credit data, thus even making the unbanked visible and governable through powerful algorithms). Depending on access to and ownership of (meta) data on economic and non-economic digital activities, lead firms could, therefore, access these data streams not only to react and understand consumer trends (Mann, 2018) but also to exploit producers under the conditions of a ‘divide in data-driven knowledge’ (Unwin, 2017: 166). The emergence of digital tools and practices in the agricultural sector of the ‘laboratory global South’ is, therefore, quite contested. Although there is agreement that the digitization of practices at the upstream end of value chains can generally add value, it is still unclear as to whether this added value can be appropriated at the producer segment or if smallholders might become even more marginalized. Hence, the incipient debate on the impacts of Internet-based ICT at the upstream end of value chains, therefore, rests between optimistic hopes for a technological fix vis-à-vis reserved scepticism and clear warnings about increasing digital power and control by those who define how and for whom the digital works.

Concluding our literature review, we must foremost register the complexity and interplay of different value chain forms which need to be navigated by smallholders. Crucial for this navigation is the access to specific knowledge as well as the capacities and capabilities to coordinate marketing activities. The extensive evidence on phone-based practices indicates varying effects on ICT-based practices for different chain arrangements without referring to a transformational impact. With the ongoing adoption of smartphones and, thus, the shift towards Internet-based connectivity, a reconfiguration of established practices and the emergence of new practices amongst producers is, however, expected and attached to high hopes and cautious scepticism. Here, Ojo (2018: 2) sees ‘a dearth of research’ that provides ‘practical understanding of ICTs as a set of economic practices in African contexts’. Therefore, in the following section, we put a perspective on the shift towards smartphones by answering to what extent smallholders adopted the Internet, which digital practices in relation to value chains they apply, and lastly, how this affects the inter-firm coordination at the upstream end of value chains. The illustration of our conceptual framework (Figure 1) summarizes the different digital practices in value chains that are discussed in relation to quantitatively and qualitatively increasing ICT usage.

OPEN IN VIEWER

Figure 1.

Conceptual framework for digital practices in value chains (authors’ design).

Case study: Smartphone adoption around Mt Kenya

In the last few years, Kenya’s broadband Internet has been massively modernized and has become affordable for the broader public (Graham and Mann, 2013; Graham et al., 2015). As indicated in Table 1, Kenya has undergone stark growth in mobile ICT diffusion with the increasing relevance of mobile Internet which goes hand in hand with decreasing costs for ICT devices and data usage (World Bank, 2011). This makes Kenya a vanguard of Internet adoption in sub-Saharan Africa. ²

OPEN IN VIEWER

Table 1.

Development of mobile telephony and data subscriptions in Kenya since 2013 (data: Communications Commission of Kenya 2013, 2018).

	June 2013	September 2017	Trend
Mobile subscriptions (texts, calls)	29,849,000	41,028,000	+38%
Mobile data subscriptions (Internet)	12,340,000	30,891,000	+150%
Population in Kenya	43,693,000	47,251,000	+9%

Smallholder farming in Kenya is far advanced in terms of commercialization and export orientation (Muriithi and Matz, 2015). Particularly, the selected case study region around Mt Kenya is known for the integration of smallholders in GVCs. Due to favourable agro-climatic conditions and decent infrastructure to the export gateway Nairobi, smallholders are pervasively able to produce horticultural produce for regional export firms (cf. Dannenberg and Nduru, 2013; Dolan and Humphrey, 2000; Humphrey, 2008; Mithöfer et al., 2008; Ouma, 2010; Ulrich, 2014).

Both features, the advanced export chain integration and the latest proliferation of mobile Internet in Kenya, justify the assumption that Kenya might take a pioneering role in terms of Internet adoption and the interplay with the GVC integration of smallholders. Therefore, we selected this research area, as it seemed specifically suitable to derive some of the trajectories of increasing digital connectivity.

Methods, data and analysis

This article draws from a two-wave cross-sectional survey with smallholders. Both survey waves from mid-2013 and late-2017 followed the same cluster sampling approach (Bernard, 2006) targeting 10 sites around Mt Kenya ³ that have similar agro-climatic conditions and socio-economic structures. Our selection criteria for interviewed farmers were: (a) cultivating horticultural produce, (b) export-oriented production and (c) cultivating on up to two hectares. For each farm, we interviewed the household member who was primarily responsible for production and marketing. Apparently, a longitudinal/panel survey framework would have been a much more robust methodological framework. However, since the first survey in 2013 was initially not designed to be repeated later, it was not possible to revisit the same sample.

Our initial survey from mid-2013 includes n = 192, the follow-up survey from late-2017 n = 248 smallholders. As tested for in Table 2, the major characteristics of both surveys’ samples do not deviate substantially with regard to socio-economic characteristics such as gender, education, farm income and export chain models do not differ to a significant degree. Therefore, we argue that our chosen methodology led to a data set as reasonable and reliable as possible under these conditions and sufficient for basic comparative analysis.

OPEN IN VIEWER

Table 2.

Characteristics of sampled farmers in 2013 and 2017 (data: survey 2013, 2017).

	Survey 2013 (N = 192)		Survey 2017 (N = 248)		Chi-square statistic
	N	%	N	%
Gender of farm manager
Male	117	62	154	62	0.012
Female	73	38	94	38
Age (years)
18–40	71	37	112	45	2.98*
41+	121	63	136	55
Education
Prim. school and lower	98	51	114	54	0.39
Sec. school and higher	94	49	134	46
Income horticulture (monthly)
Low (<=15k KES)	136	73	159	68	1.48
High (>15k KES)	50	27	76	32
Export chain integration
Captive	126	66	178	72	1.92
Market based	66	34	70	28

*P < 0.1.

For analysis, we first compare both surveys to juxtapose the current ICT-usage types with the situation in 2013. The second step of analysis aims at delineating the digital practices of smallholders in relation to their value chain integration. For the examination of these practices, we differentiate between ICT-based practices relevant for vertical and horizontal knowledge flows as well as market coordination. Both layers of analysis are based on a basic statistical apparatus of descriptive and comparative research methods (frequency tables, Chi-square).

Case study results

Internet adoption around Mt Kenya

Our data confirm the rapid adoption of Internet-based ICT amongst smallholders. Since 2013, Internet users have increased from 7% to 30% (Table 3). This growth has even surpassed the rise of mobile data subscriptions during the same period in Kenya (see also Table 1). It shows that the surveyed smallholders are quickly catching up to nationwide connectivity rates. This trend is further reflected by device ownership. Simple phone use has almost reached a point of full saturation (93–99%) and smartphone ownership is increasingly common (14–37%). Although some farmers own computers (6% in 2017) or use public computers (<1% in 2017), smartphones are the most important devices to access the Internet (Table 3). This dynamic can be further derived from isolated 2017 data. Interviewees were asked to estimate how long they had been using the respective ICT type. Two phases of ICT adoption can be traced back when cleaning the data for demographic effects. A first phase of massive phone adoption is long finished, whereas adoption of the Internet can be considered as being in a stage of increasing digital connectivity (Figure 2).

OPEN IN VIEWER

Table 3.

Mobile ICT adoption 2013–2017 (data: survey 2013, 2017).

	Survey 2013 (N = 192)		Survey 2017 (N = 248)		Chi-square
	N	%	N	%
ICT usage
No ICT	13	7	3	1	41.68***
Call–text	165	86	170	69
Call–text–Internet	14	7	75
Ownership mobile ICT device
No device	13	7	3	1	33.83***
Simple phone	152	79	154	62
Smartphone	27	14	91	37

***P < 0.01.

OPEN IN VIEWER

Figure 2.

Adoption of phone-based and Internet-based ICT around Mt Kenya (N = 248, data: survey 2017).

Internet-based practices in value chains

Since a considerable number of the sampled smallholders are connected to the Internet today, the 2017 data can provide descriptive insights on the digital practices amongst early adopters. Starting with the different Internet-based services that are used on smartphones, a clear delineation between private and business use cases is evident. Despite the prevailing private purposes, all surveyed services are, however, used for agriculture at least to some extent (Figure 3). The highest importance is given to means of knowledge access: by number, web searches for agricultural knowledge are the most used services (59%). Moreover, messenger services such as WhatsApp and Telegram (37%), social networks such as Facebook (36%) as well as mailing (11%) are used to share knowledge and/or coordinate business activities. Lastly, e-banking ⁴ (37%), farming apps (17%) and electronic tracking tools (<1%) are services related to digital value chain practices.

OPEN IN VIEWER

Figure 3.

Internet services regularly used for private and/or agricultural purposes by those farmers who are connected to the Internet (N = 75, data: survey 2017).

For each of the outlined services, interviewees were asked to describe qualitatively how, for what and in relation to whom these services were used. We categorized these descriptions in order to derive Internet-based practices with regard to accessing and sharing knowledge as well as coordinating with business partners.

Knowledge flows: Vertical diversification and horizontal networks

The majority of Internet users in 2017 (64%) stated they consult digital sources for agricultural knowledge online (see Table 4). Farmers strategically search for specific agricultural knowledge on private web portals or apps (e.g. Mkulima Young, FarmLINK Kenya and Infonet biovision), governmental information sites or e-extension services (e.g. NAFIS ⁵ and KEPHIS ⁶ ), and national or international input suppliers (e.g. KenyaSeed and Syngenta). These sources have in common that they are free of charge and easily accessible whilst being limited to a uni-directional, top-down knowledge transfer. Informal discussions framing the survey process showed that farmers value the opportunity to access knowledge online which is usually only accessible ‘in time’ or without a certain effort as it is bound to a distinct circle of intra-chain knowledge sources (e.g. export firms). Therefore, drawing from digital knowledge sources is seen as a supplemental or alternative practice to access external knowledge that goes beyond the knowledge which can be accessed within captive GVCs or is diffused through subsequent horizontal flows.

OPEN IN VIEWER

Table 4.

Classification of digital practices amongst Internet users in 2017 (N = 75; data: survey 2017).

Digital value chain practices	N	%
Vertical knowledge transfers	48	64
Web search	44	59
Farming App	13	17
Horizontal knowledge transfers	40	53
Bilateral only (farmer to farmer)	7	9
Bilateral and multilateral (farmer networks)	33	44
Market coordination	43	57
Coordination with buyers	24	32
Digital marketing platforms	18	24
Formal platforms only	8	11
Informal platforms only	7	9
Formal and informal platforms	3	4

For horizontal knowledge transfers, calling and texting remain crucial for inter-firm interaction. Whilst in 2013, roughly half of all phone users (54%) stated they regularly call or text other farmers, the importance of bilateral, call- or text-based interaction further increased: 76% of the 2017 interviewees call or text other farmers to consult about agricultural topics. ⁷ Emphasizing Internet-based horizontal knowledge transfers, today, applications such as mailing, messenger services or social networks are also widely used for bilateral interaction from farmer to farmer (53%). This interaction evolved as an alternative practice of horizontal knowledge exchanges and extends the former phone-based communication with different technological advances such as sharing data (pictures, videos and links). Whilst this is unsurprising, the major novelty is certainly the emergence of digitally enabled multilateral knowledge networks. Multilateral communication channels are accessed by the clear majority of Internet users, who stated they use the smartphone to share knowledge and experience with other farmers and people from ‘the region’ (Table 4). Farmers particularly connect via group chats (Messenger services such as WhatsApp and Telegram), and sometimes chat groups in social networks. Whilst some of these groups link a small circle of local farmers or serve as platform to coordinate established farming groups (e.g. self-help groups or producer cooperatives), other groups have distinct foci on certain products (e.g. French beans farming group, horticultural crops group and dairy group) or regions (e.g. Star Rays Agrinfo – Nyeri). The largest groups connect thousands of farmers all across Kenya. For instance, the Telegram-group Farmers +254 connects ∼3100 members. This group is used to identify pests and diseases and discuss their treatment, evaluate the general status of cultivated plants, request where and for what prices specific inputs and varieties are available and further to market produce. Regardless of size and reach, the emergence of a multilateral component in horizontal knowledge networks means substantial changes in relation to prior phone-based practices of exchanging knowledge horizontally. The major limitations of phone-based practices to access and especially share local knowledge (bilateral communication, limited reach and condition of established contacts) are not only remedied in theory but evidently used and appreciated by a solid share of smallholders today.

Value chain coordination: Digital coordination and digital markets

Common modes of market coordination in value chains are also affected to some degree. More than half of the Internet users (57%) use digital modes to coordinate. About one-third (32%) coordinate marketing processes by bilateral communication with buyers via messenger services, mails or social networks (Table 4). Such bilateral coordination usually includes negotiations about price and quantity, the mode of transport and agreement on payment formalities. When investigating coordination in different chains in detail, the Internet’s implications for intra-chain coordination become more visual. Table 5 differentiates the coordination with direct buyers, intermediaries and exporting lead firms. Apparently, the use of simple phones remains the major mode to coordinate with business partners. Especially, for market-based interactions with direct buyers (89% of phone users) and intermediaries (93%), calling and sometimes texting serves to arrange transactions which is also the case for interactions with exporters – although to a lower extent (65%).

OPEN IN VIEWER

Table 5.

ICT-based coordination in value chains (data: survey 2017).

ICT-based coordination	Direct buyers (market-based, RVC)	Intermediaries (market-based, RVC and GVC)	Exporting firm (captive, GVC)
Selling to … (n = 248)	108	179	178
Phone-based coordination
Phone users selling to … (n = 245)	107	177	175
Coordinated via phone …	95 (89%)	165 (93%)	114 (65%)
Calls	94 (88%)	155 (88%)	113 (65%)
Texts	31 (29%)	42 (24%)	49 (28%)
Mobile payment (M-Pesa)	52 (49%)	109 (62%)	–
Internet-based coordination
Internet users selling to … (n = 75)	35	54	55
Coordinated via Internet	10 (29%)	15 (28%)	20 (36%)
Mail	1 (<1%)	–	2 (<1%)
Messenger	10 (29%)	14 (26%)	7 (13%)
Social network	2 (1%)	8 (15%)	1 (<1%)
Electronic bank transfer	–	–	16 (29%)

The Internet, however, has little relevance so far. Although roughly a third (29%) of the Internet users in captive chains stated they use e-banking to receive and check payments, there is certainly no ample evidence of Internet-based coordination in captive GVCs. The few respondents who stated they use messengers (13%), mail or social networks (both <1%) to coordinate with exporters merely did so since they knew individual staff members rather than because of obligatory conditions set by exporters. Considering the discussion on the implementation of instruments to directly track and document smallholder activities and performances, our data further do not uphold evidence of these trends for the current situation (Figure 3). Although payments are increasingly organized via e-banking services, captive GVC integration does not require smallholders to connect digitally and be tracked (so far).

In market-based chains, Internet-based coordination is much more common and to some extent valued as a practice to decrease transaction costs as well as increase market reach. Some respondents stated that they prefer Internet-based coordination with direct buyers and intermediaries for the opportunity to send pictures before meeting and even before harvesting to bargain and agree on deals online, thus making some meetings obsolete. Nevertheless, in general, also for market-based chains, Internet-based coordination is so far not the predominant practice.

Lastly, the emergence of incipient practices of digital marketing is not yet dominant but rather remarkable. Roughly 24% of Internet users participate on digital marketing platforms in order to sell some of their produce online or search for specific inputs such as seeds or machinery (Table 4). Respondents stated they access formal market platforms (15%), such as the popular Kenyan e-commerce platforms Jumia, OLX or – fully devoted to agricultural products – MkulimaYoung and M-Farm. Simultaneously, some smallholders practice less formalized marketing strategies (13%) using social networks such as Facebook and messenger-based group chats to advertise, sell or buy products mainly to their network of established online contacts. However, most interviewees evaluated both formal and informal digital markets as being ‘underdeveloped’ and rarely leading to an actual deal. Indeed, rather than selling via digital platforms, the opportunity to find inputs, which are not available on local markets, are often seen as the main motivation of some smallholders to browse markets digitally (e.g. mushroom cultivation equipment, rare plant breeds and machinery). Whilst digital markets remain immature, our results indicate that formal and informal digital markets address solely the marketization of domestic products so far. We could not identify any digital marketing practices with clear export-oriented intention.

Discussion

How can the case study be embedded in the broader context of the connectivity debate? This final section aims to pigeonhole our results by discussing the findings alongside our introductory research questions.

First, referring to the evolution and extent of Internet adoption in the research area, our study can clearly document the rapid shift towards smartphone use and, thus, Internet-based connectivity amongst smallholders. This is not unexpected; however, due to our dynamic approach, our results trace back the speed and extent of the two phases of phone and smartphone adoption. We will most likely see similar developments in similar producer regions sooner or later.

Second, and contrary to the case study on digital practices in value chains by Foster et al. (2018), our data highlight that, amongst the first Internet users, digital practices with agricultural relevance have widely manifested in a region exceptional for its widespread export integration of smallholders. These practices are characterized by their different implications for value chain integration especially with respect to process upgrading (Humphrey and Schmitz, 2002). Today, a solid share of Internet users accesses agricultural knowledge online, thus broadening the variety of established vertical knowledge sources. Simultaneously, inter-firm cooperation is to some degree extended by digital multilateral knowledge transfers between farmers. The diversification of vertical knowledge transfers, as well as the emergence of multilateral digital networks amongst farmers, means an unprecedented practice compared to simple phone use for the question of what knowledge can be accessed from whom and how. Since our data show that farmers indeed use Internet-based sources as alternatives to support their agricultural activities, this certainly resembles emancipation from established gatekeepers of knowledge to some extent. Farmers no longer necessarily solely rely on direct (through captive chain integration) or indirect (through horizontal, inter-firm spillovers) sources of knowledge but can – and do – supplementarily access and share agricultural knowledge and experience online.

In addition to these knowledge-oriented, digital practices, smallholders have further adopted some initial practices affecting how they can coordinate and navigate the market – or in our case, different value chain options. Here, the results are however less clear and best discussed in relation to inter-firm coordination as addressed with our third research question.

From a ‘smallholder’s perspective’, inter-firm coordination is affected with increasing Internet adoption by varying degree. As stated in our introductory literature review, the debate on digital connectivity in GVC is polarized between optimistic expectations about the Internet breaking entrenched power asymmetries and dark scenarios of digitally fostered control and increased extraction of labour and/or value. Here, our case study can add some important observations. Generally, in captive GVC arrangements, firm-to-firm coordination between smallholders and subsequent chain actors does not occur through digital tools and practices. We neither found tools to generate value (e.g. reduction of transaction costs) nor to exert direct, digital control (e.g. tracking of practices and performance, measures to simplify switching between producers). Long-established modes of governing GVCs at the interface of producers and exporters widely remain offline. Therefore, digitally driven addition of value as well as its appropriation in the favour of lead firms does not seem to occur at the chain’s upstream end at this stage of Internet adoption.

This lack of ‘digital’ and all its implications for captive GVCs does, however, to no extent diminish the relevance of understanding value creation and value appropriation (through digital control) along the full spectrum of GVCs. Rather, it extends the question of if by where and by whom. These questions are more likely to be answered when also exploring indirect, less tangible measures at the mid- and down-stream level of GVCs as exemplified by Foster et al. (2018). Moreover, albeit these findings, the current condition might change quite quickly when digital integration becomes a compulsory requirement for smallholders. Hitherto non-existence of direct digital coordination and control might, therefore, just be a matter of time and shifting lead firm imperatives.

The case study further indicates trends that are at first sight disconnected from captive GVC arrangements. Whether it is for RVC or for intermediated GVC arrangements, a solid share of the interviewees engages in digital practices on a continuum of incipient experimentation up to daily routine not only to upgrade their production processes through new knowledge sources or channels, but also via new modes of coordinating inter-firm transactions.

Especially in RVCs, digital marketing platforms are not only available; they are evidently also used by some smallholders. Although it remains open to what extent such platforms can scale up, digital marketing practices enable farmers to experiment with the disintermediation of middlemen and widen their field of potential buyers. Such changes directly feed into the notion of functional upgrading from the value chain literature in letting smallholders partly take over the functions of intermediaries (Humphrey and Schmitz, 2002). For further work, it will be important to see how digital markets are used and (re-)made as the question remains not only if, but where (in a relational value chain manner) and by whom digital markets are accessed and controlled. Here, research on digital intermediation in other sectors, such as the business process outsourcing industry (Graham et al., 2017; Mann et al., 2014) or with uber in the transport sector (Rekhviashvili and Sgibnev, 2018), shows that initial benefits for workers and producers can quickly lead digital re-intermediations and, thus, value appropriation by platforms, as soon as they and become obligatory rather than optional.

Inter-firm coordination via digital practices, therefore, occurs and impacts differently depending on the type of chain that they are devoted to. Relating to Krishnan (2018) and Barrientos et al.’s (2016) arguments on diversification options under multiple chain forms, these impacts should, however, not be looked at in an isolated manner. We argue that the extended and simplified alternatives (in RVCs and intermediated GVCs) have to be interpreted with regard to their interrelated opportunity costs. Digitally enabled upgrading trajectories for producers in one chain arrangement simultaneously challenge its alternatives by creating agency for the negotiation whether and under which conditions chain integration takes place elsewhere. Likewise, digitally driven downgrading trajectories in one chain alternative do not only affect this alternative itself, but can also induce a general race to the bottom as well. Here, our study suggests quite promising effects in the form of niches that are occupied by some smallholders in order to increase their agency in market-based interactions, but also (indirectly) the negotiation about the take it or leave it deal, that is captive GVC participation. These niches should however not be interpreted over-optimistically as a digital miracle. Merely, they represent a snapshot of the gradual process towards a future of digital connectivity, that is still open and remains contested along on the one hand, the social strata of producers and, on the other hand, different chain actors.

Conclusion

Our dynamic perspective on export-oriented, horticultural smallholders highlights the rapid shift towards the use of the Internet amongst smallholders producing for agricultural value chains. Our case study shows that smartphones are more than an artefact of social interaction and entertainment and used for so far unprecedented digital practices of accessing vertical knowledge flows, horizontal knowledge networks, and incipient digital markets amongst smallholder farmers. These practices can be linked to intrinsic process and functional upgrading opportunities. Particularly in market-based value chain options (RVCs and intermediated GVCs), some users access alternative knowledge channels and marketing practices to identify and use production and marketing trends. Other than often suggested in the debate about digital governance in value chains, we find, however, no evidence for lead firm driven, direct exertion of digital control in captive GVC arrangements. Rather, long-established analogue practices of governing the upstream end of captive GVCs persist widely. Despite these unchanged logics of captive GVC coordination, our work, therefore, indicates the potential for some smallholders to digitally navigate the complex nexus of multiple value chain options, thus creating leverage for the negotiation whether to integrate in captive GVCs or choose alternative production and marketing arrangements.