The following is the established format for referencing this article:
Magliocca, N., E. A. Ellicott, M. Ingalls, M. Epprecht, C. Hett, V. Nanhthavong, and A. C. de Bremond. 2022. Spatio-temporal unevenness in local land system regime shifts caused by land deals in Lao PDR. Ecology and Society 27(4):7.ABSTRACT
Extensive land-use “regime shifts” have been observed as rapid transitions from natural land cover or subsistence-oriented land use to intensified and/or expanded commodity production. However, it is often unclear whether these land-use changes are part of broader land system regime shifts in which pre-existing production systems and livelihood strategies are fundamentally transformed along with observable land-use changes rather than simply displaced or eliminated. This is a critical social-environment question given that regime shifts are often a desired and intended outcome of national rural development and market-liberalization policies but must also be attentive to environmental conservation and/or climate change mitigation goals. We investigated whether nationally extensive land-use changes implemented through large-scale land deals in Lao People’s Democratic Republic resulted in full, partial, or no village- and landscape-level regime shifts in and around land deals. Overall, land deals triggered a wide variety of full, partial, and no regime shift outcomes. Land deals with both domestic and foreign investors produced positive and negative outcomes, although foreign land deals for the production of rubber led to significantly higher rates of indirect land-use change in impacted villages than domestic and/or non-rubber land deals. Also, financial compensation alone was insufficient to improve community well-being because it could be reinvested to perpetuate previous land uses without the desired transformation in livelihoods and rural development. Land deals that exhibited greater social embeddedness (i.e., provided adequate compensation complemented by job-creation and improved access to infrastructure and/or services) were more likely to lead to positive regime shifts. Our findings demonstrate that any land system regime shift will produce both winners and losers, and thus it becomes necessary to critically analyze the localized distribution of social and environmental costs and benefits within broader-scale land-use regime shifts.INTRODUCTION
Land system change is a global sustainability challenge. Environmental impacts from land-use change, particularly from forested to cultivated landscapes, have profoundly altered the structure and function of the Earth’s system (Ellis et al. 2020, Turner et al. 2020). Concurrently, average material consumption levels are at their highest in human history, concomitant with increased trade in agricultural commodities, which is associated with widespread conversion of natural land cover to production landscapes (Steffen et al. 2011, Arbault et al. 2014, Turner et al. 2020). Rapid transitions from natural land cover or subsistence-oriented land use to intensified or expanded commodity production have been described as land system “regime shifts” (Meyfroidt and Lambin 2009, Aide et al. 2013, Müller et al. 2014). In many cases, these regime shifts are a desirable and intended outcome of national rural development and market-liberalization policies (Vongvisouk et al. 2016, Ingalls et al. 2018, Junquera and Grêt-Regamey 2019, Nanhthavong et al. 2020). However, land use regime shifts at the aggregate, national level may conceal stalled or even deleterious socioeconomic change, or sub-national regime shifts at the local level when a transition to commodity production leads to uneven rural development, unequal access to non-land livelihood opportunities (e.g., wage labor), or geographically concentrated environmental degradation in and around production regions (Meyfroidt et al. 2013, Godar et al. 2016). We investigate whether nationally extensive land-use changes implemented through large-scale land deals in Lao People’s Democratic Republic (henceforth Lao PDR, or Laos) rooted in state strategies to promote rural development, result in full (i.e., poverty reduction and no indirect land-use change (iLUC)), partial (i.e., poverty reduction or no iLUC), or no regime shifts associated with land deals.
Regime shifts are an emergent, intrinsic characteristic of complex adaptive systems (Levin et al. 2013), of which land systems are a type. Regime shifts are a form of punctuated equilibrium in which dramatically large, persistent, and often unexpected changes can take place (Scheffer et al. 2001, Müller et al. 2014). Regime shifts in land systems can be triggered by the sudden onset of stochastic events or by more gradual changes due to loss of resilience through longer-term and/or exogenous system changes (e.g., climate change or global market integration; Folke et al. 2004, Dearing et al. 2010, Brook et al. 2013). The accumulation of many relatively small but synergistic changes, such as new individual decision-making logics or livelihood opportunities that lead to land-use changes, can also endogenously trigger a regime shift and prompt land governance policy changes (Rudel 2007, 2013). However, regime shifts may not be evident at all levels of a land system (i.e., local to regional to global) or for all land users and locations in a large-scale land system (Müller et al. 2014, Filatova et al. 2016, Ramankutty and Coomes 2016). The possibility of partial or scale-dependent regime shifts poses significant conceptual and analytical challenges for land systems (Müller et al. 2014, Ramankutty and Coomes 2016).
Regime shifts challenge current efforts to theorize and understand land system change (Turner et al. 2020), and the type of regime shift studied or detected depends on how the land system is defined. Verburg et al. (2015:29-30) offered an encompassing definition of land systems as “. . . all processes and activities related to the human use of land, including socioeconomic, technological and organizational investments and arrangements, as well as the benefits gained from land and the unintended social and ecological outcomes of societal activities.” Causally linking these diverse processes and activities to land system change is a long-standing challenge, particularly in the context of telecoupled land systems linking distant production and consumption regions (Friis and Nielson 2017, Munroe et al. 2019). Regime shift analysis is often made tractable by treating land systems as ontological objects with static boundaries (Hertz et al. 2020). This is due in part to methodological constraints, such as the reliance on remote sensing tools to characterize quantities and trends in land-use and land-cover change over a given spatial and temporal extent (Hurni et al. 2017). However, mechanisms or pathways for land system change are often context dependent (Meyfroidt et al. 2018, Schlüter et al. 2019), and defining what constitutes a land system largely depends on the analyst and context being investigated (Ingalls and Stedman 2016, Haider et al. 2018). Thus, what might be deemed a regime shift for a given system definition (e.g., national-level shift from deforestation to reforestation) may appear otherwise under broader-scale (e.g., displacement of deforestation to neighboring countries; Meyfroidt and Lambin 2009, Ingalls et al. 2018) or more localized (e.g., displacement of poor households’ shifting cultivation by commodity production; Meyfroidt et al. 2013) system definitions.
The complexity of contemporary, and often telecoupled, land systems challenges more closed land system definitions. For example, national-level forest conservation policies in Vietnam produced a dramatic increase in forest cover likely representing a new, stable land-use regime, but displacement (or leakage) effects were observed as forest loss increased in neighboring countries to meet global demand for forest products (Lambin and Meyfroidt 2011, Müller et al. 2014) and land for investment (Ingalls et al. 2018). Cascade effects can also produce both desirable and undesirable land system regime shifts through indirect land-use changes. For example, the stated goal of many land deals in Lao PDR is to increase agricultural productivity and investment in rural communities. Realization of this goal depends on whether land deals offer benefits to both the investor and surrounding communities. Such co-benefits were realized when latex processing facilities were established to support large-scale rubber plantations, which also lowered the costs of production and increased profitability of smallholder rubber producers in northern Laos (Vongvisouk et al. 2016, Dwyer and Vongvisouk 2019, Junquera et al. 2020). In contrast, land dispossession and land-use change related to large-scale rubber investments also triggered indirect land-use changes through smallholder displacement to distant locations (Dwyer and Vongvisouk 2019, Nanhthavong et al. 2020). In this case, pre-existing land-use and livelihood logics are displaced to adjacent locations where they persist (often times on lower quality land and/or more remote locations, increasing the precariousness of those land users; Magliocca et al. 2019, 2020), creating “stickiness” or rigidity in the current land system state (i.e., poverty traps; Barrett and Swallow 2006, Radosavljevic et al. 2021). More fully accounting for the fluidity and multi-dimensional nature of land system responses to economic globalization requires an open definition of land systems, and deeper understanding and theorizing of land system regime shifts as social-ecological systems (Turner et al. 2020).
Precisely defining a land system, and thus bounding what constitutes a land system regime shift for this analysis, is a complex yet critical task. Land-use patterns are manifestations of the dynamic interplay between external factors and underlying decision-making processes and local context (Verburg et al. 2015, Turner et al. 2020). Although changes in the distribution or spatial extent of a given land use may be an indication of a land system regime shift, there are always trade-offs associated with any land-use change (Leach et al. 2018, Ehrensperger et al. 2019). In this way, large-scale landscape transformation may conceal uneven social-ecological changes (e.g., actor well-being, soil degradation) that generate complex mosaics of heterogeneous land ownership patterns, land-use motivations, and land system impacts. Thus, increased land-use change does not necessarily indicate a land system regime shift, particularly if existing land-use and livelihood decision-making logics and/or land ownership patterns are simply reconfigured in situ or reproduced elsewhere (rather than replaced). To accommodate this possibility conceptually, we adopt a broader definition of land systems and a more precise definition of land system regime shifts. We conceptualize land systems in more relational terms, rather than spatially or as shifts between land-use categories, to include locations that share mutual influence. For example, displacement of smallholder production from one location to another by large-scale agriculture causally links land-use change in those two locations, which can be considered part of the same land system. We then define a “full” land system regime shift as a transition to new land uses “and” the decision-making logics that produce them with social-ecological consequences distinct from those of previous land uses. These working definitions make it possible to observe a broad-scale shift in land-use patterns while acknowledging that processes of land-use change may not have fundamentally changed for many land users within the land system.
Land deals in Laos
Commodity crop booms have been a focal point of land system regime shift research because of their potential to induce rapid and extensive changes in land cover, ownership, and use (Müller et al. 2014). Commodity crops have long been a consistent feature of Southeast Asian landscapes (Ziegler et al. 2009, Li and Fox 2012, Mahanty and Milne 2016, Hurni et al. 2017), and recent commodity crop booms have been driven by both smallholders (particularly for rubber) and large agribusiness farms and plantations (Byerlee 2014, Cramb et al. 2017, Ornetsmüller et al. 2019). Land deals have been a primary vehicle for commodity crop booms, and when involving well-capitalized investors in large-scale land acquisitions (LSLAs), land deals are used by national governments to attract investment in rural economies and increase agricultural productivity and modernization (Anseeuw et al. 2012, Messerli et al. 2014). Large-scale land acquisitions have also been promoted as a means for alleviating poverty by introducing employment opportunities and/or improving access to markets for the communities in which the land deals are embedded (Mahanty and Milne 2016, Nanhthavong et al. 2020). However, there are many examples of land deals failing to fulfill such promises and even exacerbating rural under-development through land dispossession, displacement of local populations, and instigation of social and political conflict (Baird and Fox 2015, Oberlack et al. 2016, Liao and Brown 2018, Magliocca et al. 2019, Liao et al. 2020). Land deals have also accelerated deforestation directly through extensive forest clearing associated with large-scale production activities (Davis et al. 2015) and indirectly through displacement and cascade effects associated with land speculation, displaced smallholder agriculture, and land tenure conflicts (Zaehringer et al. 2018, Magliocca et al. 2019, 2020). As an exogenous and catalytic force for land system change, land deals represent a targeted means to investigate multi-scalar influences on land system regime stability and change (Magliocca et al., this issue). In contexts in which more complete investor and location information exists, such as Lao PDR, land deals enable rich spatio-temporal analyses of land system change and provide a more precise lens through which to investigate interactions between endogenous and exogenous factors that contribute to land system stability or regime shifts.
Over the last two decades, the government of the Lao PDR (GoL) pursued a model of agricultural modernization in which large-scale land deals (i.e., land concessions, leases, or contract farming arrangements to/with investors) are put forward as a way to increase investment and productivity in the agricultural sector, support rural development, and alleviate poverty (Deininger and Xia 2016, Ali et al. 2019, Nanhthavong et al. 2020). Between 2005 and 2015, the national poverty rate declined 16% (Nanhthavong et al. 2020), whereas land-based investments accelerated in the early 2000s and peaked between 2006-2008. Between the early 1990s and 2017, approximately 1.02 million hectares were granted for 1797 land deals for agricultural, mining, and hydropower development (Hett et al. 2020). During that time, many of these land deals coincided with crop booms, particularly for rubber, which led to widespread foreign investment and further land conversion in many Southeast Asian countries (Byerlee 2014, Hurni et al. 2017, Junquera and Grêt-Regamey 2019, Junquera et al. 2020). Land deals associated with commodity crop booms in Southeast Asia have generally followed the model of commodity frontier expansion (Meyfroidt et al. 2014, le Polain de Waroux et al. 2018, Junquera et al. 2020) by locating in relatively remote areas with abundant natural resources, low land prices, thin labor markets, poorer local communities, and weak land governance. However, land deals associated with booms for commodity crops in Laos, such as maize (Ornetsmüller et al. 2019) and rubber (Baird 2010, 2020, Hurni et al. 2017, Fox et al. 2018), diverged from this general model. Earlier investments favored areas with greater accessibility where poverty tends to be comparatively lower and only investments established later in crop boom cycles, particularly by foreign investors, occurred in increasingly remote locations (Nanhthavong et al. 2020). The rubber boom in Laos also differed from neighboring regions in that a substantial portion of rubber plantations in northern Laos were established by smallholders rather than outside investors (Baird 2010, Fox and Castella 2013, Lu 2017, Junquera et al. 2020). Smallholder rubber establishment was also a strategy to claim and secure private land holdings, which is a strategy that has been repeatedly observed among smallholders participating in crop booms in shifting cultivation landscapes (Mahanty and Milne 2016, Junquera et al. 2020). The diversity of pathways for land deal establishment, implementation, and subsequent land-use and livelihood impacts throughout Laos have produced a mosaic of distinct land systems (Ornetsmüeller et al. 2016) that may or may not constitute localized regime shifts within national-level development trends.
Conceptual framework
We propose a conceptual framework to analyze the degree to which full, partial, or no land system regime shifts have occurred in villages impacted by land deals throughout Lao PDR (Fig. 1). The degree to which a land system regime shift occurred depends on presence/absence of poverty reduction and iLUC. Hypothetically, a full land system regime shift would occur if poverty was reduced and no iLUC occurred following land deal implementation, because pre-deal production logics were replaced by market-oriented production and were accompanied by income increases. By contrast, no land system regime shift would be observed if poverty remained largely unchanged and extensive iLUC resulted from displaced communities perpetuating pre-deal livelihood strategies. Partial regime shift scenarios would also be possible in which poverty reduction occurs with extensive smallholder land-use change, such as in smallholder expansion of rubber production in northern Laos (Manivong and Cramb 2008), or land deals can go relatively unimplemented having little effect on either poverty or iLUC. Using the terminology proposed by Ramankutty and Coomes (2016), we investigate triggering events, pre-conditions, and self-reinforcing processes present or operating at the village-level in the context of land deal implementation in Lao PDR.
The implementation of a land deal represents a potential triggering event at the local (i.e., village) scale. Variations among land deal implementation processes, often related to crop type, may influence consequent self-reinforcing processes (or lack thereof) that accelerate or perpetuate a regime shift. For example, the rate and extent of land deal implementation, particularly for rubber production, influenced the occurrence of social conflict and/or iLUC in Cambodia, with shorter lag times between establishment and implementation leading to more conflictual or undesirable outcomes (Magliocca et al. 2019, 2020). The origin of investor has also been cited as a source of variation in the outcomes of land deals. In the broader context of LSLAs globally, Oya (2013) suggested that land deals with domestic investors may produce more positive outcomes due to presumed greater commitment to the impacted communities. Similarly, Nanhthavong et al. (2020) found slightly stronger poverty reduction in Laos associated with land deals of domestic investors.
Salient pre-conditions for regime shifts include access to alternative livelihoods and potentially available land. The implementation of a land deal places stress on livelihoods associated with local land systems previously dominated by extensive, subsistence-oriented land uses. If no other livelihood alternatives are available, either through access to non-farm income opportunities or buyers for smallholder market-oriented production, land-use decision making may not significantly change following deal implementation. In which case, a land-use transition may occur with increased commodity production within land deal boundaries, but a full land system regime shift would not because pre-existing land-use decision making persisted outside of land deal boundaries manifested as iLUC. Similarly, a lack of available land may force the local land system into a regime shift after land deal implementation by either permanently displacing previous land users (and their land-use decision-making logics) or inducing the adoption of intensified production of already cultivated land. In cases with available land, a regime shift may be less likely because dispossessed and/or displacement land users may continue previous land uses in nearby available lands, thereby relieving pressures toward regime shift.
Self-reinforcing processes that may sustain or accelerate a potential land system regime shift triggered by land deal implementation include the degree of socioeconomic embedding of the deal and subsequent migration flows to the deal locations. Socioeconomic embedding, a concept drawn from economic anthropology to describe how social actors exist within relational, institutional, and cultural contexts, can be measured by the extent to which land deals support local communities. Examples include compensation, employment of local community members, contract framing, and positive spillover effects (e.g., technology transfer). Importantly, social embeddedness can occur along a spectrum ranging from minimal community interaction, such as community consultation before land deals were established, to full engagement, such as construction of roads, schools, and other beneficial infrastructure (Nanhthavong et al. 2020, 2021). Some forms of social embeddedness, such as financial compensation for land deal impacts, may unintentionally reinforce pre-deal livelihood strategies and poverty levels by, for example, providing short-term capital necessary to bring new land into production (Vongvisouk et al. 2014, Baird and Fox 2015, Junquera et al. 2020). Immigration to deal locations may support LUCs in a number of ways. Deforestation and subsequent cultivation may result from the influx of non-local labor as deals go into production and labor demands grow (Baird and Fox 2015). The influx of people and capital that results from land deal implementation can also fuel iLUC caused by land speculation or homestead establishment by migrants (same cites). Conversely, dispossession and/or displacement that results in substantial emigration by local communities may reinforce a regime shift because the previous land-use decision-making logic is absent.
Based on this conceptual framework, we address the following questions and hypotheses:
- Have land deals in Laos for agriculture and forestry/plantation accelerated a land system regime shift at the village-level to commodity crop production and lower poverty, or reinforced a land system regime of pre-existing smallholder land use and livelihood decision-making logic associated with persistent or potentially increasing poverty levels?
- How do land-use and livelihood changes vary based on the characteristics of the land deal as a triggering event (i.e., foreign vs. domestic, crop type, extent of implementation) and the pre-conditions and self-reinforcing processes present in the impacted villages?
- Hypothesis: Land deals producing rubber will be associated with more negative outcomes in impacted villages than land deals producing other crops.
- Hypothesis: Villages impacted by land deals with domestic investors have greater access to markets and proximity to areas with higher population densities and will thus experience less iLUC and improved incomes. More remote villages impacted by foreign investments will experience more iLUC and no change in income due to more frequent displacement by land deals and new iLUC to continue previous land uses.
METHODS
A combination of methods and data sources were used to identify the village-level conditions under which land deals interacted with socioeconomic development to accelerate land system change toward a regime shift, or reinforced current land-use and livelihood decision-making logics (Fig. 2). Our analysis integrated qualitative and quantitative data and analyses to link indirect forest loss (or not) and changes in village-level poverty resulting from the land deal. Occurrence of iLUC was independently coded from and cross-referenced between forest-loss mapping and survey responses. Annual forest-loss mapping identified areas of new land-use changes in villages impacted by land deals, and a quasi-experimental matching approach was used to investigate differences in iLUC among villages impacted by land deals. Forest-loss outcomes observed in the matching analysis were then interpreted through a smaller set of impacted villages for which survey data were available. Survey responses from villages impacted by land deals were systematically coded for the types and extents of impacts, particularly income changes resulting from land deals and causally linked to land-use trends and land deal characteristics with qualitative comparative analysis (QCA). Qualitative comparative analysis provided insights into the local contingencies influencing localized impacts on poverty and land-use change and allowed for more nuanced interpretation of the broader iLUC trends detected by the matching analysis.
Such a mixed methods approach was essential due to multiple datasets of different data types (e.g., remote sensing of forest loss quantities, coded survey responses), different geographic units (i.e., land deal boundaries, village area boundaries), and incomplete information (e.g., establishment and/or implementation dates missing), and across dataset observations that only partially overlap. The last characteristic was particularly challenging because not all land deals included in the quality of investment survey were also represented with georeferenced boundaries in the 2017 Lao National Land Concession Inventory (LCI). Thus, we used the broader set of agricultural or forestry/plantation land deals from the LCI with georeferenced boundaries to assess direct forest loss within deal boundaries and indirect forest loss within impacted village areas since land deal establishment and/or implementation times. Inevitably, integrating such diverse data sources presented limitations, which we discuss below and in more detail in Appendix 1.
Land deals and impacted villages data sources
Land concession inventory (LCI)
Land deal data were accessed from the 2017 Lao National LCI database, which was compiled by the GoL with technical and conceptual support from the University of Bern’s Center for Development and Environment between 2010 and 2017. Data were compiled from sectoral sources and ministerial databases, as well as through local consultation with government authorities and participatory analysis of remotely sensed imagery in each of Laos’s 148 districts (Hett et al. 2015, 2020, Nanhthavong et al. 2020, 2021). The complete LCI database contains 1797 records covering all land concession sectors, land deals established or implemented between 1993-2017, and one to many records for each deal or segment of deal. The LCI database contains information about land deal characteristics, including deal size, year(s) of establishment and/or implementation, location (provincial level), country of investor origin, intended use (i.e., agricultural product), stage of operation, investment (e.g., foreign, domestic, shareholder), and contract type (e.g., concession, lease, contract farming). Of the total 1797 records, the majority (62%) were categorized as domestic or shareholder (i.e., foreign investor with domestic partner) deals (n = 1177) and 33% (n = 591) as foreign investor deals. Georeferenced boundaries were available for 465 unique land deals from the full LCI database across agricultural and forestry/plantation sectors. We removed any land deal with an establishment (or implementation date if the former was not available) data before 2000, which reduced the number of applicable land deals from 465 to 424. This subset was used for subsequent remote sensing-based land change mapping. For further details about the development of the LCI database, please see Hett et al. (2015, 2020).
Quality of investment survey
Information about the villages impacted by land deals was extracted from the quality of investment (QI) assessment, which was conducted in conjunction with the LCI (Hett et al. 2015, 2020, Nanhthavong et al. 2020, 2021). The QI dataset was created, among others, through the implementation of household and group interviews, with the latter composed of community members from the same village. The QI database provided baseline information about household livelihood strategies in the form of priority and sources of income, and details about the land deal characteristics (from LCI) impacting each village. In addition, a range of responses were solicited to describe changes that occurred from land deals affecting each village. Responses were solicited for positive and negative impacts to income, food security, and livelihoods. Specific questions addressed the immediate impacts of land deals, including: the number of households displaced, losing land, and/or receiving compensation; previous uses of land granted to land deals; channels that households used to access new land; and the origins of workers (i.e., from within or outside of village) for the land deals. The QI dataset covered 176 unique land deals affecting 294 villages. Nearly half of villages were affected by only one deal (n = 149), whereas the remaining were affected by multiple. Land deals affected between 1 and 68 villages (Hett et al. 2020). When a land deal affected multiple villages, approximately 30% of those villages were assessed (Hett et al. 2018). Considering only land deals that were contained in the QI and that also had georeferenced deal boundaries reduced the set of deals used in the QCA to 84. For a detailed description of the methods used to create the QI dataset, including interview protocols, please see Hett et al. (2018,2020).
Analyses
Annual forest loss mapping
Forest loss was used to measure both direct and indirect LUC associated with land deals. We considered five existing LUC data products for Laos. Classified land-use data developed by Hurni et al. (2013, 2017) focused on boom crops for mainland Southeast Asia. The final dataset was created using moderate resolution imaging spectroradiometer (MODIS) with a nominal spatial resolution of 250 m. Although the dataset effectively differentiated various crops, the spatial resolution was too coarse to identify rotational and smallholder agriculture shifts. In addition, the dataset was centered around a single year (2013) except for the boom crops for which the time of change was estimated in three-year blocks. The dataset created by the Japanese space agency (JAXA) for the Lao government was in vector format. The dataset was divided by province (n = 17) and included 22 potential classes. Unfortunately, this dataset was available for only 4 epochs spaced in 5-year increments (2000, 2005, 2010, 2015). We attempted to fill the gap years using the JAXA dataset to train a classifier (e.g., support vector machine) and conduct land cover classification. However, visual inspection against Google Earth’s time lapse, Planet, and Landsat images found numerous cases in which land cover was misclassified or did not change from one epoch to the next even when visual inspection showed clear evidence of change. Another dataset examined was from SERVIR, which provided annual, wall-to-wall coverage of the Mekong basin, 18 landcover classes, and Landsat resolution (30 m). Visual inspection revealed misclassification and, like the JAXA data, cases in which no land-cover change was recorded but was clearly evident in visual interpretation. A land system classification for Laos was produced by Ornetsmüeller et al. (2018a) but was not suitable for identifying iLUC because of coarse spatial resolution (2 km) and limited temporal coverage (2010-2011). Of these four available datasets for Laos, only three had wall-to-wall coverage, and none were able to consistently identify the land use associated with instances of iLUC. Due to these limitations, we chose to use the Hansen et al. (2013) Global Forest Change (GFC) dataset for our study period of 2000-2018. This dataset was chosen because it provided full temporal coverage for the study period, wall-to-wall classification, and 30 m resolution. In addition, many land deals enclosed at least partially forested areas, which provided time series measurements of LUC at relatively high resolution across the country. The GFC product is not without its limitations. Bias as a result of input data (e.g. Landsat ETM+), preprocessing, and interpretation and classification methods can all lead to some level of uncertainty. However, a robust validation process, conducted independent of forest-cover mapping, showed good agreement with accuracy well over 90% (Hansen et al. 2013, Cunningham et al. 2019).
Direct land-use change, as a result of LSLAs, was measured within the georeferenced land deal boundaries, provided by the LCI, using the GFC annual forest-cover change data for 2000-2018. We extended the area analyzed to include a 500-meter buffer around the deal boundaries because production areas often exceeded georeferenced boundaries with contiguous row crop, pasture, or plantation land use (Magliocca et al. 2019, 2020). Land-use change was measured from the establishment year (or implementation year if establishment was not available) forward in time inclusive, again, with the 500-meter buffer. In cases in which two or more land deals overlapped the same location, annual forest-loss areas were attributed to the land deal most recently established in a given location. The iLUC was measured within each affected village’s projected boundary (absent boundary demarcation for most villages in Laos, the effective boundaries were projected from “influence area,” based on equal travel time between any two neighboring village centers; Hett et al. 2015) excluding the area enclosed within the 500-meter buffered land deal boundary.
Quasi-experimental matching to assess land-use change (LUC) trends based on land deal characteristics
A quasi-experimental matching approach was used to estimate the average treatment effect on the treated (ATT). Several land deal characteristics, i.e., investor origins, rubber vs. non-rubber production, and level of implementation, were distinguished between treatment and control villages, and their effects were measured as iLUC after the establishment and implementation of land deals that intersected their areas. The ATT on village iLUC was estimated for four different time spans: (1) within three years of a land deal’s establishment dates and (2) within three years of a land deal’s implementation dates, (3) total iLUC since a land deal’s establishment, and (4) total iLUC since a land deal’s implementation dates. Villages were chosen as the unit of analysis to be consistent with the data reported in the QI dataset. Most villages were impacted by a single land deal, however there were instances of multiple land deals in different locations and/or at different times impacting a single village (Fig. 3). The maximum number of land deals impacting a single village was 11, and 29 relatively small villages were completely contained within the boundaries of 1 or more land deal(s). After identifying impacted villages, we performed a spatial join in ArcGIS between villages and deals, with a one-to-many relationship. Thus, although there were 1484 unique villages impacted by land deals, the matching analysis compared areas of iLUC among an analytical set of 2118 impacted villages in which a given village could be included more than once if impacted by multiple land deals. Joins were aggregated by establishment or implementation year, with preference given to the latter if available. We then sequentially, by year, erased the land deal areas from village areas. This allowed estimation of forest cover change in each village, by year, without including the influence of direct production within the buffered land deal boundaries. Villages were categorized as treated versus control based on land deal characteristics, i.e., foreign vs. domestic investor, greater vs. less than 20% implementation, rubber vs. non-rubber, etc., to estimate the effects of land deals on the likelihood of LUC or iLUC occurring. Additionally, odds ratios were estimated to indicate the relative magnitude and direction of treatment effects. Confidence intervals (95%) for odds ratios were calculated using 1000 bootstrapped samples.
Treatment and control villages were matched based on both propensity scores (using log probit binomial regression and covariate distance) and using Mahalanobis distance. The method producing the most balanced and unbiased matching was chosen. Villages were matched to control for all or a subset of contextual village characteristics that likely influenced agricultural expansion or deforestation: village area, forest cover area in 2000, population density, market access, terrain roughness, headcount of population below the poverty line, cassava yields, land system composition, prevalence of ethnic minorities, and percent of suitable soils (Table 1). Each treatment village was matched one-to-one with the most similar control village clustered geographically at the provincial level.
Quality of matching was evaluated with median of standardized biases (MSB) estimated for each village characteristic. A clear threshold for acceptable MSB does not exist, but we adopted a statistic of less than 10% as an indication of quality matching (Caliendo and Kopeinig 2008, Blackman et al. 2015). Appendix 2 provides detailed results of the MSB assessments for each matching analysis comparing propensity score matching with the common alternative approach of covariate matching based on Mahalanobis distance. Propensity score matching outperformed covariate matching, produced paired treatment and control villages with sufficiently low MSB, and reduced variations in paired treatment and control covariate means. We also calculated Rosenbaum bounds (Keele 2010) to check for sensitivity to unobserved factors that might bias selection into the treatment group (Rosenbaum and Rubin 1983, DiPrete and Gangl 2004, Blackman et al. 2015).
Matching was first conducted with all of the covariates listed in Table 1. Variables were removed in stepwise fashion until a satisfactory MSB was achieved. Population density was excluded as a covariate when matching domestic plus shareholder (control) and foreign (treatment) land deals because the population density was statistically significantly higher for domestic plus shareholder than foreign deals regardless of the matching method used and a sufficiently low MSB could not be achieved. These population density differences were consistent with findings of Nanthavong et al. (2020), which noted the domestic deals were preferentially located in less remote locations and in areas with larger populations. However, population density is a well-established underlying cause of land-use change. For the purposes of matching domestic and foreign deals, we focused on other variables that collectively reflect the influence of population density on land systems. Thus, we retained the land system composition (2010) variables and initial percent forest cover (2000), which were partially or fully excluded in other matching analyses. Combined, these variables proxied the effects of population density and directly affected the amount of possible forest lost. Similarly, population density remained statistically significantly different between rubber (treatment) and non-rubber (control) matched villages, regardless of the matching method. We used a similar approach of including proxy variables, but population density was retained as a matching covariate because it produced a low MSB and the treatment effect was robust to relatively large uncertainties in treatment and control group assignment.
Qualitative comparative analysis (QCA)
We conducted qualitative comparative analysis (QCA) to extract information related to social, economic, livelihood, and land-use impacts from surveyed villages affected by land deals (n = 84). Qualitative comparative analysis is a case-oriented method that uses Boolean logic to establish conditions causally associated with an outcome (Rihoux and Ragin 2009). Qualitative comparative analysis was chosen for two reasons. First, QCA has been used widely to support causal inference about regional and global change, and it has the flexibility to accommodate causal factors at multiple scales (Rudel 2008, Schneider and Wagemann 2010). Second, QCA is a robust and still growing research area (Schneider and Wagemann 2010) supported by many open-source platforms, such as R packages and dedicated QCA software (Rihoux and Ragin 2009, Thiem and Duşa 2013, Thomann and Wittwer 2016). We used the QCApro package for R rather than the more popular QCA package because of recent advances in QCA methodologies for causal interpretations based on minimally necessary conditions. Rather than excluding conditions a priori as unnecessary, the updated algorithm searches all possible conditions (excluding redundancies) for necessity and sufficiency, and only the parsimonious solutions are used to make causal inferences from empirical data (Baumgartner and Thiem 2017, Thiem 2017).
Each case in the QCA analysis constituted a single land deal and any villages impacted by that deal as reported in the QI database. Cases included in the analytical set for QCA were selected based on several criteria. Land deals were included only if survey responses were reported in at least one impacted village for each of the coded variables in Table 2. Land deals with no data responses for any of the variables were excluded. Any land deals with establishment or implementation dates (whichever came first) prior to 2000 were also excluded. Additionally, for land deals with no reported establishment or implementation date in LCI, we estimated implementation year based on first visible land-use change within 500-meter buffered land deal boundaries using visual interpretation of Google Earth Pro Time Lapse imagery. Village survey responses were coded and assigned to specific land deals using the codebook and procedure listed in Appendix 1, Table A1.3.
Generally, as the number of causal variables increases relative to the number of cases, the number of possible solutions per outcome explodes and interpretation becomes difficult (Schneider and Wagemann 2010, Baumgartner and Thiem 2017, Thiem 2017). To avoid this issue, some coded survey responses were aggregated to create composite causal variables for use in the QCA. Instances of dispossession or displacement were combined into a single “Displacement” variable in which the occurrence of either was coded as “present.” Employment and compensation were combined into a “Financial” variable with three levels of none, either, or both constituent variables present. Social embeddedness was calculated as the net count of positive (+1) or negative (-1) instances of community impacts for all villages impacted by a given land deal. Finally, the presence of alternative livelihoods was quantified using the average Shannon’s Evenness Index calculated for all possible livelihoods recorded in the survey across all impacted villages for a given land deal. In other words, the greater the number livelihoods reported, the higher the evenness index value and more diverse the livelihood options available.
Outcomes of poverty reduction and iLUC were also coded from the survey data. Changes in village-level poverty was measured directly with reported changes in household income. Similar to social embeddedness, poverty change was calculated as the net count of positive (+1) or negative (-1) instances reported across all villages impacted by a given land deal. The presence or absence of iLUC was coded based on survey responses related to bringing new or previously fallowed land into production (Appendix 1, Table A1.3 for details). These coded responses were compared with the forest loss mapping results, and only cases in which both the remote sensing analysis and survey responses agreed were coded for the presence of iLUC.
All binary causal variables were coded into simple presence and absence, whereas composite causal variables were calibrated with three levels based on thresholds set to natural distribution break points. Truth tables, a central analytic device in QCA, were then constructed by coding the survey response variables using crisp set membership values. Applying Boolean logic then reduced the data in the truth table to the simplest set of causal and outcome conditions into causal configurations. Only parsimonious solutions were used consistent with the aim of causal inference (Thiem 2017). To ensure robust final solutions, we adjusted crisp set membership scores for causal conditions until the parsimonious solutions reached high consistency (i.e., above 0.9; Schneider and Wagemann 2010, Thomas et al. 2014) and validated the correct membership of individual cases to each final solution.
RESULTS
Deal characteristics
We first considered differences between land deals with domestic plus shareholder and foreign investors that might impact poverty reduction or land-use change. Foreign land deals had earlier median dates of establishment and implementation than domestic plus shareholder deals, and the timing of foreign deals had a narrower interquartile range than domestic plus shareholder deals (Fig. 4). Land deals by foreign investors generally occurred earlier than those by domestic plus shareholder investors.
Differences in mean rates of direct LUC in the form of forest loss were also observed based on the origin of investor and rubber or non-rubber production. The mean percentage of forested deal area cleared was higher for foreign (28.25%) than domestic plus shareholder (21.27%) land deals (Fig. 5a). Similarly, the mean percentage of forested deal area cleared was higher for rubber (32.9%) than non-rubber (21.5%) land deals (Fig. 5b). It was unsurprising that these trends were similar given that the majority (57%) of land deals for rubber were also by foreign investors.
Matching analyses
Matching analysis was first applied to estimate the average difference in iLUC (i.e., ATT) in villages with domestic plus shareholder (control) versus foreign (treatment) investors. Matching was based on the subset of covariates that produced paired control and treatment deals with no statistically significant differences (Table 3). Standard difference reductions and median standard biases are reported for each covariate in Appendix 2, Tables A2.1.1-A2.1.3. Sensitivity analysis with Rosenbaum bounds (Appendix 2) showed that estimated ATT were insensitive to the exclusion of population density and terrain roughness because observed differences in iLUC outcomes would remain significant at the 10% level even with matched pairs differing in their odds of treatment by as much as 20% (Appendix 2, Tables A2.2.1-A2.2.4). Propensity score matching produced a lower overall median standard bias (0.0561) than covariate matching with Mahalonobis distance (0.1694; Appendix 2, Table A2.1.3).
Average treatment effects based on the origin of investors differed depending on the time span over which iLUC was measured. Foreign land deals had a slight, non-statistically significant dampening effect on forest loss within three years since land deal establishment or implementation (Table 4). Overall, relatively little forest loss occurred in impacted villages within three years of either date. Extending the timeframe of the analysis to consider total indirect forest loss since establishment or implementation date indicated a strong, positive effect of foreign compared to domestic plus shareholder land deals on iLUC. On average, 65 and 69 ha more forest was lost in villages impacted by foreign land deals than domestic plus shareholder land deals (Table 4). Odds ratios indicated that indirect forest loss was approximately three times more likely in villages impacted by foreign than by domestic plus shareholder land deals.
Matching was next applied to rubber (treatment) versus non-rubber (control) pairings. A treatment sample that did not have a statistically significant different average population density from the control sample could not be found (Table 5). Pre- and post-matching standard difference reductions for each covariate are reported in Appendix 2, Tables A2.1.4-A2.1.5. However, median standard bias below 8% was produced with propensity score matching for all covariates, and the resulting median standard bias for population densities between treatment and control groups was only 2.3%. Propensity score matching produced a lower overall median standard bias (0.0360) than covariate matching with Mahalonobis distance (0.2084; Appendix 2, Table A2.1.6). The inherent differences in population densities between the treatment and control samples warranted caution when interpreting differences in ATT. Specifically, ATT estimates for iLUC within three years of establishment or implementation date were relatively sensitive because the difference between treatment and control outcomes became non-significant above a 10% difference in odds of treatment (Appendix 2, Tables A2.2.5-A2.2.6). Thus, we did not consider these ATTs significant in light of the differences in population densities among the samples. In contrast, the differences in total indirect forest loss in impacted villages since deal establishment or implementation were robust with ATT estimates remaining significant at the 10% level even with matched pairs differing in their odds of treatment by as much as 50% (Appendix 2, Table A2.2.7-A2.2.8).
The ATT of land deals producing rubber on iLUC in impacted villages was positive for all time horizons investigated (Table 6). Although odds ratio estimates suggested a slight positive and statistically significant effect of rubber production on forest loss after three years of land deal establishment or implementation, the irreconcilable population density differences between treatment and control villages based on rubber production casts doubt on whether these differences were meaningful. In contrast, the estimated ATTs of rubber production on total forest loss in impacted villages were sufficiently strong to confidently draw conclusions. Villages impacted by rubber-producing land deals lost an average of 138 ha more forest cover following land deal establishment or implementation than villages impacted by non-rubber producing land deals. Odds ratios indicated that total indirect forest loss was 7.4 and 5.1 times higher in villages impacted by rubber-producing land deals since establishment and implementation, respectively.
Finally, matching was applied to low (< 25%; control) versus high (> 25%; treatment) percent of the area of the land deal implementation (i.e., direct forest loss). A set of covariates were found for which no statistically significant differences between paired control and treatment deals persisted (Table 7). Pre- and post-matching standard difference reductions for each covariate are reported in Appendix 2, Tables A2.1.7-A2.1.8. Propensity score matching produced a lower overall median standard bias (0.0365) than covariate matching with Mahalonobis distance (0.2269; Appendix 2, Table A2.1.9). Sensitivity analysis with Rosenbaum bounds showed that estimated ATT were insensitive to the exclusion of village area, initial forested area, or percentage of forest, swidden, and plantation area because observed differences in iLUC outcomes would remain significant at the 10% level even with matched pairs differing in their odds of treatment by as much as 20%, 50%, and 60% for forest loss within three years of implementation and total forest loss since establishment and implementation, respectively (Appendix 2, Tables A2.2.9-A2.2.12).
The ATT of land deals with high (> 25%) forested area lost was positive and statistically significant for three years since implementation and total forest loss since either establishment or implementation (Table 8). Within three years of implementation, villages impacted by land deals undergoing active production experienced nearly 15 ha more forest loss, or slightly more than double, than that of villages impacted by land deals with low rates of implementation. Villages impacted by land deals with high levels of implementation also experienced a total of 102 ha (6.15 times more) and 95 ha (2.55 times more) of forest loss since establishment and implementation, respectively, than villages impacted by low implementation land deals.
Qualitative comparative analysis
Distinct causal configurations leading to the presence/absence of income improvements (i.e., poverty reduction) and/or iLUC were identified for villages impacted by land deals with foreign versus domestic plus shareholder investors (Appendix 2, Table A2.4). Integrating these causal configurations with the ATTs of iLUC by land deals with different characteristics enabled the synthesis of recurring causal pathways leading to full, partial, or no regime shifts in villages impacted by land deals with domestic plus shareholder (Fig. 6) and foreign (Fig. 7) investors. Several common causal relationships were observed. A clear and consistent pathway to the no regime shift outcome, regardless of origin of investor, was the presence of dispossession and/or displacement in the absence of mitigating factors (e.g., socioeconomic embedding of land deals, financial compensation, alternative livelihoods). Also, land deals for which there was an absence of social embeddedness, displacement, conflict, and financial compensation created little impact, regardless of the size of the deal, crop produced, or investor of origin; suggesting that these were key self-reinforcing processes that produced positive and/or negative impacts.
Additionally, non-rubber land deals were more often associated with income improvement regardless of the origin of the investor and pathways without iLUC were possible (i.e., full regime shift). Although land deals producing rubber also had instances of income improvement, iLUC was much more common. Outcomes with no income improvement occurred with both types of production, but the type of crop produced in these cases was not found to be causally related to the lack of income improvement. Financial compensation led to more iLUC, even in the context of income improvements, most likely because additional capital was invested in expanding existing or new land-use practices rather than shifting away from land-based livelihoods.
Land deals with domestic plus shareholder investors had several distinct causal pathways. Conflict was a primary driver of undesirable (i.e., partial or no regime shift) outcomes based on its inclusion in four distinct causal pathways (Fig. 6). The presence of alternative livelihoods mitigated the negative effects of conflict leading to partial regime shifts with income improvements and iLUC. There was also evidence that larger domestic plusshareholder investor land deals were associated with income improvements when combined with social embeddedness. Importantly, the inverse, i.e., small deals and no income change, was not observed. Small deals with absence of conflict and socioeconomic embedding had no effect on household incomes in affected villages or iLUC. Overall, there were an equal number of pathways associated with income improving outcomes as there were with no income change outcomes, and iLUC occurred in similar frequencies because it was absent.
Distinct causal pathways were also found for land deals with foreign investors. Social embeddedness and the presence of alternative livelihoods were primary drivers of improving income outcomes (i.e., full or partial regime shift) based on their inclusion in three causal pathways each (Fig. 7). When rubber and non-rubber land deals exhibited social embeddedness, both led to income improvements. The presence of diverse alternative livelihoods prevented iLUC with non-rubber land deals, whereas conflict led to iLUC in the context of rubber producing land deals. Also, financial compensation, in the forms of direct payments and/or employment, had no discernable positive effect on incomes and was associated with dispossession/displacement and conflict leading to a no regime shift outcome. However, there were more distinct pathways that led to desirable full or partial regime shifts with income improvements among land deals with foreign investors but iLUC was also more prevalent.
DISCUSSION
Our mixed methods approach combined quantified, remote-sensing-based observations of direct and indirect LUC within land deal boundaries and impacted villages with qualitative analysis of survey responses linking socioeconomic impacts to LUC outcomes. Results provided mixed support for our hypotheses. Land deals producing rubber were more frequently associated with iLUC than those producing other crops but both types of land deals were associated with outcomes of improved income. The origin of investor produced similarly mixed regime shifts with the same number of causal pathways leading to income improving regimes, but fewer causal pathways leading to partial or no regime shifts for land deals with foreign investors. Overall, land deals triggered a wide variety of full, partial, and no regime shift outcomes, which were dependent on both the characteristics of the land deal and heterogeneous pre-conditions and self-reinforcing processes across impacted villages. The observed pathways to full and desirable land system regime shifts via land deal investment are complex and highly contingent. The identification of such pathways provides a counterpoint to land- and water-grabbing narratives (e.g., Rulli et al. 2013, Dell’Angelo et al. 2018) and reinforces emerging perspectives of the potential benefits of land deals that include smallholders (Cramb et al. 2017, Liao and Brown 2018).
Several other key findings emerged. The average treatment effects of land deals with foreign investors producing rubber and with a high degree of implementation were found to have overall higher rates of iLUC than their counterparts. Displacement and/or land dispossession more often than not led to negative impacts on the communities surrounding land deals. This was particularly true when actions that investors could take to socially embed within communities, such as employment, improved infrastructure as well as access to social services, were lacking. Such actions represent self-reinforcing processes that can mitigate negative impacts on communities’ land resources, and when present always improved outcomes. Consistent with the Land Matrix Initiative’s most recent analytical report (Lay et al. 2021), these actions have >been identified as prerequisites for avoiding adverse social and environmental impacts of LSLAs. Importantly, financial compensation alone was insufficient to improve community well-being because it could be reinvested to perpetuate previous land uses (imüller et al. 2018b) without the desired transformation in livelihoods and rural development. Finally, the prevalence of iLUC in impacted villages, regardless of investor origin, suggests that avoiding environmental impacts while achieving rural development through land deal investment is difficult. This represents a major downside and trade-off to land-based external investment in light of environmental conservation and climate change mitigation goals (Liao et al. 2021). This may prove particularly problematic for local farmers. Insofar as iLUC signifies livelihood displacement (from investment areas to nearby forests), this places local farmers in a difficult position in a context in which political approaches to conserve forest carbon rely heavily on shoring up forest protection measures and more robust law enforcement. Absent true alternative livelihood options, there is a substantial risk that local livelihoods will experience increasing pressure squeezed between investment areas and protected forests.
These alternative pathways for land system change have important implications when policy interventions are designed to trigger a regime shift to more desirable land system states. Our findings show, for example, that although rubber investment produced some positive social effects, they were also, and more commonly, associated with the social costs and the expansion of iLUC; two outcomes that conflict with the stated policy goals of promoting rubber investments, i.e., rural transformation (poverty reduction and the provision of labor) and forest conservation. Despite this, there appear to be better and worse ways of developing these investments, which indicates that measures can be taken to improve outcomes at the local level within existing policy directions. Land deals that paid greater attention to social embeddedness, which provided adequate compensation complemented by job-creation and improved access to infrastructure and/or services, were more likely to lead to positive regime shifts. The complex and contingent outcomes produced by land deals in Laos also point to a high degree of non-linearity between policy decisions (e.g., the promotion of foreign direct investment in rubber) and real-world impacts. This suggests that policymakers need to be more responsive to local causal factors and preconditions, and policy processes need to be more adaptive in response to outcomes in practice. For example, financial compensation, although often inadequate, can be an important aspect of a negotiated outcome, however, (better) socioeconomic embeddedness is at least equally important, which can be promoted through adequate policies and regulations (and enforcement thereof). Blanket policy prescriptions appear to be insufficient to enable positive regime shifts. In addition to paying more adequate attention to social embeddedness, local contextual factors and alternative livelihood options, i.e., the observation that rubber was more closely associated with the negative environmental and social outcomes, suggests that the identification of strategic crops needs more careful consideration. It is difficult to assess the degree to which the collapse of rubber latex pricing after 2011 affected outcomes. Had pricing remained high such as seen during its boom, it is possible that income and wage-labor alternatives for local communities could have been more positive. Regardless, the bust seen in the rubber sector is characteristic of such booms (Hall 2011). What distinguishes rubber from many other boom crops, which are typically annual crops, is the high degree of investment frontloading and the lag time between plantation and latex harvesting. This suggests that rubber is a particularly high-risk boom crop for poorer, local communities and less agile in adapting to changing market conditions.
Finally, it is important to critically assess what is meant by a positive or desirable regime shift because the desirability of one regime over another is socially contingent (Cote and Nightingale 2012, Ingalls and Stedman 2016). Any regime change will produce sets of winners and losers and thus it becomes necessary to critically analyze not only the social distribution of costs and benefits in regime shifts but also the equitability of decision-making processes.
CONCLUSIONS
Land system regime shifts have been studied at various scales and with varying degrees of unequal integration of human and natural system components (Müller et al. 2014, Filatova et al. 2016, Ramankutty and Coomes 2016). Very few studies have integrated insights into the local land-use and livelihood dynamics driving regime shifts at the regional or national scales at which land-use change analyses are conducted. Thus, the operating assumption for studying land system regime shifts is that they are scale independent (Ramankutty and Coomes 2016), while also acknowledging that process rather than pattern is the key to understanding land-use change (Dearing et al. 2010, Ramankutty and Coomes 2016, Turner et al. 2020). Integration of qualitative and quantitative findings enabled a more in-depth analysis of interactions between land deals, pre-existing conditions, and self-reinforcing processes that did or did not lead to land system regime shifts. Analyses exclusive to land-cover and -use changes fail to capture potential livelihood transformations, which are often the target of national rural development policies, producing observed landscape changes that may or may not occur when land deals are introduced in a landscape. Similarly, purely qualitative studies are unable to systematically link the impacts of land deals on livelihood decision making to land-use changes across the diversity of local conditions present in the national context. Combining the strengths of each of these approaches identified the diversity of potential rural development pathways triggered by land deals and, more importantly, provided insights into the local conditions and self-reinforcing processes that can produce full, partial, or no regime shift outcome with varying desirability. As expected, the fusion of qualitative insights at the village level with regional, remote-sensing analyses illustrated a much more complex, scale-dependent, and contingent reality. The land system conceptualization applied here provided a holistic assessment of regime shifts that was sensitive to land-use and livelihood decision-making processes operating at the local scale that cumulatively produced landscape-scale transformations. Future research that leverages trend analysis of increasingly available very high-resolution satellite imagery combined with the contextualized perspectives of impacted households can make direct inferences about the effects of land deals. Embedding such methods in a holistic land systems framework can produce the locally nuanced information needed for actionable insights to inform national development strategies.
RESPONSES TO THIS ARTICLE
Responses to this article are invited. If accepted for publication, your response will be hyperlinked to the article. To submit a response, follow this link. To read responses already accepted, follow this link.ACKNOWLEDGMENTS
The authors acknowledge support from NASA ROSES Land Cover Land Use Change (LCLUC) project award #NNX17AI15G. This analysis was conducted within the Knowledge for Development (K4D) Project funded by the Swiss Agency for Development and Cooperation, of the GoL with the University of Bern's Center for Development & Environment. This study also contributes to the Global Land Programme (https://glp.earth).
DATA AVAILABILITY
The two main datasets used in this work, the Land Concession Inventory and the Quality of Investment survey, are not publicly available but may be obtained through a formal request to the Centre for Development and Environment, University of Bern and the Government of Laos PDR. All other data that are sharable are available publicly through referenced sources and/or the authors' personal websites.
LITERATURE CITED
Aide, T. M., M. L. Clark, H. R. Grau, D. López-Carr, M. A. Levy, D. Redo, M. Bonilla-Moheno, G. Riner, M. J. Andrade-Núñez, and M. Muñiz. 2013. Deforestation and reforestation of Latin America and the Caribbean (2001-2010). Biotropica 45(2):262-271. https://doi.org/10.1111/j.1744-7429.2012.00908.x
Ali, D., K. Deininger, and A. Harris. 2019. Does large farm establishment create benefits for neighboring smallholders? Evidence from Ethiopia. Land Economics 95(1):71-90. https://doi.org/10.3368/le.95.1.71
Anseeuw, W., M. Boche, T. Breu, M. Giger, J. Lay, P. Messerli, and K. Nolte. 2012. Transnational land deals for agriculture in the Global South: analytical report based on the Land Matrix database. Land Matrix Partnership.
Arbault, D., M. Rivière, B. Rugani, E. Benetto, and L. Tiruta-Barna. 2014. Integrated earth system dynamic modeling for life cycle impact assessment of ecosystem services. Science of the Total Environment 472:262-272. https://doi.org/10.1016/j.scitotenv.2013.10.099
Baird, I. G. 2010. Land, rubber and people: rapid agrarian changes and responses in Southern Laos. Journal of Lao Studies 1(1):1-47. https://landmatrix.org/media/uploads/baird2010bachiengrubber.pdf
Baird, I. G. 2020. Problems for the plantations: challenges for large-scale land concessions in Laos and Cambodia. Journal of Agrarian Change 20(3):387-407. https://doi.org/10.1111/joac.12355
Baird, I. G., and J. Fox. 2015. How land concessions affect places elsewhere: telecoupling, political ecology, and large-scale plantations in Southern Laos and Northeastern Cambodia. Land 4(2):436-453. https://doi.org/10.3390/land4020436
Barrett, C. B., and B. M. Swallow. 2006. Fractal poverty traps. World Development 34(1):1-15. https://doi.org/10.1016/j.worlddev.2005.06.008
Baumgartner, M., and A. Thiem. 2017. Model ambiguities in configurational comparative research 46(4):954-987.https://doi.org/10.1177/0049124115610351
Blackman, A., A. Pfaff, and J. Robalino. 2015. Paper park performance: Mexico’s natural protected areas in the 1990s. Global Environmental Change 31:50-61. https://doi.org/10.1016/j.gloenvcha.2014.12.004
Brook, B. W., E. C. Ellis, M. P. Perring, A. W. Mackay, and L. Blomqvist. 2013. Does the terrestrial biosphere have planetary tipping points? Trends in Ecology and Evolution 28(7):396-401. https://doi.org/10.1016/j.tree.2013.01.016
Byerlee, D. 2014. The fall and rise again of plantations in tropical Asia: history repeated? Land 3(3):574-597. https://doi.org/10.3390/land3030574
Caliendo, M., and S. Kopeinig. 2008. Some practical guidance for the implementation of propensity score matching. Journal of Economic Surveys 22(1):31-72. https://doi.org/10.1111/j.1467-6419.2007.00527.x
Cote, M., and A. J. Nightingale. 2012. Resilience thinking meets social theory: situating social change in socio-ecological systems (SES) research. Progress in Human Geography 36(4):475-489. https://doi.org/10.1177/0309132511425708
Coulombe, H., M. Epprecht, O. Pimhidʓai, and V. Sisoulath. 2016. Where are the poor? Lao PDR 2015 census-based poverty map: province and district level results. Ministry of Planning and Investment, Lao Statistics Bureau, Vientiane Capital, Lao PDR. https://documents1.worldbank.org/curated/en/477381468415961977/pdf/106899-WP-P146141-PUBLIC.pdf
Cramb, R., V. Manivong, J. C. Newby, K. Sothorn, and P. S. Sibat. 2017. Alternatives to land grabbing: exploring conditions for smallholder inclusion in agricultural commodity chains in Southeast Asia. Journal of Peasant Studies 44(4):939-967. https://doi.org/10.4324/9781351134873-11
Cunningham, D., P. Cunningham, and M. E. Fagan. 2019. Identifying biases in global tree cover products: a case study in Costa Rica. Forests 10(10), 853. https://doi.org/10.3390/f10100853
Davis, K. F., K. Yu, M. C. Rulli, L. Pichdara, and P. D’Odorico. 2015. Accelerated deforestation driven by large-scale land acquisitions in Cambodia. Nature Geoscience 8(10):772-775. https://doi.org/10.1038/ngeo2540
Dearing, J. A., A. K. Braimoh, A. Reenberg, B. L. Turner, and S. van der Leeuw. 2010. Complex land systems: the need for long time perspectives to assess their future. Ecology and Society 15(4):21. https://doi.org/10.5751/ES-03645-150421
Dell’Angelo, J., M. C. Rulli, and P. D’Odorico. 2018. The global water grabbing syndrome. Ecological Economics 143:276-285. https://doi.org/10.1016/j.ecolecon.2017.06.033
Deininger, K., and F. Xia. 2016. Quantifying spillover effects from large land-based investment: the case of Mozambique. World Development 87:227-241. https://doi.org/10.1016/j.worlddev.2016.06.016
DiPrete, T. A., and M. Gangl. 2004. Assessing bias in the estimation of causal effects: Rosenbaum bounds on matching estimators and instrumental variables estimation with imperfect instruments. Sociological Methodology 34(1):271-310. https://doi.org/10.1111/j.0081-1750.2004.00154.x
Dwyer, M., and T. Vongvisouk. 2019. The long land grab: market-assisted enclosure on the China-Lao rubber frontier. Territory, Politics, Governance 7(1):96-114. https://doi.org/10.1080/21622671.2017.1371635
Ehrensperger, A., A. de Bremond, I. Providoli, and P. Messerli. 2019. Land system science and the 2030 agenda: exploring knowledge that supports sustainability transformation. Current Opinion in Environmental Sustainability 38:68-76. https://doi.org/10.1016/j.cosust.2019.04.006
Ellis, E. C., A. H. W. Beusen, and K. K. Goldewijk. 2020. Anthropogenic biomes: 10,000 BCE to 2015 CE. Land 9(5):129. https://doi.org/10.3390/land9050129
Filatova, T., J. G. Polhill, and S. van Ewijk. 2016. Regime shifts in coupled socio-environmental systems: review of modelling challenges and approaches. Environmental Modelling and Software 75:333-347. https://doi.org/10.1016/j.envsoft.2015.04.003
Folke, C., S. Carpenter, B. Walker, M. Scheffer, T. Elmqvist, L. Gunderson, and C. S. Holling. 2004. Regime shifts, resilience, and biodiversity in ecosystem management. Annual Review of Ecology, Evolution, and Systematics 35:557-581. https://doi.org/10.1146/annurev.ecolsys.35.021103.105711
Fox, J., and J.-C. Castella. 2013. Expansion of rubber (Hevea brasiliensis) in Mainland Southeast Asia: what are the prospects for smallholders? Journal of Peasant Studies 40(1):155-170. https://doi.org//10.1080/03066150.2012.750605
Fox, J., T. Nghiem, H. Kimkong, K. Hurni, and I. G. Baird. 2018. Large-scale land concessions, migration, and land use: the paradox of industrial estates in the Red River Delta of Vietnam and rubber plantations of Northeast Cambodia. Land 7(2):77. https://doi.org/10.3390/land7020077
Friis, C., and J. Ø. Nielsen. 2017. On the system. Boundary choices, implications, and solutions in telecoupling land use change research. Sustainability 9(6):974. https://doi.org/10.3390/su9060974
Godar, J., C. Suavet, T. A. Gardner, E. Dawkins, and P. Meyfroidt. 2016. Balancing detail and scale in assessing transparency to improve the governance of agricultural commodity supply chains. Environmental Research Letters 11(3):035015. https://doi.org/10.1088/1748-9326/11/3/035015
Haider, L. J., W. J. Boonstra, G. D. Peterson, and M. Schlüter. 2018. Traps and sustainable development in rural areas: a review. World Development 101:311-321. https://doi.org/10.1016/j.worlddev.2017.05.038
Hall, D. 2011. Land grabs, land control, and Southeast Asian crop booms. Journal of Peasant Studies 38(4):837-857. https://doi.org/10.1080/03066150.2011.607706
Hansen, M. C., P. V. Potapov, R. Moore, M. Hancher, S. A. Turubanova, A. Tyukavina, D. Thau, S. V. Stehman, S. J. Goetz, T. R. Loveland, A. Kommareddy, A. Egorov, L. Chini, C. O. Justice, and J. R. G. Townshend. 2013. High-resolution global maps of 21st-century forest cover change. Science 342(6160):850-853. https://doi.org/10.1126/science.1244693
Hertz, T., M. Mancilla Garcia, and M. Schlüter. 2020. From nouns to verbs: how process ontologies enhance our understanding of social-ecological systems understood as complex adaptive systems. People and Nature 2(2):328-338. https://doi.org/10.1002/pan3.10079
Hett, C. V. Nanhthavong, S. Hanephom, A. Phommachanh, B. Sidavong, K. Phouangphet, J. Lu, A. Shattuck, M. Ingalls, R. Bernhard, S. Phathitmixay, C. Phomphakdy, A. Heinimann, and M. Epprecht. 2020. Land Leases and concessions in the Lao PDR: a characterization of investments in land and their impacts. Centre for Development and Environment (CDE), University of Bern, Switzerland, with Bern Open Publishing.
Hett, C., V. Nanhthavong, M. Kenney-Lazar, K. Phouangphet, and S. Hanephom. 2018. Assessing land investment quality: a methodology to assess the quality of land concessions and leases in the Lao PDR. Centre for Development and Environment, University of Bern, with Bern Open Publishing, Bern, Switzerland. https://doi.org/10.7892/boris.111041
Hett, C., V. Nanhthavong, T. Saphangthong, G. R. Robles, K. Phouangphet, W. Speller, P. Messerli, M. Epprecht, and A. Heinimann. 2015. Land deals in Laos: first insights from a new nationwide initiative to assess the quality of investments in land. In Land grabbing, conflict and agrarian-environmental transformations: perspectives from East and Southeast Asia, an international academic conference, 5-6 June 2015, Chiang Mai University. BRICS Initiatives for Critical Agrarian Studies, MOSAIC Research Project, Land Deal Politics Initiative (LDPI), RCSD Chiang Mai University, Faculty of Social Sciences, Chiang Mai University, Chiang Mai, Thailand, and Transnational Institute, Amsterdam, The Netherlands. https://www.iss.nl/sites/corporate/files/CMCP_18-_Hett_et_al.pdf
Hurni, K., C. Hett, A. Heinimann, P. Messerli, and U. Wiesmann. 2013. Dynamics of shifting cultivation landscapes in Northern Lao PDR between 2000 and 2009 based on an analysis of MODIS time series and Landsat images. Human Ecology 41(1):21-36. https://doi.org/10.1007/s10745-012-9551-y
Hurni, K., A. Schneider, A. Heinimann, D. H. Nong, and J. Fox. 2017. Mapping the expansion of boom crops in mainland Southeast Asia using dense time stacks of Landsat data. Remote Sensing 9(4):320. https://doi.org/10.3390/rs9040320
Ingalls, M. L., P. Meyfroidt, P. X. To, M. Kenney-Lazar, and M. Epprecht. 2018. The transboundary displacement of deforestation under REDD+: problematic intersections between the trade of forest-risk commodities and land grabbing in the Mekong region. Global Environmental Change 50:255-267. https://doi.org/10.1016/j.gloenvcha.2018.04.003
Ingalls, M. L., and R. C. Stedman. 2016. The power problematic: exploring the uncertain terrains of political ecology and the resilience framework. Ecology and Society 21(1):6. https://doi.org/10.5751/ES-08124-210106
Junquera, V., and A. Grêt-Regamey. 2019. Crop booms at the forest frontier: triggers, reinforcing dynamics, and the diffusion of knowledge and norms. Global Environmental Change 57:101929. https://doi.org/10.1016/j.gloenvcha.2019.101929
Junquera, V., P. Meyfroidt, Z. Sun, P. Latthachack, and A. Grêt-Regamey. 2020. From global drivers to local land-use change: understanding the northern Laos rubber boom. Environmental Science and Policy 109:103-115. https://doi.org/10.1016/j.envsci.2020.04.013
Keele, L. 2010. An overview of rbounds: an R package for Rosenbaum bounds sensitivity analysis with matched data. White Paper. Columbus, Ohio, USA. https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.408.7360&rep=rep1&type=pdf
Lambin, E. F., and P. Meyfroidt. 2011. Global land use change, economic globalization, and the looming land scarcity. Proceedings of the National Academy of Sciences 108(9):3465-3472. https://doi.org/10.1073/pnas.1100480108
Lay, J., W. Anseeuw, S. Eckert, I. Flachsbarth, C. Kubitza, K. Nolte, and M. Giger. 2021. Taking stock of the global land rush: few development benefits, many human and environmental risks. Centre for Development and Environment, University of Bern; Centre de coopération internationale en recherche agronomique pour le développement; German Institute for Global and Area Studies; University of Pretoria; Bern Open Publishing. Bern, Montpellier, Hamburg, Pretoria. https://doi.org/10.48350/156861
Leach, M., B. Reyers, X. Bai, E. S. Brondizio, C. Cook, S. Díaz, G. Espindola, M. Scobie, M. Stafford-Smith, and S. M. Subramanian. 2018. Equity and sustainability in the anthropocene: a social-ecological systems perspective on their intertwined futures. Global Sustainability 1:E13. https://doi.org/10.1017/sus.2018.12
le Polain de Waroux, Y., M. Baumann, N. I. Gasparri, G. Gavier-Pizarro, J. Godar, T. Kuemmerle, R. Müller, F. Vázquez, J. N. Volante, and P. Meyfroidt. 2018. Rents, Actors, and the Expansion of Commodity Frontiers in the Gran Chaco. Annals of the American Association of Geographers 108(1):204-225. https://doi.org/10.1080/24694452.2017.1360761
Levin, S., T. Xepapadeas, A.-S. Crépin, J. Norberg, A. de Zeeuw, C. Folke, T. Hughes, K. Arrow, S. Barrett, G. Daily, P. Ehrlich, N. Kautsky, K.-G. Mäler, S. Polasky, M. Troell, J. R. Vincent, and B. Walker. 2013. Social-ecological systems as complex adaptive systems: modeling and policy implications. Environment and Development Economics 18(2):111-132. https://doi.org/10.1017/S1355770X12000460
Li, Z., and J. M. Fox. 2012. Mapping rubber tree growth in mainland Southeast Asia using time-series MODIS 250 m NDVI and statistical data. Applied Geography 32(2):420-432. https://doi.org/10.1016/j.apgeog.2011.06.018
Liao, C., and D. G. Brown. 2018. Assessments of synergistic outcomes from sustainable intensification of agriculture need to include smallholder livelihoods with food production and ecosystem services. Current Opinion in Environmental Sustainability 32:53-59. https://doi.org/10.1016/j.cosust.2018.04.013
Liao, C., S. Jung, D. G. Brown, and A. Agrawal. 2020. Spatial patterns of large-scale land transactions and their potential socio-environmental outcomes in Cambodia, Ethiopia, Liberia, and Peru. Land Degradation and Development 31:1241-1251. https://doi.org/10.1002/ldr.3544
Liao, C., K. Nolte, J. A. Sullivan, D. G. Brown, J. Lay, C. Althoff, and A. Agrawal. 2021. Carbon emissions from the global land rush and potential mitigation. Nature Food 2(1):15-18. https://doi.org/10.1038/s43016-020-00215-3
Lu, J. N. 2017. Tapping into rubber: China’s opium replacement program and rubber production in Laos. Journal of Peasant Studies 44(4):726-747. https://doi.org/10.1080/03066150.2017.1314268
Magliocca, N. R., A. C. de Bremond, E. A. Ellicott, L. Seghezzo, C. D. Venencia, M. Moscario, and C. Nolte. 2022. Two of a kind? Large-scale land acquisitions and commodity frontier expansion in Argentina’s Dry Chaco. Ecology and Society 27(2):25. https://doi.org/10.5751/ES-13103-270225
Magliocca, N. R., Q. Van Khuc, A. de Bremond, and E. A. Ellicott. 2020. Direct and indirect land-use change caused by large-scale land acquisitions in Cambodia. Environmental Research Letters 15(2):024010. https://doi.org/10.1088/1748-9326/ab6397
Magliocca, N. R., Q. Van Khuc, E. A. Ellicott, and A. de Bremond. 2019. Archetypical pathways of direct and indirect land-use change caused by Cambodia’s economic land concessions. Ecology and Society 24(2):25. https://doi.org/10.5751/es-10954-240225
Mahanty, S., and S. Milne. 2016. Anatomy of a boom: cassava as a ‘gateway’ crop in Cambodia’s north eastern borderland. Asia Pacific Viewpoint 57(2):180-193. https://doi.org/10.1111/apv.12122
Manivong, V., and R. A. Cramb. 2008. Economics of smallholder rubber expansion in Northern Laos. Agroforestry Systems 74(2):113. https://doi.org/10.1007/s10457-008-9136-3
Messerli, P., M. Giger, M. B. Dwyer, T. Breu, and S. Eckert. 2014. The geography of large-scale land acquisitions: analysing socio-ecological patterns of target contexts in the global South. Applied Geography 53:449-459. https://doi.org/10.1016/j.apgeog.2014.07.005
Meyfroidt, P., K. M. Carlson, M. E. Fagan, V. H. Gutiérrez-Vélez, M. N. Macedo, L. M. Curran, R. S. Defries, G. A. Dyer, H. K. Gibbs, E. F. Lambin, D. C. Morton, and V. Robiglio. 2014. Multiple pathways of commodity crop expansion in tropical forest landscapes. Environmental Research Letters 9(7):074012. https://doi.org/10.1088/1748-9326/9/7/074012
Meyfroidt, P., R. Chowdhury, A. De Bremond, E. C. Ellis, K.-H. Erb, T. Filatova, R. D. Garrett, J. M. Grove, A. Heinimann, T. Kuemmerle, C. A. Kull, E. F. Lambin, Y. Landon, Y. le Polain de Waroux, P. Messerli, D. Müller, J. Ø. Nielsen, G. D. Peterson, V. Rodriguez García, M. Schlüter, B. L. Turner, II, and P. H. Verburg. 2018. Middle-range theories of land system change. Globale Environmental Change 53:52-67. https://doi.org/10.1016/j.gloenvcha.2018.08.006
Meyfroidt, P., and E. F. Lambin. 2009. Forest transition in Vietnam and displacement of deforestation abroad. Proceedings of the National Academy of Sciences 106(38):16139-16144. https://doi.org/10.1073/pnas.0904942106
Meyfroidt, P., T. P. Vu, and V. A. Hoang. 2013. Trajectories of deforestation, coffee expansion and displacement of shifting cultivation in the Central Highlands of Vietnam. Global Environmental Change 23(5):1187-1198. https://doi.org/10.1016/j.gloenvcha.2013.04.005
Monfreda, C., N. Ramankutty, and J. A. Foley. 2008. Farming the planet: 2. Geographic distribution of crop areas, yields, physiological types, and net primary production in the year 2000. Global Biogeochemical Cycles 22(1). https://doi.org/10.1029/2007gb002947
Müller, D., Z. Sun, T. Vongvisouk, D. Pflugmacher, J. Xu, and O. Mertz. 2014. Regime shifts limit the predictability of land-system change. Global Environmental Change 28(1):75-83. https://doi.org/10.1016/j.gloenvcha.2014.06.003
Munroe, D. K., M. Batistella, C. Friis, N. I. Gasparri, E. F. Lambin, J. Liu, P. Meyfroidt, E. Moran, and J. Ø. Nielsen. 2019. Governing flows in telecoupled land systems. Current Opinion in Environmental Sustainability 38:53-59. https://doi.org/10.1016/j.cosust.2019.05.004
Nanhthavong, V., M. Epprecht, C. Hett, J. G. Zaehringer, and P. Messerli. 2020. Poverty trends in villages affected by land-based investments in rural Laos. Applied Geography 124:102298. https://doi.org/10.1016/j.apgeog.2020.102298
Nanhthavong, V., C. Oberlack, C. Hett, P. Messerli, and M. Epprecht. 2021. Pathways to human well-being in the context of land acquisitions in Lao PDR. Global Environmental Change 68:102252. https://doi.org/10.1016/j.gloenvcha.2021.102252
Oberlack, C., L. Tejada, P. Messerli, S. Rist, and M. Giger. 2016. Sustainable livelihoods in the global land rush? Archetypes of livelihood vulnerability and sustainability potentials. Global Environmental Change 41:153-171. https://doi.org/10.1016/j.gloenvcha.2016.10.001
Ornetsmüller, C., J. C. Castella, P. Thanichanon, G. Lestrelin, and P. H. Verburg. 2019. Modelling the location and spatial pattern of a crop boom. A case study from Laos. Environmental Science and Policy 99:58-71. https://doi.org/10.1016/j.envsci.2019.04.001
Ornetsmüeller, C., J.-C. Castella, and P. H. Verburg. 2018b. A multiscale gaming approach to understand farmer’s decision making in the boom of maize cultivation in Laos. Ecology and Society 23(2):35. https://doi.org/10.5751/es-10104-230235
Ornetsmüeller, C., A. Heinimann, and P. H. Verburg. 2018a. Operationalizing a land systems classification for Laos. Landscape and Urban Planning 169:229-240. https://doi.org/10.1016/j.landurbplan.2017.09.018
Ornetsmüeller, C., P. H. Verburg, and A. Heinimann. 2016. Scenarios of land system change in the Lao PDR: transitions in response to alternative demands on goods and services provided by the land. Applied Geography 75:1-11.https://doi.org/10.1016/j.apgeog.2016.07.010
Oya, C. 2013. Methodological reflections on ‘land grab’ databases and the ‘land grab’ literature ‘rush.’ 40(3):503-520.https://doi.org/10.1080/03066150.2013.799465
Radosavljevic, S., L. J. Haider, S. J. Lade, and M. Schlüter. 2021. Implications of poverty traps across levels. World Development 144:105437. https://doi.org/10.1016/j.worlddev.2021.105437
Ramankutty, N., and O. T. Coomes. 2016. Land-use regime shifts: an analytical framework and agenda for future land-use research. Ecology and Society 21(2):01. https://doi.org/10.5751/ES-08370-210201
Rihoux, B., and C. C. Ragin. 2009. Configurational comparative methods: qualitative comparative analysis (QCA) and related techniques. Applied Social Research Methods Series, vol. 51. Sage, Thousand Oaks, California, USA. https://doi.org/10.4135/9781452226569
Riley, S. J., S. D. DeGloria, and R. Elliot. 1999. Index that quantifies topographic heterogeneity. Intermountain Journal of Sciences 5(1-4):23-27.
Rosenbaum, P. R., and D. B. Rubin. 1983. The central role of the propensity score in observational studies for causal effects. Biometrika 70(1):41-55. https://doi.org/10.1093/biomet/70.1.41
Rudel, T. K. 2007. Changing agents of deforestation: From state-initiated to enterprise driven processes, 1970-2000. Land Use Policy 24(1):35-41. https://doi.org/10.1016/j.landusepol.2005.11.004
Rudel, T. K. 2008. Meta-analyses of case studies: a method for studying regional and global environmental change. Global Environmental Change 18(1):18-25. https://doi.org/10.1016/j.gloenvcha.2007.06.001
Rudel, T. K. 2013. Defensive environmentalists and the dynamics of global reform. Cambridge University Press, Cambridge, UK. https://doi.org/10.1017/CBO9781139343800
Rulli, M. C., A. Saviori, and P. D’Odorico. 2013. Global land and water grabbing. Proceedings of the National Academy of Sciences 110(3):892-897. https://doi.org/10.1073/pnas.1213163110
Scheffer, M., S. Carpenter, J. A. Foley, C. Folke, and B. Walker. 2001. Catastrophic shifts in ecosystems. Nature 413(6856):591-596. https://doi.org/10.1038/35098000
Schlüter, M., L. J. Haider, S. J. Lade, E. Lindkvist, R. Martin, K. Orach, N. Wijermans, and C. Folke. 2019. Capturing emergent phenomena in social-ecological systems: an analytical framework. Ecology and Society 24(3):11. https://doi.org/10.5751/ES-11012-240311
Schneider, C. Q., and C. Wagemann. 2010. Standards of good practice in qualitative comparative analysis (QCA) and fuzzy-sets. Comparative Sociology 9(3):397-418. https://doi.org/10.1163/156913210X12493538729793
Steffen, W., Å. Persson, L. Deutsch, J. Zalasiewicz, M. Williams, K. Richardson, C. Crumley, P. Crutzen, C. Folke, L. Gordon, M. Molina, V. Ramanathan, J. Rockstróm, M. Scheffer, H. J. Schellnhuber, and U. Svedin. 2011. The Anthropocene: from global change to planetary stewardship. Ambio 40(7):739-761. https://doi.org/10.1007/s13280-011-0185-x
Thiem, A. 2017. Conducting configurational comparative research with qualitative comparative analysis. American Journal of Evaluation 38(3):420-433. https://doi.org/10.1177/1098214016673902
Thiem, A., and A. Duşa. 2013. QCA : a package for qualitative comparative analysis. The R Journal 5(1):87-97. https://doi.org/10.32614/RJ-2013-009
Thomann, E., and S. Wittwer. 2016. Performing fuzzy- and crisp set QCA with R: a user-oriented beginner’s guide. Version 2016.
Thomas, J., A. O’Mara-Eves, and G. Brunton. 2014. Using qualitative comparative analysis (QCA) in systematic reviews of complex interventions: a worked example. Systematic Reviews 3(1):67. https://doi.org/10.1186/2046-4053-3-67
Turner, II, B. L., P. Meyfroidt, T. Kuemmerle, D. Müller, and R. R. Chowdhury. 2020. Framing the search for a theory of land use. Journal of Land Use Science 15(4):489-508. https://doi.org/10.1080/1747423X.2020.1811792
Verburg, P. H., N. Crossman, E. C. Ellis, A. Heinimann, P. Hostert, O. Mertz, H. Nagendra, T. Sikor, K.-H. Erb, N. Golubiewski, R. Grau, M. Grove, S. Konaté, P. Meyfroidt, D. C. Parker, R. R. Chowdhury, H. Shibata, A. Thomson, and L. Zhen. 2015. Land system science and sustainable development of the earth system: a global land project perspective. Anthropocene 12:29-41. https://doi.org/10.1016/j.ancene.2015.09.004
Verburg, P. H., E. C. Ellis, and A. Letourneau. 2011. A global assessment of market accessibility and market influence for global environmental change studies. Environmental Research Letters 6(3):034019. https://doi.org/10.1088/1748-9326/6/3/034019
Vongvisouk, T., R. B. Broegaard, O. Mertz, and S. Thongmanivong. 2016. Rush for cash crops and forest protection: neither land sparing nor land sharing. Land Use Policy 55:182-192. https://doi.org/10.1016/j.landusepol.2016.04.001
Vongvisouk, T., O. Mertz, S. Thongmanivong, A. Heinimann, and K. Phanvilay. 2014. Shifting cultivation stability and change: contrasting pathways of land use and livelihood change in Laos. Applied Geography 46:1-10. https://doi.org/10.1016/j.apgeog.2013.10.006
Zaehringer, J. G., A. Atumane, S. Berger, and S. Eckert. 2018. Large-scale agricultural investments trigger direct and indirect land use change: new evidence from the Nacala corridor, Mozambique. Journal of Land Use Science 13(3):325-343. https://doi.org/10.1080/1747423X.2018.1519605
Ziegler, A. D., J. M. Fox, and J. Xu. 2009. The rubber juggernaut. Science 324(5930):1024-1025. https://doi.org/10.1126/science.1173833
Table 1
Table 1. Co-variates used in quasi-experimental matching analysis. Unless otherwise noted, all variables were aggregated to conform with the village areas impacted by land deals.
Variable | Description | Units | Source |
Land Deal Boundaries | Polygons of land deal boundaries buffered at 500 m and 2 km. The 500 m dataset will be used for all analyses of direct land-use changes (LUC) because the buffer accounts for production activities of the deals that may exceed the boundaries. | 500 m and 2 km buffered polygons | Land concession inventory (LCI), Centre for Development and Environment (CDE), and Gov. of Laos (Hett et al. 2015, 2020, Nanhthavong et al. 2020, 2021) |
Administrative Boundaries and Village Areas | Georeferenced administrative boundaries for Laos (levels 0 = country; 1 = province; 2 = district; 3 = village). | Polygons | Global Administrative Boundaries Database (GADM), Version 3.6 |
Market Access | Normalized index of travel time to city of at least 50,000 for the year 2000. | Index | (Verburg et al. 2011) |
Land System Classification | Classification of 17 land systems for the period 2010/2011 and calculated as percent of village area. | 2 km | (Ornetsmüeller et al. 2016) |
Crop Yields | Estimated yield for rice and cassava for the year 2000. | 5 arcmin | (Monfreda et al. 2008) |
Poverty | Headcount of population below the poverty line for the year 2015. | District | (Coulombe et al. 2016); Open Development Laos† |
Ethnicity | Geo-referenced ethnic groups of Laos in 2010. | % | Open Development Laos |
Soil Type | A geospatial dataset containing polygons of soil types Laos, according to FAO classifications. | % | Open Development Laos |
Terrain Roughness Index (TRI) | The amount of elevation difference between adjacent cells of a digital elevation model. | 30m | (Riley et al. 1999) |
Population Density | Gridded population density, Revision 11‡ adjusted for density and country totals. | 30 arcsec | Gridded Population of the World (GPW) Version 4.10§ |
Precipitation | Average annual precipitation (aggregated from monthly). | 0.25 degree | TRMM 3B43, NASA EOSDIS |
Forest Cover | Initial forest cover and annual forest loss. | 30 m | (Hansen et al. 2013) |
†Open Development Laos (https://data.laos.opendevelopmentmekong.net/en/dataset?odm_spatial_range_list=la) ‡Revision 1 is the version of the dataset that was used. §Dataset available here: https://sedac.ciesin.columbia.edu/data/collection/gpw-v4 |
Table 2
Table 2. Survey response variables coded for the qualitative comparative analysis (QCA). Note: LCI = Lao National Land Concession Inventory; LSLA = large-scale land acquisitions.
Variable | Definition | |
Deal Size | Reported area granted for concession land (ha) from the LCI database. | |
Crop Type | Primary intended use reported for the land deal in the LCI database. | |
Conflict | direct | Evidence of direct confrontation between land deals and community. Examples include reported land disputes, re-taking or stopping use of concessioned land through force or threat of force. |
indirect | Evidence of political, legal, or otherwise non-physical contestation of land deals by community members. For example, a more conflictual livelihood context (sensu Oberlack et al. 2016), contested compensation, political advocacy. | |
Displacement | Description of community displacement and/or out-migration resulting from land deal. | |
Dispossession | Dispossession of community and/or household land as a result of land deal. | |
Compensate | Some form of individual compensation described, for example, financial or land exchange. | |
Employment | full | Local community members employed in activities related to land deal. |
partial | Only some local community members employed because of insufficient employment opportunities, competition from immigrants, or by choice as form of resistance. | |
none | No employment opportunities offered to local community. | |
Immigration | Land deal resulted in in-migration, usually from migrants seeking employment. | |
Community Impacts | A composite of survey responses reflecting positive and/or negative impacts on the community, such as road construction or improved access to services. | |
Livelihood Evenness | Normalized Shannon’s evenness index measuring the diversity of alternative livelihood engaged in by households in impacted villages. | |
Direct Land-Use Change (dLUC) | Reported direct land change associated with production within land deal boundaries. For example: change from less to more intense land use, most likely a different land use as well, includes both cultivation (e.g., shifting to row crop) or capital (subsistence to plantation) intensity; increase area of existing land uses at the same intensity; or land conversion/agriculture initiated, then abandoned or stalled. | |
Indirect Land-Use Change (iLUC) | Same definitions as dLUC, but reported land change occurred outside of LSLA boundaries by actors other than the investment company (e.g., local farmers). | |
Poverty Change | Measured as changes in household income resulting from land deal. Reported as direct increases in household income and/or improvement in working conditions including better wage or salary. | |
Table 3
Table 3. Reduction in bias between domestic + shareholder (control) and foreign (treatment) land deals before and after matching based on propensity scores.
Pre-matching | Post-matching | ||||
Variable | Domestic + Shareholder Mean N=545 |
Foreign Mean N=1573 |
Two-sample t-statistic |
Foreign Mean N=545 |
Two-sample t-statistic |
Village Area (ha) | 3388 | 3043 | 1.2331 | 3922 | -1.6324 |
Market Access Index (2000) | 0.1067 | 0.0332 | 10.98** | 0.0960 | 0.9086 |
Poverty Rate (2005) | 21.16 | 27.81 | -11.91** | 20.7888 | 0.7202 |
Cassava Yield (tonne/ha) | 14.93 | 13.86 | 4.295** | 14.8038 | 0.3762 |
Forest Area (ha 2000) | 2477 | 2059 | 1.640 | 2911 | -1.5094 |
% Forest (2010) | 11.73 | 14.10 | -2.021* | 10.49 | 0.9598 |
% Swidden (2010) | 14.53 | 16.17 | -1.144 | 13.47 | 0.6642 |
% Permanent Cultivation (2010) | 39.07 | 37.55 | 0.8862 | 39.92 | -0.4048 |
% Plantation (2010) | 11.33 | 19.61 | -6.339** | 13.43 | -1.7069 |
% Ethnic Minorities | 51.20 | 61.47 | -4.4613** | 50.57 | 0.2223 |
% Suitable soils | 15.44 | 19.11 | -2.1230* | 17.75 | -1.0739 |
** p < 0.01; * p < 0.05 |
Table 4
Table 4. Average treatment effect on the treated (ATT) and odds ratios for 3 years’ and total forest loss in villages impacted by domestic + shareholder (control) and foreign (treatment) land deals since establishment and implementation years.
Establishment to Three Years | Implementation to Three Years | Total Since Establishment Year | Total Since Implementation Year | |
ATT (ha) | -16.74 | -13.75 | 65.25* | 69.46* |
ATT (%) | -24.64 | -18.84 | 25.41 | 29.46 |
Odds Ratio (95% CI) |
1.43 (0.85 - 2.35) |
1.43 (0.88 - 2.33) |
3.07 (1.37 - 8.21) |
3.24 (1.20 - 9.81) |
* p < 0.01 |
Table 5
Table 5. Reduction in bias between non-rubber (control) and rubber (treatment) producing land deals before and after matching based on propensity scores.
Pre-Matching | Post-Matching | ||||
Variable | Non-Rubber Crops Mean N = 1199 |
Rubber Crops Mean N = 919 |
Two-sample t-statistic |
Rubber Crops Mean N = 1199 |
Two-sample t-statistic |
Population Density (2005) | 34.55 | 23.62 | 6.9975* | 27.94 | 4.7741* |
Market Access Index (2000) | 0.0513 | 0.0532 | -0.3150 | 0.0503 | 0.1867 |
Terrain Roughness | 33.07 | 37.84 | -3.1588* | 30.65 | 1.7793 |
Poverty Rate (2005) | 26.21 | 25.95 | 0.5017 | 26.40 | -0.4118 |
Cassava Yield (tonne/ha) | 14.79 | 13.27 | 6.9996* | 14.70 | 0.4944 |
Forest Area (ha, 2000) | 2292 | 2002 | 1.2865 | 2595 | -1.3246 |
Swidden (2010) | 12.08 | 20.53 | -6.7568* | 12.62 | -0.5195 |
% Perm. Cult. (2010) # | 42.26 | 32.31 | 6.6286* | 39.62 | 1.8740 |
% Suitable soils | 20.81 | 14.72 | 3.9933* | 21.19 | -0.2641 |
* p < 0.01 |
Table 6
Table 6. Average treatment effect on the treated (ATT) for impacted by land deals producing rubber (treatment) and non-rubber (control) crops for 3 years’ and total forest loss since establishment and implementation years.
Establishment to Three Years | Implementation to Three Years | Total Since Establishment Year | Total Since Implementation Year | |
ATT (ha) | 4.037 | 2.499 | 138.1* | 137.5* |
ATT (%) | 8.32 | 4.51 | 64.02 | 69.25 |
Odds Ratio (95% CI) |
1.47 (1.04 - 2.15) |
1.65 (1.04 - 2.42) |
7.40 (2.52 - 30.7) |
5.10 (2.02 - 13.4) |
* p < 0.01 |
Table 7
Table 7. Comparisons between low (i.e., < 25%; control) and high (> 25%; treatment) implementation land deals, measured of percent of forested area lost within deal boundaries, before and after matching based on propensity scores.
Pre-matching | Post-matching | ||||
Variable | Low Implement Mean N = 1026 |
High Implement Mean N = 1092 |
Two-sample t-statistic |
Foreign Mean N = 1026 |
Two-sample t-statistic |
Population Density (2005) | 37.75 | 22.34 | 10.08* | 35.82 | 1.0194 |
Market Access Index (2000) | 0.0544 | 0.0500 | 0.7303 | 0.0669 | -1.8279 |
Terrain Roughness | 31.93 | 38.16 | -4.158* | 32.21 | -0.1847 |
Poverty Rate (2005) | 25.41 | 26.74 | -2.651* | 25.20 | 0.4364 |
Cassava Yield (tonne/ha) | 14.88 | 13.44 | 6.701* | 14.88 | 0.0059 |
% Permanent Cultivation (2010) | 44.41 | 31.87 | 8.475* | 47.47 | -1.9736 |
% Ethnic Minorities | 52.90 | 64.39 | -5.721* | 52.29 | 0.2938 |
% Suitable soils | 25.95 | 10.85 | 10.20* | 24.44 | 0.8769 |
* p < 0.01 |
Table 8
Table 8. Average treatment effect on the treated (ATT) for villages impacted low (control) and high (treatment) percent implementation land deals for 3 years’ and total forest loss since establishment and implementation years.
Establishment to Three Years | Implementation to Three Years | Total Since Establishment Year | Total Since Implementation Year | |
ATT (ha) | 10.49 | 14.99 | 102.7* | 95.89* |
ATT (%) | 30.46 | 40.39 | 63.66 | 62.70 |
Odds Ratio (95% CI) |
1.40 (1.04 - 1.93) |
2.14 (1.46 - 3.15) |
6.15 (2.43 - 19.4) |
2.55 (1.34 - 5.47) |
* p < 0.01 |