The politics of evidence: methodologies for understanding the global land rush

Abstract

The most recent ‘land rush’ precipitated by the convergent ‘crises’ of fuel, feed and food in 2007–2008 has heightened the debate on the consequences of land investments, with widespread media coverage, policy commentary and civil society engagement. This ‘land rush’ has been accompanied by a ‘literature rush’, with a fast-growing body of reports, articles, tables and books with varied purposes, metrics and methods. Land grabbing, as it is popularly called, is now a hot political topic around the world, discussed amongst the highest circles. This is why getting the facts right is very important and having effective methodologies for doing so is crucial. Several global initiatives have been created to aggregate information on land deals, and to describe their scale, character and distribution. All have contributed to building a bigger (if not always better) picture of the phenomenon, but all have struggled with methodology. This JPS Forum identifies a profound uncertainty about what it is that is being counted, questions the methods used to collate and aggregate ‘land grabs’, and calls for a second phase of land grab research which abandons the aim of deriving total numbers of hectares in favour of more specific, grounded and transparent methods.

Keywords:

• land grabs
• politics of evidence
• data
• methods

View correction statement:

Erratum

Introduction

In the current, burgeoning debate on large-scale land deals, numbers matter.¹ There are big economic and political stakes at play, and astonishing figures of ‘millions of hectares’ play well in media and policy debates at different levels. But how do we collect reliable data on where land deals are taking place, their size, status and state of production? How do we assess where investments might be most appropriate, offering the greatest returns, given poor existing or baseline information on land use, availability and suitability? How do we understand land deals in the context of wider agrarian transitions, shifting labour regimes and reconfigurations of rural economies? What methods are most appropriate? Can crowd-sourcing approaches be effective? How are claims validated and cross-checked? What are the wider political implications of such data, especially as they become appropriated by different actors? These are just some of the questions reflected in this JPS Forum.

A range of large-scale (global and continental) assessments in the past few years have attempted to grapple with questions of the scale and distribution of land deals, including the Land Matrix project, as well as the ongoing monitoring of land deals by GRAIN, based on media reports, but also influential reports by the International Institute for Environment and Development (IIED) (Cotula et al. 2009), the International Food Policy Research Institute (IFPRI) (von Braun and Meinzen-Dick 2009), Oxfam (2010), the Oakland Institute (Daniel and Mittal 2009, Oakland Institute 2011), the International Land Coalition (ILC 2011) and the World Bank (Deininger et al. 2011), among numerous others.² The numbers and the cases that have emerged from these efforts have framed the debate.

This second JPS Forum on Global Land Grabbing brings together four papers, originally presented at the Global Land Grab II International Conference³ organised by the Land Deal Politics Initiative (LDPI) , and debated at a lively plenary session. This follows the first JPS Forum on Global Land Grabbing (Borras et al. 2011) which, ahead of the Global Land Grab I International Conference, outlined the dimensions of a research agenda to link understandings of the current land grab with longstanding questions in critical agrarian studies: rural livelihoods and social differentiation, large versus small farm development paths, labour regimes, land tenure and resource governance, political organization and mobilization, and the location of agriculture within the wider political economy.

In this second JPS Forum, two articles offer insights into the rationales and practices of two major initiatives – the Land Matrix (Anseeuw et al. 2013) and GRAIN's farmlandgrab.org database and associated reports and tables (GRAIN 2013). Both initiatives have made important and high-profile contributions to the ongoing debate about ‘land grabbing’, but each has serious limitations and problems. Some of these are highlighted in two subsequent commentaries, one from Carlos Oya (2013), agrarian theorist at the School of Oriental and African Studies who works on Africa, and another from Marc Edelman (2013), an anthropologist working on Latin America's economic history at City University New York. Both question methodological assumptions pervasive in the land grab literature, and ask what distortions might be introduced as a result in our understanding and responses.

The Forum therefore centres on a discussion of the methodologies that have been chosen to understand patterns and processes of land investment, and the politics of evidence that arise from these choices. The focus is on the methods used to identify, count, aggregate and understand land deals at global and continental scales, and also how to link these macro-level insights to the burgeoning case study literature.

Evidence for policy

In such a charged setting we must reflect deeply on the politics of evidence, and its relationship with policy. Knowledge and policy are of course always co-constructed. There is never any neat, clean separation of fact from value and context, as suggested by simplistic calls for ‘evidence-based policy’ perspectives (Denzin 2009). As Bruno Latour explains, facts have a life in a policy world that is dependent on processes of ‘enrolment’ and ‘enlistment’ (Latour 1987). Facts have reach and influence if they have backers. But such a social constructivist stance does not mean that data quality, in-depth analysis and rigour are irrelevant. Realities out there are just as real; they just get interpreted in different ways. In the complex politics of policy debates, it is the interpretive moves that are crucial, and it is here that facts get shaped by particular contexts, or historical moments (Jasanoff 1996).

The Land Matrix and GRAIN went about the task of collecting and collating evidence in different ways. The Land Matrix partnership,⁴ initiated by the International Land Coalition (ILC) and building on its other related initiatives,⁵ adopted a crowd-sourcing approach to identifying land deals. Submissions were checked and confirmed, then entered into a large database.⁶ The GRAIN approach was to use Internet searches to compile records, largely based on media reports, of land deals internationally, and present these data on a searchable database,⁷ as well as in occasional reports (GRAIN 2008, 2009, 2010, 2011). They both became the ‘go-to’ places for anyone interested in the phenomenon, and thus their data gained considerable traction in the broader policy debate (Deininger et al. 2011).

Both the Land Matrix and GRAIN have worked over time to improve accuracy, cross-checking and verification. The Land Matrix initiative emerged from a coalition of research and funding organisations, while GRAIN is a civil society organisation. They represent diverse interests and have different skills and capacities.

None of the major initial contributions to the wider debate came from exclusively academic research. Only now, as more in-depth, slightly longer-term work on the politics of land deals begins to emerge, do we have a growing body of academic research on the topic, a substantial portion of which has been published in the pages of this journal,⁸ and arose from the two major conferences convened on the subject.⁹

The debates in the Forum reflect the ways in which different people, at different moments, need to combine in critical, engaged research on an emerging topic. If we wait for the in-depth academic research, the prospects of intervening in a fast-moving political and policy debate may be long gone. But, equally, if we ignore such research, the initial and necessarily more impressionistic results may go unchallenged, untested and unverified, and myths and misunderstandings will arise. In this paper we argue that it is important to have both types of research, conducted by a variety of different actors. But what is crucial is that they critically engage with each other, and a more constructive conversation arises, one that helps shape policy directions in ways that are informed by evidence in a productive tension. This is the aim of this Forum, and indeed has been the focus for the work of the LDPI over the past few years. Challenging the experiences of the Land Matrix and GRAIN is not an exercise in academic nitpicking, nor one where practitioners are pitched against researchers, but one where important debates are raised for everyone involved.

The land rush context

The last five years or so have been a very particular context and moment, when the phenomenon which has been dubbed ‘land grabbing’ was clearly unfolding rapidly in certain parts of the world, potentially with major consequences for both economies and livelihoods. Impacts were also potentially far-reaching and irreversible, so any intervention in the debate had to occur quickly, prompting an urgent need to inform campaigning and policy advocacy at local, national, regional and global levels. This was also a process with big political consequences, with major commercial and political interests often pitted against unorganised and voiceless land users. It was also a debate that quickly gained media exposure, bringing people together – and dividing them – in the so-called global North and South. This all presented important campaign opportunities for those concerned with the consequences of unfettered globalisation. Whatever one's views about the potential benefits or otherwise of such large-scale land deals, finding out what was happening, where and involving whom was a critical and urgent task. Given the limits of existing data, due in part to high levels of commercial secrecy around such deals, this was not easy.

The moment therefore brought together a number of actors – researchers, practitioners, investors, activists, policy-makers – all eager to find out what was going on. It was thus a prime context for engaged research. But, as always, those involved came with positions, values and politics. Some were explicit about their positionality; others less so. GRAIN's position is very clear:

Let us state this plainly: GRAIN's aim is not to do neutral research. With an explicit political agenda guiding what we choose to focus on and how we use the information, our aim is to gather the best and most useful information that can support responses by local communities and activist networks. (GRAIN 2013)

Anseeuw et al. (2013) describe the role of the Land Matrix partnership as being ‘to promote transparency and open data in decision-making over land and investment, as a step towards greater accountability’. Their argument is that with greater transparency in data, and public sharing of the database, this will help inform the debate, and help hold those with more power – whether investors, national governments, financiers, or aid donors – to account. Again, the political role of the research is apparent, even if not explicitly expressed in the way GRAIN does.

Academics, particularly in the social and policy sciences, have less of a tradition of registering their interests and positions in relation to their work. Despite the long tradition of academic critique of a positivist position (Fischer 1998, 2003), many academics and policy makers continue to express faith in ‘evidence-based policy’. They adopt the view that truth is expected to speak to power (Wildavsky 1987), somehow mediated through the presentation of peer-reviewed evidence, rather than through a more argumentative process (Fischer and Forester 1993, Hoppe 1999).

But how do data and other sources of evidence actually engage with policy in practice, particularly in the context of a highly charged, and often emotive, debate? The next section reflects on this relationship.

Engaging with policy: the role of data

How have data been presented in the policy debate? We have already mentioned the media catalogue produced by GRAIN and the database by the Land Matrix, but other reviews have made use of these sources, and presented them as an authoritative picture, with the headline figures in particular often grabbing the headlines.

Often on the basis of the same, limited sources, very different conclusions are drawn. The World Bank study found that, during 2009, investors expressed interest in 56 million hectares of land in less than a year (Deininger et al. 2011, xxxii), while the ILC (2011) indicated the figure was 80 million (White et al. 2012, 620). In another publication the same year, Oxfam (2011, 2) stole headlines with the claim that 227 million hectares was leased or sold in developing countries (although their calculation began in 2001). Burrowing into the footnotes, reasons for the discrepancies can be found: for example, in some cases all ‘reported’ cases were included, while in others only ‘confirmed’ ones were. Few assess whether anything is actually happening on the ground, and most reports offer very few insights into overall impacts, either positive or negative.

The World Bank report (and later Deininger and Byerlee 2012) combined these global assessments of land investment with an analysis of land availability and suitability based on satellite imagery developed by the International Institute of Applied Systems Analysis (IIASA) based in Vienna (Fischer and Shah 2010). This offered a classification of countries according to four types: ‘little land for expansion, low yield gap’ (Type 1); ‘suitable land available, low yield gap’ (Type 2); ‘little land available, high yield gap’ (Type 3); ‘suitable land available, high yield gap’ (Type 4) (Deininger et al. 2011, xxxv–xxxvii). Flawed assumptions about land uses, and reliance on unverified land deal data, combined one disputed set of data with another, to come up with a typology with an illusion of validity – the ‘syndrome of false precision’, as Oya (2013) calls it – informing a set of quite concrete recommendations, which gained traction through the authority of the report's origin, the World Bank.

As both Edelman and Oya note in their commentaries, these sources of contradictory evidence, based often on shaky data sets, made any considered debate about policy implications very difficult. This remains the case, as we actually still don't know how many land deals have been entered into, where and with what consequences. Despite the ‘literature rush’ that Oya identifies, and the proliferation of case studies, media reports, databases and so on, how can we still be in such a position? The papers in this Forum point to some of the reasons.

First is the fixation on ‘the killer fact’: the number that sways the debate, gains the media profile and is in the top line of the press release. This is understandable but clearly problematic, especially when such ‘facts’ diverge so dramatically – for example, when estimates of hectares being transacted vary from 43 million to four times that, and are often compared, even though they refer to different types of transactions, with information obtained through different methods, and relate to different time periods. While it is inevitable that the media will pick up on such elements of a story, no matter how many caveats, researchers need perhaps to be more aware of the consequences of their fact-building enterprises in this 24-hour media world. Oya (2013,) quotes Oxfam's policy advisor, Duncan Green, who cautions, ‘don't “[u]se a killer fact that is not credibly sourced, even if it fits your message. It is not worth damaging your credibility for a quick hit. And remember – if in doubt, leave it out!”’ This warning has not always been heeded.

Second is the way inappropriate inferences are derived from ‘data’. Apparently ‘empty’, ‘underutilised’ land viewed by satellite image analysts in Austria may look very different when looked at from the ground, where land is often occupied, used and governed in ways not visible via the gaze from space. Equally, land deals are not synonymous with productive use; data from case studies suggest that many have not been put into production by the investor, although in other cases, land deals do result in productive use, with employment and other economic linkages being generated, benefiting substantial numbers of people and/or displacing others. The problem, again, is that data on registered interest in land are not good indicators of use and impact.

Third is the issue of sources and their quality. As Edelman notes, every researcher should be ‘obligated to account for the existence of the source she or he employs. Who created it and why? What were the circumstances and context of its production? What accounts for its preservation, its location in an archive or its diffusion? What does it say and what are its silences?’ (2013,) This sort of rigorous reflection has often been absent. Ward Anseeuw et al. (2013,) reflect on this in respect of the Land Matrix:

the new version of the Land Matrix will show the negotiation and implementation status of each deal, also allowing filtering by status. In addition, it will include separate categories like “expression of interest” in land, “negotiations failed” and “project abandoned”, giving an indication of the scale of interest in land acquisition and allow tracking of changes to deals over time’.

Each piece of information will directly be related to its source, enabling a reader to filter deals by the type of source (e.g. media report, research paper, company source, crowd-sourcing etc.). Thereby, users can judge themselves whether they consider the information reliable.

Fourth are selection biases. Due to the differential availability of information and the flocking instincts of researchers, NGOs and media commentators alike, there are often severe selection biases – by region (with an apparent bias towards Africa), by country (perhaps Ethiopia with lots of reports and high-profile cases), by investor country (seemingly always China, at least in the Western press), by origin (foreign investors highlighted over domestic partnerships), and by scale (with a bias towards counting a few ‘big’ deals over many small deals). This imposes biases as data are aggregated up, and may generate further misunderstandings. If the aggregate data is skewed, it means that any statistical analysis is meaningless, and the database is no basis for choosing cases, or defining a sampling frame – or for donor or policy interventions.

Fifth are issues surrounding the review process. Rigour, authority and reliability are supposed to be checked by claims being subject to review, either by a formal peer review process as in the academic literature, or more open styles of review and response. Oya comments that a number of publications offer ‘a mix of actual facts, perceptions, intentions, rumours, guesstimates’ (2013). Sources are variable too, mixing media reports, crowd-sourcing and more detailed case studies and field enumeration. In terms of the limits of formal review, Edelman comments on a recent piece in the prestigious Proceedings of the National Academy of Sciences (Rulli et al. 2013), that claims to assess the extent of water resources appropriated through land grabs. But, despite the supposedly rigorous review, closer examination discloses serious flaws. As Fred Pearce comments in the New Scientist, ‘The findings would be best ignored, except they are the only peer-reviewed global water grab assessment in existence and are already being quoted’ (Pearce 2013) – and will be again now! Whatever the review process, there needs to be some level of assurance that the data and analysis presented is rigorous and reliable. In a context where data are shaky and analysis heavily conditioned by authors' stances, this is challenging, but not impossible, and requires perhaps greater attention than has been afforded to date – in top-rank peer reviewed journals and non-governmental organisation (NGO) and donor reports alike. These patterns show the extent of academics' reliance on NGOs for data in this recent literature rush on land deals.

Sixth, the rapidity of easy access to ‘data’ and the dangerous allure of Google have facilitated the recycling of facts long after their sell-by date. Reliance on often outdated web sources has led to a circularity of referencing, producing a meta-discussion of land deals quite ungrounded in on-the-ground verification. The result has been circular referencing, reproduction of discredited data and double-counting of deals, as exemplified in Locher and Sulle's (2013) meticulous unpicking of the dodgy data on Tanzania. With a call for ground-truthing, new norms of presenting traceable data and making available sources and specifying effective dates of data-gathering, they caution against academics (or anyone else for that matter) relying on shoddy modes of knowledge production while showing just how widespread this has been.

Seventh, and most profoundly, is the unresolved question: what is a ‘land grab’? A register of interest, a land deal proposed and concluded, an actual enclosure and resulting dispossession, the actual conversion of enclosed land to new uses, the consequences of some or all these developments for employment, incomes and community life? Each of these stages in the process of land grabbing needs to be documented, as carefully as possible. The problem is that data on different steps have often been aggregated together. Whatever protocols and improvements are generated for data gathering and verification will be meaningless if non-equivalent data is aggregated because we are not agreed on what is being counted.

Generating policy narratives

Accuracy, rigour, cross-checking and being honest about sources is essential for any research, from whatever provenance. But we also must remember that much of this research is not research for its own sake. It is very much geared to influencing policy, raising the profile of an issue, generating debate, sharing information, and influencing decisions. The growing body of formal and informal literature, from all sources, has indeed had a major impact, and over quite a short period, notably on the G8's launch of a New Alliance for Food Security and Nutrition (White House 2012).

For example, there is now wide concern about transparency and accountability in land investments, rising to the top of the G8 Agenda this year. Also, the Voluntary Guidelines on Responsible Governance of Tenure of Land, Fisheries and Forests in the Context of National Food Security were approved at the Committee on Food Security in May 2012 (FAO 2012). Given the divergent positions of those involved, this was an extraordinary achievement, made possible by the strength of shared concerns for a more effective policy framework. And in national and regional settings across the globe, governments, civil society groups and indeed investors are debating how to address the challenges raised by this work.¹⁰

All research, from whatever source, carries with it a politics. Policy narratives – storylines about the world that frame the problem and suggest a solution – are always underpinned by particular knowledge claims, supported by evidence and data. Here science and policy become deeply intertwined, one mutually informing the other. But narratives are not neutral statements; they are associated with particular interests, and deployed towards particular ends. Evidence can be captured, spun, selectively presented and interpreted in certain ways.

Thus NGO campaigns may need a media-grabbing number, usually one that is sufficiently large to attract attention. This in turn may create a narrative about how land grabs are ‘as big as x [small European] countries combined’ (or some other equivalent unit), and this generates the story. This generates interest among politicians, generates concern among the public, and mobilises activists, and of course the public support that helps keep the organisations going. The ‘fact’ may or may not be true, but the narrative gains power almost independently.

This ploy is not restricted to NGO campaigners. The same goes for investors, who equally deploy different versions of a ‘scarcity’ narrative to their own ends.¹¹They argue that the world is running out of food, and that certain areas of the world have plenty of land, and that this justifies investment. They also profit from this ‘manufacture’ of scarcity, as land prices rise through speculation (McCarthy and Wolford 2011, Merian Research and CRBM 2010, Oxfam 2011). Others too may benefit from narratives that point the blame, sometimes selectively or inaccurately, towards others. China, for example, has been singled out as a rapacious land grabber, when in fact this is not true (Brautigam and Ekman 2012). Although investment plans exist, and major infrastructural developments are underway, China has not ‘grabbed’ large tracts of land for investment in Africa. Such narratives of blame serve real interests, including those of domestic elites in alliance with other investors, for whom such narratives serve as a major public relations boost, diverting attention elsewhere.

Fact building, and the creation of associated narratives, must therefore be seen as a social and political process, whereby certain people, institutions and networks are enlisted. As such processes of enrolment occur, a particular narrative can hold sway in policy until it is upset by competing alternatives. The struggle over which narrative is at the centre is thus intensely political, and so reflects interests and power differences between actors. In some cases, unlikely alliances and unusual compromises can be brokered, as, for example, in the agreement over the Voluntary Guidelines, but such a document has to opt for an inclusive, compromise narrative, whereby disputes are set aside and language neutralised (Stone 1988, Shore and Wright 1997).

As sociologists of science have long argued (Jasanoff 1996, Jasanoff and Wynne 1998), all knowledge building must be understood in this way, and the interests associated with different forms of evidence must be taken into account – whether a database by an NGO, a report with the World Bank imprint, or a peer reviewed article by a professor from London or New York. This does not abandon a concern with reality, but accepts that interpretations are inevitably and always conditioned and socially constructed.

Framing assumptions

As both Oya and Edelman point out in their contributions, there are many problematic framing assumptions involved in the land grab debate, all of which colour interpretations – and ultimately actions – in important ways. Being aware of these, and challenging them, is an important part of any rigorous analysis.

First is the focus on land areas, and the numbers of hectares. A concentration of data collection on this frames the problem in a particular way. But what about issues of land quality, value or location, for example, rather than just extent? And who has an incentive to tell a researcher about their land holdings? Some may wish to hide their land holdings, fearing taxation or expropriation; some may be part of formal registration schemes, while others, with scattered and small land holdings, are not; others may simply not know or care about how many hectares they ‘own’, as it is family or communal land, held by numerous people. The reality is intensely ‘messy’, as Edelman points out with fascinating examples from Costa Rica.

A focus on land area of course frames the debate in terms of ownership, tenure and title, and so creates a politics of measurement, legibility and control (Scott 1998). Measuring, defining and registering hectares through processes of cadastral surveys, land registration and land title have a differential impact on the rich and the poor. As Edelman notes:

If the poor may cultivate illegibility about the land areas they control, the wealthy seek to make their own areas legible for the purposes of obtaining legal guarantees (registration, leases, etc.) while simultaneously cultivating illegibility about ownership (as noted above) through the use of corporate holding companies, subsidiaries and cut-outs. Both of these zones of illegibility pose significant challenges to land deal researchers. (2013,)

At the same time, quantitative renderings of land deals suggest an authority, a mathematical precision, belied by the messiness of their collection. Statistics, particularly around something as sensitive as land, exert a form of governmentality, ordering the world in favour of a particular hierarchy of power. Statistical services, cadastres and tax offices have long been central to state control and ordering (Hacking 1990, 1999, Scott 1998). And today, perhaps inadvertently, the new statistical databases, available on the Internet, and produced by very different players, have a similar effect, exerting power through measurement. As Edelman notes, ‘Every dataset has an implicit epistemology behind it. Different kinds of datasets are created for different administrative, bureaucratic, political or other purposes and always contain systematic biases’ (2013,).

Another way the debate is framed, influencing in turn the way data are collected and evidence is processed, is through dichotomous contrasts, as Oya points out. These may miss the point, and obscure some of the key issues. So, if simply framed in terms of returns to land (as in the old inverse size-productivity relationship debate), for example, the contrast between ‘small farms’ and ‘large farms’ may miss the consequences for employment, livelihoods and off-farm economies of a mix of farm sizes in an area, and the real benefits of a commercial sector to a regional economy. Equally, the focus on the own-farm ‘peasant’ producer may miss out on wider questions of labour where, particularly if a gender lens is added, paid employment on larger farms may offer greater livelihood opportunities than toiling on a small plot under exploitative, patriarchal conditions. And an emphasis on ‘foreign’ land grabbers may similarly deflect attention from the role of domestic elites, and the way capital flows between overseas and domestic business and political interests.

These frames therefore build on the biases that emerge from data collection and presentation. By not looking at regional economic linkages, or avoiding the labour question, research may end up failing to address impacts, beyond indicating a certain number of hectares that have been appropriated (or were under negotiation), and a certain number of households that have been dispossessed. These data of course are still valid, but tell only part of the story and feed only a certain narrative. The result, Oya argues, is that a ‘neo-populist’ pro-smallholder farmer, anti-labour, anti-foreign investment narrative evolves, and alternative frames do not get a look in.

Styles of research: what is appropriate?

Work on ‘land grabbing’ has reached a critical juncture. The early urgency of the 2008–2012 period has perhaps passed, the debate has definitely risen up the political agenda, and now there is a need to reflect, challenge and reframe, nuancing and sometimes confronting existing narratives.

Both Edelman and Oya argue strongly for the need for long-term, detailed, empirical research, based on painstaking, rigorous analysis of the data. They point out the real challenges of researching land issues, and the controversies that surround naming land sizes or discussing tenure. They argue for establishing baselines, counterfactuals, comparative frames and careful sampling to gain rigorous data that are credible and authoritative. They contrast this research mode with ‘quick and dirty’ research, involving fast fact-finding missions and rapid assessments. They also critique the practice of putting unchecked data into the public domain, with verification expected to come later. They would prefer, it seems, a much more slow, deliberate and thorough approach. Oya argues for

being patient and spending more time to collect high-quality evidence on process, actors and impact and systematically dealing with biases, lies, imprecise figures and mistakes that are unfortunately common in any research dealing with land use, labour and production in developing countries. (2013,)

Certainly there are real dangers of false precision (where ‘facts’ are presented as concrete and undisputed, yet their basis is dubious) and there are even bigger problems with straight inaccuracies, mistakes and (worse) active distortions. At the same time, we must be cognisant of the way in which debates in the fast-moving, real world are actually shaped. The contributions of GRAIN, the Land Matrix and others, despite their limitations, have been substantial and important. Rough-and-ready, quick-and-dirty work of this sort is, under certain conditions, necessary.

This methodological discussion reflects a long-running debate in applied development studies, where calls for appropriate imprecision, optimal ignorance and proportionate accuracy (Chambers 1981, 1983) are pitched against the alternatives which are deemed to be ‘long and lost’ anthropology and the type of ‘survey slavery’ where the results never see the light of day. In 1981, Robert Chambers argued for a ‘fairly quick and fairly clean’ approach, which then became known as ‘rapid rural appraisal’ (Chambers 1981). This was an attempt to find a balance between extended multi-year field immersions, detailed and very expensive monographs, and floods of arcane data and statistics which dominated academic outputs then as now, and the rapid roadside consultancy with all its rural biases (Chambers 1983).

In some ways we see a similar contrast being played out now. But the resolution of the earlier discussion was not an either/or choice, but a mix. Rapid appraisal had its place, as did well-designed surveys and long-term qualitative fieldwork. The same applies today. There is a need for complementary efforts and, crucially, for the different sources of evidence – and their associated frames – to speak to each other. So, yes, the rapid collection and collation of data have very important uses, especially when we need to know about things quickly. But their level of imprecision must be appropriate, and the caveats need to be clearly highlighted. Triangulation and checking have always been hallmarks of good rapid appraisal, and should be part of the current efforts too. And such processes need to give a broad, and necessarily rough, picture, but should avoid boiling down to singular facts (or ‘factoids’). Researchers also need to be more careful in engagements with the media and policy, to avoid the emergence of simplistic – or simply inaccurate – narratives. Long-term, detailed case study work is certainly an ideal complement. It allows us to dig deeper, sample more effectively, triangulate better, and develop more comparative insights. But, perhaps most especially, such work allows us to ask new questions which can help reframe the debate, and recast the rapid data collection exercises. Therefore, one approach is not necessarily wrong, just insufficient.

In addition to the discussion about the types and pace of data collection, another long-running methodological debate is also raised: the question of what type of evidence is appropriate. This goes beyond the hackneyed qualitative versus quantitative debate (of course a mixed method approach always makes sense), to thinking about the form and medium for evidence collecting and portrayal. In addition to the more conventional forms of data, how do personal testimonies, video and photographic records, case studies and histories fit in? Rather than a tyranny of statistics produced, processed, analysed and verified by outsiders, how can local participants have a say? Can those involved in land deal processes be part of the knowledge building? Advocates of participatory action research (Fals Borda and Rahman 1991), argue that this is an essential route to gaining authentic, subaltern perspectives, with research linked to action. This is not a passive form of participation, whereby villagers provide information to outsiders, but one where those not normally accredited as ‘researchers’ become more centrally involved. While there has been much talk of ‘crowd-sourcing’ data, this perspective takes a different angle, one not really pursued in the discussions in this Forum, but something worth considering – namely, researchers being conduits for local voices rather than replacing them.

Today, with greater levels of education and access to the new media, many more people can be ‘researchers’. The elitism of academia is being challenged, and it is perhaps incumbent on those with the experience and training in research approaches to engage more fully with this wider diversity of research players (Edelman 2009). If we keep to our ivory towers, and our slow and patient processes of fact building, we may be left hopelessly behind. A new story or fact can be tweeted from a remote location, and be promoted, amplified, and spread through the new media at an extraordinary pace. Interactions and conversations between different players are essential: all are part of a new, potentially highly productive mix.

Critical, engaged research

What next steps are suggested by the debate generated in this Forum? What should ‘critical, engaged research’ on land investments look like?

All those involved need to be explicit about politics and framings. We must be transparent about our data and accountable for our findings. We must reach out to different researchers and diverse audiences. We must facilitate critical dialogue and debate about findings, between all parties. In sum, we must not naïvely expect facts simply to speak to power, but become involved in the policy debate, maintaining rigour while facilitating debate.

But also, such engagement requires the important quality of reflexivity: around position, identity, politics, ideology and other biases. Much of this again centres on methodological choices, and the ways that knowledge is produced. By claiming authorship, we claim power, and we need to be clear about the responsibilities of this. Reflexivity also requires open conversation and respectful debate: challenge and confrontation, as well as dialogue and deliberation. The Cornell conference, and the panel from which the papers in this Forum arose, was one such space, and there need to be many others where academics, practitioners, policy-makers, investors and others can meet, listen, debate, challenge and learn.

Without doubt, in the last five years, the collective endeavours of a huge array of researchers from a diversity of organisations, both within and outside academia and with important alliances between groups, have raised the profile of an issue, presented data, debated the consequences and had substantial impact on the policy debate, at both global and national levels. In a sense, this has indeed been critical, engaged research at its best. Yet, we argue, we are at the end of the era of the ‘killer fact’; aggregating total hectarage to describe the scale of land deals has been crucial to placing the issue of land grabs on the public and policy map, but this exercise has largely run its course.

The past five years have represented, in important ways, new pathways of knowledge building. Research has had to respond to a fast-moving context, operating in real time. Information had to be publicly available soon after the research took place, to allow for response, but also triangulation, correction and validation. This offered opportunities, therefore, for a wider, more inclusive process of peer review, including by activists and others on the ground. Conferences and workshops have provided opportunities for discussion of results, and wider synthesis, and these have often been characterised by a diversity of participants, including academics, field practitioners, activists, and policy-makers, all in the same room, discussing, reviewing, critiquing and making sense of the data. This has not been the slow, painstaking generation of evidence, academic peer review and publishing where field data may appear years after it was collected. That mode, where the routines are fixed, the pace slow and the peer review community narrow, seems inappropriate and outdated for this sort of issue. Instead, research on land deals has been fast, intense and grounded in a commitment to open data. With a philosophy of data transparency, accountability and sharing, much of the material has been available open source via the Internet. Some academic journals too – including most notably this one – have recognised the importance of this, perhaps especially for this debate, and have offered free access to articles, where data normally sits behind paywalls for long embargo periods.

Different modes of research are appropriate for particular moments. All are co-constructed with policy and politics, and when issues are urgent, fast-moving and requiring a response, then one approach is needed; as we reflect, analyse and produce deeper understandings, new approaches may be more appropriate. The functions of academic research are vital, but must not be obscured by language, access or rarefied and narrow debates, and above all must engage in a critical dialogue, if injustices are to be challenged and improved policy frameworks are to be forged.

A new phase of land grab research is now needed, which builds on the first phase discussed in this Forum, refines methods, concepts and criteria, and establishes new norms and systems for sampling, recording and updating information. Such a new, open, engaged process of collaborative research and review will require new research ethics, processes, protocols and checks. These have yet to be defined, but examples are already emerging of innovations in this area, the second phase of the Land Matrix being an example. As research on the ‘land rush’ continues and deepens, more critical dialogue is needed among NGOs, other civil society formations and academics on conceptual, political and methodological questions that can underpin the co-production of data.

Additional information

Ian Scoones is a Professorial Fellow at the Institute of Development Studies at the University of Sussex. He is co-director of the STEPS Centre (www.steps-centre.org), joint coordinator of the Future Agricultures Consortium (www.future-agricultures.org) and a founder member of the Land Deal Politics Initiative (www.iss.nl/ldpi). He is co-editor of the Journal of Peasant Studies issue on ‘Green Grabbing: A New Appropriation of Nature?’ (2012). Email: [email protected]

Saturnino M. Borras Jr. is Associate Professor at the International Institute of Social Studies (ISS), The Hague, Adjunct Professor at the China Agricultural University, Beijing, and Fellow of the Amsterdam-based Transnational Institute (TNI) and California-based Food First. He is co-guest editor of Globalizations journal special issue on ‘Land grabbing and global governance’ and of the Canadian Journal of Development Studies special issue on ‘Land grabbing in Latin America’. Email: [email protected]

Ben White is Emeritus Professor of Rural Sociology at the International Institute of Social Studies (ISS), The Hague. He is a founder member of the Land Deal Politics Initiative. He is co-editor of the Journal of Peasant Studies special issue ‘The new enclosures: critical perspectives on corporate land deals’ and the Development and Change special issue ‘Governing the global land grab: the role of the state in the rush for land’. Email: [email protected]

Wendy Wolford is the Robert A. and Ruth E. Polson Professor of Development Sociology at Cornell University; she is also the Associate Director for Economic Development at Cornell's Atkinson Center for a Sustainable Future. Wendy has worked on issues of land distribution, social mobilization, and agrarian politics for many years, primarily in Latin America. She is one of the founding members of the Land Deals Politics Initiative. Email: [email protected]

Notes

¹The authors are the co-convenors of the Land Deal Politics Initiative (LDPI). See http://www.iss.nl/ldpi

²It is notable that the ‘literature rush’, as Oya (2013) terms it, is a largely Anglophone phenomenon, driven by institutions working in English, with relatively few major reports in other languages.

³The Global Land Grab II International Conference was held at Cornell University, 17–19 October 2012. http://www.cornell-landproject.org/program/

⁴The Land Matrix partnership members are the International Land Coalition www.landcoalition.org; Centre for Development and Environment (CDE) www.cde.unibe.ch; Centre de coopération internationale en recherche agronomique pour le développement (CIRAD) www.cirad.fr; the German Institute of Global and Area Studies (GIGA) www.giga-hamburg.de; and the Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) www.giz.de.

⁵Prior initiatives include the Land Reporting Initiative, the Commercial Pressures on Land project, the Land Portal, and the Land Observatory.

⁶ http://landportal.info/landmatrix

⁷ http://www.farmlandgrab.org/

⁸See: Wolford et al. (2013); Margulis et al. (2013) ; Borras et al. (2013) ; White et al. (2012) ; Mehta et al. (2012) and Fairhead et al. (2012).

⁹Global Land Grab (University of Sussex, 4–6 April 2011) and Global Land Grab II (Cornell, 17–19 October 2012).

¹⁰An example of such processes was an awareness-raising workshop with government officials, civil society organisations and academics from across east and southern Africa on the voluntary guidelines, held in Kigali in February 2013.

¹¹See for instance Afgri (2013) and InvestAg Savills (2011), as examples of such narratives.