Minding ‘Productive Gaps’: An Appraisal of Non-operational Land Deals in Seven Sub-Saharan African Countries

Abstract

One of the dominant global development agendas for rural Africa in the past two decades has cast large-scale agro-industrial investments as a solution to achieve more efficient land use, higher crop yields, enhanced food security, and poverty reduction, among others. However, mounting evidence shows that this agenda has not fulfilled its promises: most land deals for agricultural production have not materialised as planned and their socio-economic development objectives often remain unreached. Despite the often severe impacts of non-operational projects, knowledge about why they fail to take place and operate remains fragmentary. Based on an extensive literature review of contemporary land deals in seven sub-Saharan countries, this paper sheds light on two ‘productive gaps’. First, the article delves into the ‘productive gap’ of land deals themselves, identifying key drivers of non-operational land deals. The reviewed literature points to local opposition and financial difficulties as significant factors impacting agricultural operations. Local opposition, in turn, stems largely from flawed land acquisition processes and unfulfilled investors’ promises. Second, this article offers a critical appraisal of the biases and oversights in the knowledge the land grab scholarship has ‘produced’.

Keywords:

• Land rush
• large-scale agro-industrial investment
• local opposition
• failed land deals
• sub-Saharan Africa
• comparative analysis

1. Introduction

Large-scale agricultural projects in sub-Saharan Africa (SSA), for all the ‘high modernist’ dreams (Scott, 1998) they have conjured up, have historically been prone to failure. Colonial administrations and concessionary companies attempted to run agricultural activities that, most of the time, could only function due to their highly repressive nature. In the 1960s, postcolonial governments also set up state-run farms that frequently fell into disarray due to, among others, a lack of funding, poor feasibility studies, and corruption problems (Coulson, 2013). There is accumulating evidence that the contemporary wave of land acquisitions follows a similar trend.

One of the dominant global development agendas for rural Africa in the past two decades has cast large-scale agro-industrial investments as a solution to achieve more efficient land use, increased crop yields, enhanced food security, and poverty reduction. But across a range of African countries whose political contexts, social structures, and land governance systems vary considerably, a large share of land deals in the agricultural sector are scaled back, perform poorly, go bankrupt, or are abandoned (Nolte, 2020; Lay et al., 2021). The surface areas that investors effectively farm are often significantly smaller than the size of the acquired land. The purported development effects of these projects are often unmet, or when they are, negative externalities frequently outweigh their positive contributions (Hufe & Heuermann, 2017; Rulli, Passera, Chiarelli, & D’Odorico, 2018). Land projects also foster local land competition and conflicts (Haar, van Leeuwen, & de Vries, 2020; Bekele, Dries, Heijman, & Drabik, 2021) and trigger protests and community resistance (Temper, 2019; Gagné, 2020). These projects—whether operational, failed, or scaled-back—drive rural agrarian changes that accelerate land inequality and social differentiation dynamics through privileged access to land and formal proofs of ownership (Antwi-Bediako, 2021; Bélair, 2021; Wangu, Mangnus, Westen, & Vocht, 2021). Furthermore, many projects induce lasting landscape transformations, exposing affected populations to biodiversity loss and increased environmental risks (Brüntrup et al., 2016; Lay et al., 2021).

Despite the severe implications of non-operational projects for local populations, but also for investors and host states, our understanding of why so many projects are scaled back, perform poorly, or are abandoned, while others manage to start and sustain production, remains fragmentary. Single case studies dominate the literature; comparative reviews of land deals are few, although middle-size scoping studies about their livelihood and food security impacts have recently been published (Hall, Scoones, & Tsikata, 2017; Hufe & Heuermann, 2017; Dekeyser, 2019; Fitawek & Hendriks, 2021; Oberlack et al., 2021; Zaehringer et al., 2021). For their part, larger comparative studies focus solely on non-operational and closed land projects (Borras et al., 2022), are based exclusively on aggregated data from the Land Matrix (Lay et al., 2021; Nolte, 2020) or need to be updated in light of changing dynamics on the ground (German, Cavane, Sitoe, & Braga, 2016; Oya, 2013).

Based on an extensive literature review of contemporary land deals in seven sub-Saharan countries, this paper sheds light on two ‘productive gaps’. First, the article delves into the ‘productive gap’ of land deals themselves, identifying key drivers of non-operational land deals. This article explores what the scholarship has to say as to why some land deals begin and maintain productive activities while others do not. Second, it offers a critical appraisal of the biases and oversights in the knowledge the land grab scholarship has ‘produced’. Our cross-country study shows that this scholarship overly focuses on certain kinds of crops, types of investors and investments, and geographic sites. Our paper analyses 99 academic publications that cover 90 land deals in seven SSA countries. By building an original database containing 45 distinct parameters, the paper offers a solid analysis of the stated drivers behind land deals’ varied trajectories towards failure or success. The methodology adopted permits a systematic comparison of previously scattered empirical observations to understand why a large share of land deals in SSA are non-operational.

The paper acknowledges that corporate failure and success in the agricultural sector are contested concepts. For instance, failure to start farming operations on the part of investors does not mean that other unstated objectives are not achieved (e.g. increased political control over people and resources, land commodification, speculative gains, etc.) and failure for some may entail success for others, and vice versa (Borras et al., 2022). Moreover, conceptualising non-operational projects as ‘failed’ can conceal their adverse consequences for local communities (e.g. land dispossession, loss of access to the commons, landscape transformation, environmental degradation, the loss of means of production, etc.) (Chung & Gagné, 2021; Schlimmer, 2020).

Following the terminology of Borras et al. (2022, p. 2), we hereafter use the more politically neutral terms of non-operational and operational land deals. However, our conceptualisation is more stringent in terms of production activities:¹ we define an operational deal as a deal wherein the investor cultivates at least 50 per cent of the acquired land. We distinguish two types of partially operational land deals: those where between ≥ 20 and <50 per cent of the concession is farmed and those where <20 per cent is farmed. Non-operational deals are those that ceased operations, never started farming, or only undertook preliminary steps. Our contention is that such a conceptualisation allows us to compare projects along a very specific dimension: their materialisation (or absence thereof) in the landscape.

The article is structured as follows. First, we briefly situate our paper in the existing literature and then present our methodological approach, followed by a presentation and discussion of findings and, finally, concluding remarks.

2. Non-operational land deals in sub-Saharan Africa: what do we know?

Initially, the literature was primarily concerned with changes in the global economy that led to the ‘de-territorialization and commodification of sovereign national territory across much of the globe’ (Margulis, McKeon, & Borras, 2013, p. 3) and the consequences on rural livelihoods of this heightened demand for land, especially in SSA, by powerful outsiders. Rapidly, however, scholars broadened their focus beyond global forces and turned their gaze to the influence of national states, elites, and institutions as enablers of land deals. This second wave of work, as many have aptly remarked, helps better understand how land deals materialise and unfold differently depending on the domestic historical and political contexts in which they take place (White, Borras, Hall, Scoones, & Wolford, 2012; Wily, 2012; Baglioni & Gibbon, 2013; Cotula, 2013).

More recently, a third analytical perspective has emerged, with a growing number of scholars investigating factors that hinder these projects, in particular in SSA where non-implementation rates appear higher than elsewhere in the world. Most of the reviewed studies included in our database belong to this stream of literature.

Knowledge about the root causes of non-operational land deals is more advanced on some fronts than others. In particular, the literature has devoted an appreciable degree of attention to understanding the spectacular collapse of the biofuel sector. Studies on what we term the ‘biofuel flop’ largely centre on explanatory factors that are not relevant in explaining the failure of food investments: the lack of agronomic knowledge about jatropha curcas; the absence of structured commodity chains; the fall of oil prices on global markets in 2011; and the introduction by the European Union of stricter sustainability criteria for renewable energy (Engström, 2009; Romijn, Heijnen, Rom Colthoff, De Jong, & Van Eijck, 2014; Wendimu, 2016; Nygaard & Bolwig, 2018; Ahmed, Campion, & Gasparatos, 2019; Gagné & Fent, 2021).

The scholarship about non-operational projects for food production is more eclectic and largely consists of single-country studies, with some exceptions (e.g. Hall, Gausi et al., 2015). For instance, scholars have pointed out that although the land has the potential to be a productive and financial asset, capitalising on land appreciation in the short term has been more difficult than expected, leading many investors to abandon or rescale their projects (Knuth, 2015; Magnan, 2015; Visser, 2015; Sippel, Larder, & Lawrence, 2017; Kaarhus, 2018; Salverda & Nkonde, 2021). More broadly, speculation, money laundering, tax evasion, or access to subsidies seem to be a frequent factor explaining the partial implementation of so-called agricultural projects (McCarthy, Vel, & Afiff, 2012; Keene, Walsh-Dilley, Wolford, & Geisler, 2015; Bélair, 2022).

The local context has also been invoked as a factor critically shaping projects’ outcomes. For example, Schoneveld (2017) identifies various governance issues and legislative deficiencies as contributing to the cancellation of farmland projects in Ethiopia, Ghana, Nigeria, and Zambia. Gagné (2020) and Temper (2019) point out that local resistance and contention play a role in projects’ termination and/or suspension. Gagné and Fent (2021) and German et al. (2016) show that business models and the nature of partnership between local actors and investors contribute to explaining the uneven progress of land deals in Mozambique and Senegal. This includes factors, such as the degree to which customary rights are respected and how land is acquired. In Tanzania, Brüntrup et al. (2016) point to issues of plural legalism in land ownership but also inadequate agricultural policies and an uncertain business environment.

Lay et al. (2021) and Nolte (2020) provide the most comprehensive cross-country overviews of drivers behind non-operational land deals in recent years, to our knowledge. Nolte uses Land Matrix data and reports that larger land deals, deals targeting land formerly used by smallholder farmers, and biofuel deals are more likely to fail worldwide. She also finds, perhaps a little counter-intuitively, that ‘political stability, government effectiveness, and corruption control are related to more project failures, whereas high land tenure insecurity and business-friendly environments reduce project failure’ (Nolte, 2020, p. 9). However, she recognises the ambiguity of her findings, calling for greater attention to ‘more localized institutional variables to better understand the role of institutions for project failure’ (Nolte, 2020, p. 1). Lay et al. (2021), who also rely on Land Matrix data, find that conflicts and protests are the most common drivers of failed deals, followed by financial problems, in all regions of the globe. The use of Land Matrix data in these two studies is, however, problematic since the Land Matrix dataset is not exhaustive and contains many inaccurate entries, as several scholars have observed (Cotula, 2012; Woodhouse, 2012; Anseeuw, Lay, Messerli, Giger, & Taylor, 2013; Edelman, 2013; Oya, 2013; Scoones, Hall, Borras, White, & Wolford, 2013). In sum, although the contemporary wave of large-scale land acquisitions has ushered in a ‘literature rush’ (Oya, 2013), ‘we (still) lack good data on the number, quality, and effects’ of these deals, thus highlighting ‘the ongoing need for research’ (Wolford et al., 2024).

3. Building an original database to analyse the existing land grab scholarship

To address the methodological shortcomings mentioned above, we started in 2021 to conduct an extensive literature review of empirical studies on land deals in Ethiopia, Ghana, Kenya, Mali, Mozambique, Senegal, and Tanzania. All these countries have been highly targeted by new investors since the beginning of the land rush. In addition, our country selection reflects a broad diversity of colonial legacies and political contexts that potentially influence how land deals evolve. Three researchers² specialised in land investments in sub-Saharan Africa conducted the review³ through a broad search including the words ‘farmland’, ‘land deal’, ‘land grab’, ‘large scale agro-investment’, ‘agricultural investment’, and ‘failure’ to capture all potentially relevant articles in the targeted countries. We only included papers presenting first-hand empirical data published between 2010 and April 2022. Our focus on primary evidence ensured greater empirical accuracy and analytical depth and aimed at avoiding that incorrect data is reproduced without being validated (see Edelman, 2013). Without claiming to be exhaustive, this research process allowed us to retrieve most of the pertinent scientific literature featuring primary empirical evidence. In total, we reviewed 99 publications⁴ that documented 90 farmland deals.

We employed an iterative approach to construct our database. We first read the papers to get a general sense of the literature and identify key factors mentioned to explain the trajectories and outcomes of land deals. We then extracted and coded the findings of these studies in a database comprising 45 analytical parameters (listed in Supplementary Appendix 1). The process of assembling rich in-depth, yet dispersed, qualitative studies in a systematic fashion allowed us to decipher broader patterns, both across and within countries. Nevertheless, it is impossible to determine the extent to which our database is representative of the true population of land deals in each country, notably due to the lack of reliable and publicly available data on land investments in most SSA countries. We will elaborate further on the implications of data shortcomings after we have presented our findings.

4. Key drivers behind non-operational land deals

Among the 90 land deals we reviewed, 51 are non-operational (i.e. they never started production, were stalled or cancelled); 12 are partly operational and 27 are operational. Our findings indicate that the rate of non-operational deals is similar across business models: about 63 per cent for large-scale industrial operations in comparison to 59 per cent for nucleus estate out-grower models. Out of 51 non-operational projects, 24 are biofuel projects (47%). As mentioned, we believe that ‘biofuel flops’ stand apart analytically from other cases of non-operational deals because they terminated production for a specific set of reasons, similar across countries and circumstances, that are not necessarily relevant to understanding non-operational investments for food production. If we exclude biofuel projects from the analysis, we are left with 59 land deals, of which 24 are operational (41%), and 35 are non-operational or partly operational (59%), as visible in Figure 1. The rate of ‘failure’ is quite high considering that our threshold for defining success (cultivating more than 50% of the land) was not especially high. This figure is substantially higher than Nolte’s (2020) estimate of 20 per cent but more in line with Lay et al. (2021) estimation that 30–60 per cent of the land acquired for investment is not under production.

Figure 1. Status of projects (excluding biofuel projects).

In the literature we reviewed, the two factors most commonly cited as drivers of non-operational land deals in the food sector are land conflicts and other forms of local resistance (hereafter termed ‘local opposition’)⁵ and investors’ financial difficulties (Figure 2).⁶ Our findings confirm previous analysis by Lay et al. (2021). Other drivers of failure include investors’ lack of environmental and/or farming knowledge; ecological vagaries (such as droughts or floods) and environmental challenges (such as barren land); and speculation or other motivations for acquiring land (e.g. access to loans). Due to space constraints, we only discuss the two most frequent drivers of failure.

Figure 2. Main identified causes of failure in non-operational projects (excluding biofuel projects).

4.1. Driver 1. Local opposition

In 54 per cent (19 out of 35) of non-operational land deals (excluding biofuel projects) included in our database, the authors reported that local opposition contributed to derailing projects. We identified three main drivers behind local opposition against non-operational projects in the literature: (i) flaws in the land acquisition process; (ii) loss of land access for local communities as a result of the project; (iii) unfulfilled promises by the investor and/or government. While presented here as distinct, in practice the above drivers are interlinked. For instance, local grievances associated with exclusionary dynamics can be exacerbated when companies fail to deliver on their promises. Below, we discuss the three most important causes of local opposition.

4.2. Flawed land acquisition processes

Flawed land acquisition processes represent the most common factor mentioned by authors to explain local opposition. The lack of free, prior, and informed consent (FPIC) of current land users as well as unfair or absent compensation for land loss are important shortcomings in land acquisition processes. This finding aligns with the observation by Lay et al. (2021, p. 12) that only 15 per cent out of 250 land deals duly sought FPIC while almost 45 per cent reported no consultation at all. In a similar vein, Hall, Gausi et al. (2015, p. 4) identify the lack of consultation with local communities as a key factor behind non-operational land deals in Malawi, Namibia, and Zambia.

Importantly, nearly half of the deals in our database that effectively obtained free, prior, and informed were also non-operational. This finding illustrates that securing local approval does not always suffice to ensure a project’s longevity. In a telling case in Senegal, a company that enjoyed robust community support ended due to the central state’s indifference and lack of oversight, leading regional administrative authorities opposed the company to override the investment (Gagné, 2020). However, it might also be that free, prior, and informed consent is an ambiguous, if not hackneyed concept, subject to interpretation and highly variable in its implementation (German 2022). Local communities generally have limited knowledge about land laws and regulations, available compensations for the loss of their land, and long-term consequences of land concessions. Their capacity to negotiate beneficial business agreements with investors typically remains low (Locher & Müller-Böker, 2014; German et al., 2016).

Importantly, the literature emphasises that local communities are not homogeneous and that FPIC most likely does not reflect the consent of all members of a community. For example, Bruna (2022) and Gyapong (2021) show that opposition by marginalised groups was brushed aside by more powerful community members who found the deal advantageous. Finally, even if communities were properly informed and consulted before the commencement of a project, they can eventually withdraw their consent due to unfulfilled expectations. As Antwi-Bediako (2021) exemplifies in Ghana, local conflicts fuelled by competing claims often surface after productive activities have started.

4.3. Unfulfilled promises

The second largest driver of local opposition mentioned in the literature pertains to unfulfilled promises by investors and/or governments. These promises can initially incentivise local people’s consent. Many authors report that, at the onset, communities welcomed new investors and were optimistic that their projects would foster local socio-economic development. However, failure to deliver on commitments may generate opposition from disappointed local populations, leading in some cases to the closure of projects. For instance, Gyapong (2020, p. 5), discussing an oil palm investment in Ghana, reports that many affected families ‘expressed regret agreeing to the land transfer, but their major worry has not been because of dispossession rather the failed promise of decent jobs’. Another illustrative example is provided by Guta (2016) in an analysis of the SEKA-Agro Processing project in Ethiopia, over which local grievances grew over time, especially when the government continued defending unproductive investors at the expense of the community. In other cases, as Chung (2021) and Engström (2020) show with the SEKAB project in Tanzania, local opposition occurs earlier in the project cycle, when rumours of impending eviction spread.

In fact, our comparative analysis illustrates that one of the most noticeable differences between operational and non-operational land deals is the extent to which they construct infrastructure and create jobs,⁷ two variables that are lower for non-operational land deals. In our data, 63 per cent of operational projects have developed local infrastructure against 26 per cent of non-operational land deals. Although we coded information on employment and infrastructure, the articles in our corpus use different assessment methods to qualify these impacts, making a solid comparison across cases impossible. Similarly, all operational projects have created jobs, while only 66 per cent of non-operational projects have done so. As shown in smaller-scope studies, employment effects in commercial farms are commonly lower than promised and local job opportunities are also often limited to low-skilled and casual jobs (Hall et al., 2017; Fitawek & Hendriks, 2021; Zaehringer et al., 2021). Our review adds to their analysis, showing that unfulfilled promises represent a key driver of local opposition leading to non-operational land deals. Similarly, our data substantiate claims about the lack of local infrastructure development.

4.4. Loss of land access

The third most frequent driver of local opposition mentioned by the authors is that land deals often dispossess communities of their land. Importantly, in most cases, we reviewed—apart from the studies in Ghana—once the land is attributed or transferred to an investor, it is rarely returned to the original landowners when a deal is cancelled. Local communities typically have no legal recourse to regain their land, although there are cases wherein investors lease unproductive land to local farmers (e.g. Greco, 2015; Gagné, 2020). In Tanzania, even though the state reclaimed land from unproductive investors, confiscated parcels were not given back to villagers but rendered available for new investors (Bélair, 2022).

Another finding that emerges from the review is that the nature of investor-community relationships differs in operational and non-operational land deals. Indeed, among the 13 reported cases of collaborative investor-community relations, nine are operational land deals. In contrast, 15 of the 24 land deals characterised by conflict are non-operational land deals (Figure 3). One driving factor seems to be how local people perceive the balance between the investment’s positive and negative effects on their livelihoods. However, understanding which of the above factors matters most in shaping these investor-community relationships would require a more systematic investigation.

Figure 3. Relations between investors and local communities by land deal status (excluding biofuel projects).

Lastly, while local opposition is an important driver of non-operational land deals, we also see in our data that they frequently occur in operational projects. Indeed, the corpus we reviewed alluded to land conflicts in nearly half (46%) of operational investments. Environmental impacts, for instance, are likely to become more visible when the company starts operations, as a result of land clearing and farming. This could partly explain why we frequently see conflicts in operational investments too. For instance, 85 per cent of respondents in one Kenyan case reported adverse environmental impacts, such as loss of water availability because mechanised ploughing brought more sediment into river water, in-migration created pressure on water use, and water was diverted to plantations (Zaehringer et al., 2021). This finding is similar to Brüntrup et al. (2016, p. 14), who observe that ‘from the moment the investments’ impacts materialised on the ground […], the potential for conflicts proved high’. More broadly, these findings suggest that, although some degree of conflict seems inherent in all large-scale ventures, opposition by local communities does not necessarily result in project termination.

4.5. Driver 2. Investors’ financial difficulties

Investors’ financial difficulties have been discussed in the wider land grab literature. Scholars have argued, for instance, that borrowing money from global development agencies or bank institutions usually requires compliance with international sustainability standards that are time-consuming and complex to implement (Engström & Hajdu, 2018). The revocation of investment licenses or the introduction of new moratoriums on land deals by host governments also represents important financial risks for investors (Chung & Gagné, 2021).⁸ Setting up projects on the ground commonly implies high costs due to inadequate road and market infrastructure. A lack of investors’ understanding of local governance systems may also delay the start of production and involve additional expenses (Engström, 2020; Gagné & Fent, 2021). The economic viability of farmland projects is often further endangered by investors’ unrealistic business plans and underestimation of their operating costs (Goetz, 2015; Ahmed, Campion, & Gasparatos, 2017; Engström, 2018), a lack of proper environmental and farming suitability assessments (Anseeuw & Boche, 2015; Di Matteo & Schoneveld, 2016; Ahmed et al., 2019) and insufficient environmental and farming knowledge (Gill, 2016; Antwi-Bediako, Otsuki, Zoomers, & Amsalu, 2019).

In our database, financial problems are cited as an important factor that impeded the pursuit of activities in nearly half of non-operational land deals in the food sector (12 out of 35). Indeed, it appears that many investors are ill-equipped to implement ambitious farmland investments in contexts, such as the ones we find in many SSA countries.

However, financial difficulties are often mentioned in passing, without due elaboration. When these difficulties are assessed in greater detail, they are almost always explained by shortcomings in the local context at the subnational level, or investors’ limited knowledge of this context. For example, Kamski (2016, p. 576) explains that the Ethiopian Sugar Corporation faced financial difficulties in their Kuraz Sugar Development Project (KSDP) because of ‘the remoteness of the project site and associated lack of transportation infrastructure, combined with the immense initial investments needed’. In Mozambique, Arnall (2019, p. 866) reports that the company struggled to start production because it was allocated land in a ‘densely populated area and left without government’s help or guidance into how to displace and compensate affected farmers’.

This empirical gap about investors’ financial difficulties is problematic for our understanding of non-operational land deals: Investors as political actors and the complex investment structures and processes underlying those investments should be unpacked. Although a line of inquiry more attuned to the financial drivers and consequences of agro-industrial investments has emerged (Fairbairn, 2014; Ouma, 2014; Brüntrup et al., 2016), our comprehension of investors’ financial strategies, investment structure, and experience in undertaking agricultural activities can be improved. These parameters should be investigated more broadly to understand what determines the economic viability of those projects, especially given that we cannot expect any real ‘positive developmental effects from an economically unviable land deal’ (Brüntrup et al., 2016, p. 18).

5. Biases and gaps in the existing land grab literature

Building the original database enabled us to identify trends and gaps in the existing land grab literature. Although our sample aimed at capturing all relevant works and is arguably representative of the scholarship, it is not exhaustive of the true population of land deals in each country. It is impossible to determine how representative it is due to the lack of reliable data on land investments in most sub-Saharan African countries. As several have already observed, the scholarship on land deals is skewed towards mediatised, foreign deals that result in negative outcomes (e.g. Abubakari, Twum, & Asokwah, 2020; Cochrane & Legault, 2020). An emblematic example is the Pro-Cana project (previously operated by the Central African Mining and Exploration Company-CAMEC) in Mozambique, analysed in eight academic papers included in our database. Our review supports the observation that investments by foreign investors have received more scrutiny than those of domestic investors: out of the 88 projects for which the investor’s origin was documented in our database, 59 per cent were foreign; 19 per cent were domestic, and 22 per cent were joint ventures between foreign and domestic businesspeople or companies. Knowledge about domestic agricultural projects is thus less advanced although these projects are probably more numerous and cumulatively cover larger surface areas than foreign ones (Cotula et al., 2014; for Ghana, see Anku, Andrews, & Cochrane, 2022; for Ethiopia, see Cochrane & Legault, 2020; for Tanzania, see Brüntrup et al., 2016). However, if we disaggregate our data, country-specific trends emerge. The share of studies on domestic investors greatly varies from country to country. For instance, four case studies out of five in Mali examine land deals by domestic investors while none of the studies in Mozambique do so. Better databases are needed to understand if these studies are representative sets of cases.

In addition, our sample included a majority of deals that are non-operational or partially operational (i.e. cultivating <50% of the land acquired). One of our initial research goals was to better grasp the features of operational land deals, but the bias towards non-operational deals has prevented us from advancing knowledge on ongoing projects as much as we had hoped to. It also complicates the statement by Borras et al. (2022) and Wolford et al. (2024) that most studies have focused on operational land deals. Carrying out agro-industrial activities on the African continent can be a tough business, and we suspect that the high rate of failure reflects dynamics on the ground. But it could also express greater scholarly interest in contentious mega-land deals over smaller projects that quietly carry out their operations with no visible or organised opposition.

We also found biases in the geographical coverage in the existing scholarship, both between and within countries. Some countries have been significantly more studied than others. We found as many as almost three times more publications for Ghana, Tanzania, Senegal, and Ethiopia than we have for Mali, Kenya, and Mozambique. In addition, in some countries, some regions have garnered more coverage than others. As Cotula et al. remark (2014, p. 911), it might indicate that land deals are concentrated in specific areas, but we should be cautious with this assertion. In Ethiopia for example, most studied land deals were in the Gambella and Oromia regions, while they respectively represent only 10 per cent of the total farmland investments in the country (Ali, Deininger, & Harris, 2017, p. 8). In Mali, the discussion on new farmland investments focuses almost exclusively on the Office du Niger zone, which stands apart from the rest of the country, both historically and legislatively, leaving us with limited knowledge about other areas.

Methodological issues also arise in the scholarship we reviewed. Case studies generally offer a ‘snapshot’ of a given land deal at a specific point in time. Similar to what Oya (2013) found in his review of 176 publications on land grabs in Africa between 2005 and 2013, very few studies use a ‘rigorous baseline’ and a ‘before-after comparison’ to assess impacts (see also Cotula et al., 2014). In addition, the findings of different scholars on the same deal are sometimes contradictory. These empirical inconsistencies may be due to different theoretical lenses, research questions asked, and methods employed to examine a project, in brief to different epistemological postures. When data was collected also matters. For example, the status and dynamics of a case described in 2012 are likely to differ in 2016. Sometimes, we were unable to confirm the most recent status of a land deal—perhaps it went bankrupt after the paper we reviewed was published. Therefore, although our database and analysis represent a significant improvement on previous attempts at cross-country comparisons, it remains imperfect.

6. Discussion and concluding remarks

This paper analysed the existing scholarship on large-scale agro-industrial investments in sub-Saharan Africa, focusing on the factors accounting for non-operational and operational land deals. We conducted an extensive literature review to build an original database covering 90 land deals in seven countries and mapped 45 distinct parameters for each land deal. Our analysis shows that although the land grab literature has grown exponentially in the past decades, it remains beset with biases that limit our understanding of the determinants and effects of contemporary land deals. This section outlines five important gaps we identified in the scholarship we reviewed and recapitulates our findings about the drivers of non-operational and operational deals.

6.1. Minding the gaps!

The first gap we identified is that few studies have documented operational projects. In contrast with the assertion by Wolford et al. (2024) that these projects ‘have been the focus of most studies’, we find that the conditions enabling continued agricultural production and fostering sustainable local development are poorly understood. Operational projects that have been examined can receive mixed appraisals, illustrating the need to engage seriously with the characteristics and impacts of these projects. For instance, some hail ongoing investments in the Senegalese export horticultural as reducing poverty (Colen et al., 2012) and improving food security (Van den Broeck, Van Hoyweghen, & Maertens, 2018). However, other scholars consider that these land deals rely on disciplining strategies to control labour as well as women’s broader position of subordination in society (Baglioni, 2018) or that they reinforce class and gender inequalities (Salla Dieng, 2019). That different authors appreciate similar projects in an almost opposite manner shows that we cannot assume that land deals are necessarily good or bad.

Second, our corpus of studies illustrates that domestic investments are under-reported in the scholarship although the need to further our understanding of these investments has repeatedly been called for. This is an important limitation since foreign and national businesspeople do not face the same constraints and opportunities. In addition, the cumulative effects of smaller domestic land deals on local agrarian structures and farming environments can be far-reaching, as we touched upon above.

Third, most case studies we reviewed are temporally bounded. However, long-term research is needed to grasp how the interplay between different parameters shapes land deal trajectories, including their impacts on the local economy and the changing nature of investor-community relationships over time. A longer outlook is also required to understand if and how investors can adapt to evolving circumstances (Gagné, 2020). A recent article by Salverda (2022) illustrates that ‘temporalities’ influence the extent to which investors are held accountable: A progressive decline in critical attention allowed a company in Zambia to renege on its promises without facing consequences.

Fourth, the reasons behind investors’ financial difficulties have not received sufficient scholarly scrutiny. The frequent mention of such difficulties as drivers of non-operational deals provides an important confirmation that corporate agriculture may not be ‘as profitable an outlet for capitalist accumulation as originally envisioned’ (Gagné & Fent, 2021). Future research ought to consider more fully how the agricultural background, business motivations, and political connections of investors influence their capacity to raise funds, implement farming operations, and get a return on their investment. Empirical evidence in Tanzania shows that investors overly rely on high-level political contacts and patronage relations to ensure their economic profitability (Bélair, 2021).

Fifth, the corpus we reviewed is primarily focused on documenting aspects of local contexts in relation to land deal outcomes, in particular impacts of land allocation processes, exclusion/inclusion processes, local socio-economic development (employment and local infrastructure), and local governance. Although it shows that scholars have answered Borras and Franco’s (2013) call to document ‘the reactions from below’, this bias indicates that other aspects remain less understood (e.g. environmental impacts; health impacts; sustainability of investment/agricultural models; agricultural impacts on commercial balance, etc.). Navigating local circumstances is indeed a challenge for external investors. However, scholars should be cautious not to treat local contexts as the unique reason for projects’ failure.

6.2. On the scholarship findings: what drives non-operational projects?

Although the scholarship analysed here has limitations, our database yields important insights. According to the authors of the publications included in our database, local opposition stands out as a key driver of non-operational deals. Our analysis therefore delved into the reasons prompting this resistance. We found that flawed land acquisition processes, loss of land by local communities, and unfulfilled investors’ promises feed opposition to non-operational land deals. However, the rate of opposition is similar in both operational and non-operational land deals. Therefore, future research should explicitly tackle the question of why resistance leads to the failure of some projects but not of others.

As our findings indicate, the relationship between investors and local communities shapes land deal trajectories. As Hall, Edelman et al. (2015) remark, there is thus a need to pay careful attention to the political agency of rural people. Not only should we assess whether and why they are ‘resisting, acquiescing or being adversely incorporated’ (Hall, Edelman et al., 2015) but also how they negotiate their inclusion or exclusion from land deals over time, and how these dynamics influence the outcomes of land deals. Sometimes, investors can devote conscious efforts to mend broken relationships with impacted populations (Abubakari et al., 2020; Salverda & Nkonde, 2021). Local farmers may also learn to better negotiate the terms of their inclusion (Schlimmer, 2020; Veldwisch & Woodhouse, 2022). In other cases, governments may muzzle local resistance to investments, enabling investors to initiate and sustain production against the will of the people. In addition, as argued by German et al. (2016), positive effects might arise at first, while negative effects accumulate incrementally and thus become manifest when the project is fully operational. For example, the consequences arising from land loss (e.g. emigration to adjacent areas) or negative environmental impacts can become higher over time. Indeed, ‘companies can become entangled in pre-existing conflicts’ at the local level, which their presence often contributes to exacerbating. These conflicts can backfire in unexpected ways, thus undermining investors’ capacity to control land and make profits (Gagné, 2022).

These reflections link back to issues of land governance and the role played by state actors in processes of dispossession and capitalist accumulation associated with land deals (Sikor & Lund, 2009; Wolford, Borras, Hall, Scoones, & White, 2013). Land deals are embedded in broader historical processes underpinning state power and participate in its territorial logic (Lavers, 2012; Wily, 2012; Bélair, 2022; Belay, 2023). Thus, unpacking the political configurations in which land deals take place is crucial to understanding the outcomes of the contemporary land rush.

Concomitantly, the question of what happens on land lying ‘in limbo’ (Chung & Gagné, 2021) due to delayed investment also deserves further attention. This is important because aborted, cancelled, or scaled-back farmland deals can have negative impacts on smallholder farmers and herders. As shown by Engström and Hajdu (2018), however, non-operational investments tend to be treated by practitioners as ‘non-events’, thus escaping development agencies’ regulations on monitoring and evaluation.

These ‘under the radar’ land projects are unique in two respects. First, there exist few accountability structures or mechanisms (e.g. binding investor-community contracts) to compensate those who have been negatively impacted by something ‘that never happened’. Additionally, the sustainability frameworks implemented by most global and bilateral development agencies include no guidance for monitoring and accountability when projects fail to materialise (Chung, 2020; Engström, 2020). Moreover, when redress mechanisms do exist, the complaints filed by communities are not necessarily taken seriously or dealt with appropriately due to institutional weaknesses. Second, most investors lack proper exit strategies to avoid detrimental consequences after their departure. As a result, local communities harmed by non-operational land deals rarely, if ever, have the means to seek and receive compensation.

On a final note, international institutions, such as the World Bank have pushed corporate agriculture under the assumption that it would outperform family farming. As our analysis shows, however, many land deals in SSA turn out to be non-operational. Decision-makers, including global development banks, bilateral donors, and national states, now need to act upon the fact that large-scale, mechanised farming is not necessarily more productive than smallholder agriculture. The lives of rural people are too precious to pursue these ‘experiments’ without critically engaging with the impacts of both operational and non-operational land deals.

Author contributions

Joanny Bélair: original ideation, conceptualization, methodology, investigation, analysis, original drafting and writing, review, and editing; Linda Engström: conceptualization, methodology, investigation, analysis, writing, review, and editing; Marie Gagné: original ideation, conceptualization, methodology, investigation, analysis, writing, review, and editing.

Acknowledgements

We wish to warmly thank Assistant Professor Melisew Dejene and two anonymous reviewers for their valuable comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Canadian Social Sciences and Humanities Research Council (756-20190-040), the Swedish Research Council (2021-04587), Global Challenges University Alliance (SLU.ua.2021.4.2-3962), and the Fonds de recherche du Québec-Société et culture (291780 and 322644).

Notes

1 Our conceptualisation is however narrower than the one proposed by Brüntrup et al. (2016, p. 5), who include considerations of financial viability in defining project success.

2 One researcher covered five countries (Tanzania, Mozambique, Ethiopia, Ghana, and Mali) while the two others covered one country each (Kenya and Senegal). The literature review focused on English databases. For francophone countries, references in French were included on an ad hoc basis. Two of the researchers have conducted fieldwork on land investments in Tanzania and one of the researchers, in Senegal. Their respective publications have been included in the review when they met inclusion criteria.

3 We sourced references from Google Scholar and other key academic databases.

4 The analysis included 73 papers published in peer-reviewed journal, one book, three book chapters, 10 doctoral and two master’s theses, and 10 research reports (those were included when at least one academic researcher contributed to the paper). To adjudicate between conflicting information or account for the latest developments of a given project, the analysis was sometimes supplemented with corporate materials and media articles. See Supplementary Appendix 2 for the full list of reviewed publications.

5 Local communities are not homogeneous: conflicts may concern only part of the community, and sometimes, intra-community conflicts may emerge alongside investor-community conflicts.

6 Our coding scheme allowed multiple entries for each case to reflect that failure may result from several factors that often cannot be singled out.

7 Job creation and local infrastructure development are common promises made by investors and/or governments. If realized, they have the potential to improve the life of local communities (Brüntrup et al., 2016).

8 For example, new moratoriums on land deals were introduced in Tanzania, Ethiopia, Mozambique, and Kenya.