Causes and Consequences of Displacement Decision-making in Banhine National Park, Mozambique

Many of the world’s most biologically diverse places are also home to some of the world’s poorest, most politically marginalised, and most directly natural capital-dependent people. This type of overlapping ecological and social signifi cance often occurs in places now designated as national parks (and other similar protected areas). These areas—usually referred to as IUCN Categories I and II—are generally places where human occupancy and use is legally prohibited. The presence of people living within the boundaries of such strict protected areas and their use of the natural capital located within them has spurred contentious debates about past, present and future relationships between human resource-use and biodiversity and between poverty reduction and conservation efforts (e.g., Conservation and Society Volume 7 Issue 1, 2009). Central to these debates are decisions about whether protected area residents should be physically removed from these areas

experience in the social dimensions of natural resources, and, often with few resources to properly compensate or adequately mitigate the impoverishment risks of those displaced. International development projects, including designation, planning and management of protected areas are funded by a wide variety of organisations, including the World Bank, that have established policies ''safeguarding'' the interests of those potentially affected by decisions to displace and resettle people. While such policies are integral components of loans and grants agreed to by recipient governments, implementation is often challenged by the specifi c capacities of the bureaucracies and administrators involved, the dominant policies and paradigms framing a country's natural resource management, and past experience with displacement and resettlement. In this article, we examine the factors infl uencing a decision about the future of a group of individuals who currently live within a national park but are subject to a decision to be displaced from the area. The research on which this paper reports investigated two interconnected questions: what are the political factors infl uencing displacement decision-making and how do these factors infl uence resettlement outcomes (the consequences of displacement and resettlement). The specifi c case on which we focus is displacement decision-making in Banhine National Park, Mozambique.

PROTECTED AREA DISPLACEMENT IN A BROADER CONTEXT
Defi nitions of displacement and the use of different termsincluding involuntary resettlement, forced migration, forced removal, dispossession, dislocation, relocation, exclusion, eviction, and others-vary from author to author. Some authors limit defi nitions of displacement to the physical removal of people from their homes (e.g., Agrawal and Redford 2007) while other authors defi ne displacement as physical removal and/or restriction of access to livelihood resources (Cernea 2005). The implication in the latter defi nition is that livelihood displacement can occur without physical displacement. While debates in academic literature linger, defi nitions of displacement in policy are, as Krueger (2007: 99) explains, moving 'towards…consensus that restricted access is a form of displacement.' Recent policy changes within the World Bank, Asian Development Bank, African Development Bank, Inter-American Development Bank and the Organization for Economic Cooperation and Development are indicative of this trend (Cernea 2005, Krueger 2007). Consistent with this trend, in this article we use the term displacement to refer to either or both physical removal and restrictions on access to resources needed to maintain current livelihoods. War, confl ict, violence, natural disasters, and environmental change displace millions of people every year (UNHCR 2001). While displacement resulting from, for example, the war in Iraq, Hurricane Katrina, and global climate change garner much needed attention, the cause of the largest annual number of displaced people is development (Cernea 2000, Koenig 2006. By development, we mean projects or programmes that are State or private-sector led, which are legally justifi ed by eminent domain and which are politically justifi ed as being for the greater good of society. This category of displacement is referred to in literature as Development-Induced Displacement and Resettlement (DIDR) and includes dam, irrigation, transportation, utility, urban development, as well as protected area management projects. While it is common practice in DIDR literature to situate protected area displacement within it, this is not common practice in protected area-specifi c literature (Agrawal and Redford 2007). We situate our investigation of protected arearelated displacement within the larger DIDR literature for fi ve primary reasons: 1. Justifications for and tensions regarding DIDR and protected area projects relate to the imposition of costs on a few for the greater good of a larger society; displacement is justifi ed as being in the 'public interest.' 2. The people directly affected by DIDR and protected area displacement are often rural, poor, and politically marginalised. 3. DIDR literature has theoretical and descriptive depth that can aid protected area debates. 4. The impoverishing consequences of DIDR, protected area displacement, as well as other types of displacement share many similarities (Cernea 2000, Ohta 2005). 5. The World Bank and other international organisations prominent in the literature and practice of development and displacement play a large and increasingly important role in protected area projects around the world.

The World Bank and development-induced displacement
As a leading player in post-WWII development and modernisation, the World Bank has fi nanced thousands of large infrastructure projects that, according to McMichael (2004), displaced and further impoverished millions of rural poor around the world. Consequently, the World Bank has also been the target of decades of local, regional, and international protest. Subsequent pressure for reform exerted on the World Bank from external and internal groups resulted in substantive organisational changes with regard to the organisation's involvement in displacement (Fox and Brown 2000). Most prominently, the World Bank adopted an organisational safeguard policy to address displacement. Also important is that Michael Cernea, the World Bank's Senior Resettlement Specialist at the time, developed a complementary analytical framework to inform the implementation of the policy. The World Bank adopted its fi rst version of an involuntary resettlement safeguard policy in 1980. The new policy, kept confi dential until 1988, triggered contentious debates within the World Bank, and between the Bank and its borrowers. Additional debates involved the Bank, its client governments, and an increasingly organised movement resisting displacement and challenging development policies and projects. The policy was revised in 1986, 1988, 1990, 1994, and 2001. Key objectives and principles of the policy, to which both the World Bank and borrower governments are legally bound, include: • Involuntary resettlement should be avoided where feasible, or minimised, exploring all viable alternative designs. • Where it is not feasible to avoid resettlement, resettlement activities should be conceived and executed as sustainable development programmes, providing suffi cient investment resources to enable the persons displaced by the project to share in project benefi ts. • Displaced persons should be meaningfully consulted and should have opportunities to participate in planning and implementing resettlement programmes. • Displaced persons should be assisted in their efforts to improve the livelihoods and standards of living or at least to restore them, in real terms, to pre-displacement levels or to levels prevailing prior to the beginning of project implementation, whichever is higher. Cernea, as a key actor involved in the development of the policy and as an active agent of change within the World Bank, developed an analytical framework-the Impoverishment Risk and Reconstruction (IRR) framework-to aid in the implementation of the safeguard policy. Based on analysis of nearly two hundred World Bank-fi nanced development projects involving displacement, Cernea identifi ed eight means by which displaced people become (further) impoverished. Cernea labelled these components 'impoverishment risks' and includes in them landlessness, joblessness, homelessness, marginalisation, food insecurity, increased morbidity and mortality, loss of access to common property, and social disarticulation (Cernea 1997). In cases of displacement, Cernea (2006: 10) explains, these risks 'materialise into actual, real processes of impoverishment because they are not preempted or reduced through up-front counter-risk strategies and reconstruction plans, before displacement even begins.' A primary purpose of the IRR framework, therefore, is to aid in targeting and countering the risks of impoverishment. Cernea's primary point is that the IRR framework is most useful 'not when it is confi rmed by adverse events, but, rather, when, as a result of its warnings being taken seriously and acted upon, the risks are prevented from becoming a reality, or are minimised, and the consequences predicted by the framework do not occur' (Cernea 2000: 33). The focus of research through the IRR framework, therefore, is to assess how these 'risks are arrested and preempted, or of how they sharpen and materialise into real negative impacts' (Cernea 2006: 10).

THE POLITICS OF DISPLACEMENT DECISION-MAKING AND THE INADEQUACIES OF RESETTLEMENT
Despite advanced understanding of impoverishment risks and how to mitigate them, and despite safeguards and other progressive policies and practices, people continue to be impoverished through displacement. Resettlement and reconstruction efforts continue to be inadequate and many academics, development practitioners, social justice advocates, and most importantly, the people impoverished by displacement, are left asking why (World Commission on Dams 2000). As many contributors to DIDR literature have pointed out (deWet 2006, Koenig 2006, displacement and resettlement are imbued with power and politics and to better understand why policy on such issues continues to be inadequate we need to understand the political factors influencing both displacement decision-making and resettlement efforts as well as the consequences of these decision-making processes. Understanding the political and other causes as well as consequences of displacement decision-making was the purpose of the research that contributed to this article. To address the consequences of displacement and resettlement we used IRR to understand impoverishment risks and the ability of resettlement efforts to mitigate these risks. To understand why these consequences occurred, we embedded IRR within an investigation of the political factors infl uencing displacement decision-making. We framed this layer of research using political-economic, actor-centred, and poststructural theoretical perspectives on power. Finally, we embedded both consequences and causes within a particular socio-ecological and historical context ( Figure 1). Our goal in embedding investigations of consequences, causes, and context was less to understand each layer of this framework and more to understand the connections among the layers. In the following sections, we briefl y discuss the study methods and the socio-ecological context regarding the area known as Banhine National Park. We will then briefl y note the major fi ndings from the IRR analysis. The bulk of the fi ndings presented here focus on the political factors influencing displacement decision-making and how these factors relate to the subsequent consequences of displacement and resettlement.

STUDY METHODS
The research employed qualitative, inquiry-guided methods  (Mishler 1990) aimed at understanding BNP-area resettlement issues from the perspectives of various actors and actor groups as situated within the broader historical and political context. By focussing on research participants representing various political scales, the methods employed in this study are able to capture a range of perspectives useful to better understand the phenomenon under study (Belsky, 2004). Field data were gathered in ten BNP-area communities. In-depth investigation focussed on two of these ten communities. Data were also gathered in district, provincial, and national capitals and other areas in the region, and in Washington DC. Research also involved analysis of planning, policy, legal contracts, and other documents. Field data collection methods in and around BNP-area communities included individual and household interviews (n=44), focus groups (n=4), large community meetings (n=6), dozens of observational walks and drives led by BNP-area residents, and dozens of informal conversations with and observations of BNP-area residents and their local governmentappointed and traditional leaders. These techniques were aimed primarily at understanding, 1) local histories of and current conditions regarding inhabitation, displacement, and resettlement, 2) the relationship between local livelihood strategies and park resources, and 3) the potential effects of displacement and resettlement on BNP-area residents. Sampling was purposeful and aimed at achieving diversity within BNParea residents with regard to gender, class (specifi cally relating to livestock ownership), spatial distribution (including residents in the park, buffer zone, and beyond), and degree of resistance to displacement (including those who openly supported and those who openly opposed displacement and resettlement). Semi-structured interviews were conducted with park staff from the lowest to the highest levels (n=12), provincial and national employees of the Ministries of Tourism, Environmental Coordination, Planning and Development, and the numerous directorates associated with these ministries (n=24). Interviews were also conducted with employees of Chigubo district and the administrative posts with jurisdiction in the BNP area (n=10). Other interviews were conducted with relevant consultants (n=7), NGO staff (n=8), and World Bank employees in both Mozambique and Washington DC (n=5). Engagement with these and other research participants also involved many informal conversations, interactions, and observations. Interviewees were purposefully selected based on their affi liation with government agencies or other organisations most relevant to displacement decision-making. Documents analysed included government and donor project planning and evaluation documents, consultancy reports, historical documents, national policies, strategic plans, donor policies, and legal contracts. Analysis of documents focussed on identifying and conceptually mapping the formal structures, processes, rules, and actor groups involved with displacement decision-making. As part of the methodological approach, detailed fi eld notes, which distinguished between direct observations, researcher inferences, and analysis, were produced. Formal analysis of fi eld notes involved coding themes and conceptually organising relationships between themes both within a single interview, focus group, set of observations, etc. and across them. QSR Nvivo software was used to help organise and analyse fi eld notes.

SOCIO-ECOLOGICAL CONTEXT
Banhine National Park (BNP), like nearly every other national park and protected area in Mozambique, has long been inhabited and its resources relied upon to support local livelihoods. Current residents of the park describe a history of inhabitation, displacement, and resettlement dating back at least to the time of the Gaza Kingdom in the early-to-mid 19 th century. The 7,000 km 2 area now known as BNP was designated as a hunting reserve (Coutada 17) in 1969 and then a national park in 1973. Enforcement of restrictions on resource use was minimal in the brief time before independence in 1975 and did not involve the physical relocation of park residents (Tinley, pers. comm. 10 July 2006). Most BNP-area residents, however, were displaced and many died as a result of the post-independence FRELIMO-RENAMO war. After the war and in accordance with a massive post-war repatriation and resettlement effort in the mid-1990s, many displaced BNP-area residents returned to their 'places of origin' inside and around BNP. There are approximately 2,000 to 3,000 people living within the park and thousands more living outside the park but using resources inside the park. The biophysical feature of BNP that is of primary importance for conservation and tourism efforts in BNP as well as for the livelihoods of BNP-area residents is a dynamic wetland system. The wetland is charged by periodic cyclones that fi ll the basin which then slowly drains over the course of years. This hydrological regime results in high degrees of biological diversity. Prior to the extirpation or near extirpation of many species in the latter part of the twentieth century, the wetland and surrounding areas were home to a wide variety of fauna including what are now rare antelope. At the time of the fi eld research reported here, few large wild mammal species were present in the park. Although there were no formal plans for wildlife reintroduction, numerous consultancy reports mention and government offi cials openly discuss the potential in the park for re-establishment of wildlife populations. BNP-area residents live spatially dispersed in socially operational communities on the edge of the wetland's fl oodplain both inside and outside of the park. By spatially dispersed, we mean that there are no village structures. Rather, residents live near their farms, fallow fi elds, livestock kraals, grazing areas, and water sources which are themselves spatially separated from each other. Residency in these communities is year round except, as described below, during periods of drought. By socially operational, we mean that community boundaries and formal and informal governance arrangements within those boundaries are understood and respected by most community members. BNP-area communities are very remote (~eight-hour drive from the provincial capital) and lack or have limited access to most basic services including potable water, sanitation, health clinics, roads, and schools. People in BNP-area communities are dependent on the fl oodplain and wetland in different ways and to different degrees throughout the wet and dry cycles of the hydrological regime. Generally, however, people are most dependent on fl oodplain resources during drought, crop failure, and famine. In these times, people will temporarily migrate to the fl oodplain to harvest famine foods, water livestock, and collect water from emergency wells for their consumption. Management of BNP is part of a much larger conservation and tourism development project fi nanced by the World Bank and Global Environment Facility (World Bank 2005). Because displacement was a possibility in this project, the World Bank's safeguard policy on involuntary resettlement was triggered. What this means is that the World Bank and the Mozambican government entered into a legal agreement which detailed the project plans and the safeguard policies to which both parties are legally bound and which would infl uence the fate of park residents. A project implementation unit was created within the government of Mozambique to coordinate with various government and non-governmental entities to implement the project plan and safeguard policies. In this report, we highlight the relationship between the project implementation unit and the district governments whose jurisdiction overlaps the project area. In particular, we focus on the district of Chigubo (Figure 2). Chigubo district plays an especially important role in this case because it overlaps a large portion of the park and includes most of the wetland as well as communities near the wetland (Figure 3).

DISTRICT RESETTLEMENT INTENTIONS
Beginning in early 2006, employees of Chigubo district visited BNP-area communities and communicated to leaders and community members that it was the district's desire that residents organise themselves in aggregates or villages in areas that had basic services or where basic services could be provided. Doing so, district employees explained, would give people improved access to available services. It would also allow the government or NGOs to more easily provide services or other assistance than if residents continued to live dispersed. District employees also communicated to BNParea residents that the government would not provide services within the boundaries of the park and that it was the district's desire that those people living inside BNP resettle to areas outside the park. What this meant was that people inside and outside of the park were being told to resettle; however, those inside the park were treated differently, because, in addition to being pressured to resettle into villages, they were being pressured to resettle outside of the park and further away from the wetland. This explanation of events was consistent across nearly all research participants including those employed by the district and those from affected communities.

DISPLACEMENT + INADEQUATE RESETTLEMENT = IMPOVERISHING CONSEQUENCES
The following briefl y summarises the fi ndings of an IRRbased analysis conducted in this case. Details of this analysis are included in Dear (2008). This summary of fi ndings is useful to contextualise the factors infl uencing displacement Displacement decision-making / 107

Figure 3 Overlap of BNP and Chigubo district within Greater Limpopo Transfrontier Conservation Area
decision-making and to lay the groundwork for a discussion of the connections between the causes and consequences of displacement. Efforts by the Chigubo district administration to displace BNParea residents exposed residents to a web of impoverishment risks. But more importantly, resettlement and reconstruction efforts were inadequate to address those risks. While district offi cials did not explicitly tell BNP-area residents where they should move, district offi cials did identify four sites outside the park where residents could resettle themselves in village structures. Two of these village sites did not have a borehole or another facility providing secure access to water. A recently constructed borehole in a third village provided only brackish water unsuited to drink or to irrigate fi elds. There was no assistance provided to displaced people to build new homes or to clear new fi elds. District employees made it clear that provision of services in resettlement areas was to follow rather than go before resettlement and that the district could not promise services anytime in the near future. These resettlement actions were insuffi cient to prevent or redress the various impoverishment risks to which BNP-area residents were being exposed.

FACTORS INFLUENCING DISPLACEMENT DECISION-MAKING
While various policies are designed to ensure that those displaced are not faced with unnecessary risks to their livelihoods, we argue that how policy is supposed to be carried out and how it is actually implemented are signifi cantly different. These differences are particularly important in contexts that are politically 'turbulent', characterised by vacuums in leadership or depicted by confusion and uncertainty about roles and responsibilities. In these situations, policy implementation is infl uenced by the various mental models and paradigms carried by the principal actors as well as situational factors that create noise in the communication channels needed for policy implementation. Our research was directed at understanding what factors and models were important in making decisions that would lead to an involuntary or induced displacement of residents from the Park area. Using the methodological approach described earlier our research identifi ed four interrelated sets of factors that infl uenced displacement decision-making. We also explain why these factors are important to understand the inadequacies of resettlement efforts.

Lack of coordination between conservation and district offi cials
As noted earlier, communication between the capital and district offi ce is diffi cult, and this diffi culty is partly at the foundation for a lack of coordinated and informed decisionmaking between the project implementation unit and the Chigubo district government. In addition, the integrated planning process intended to facilitate this coordination was scaled back to two pilot districts because of capacity constraints. Chigubo district was not one of the pilot districts. Implementation unit employees acknowledged that there had been little communication between the unit and Chigubo district offi cials. As a result, and as evidenced through ten interviews with various officials representing Chigubo district, district offi cials showed only a basic awareness of the conservation and tourism project plan and little to no awareness of the safeguard policy on involuntary resettlement. While project plans and safeguard policies were not infl uencing district officials, there were many other factors that did influence displacement decision-making. While many of these factors can be proximately traced back to higher levels of the government, as Figure 4 suggests, we emphasise that the origins of these factors go well beyond the Government of Mozambique.

Resettlement was conceptualised as necessary for development and poverty reduction
A set of interconnected influencing factors included the dominant ideas (or mental models) held by key actors about people, parks, and poverty and the justifi cation of these ideas within powerful international poverty reduction agendas. As this district employee and many others explained, BNPrelated displacement and resettlement was conceptualised by government offi cials as 'development and poverty reduction.' Resettlement in the district is part of a general government programme of development and poverty reduction.…We assessed life conditions. If people are suffering, they may be resettled so that water, schools, clinics, and roads can be provided. (district offi cial).
As this district offi cial stated and many other government offi cials at district, provincial, and national levels as well as key consultants and NGO employees explained, services could not be provided to residents inside the park. This was primarily because (although there were no specifi c plans to do so) district

Figure 4 Factors infl uencing displacement decision-making
offi cials contended that wildlife would be reintroduced to the park, that human-wildlife confl icts would be inevitable, and that people would therefore not be able to live in the park. "Parks are for animals," district offi cials consistently repeated, "not for people." Park residents, according to this logic, must move out of the park if they are to have access to basic services and become less impoverished. This sentiment was very frankly encapsulated by this government offi cial who also linked the resettlement to an international development agenda.
People in parks [without access to services] are poor, and if they remain inside parks [where services will not be provided] they will always be poor. If they are poor, there is less chance that Mozambique will achieve the Millennium Development Goals. (Project Implementation Unit offi cial).

'Voluntary' villagisation was conceptualised as necessary for service provision
A closely related factor was a dominant idea among those in various sectors and levels of government that concentrating or villagising dispersed rural populations was necessary to provide services and therefore promote development and poverty reduction. This idea is illustrated below in quotations from a national-level offi cial and a district-level offi cial: It is a mistake to think that we can achieve development with people living in such a dispersed manner…There is a need to concentrate people with regard to public investment. (offi cial from National Directorate of Rural Planning and Development).
We are not aggregating people because we like people aggregated. We are trying to fi ght poverty. It is easier to fi ght poverty when people are living together. (district offi cial).
Mozambique has a contested history of forced villagisation efforts. In the socialist era immediately after independence, the new government of Mozambique involuntarily resettled dispersed rural populations into villages for the expressed purposes of promoting collective agriculture, education, and national identity (Isaacman and Isaacman 1983). During the seventeen-year post-independence war, the government involuntarily resettled dispersed rural populations into what were termed 'security villages' (Bowen 2000). Over the past thirty years, the government has also involuntarily resettled rural populations out of drought and fl ood-prone areas (Bowen 2000). When asked to explain why Chigubo district employees were pressuring BNP-area residents to resettle themselves into villages outside the park, a key district-level decision-maker clearly justifi ed BNP-area displacement and resettlement as being consistent with previous government resettlement efforts. "It is a government principle to organise people into villages," explained this offi cial. Government officials at the district and higher levels of government, however, almost unanimously spoke of the operational failures of past villagisation efforts. But offi cials maintained that the idea of villagisation was a sound one. This notion was illustrated by this district employee, who said, "We must take the positive aspects of past [villagisation] policies but implement it in a better way." The difference today, offi cials claimed, was that current efforts are voluntary. "We cannot do [resettlement] like in the past. It must be in a voluntary manner." (offi cial from National Directorate of Territorial Planning). That resettlement was 'voluntary' was another key justifi cation for it. When asked to elaborate on the idea of voluntariness, many government officials not directly involved in the World Bank/GEF project explained resettlement as a process of 'sensibilising and mobilising' rural populations. Some offi cials further defi ned this as 'a process of convincing' or 'a process [of] chang[ing] people's mentality.' In describing the voluntariness of resettlement, these government offi cials did not use concepts resembling 'informed consent' or 'power of choice' which are key criteria of voluntariness the World Bank safeguard policy on involuntary resettlement.

Decentralisation pressured district decision-makers
A fi nal factor focusses on why the district played such a prominent role in decision-making. Mozambique is in a process of rapid decentralisation of governance and districts are increasingly pressured to take the lead in poverty reduction. This district offi cial clearly explained the prominent role of the district in displacement and resettlement decision-making.
The districts decide how to implement [the government plan for poverty reduction]…the districts are the poles of development…the ones who are pushing [resettlement] is the district government.
District offi cials emphasised, however, that while they were pressured to make decisions and take actions, they lacked the fi nancial or institutional resources to do resettlement and poverty reduction in a manner that reduces resettlers' exposures to risk. As example, district offi cials explained that they would have preferred to provide services before resettling people, but that they did not have the means to do so.

DISCUSSION
Four factors operating in a context framed by decentralisation, international poverty reduction agendas and shifting notions of development led to a decision to displace park residents. Plans to provide services in resettlement areas were not implemented, but even if they were, they would not have been adequate to address the livelihood, social, cultural, and psychological dimensions of these risks. The important point is not that district offi cials were focussing on services at the Displacement decision-making / 109