In March 2004, the Government of Kenya issued the terms of reference for a national level committee comprised of governmental officials, NGO representatives, private sector members, university faculty, and civil society groups.[1] The mandate of the committee was to resolve the country's land administration and management problems through the drafting of a National Land Policy. Key concerns highlighted for discussion by the National Land Policy group included: insecure land tenure for vulnerable groups such as women, pastoralists and the urban poor; poor land administration; weak dispute resolution mechanisms; and continued land fragmentation. 

That customary tenures are even a point of discussion in Kenya is in itself striking.  In the mid-1960s when the newly independent Government of Kenya was determining its land reform policies, the question of what to do with community-based tenures was not open to debate - all customary rights and interests in land were to be extinguished.[3] Unabashedly market-oriented, the Kenyan government transformed community-based tenures to individual tenures through a protracted process of adjudication, consolidation, survey, registration, and titling. The rationale for the land tenure reform was simple: indigenous, community-based tenure forms were viewed as inhibiting economic growth because they provided insufficient "security of tenure" to allow for substantial investment in land necessary for agricultural production.[4]

By the late 1980s, after decades of governmental effort to convert tenures to leasehold and freehold, individual ownership of land appeared an unchallengeable method for registering ownership rights and providing security of tenure to Kenyans. Hence the decision in 1993 of a group of "squatters" illegally residing upon government land in Voi municipality to hold land together through a group ownership model known as the Community Land Trust (CLT) model came as a surprise both to local and central government officials and the expatriate technical advisors implementing a settlement improvement or upgrading project there. In contemporary Kenya land is a scarce resource and obtaining a title deed is for many Kenyans but a distant dream. Purchasing land on the market is prohibitively expensive, while the probability of obtaining government land under concessionary terms is highly unlikely, particularly given rampant land grabbing in the 1990s.[5] Why then, when given the opportunity for individual ownership, would community members decline this coveted offer?

The decision by residents of the Tanzania-Bondeni settlement in Voi, Kenya to formulate a community-based institution for land ownership based on the Community Land Trust (CLT) model is the subject of this paper. Specifically, this paper examines the decision from the context of economic theory and institutionalism. The paper argues that Voi residents' decision to hold land together reflected their perception of themselves as being a powerless group vulnerable to losing land to outsiders. The community, moreover, had a history of shared action to defend their land holdings that served to establish a threshold level of trust which made the group tenure a possibility. The paper concludes that the decision of the settlement's residents to hold land together was rational - a collective institution better served to protect their individual self-interest than the individual institution predicted by the Evolutionary Theory of Land Rights, ETLR. The paper concludes that "history matters" in institutional analysis - to really understand institutional change we must understand the embedded context of the decision-makers. The study also supports the policy perspective that there is no one-size fits all approach to land tenure. Policymakers should strive to provide a range of tenure options that can fit the context of the specific community. To achieve this, African land policies must include mechanisms for full community participation in crafting tenure regimes and making decisions that affect the ownership and use of land.  

Following this introduction, the paper is split into four sections. The first section reviews an influential theoretical perspective on African customary land tenure and human decision-making embodied in the institutional and development economics literature. Section two presents the Tanzania-Bondeni settlement and the upgrading project. Section three outlines the methodology used for the case study and presents the findings of interviews conducted with residents of the settlement. The final section relates findings from the residents' interviews to economic theory and land tenure policy. 

CUSTOMARY LAND TENURE AND DECISION-MAKING IN ECONOMIC THEORY

The discussion of customary tenures in African land policy reforms reflects one long-standing theoretical debate in the social science literature over the role of property rights in African development.[6] Essentially the debate is over whether customary or community-based tenures represent an obstacle to economic development and if African countries should implement reform programs to transform customary regimes to individual tenures.[7] On one side of the debate are adherents of the ETLR, whose ranks include neo-classical economists, Public Choice theorists, and some Neo-Institutionalists, who believe that property rights in any society evolve due to scarcity.[8] This theory predicts that as population pressure increases and land becomes an increasingly scarce resource, rights to land will individualize until private property exists.[9] This move to individual tenure is economically advantageous for both the land owner and the state since bestowing all property rights and decision-making powers in one person overcomes key economic inefficiencies such as transaction costs and free ridership.[10] For adherents of the ETLR, the move to individual tenures is inevitable since institutions are seen as evolving in order to maximize benefits and minimize costs. Institutions, moreover, have been described as moving toward greater efficiency over time.[11] The policy implications of the ETLR are clear: development-oriented governments should assist this evolutionary process by formulating and implementing individualization reforms.[12]    

Those arguing for the retention of customary or community-based tenures, in contrast, represent a more multi-disciplinary group of scholars, including economic sociologists, anthropologists, and mainstream institutional economists. They make two primary arguments for community-based tenures. First, they contend that the function of land in African society is much more complex than granted by Western economic thought.[13] In Africa, they observe, land serves important social and political functions not common in the west. Land is the cultural basis of power and belonging.[14] The granting of land is a primary mechanism for structuring society and gaining political power and allegiance. The holding of land is the primary indicator of societal belonging. Second, these scholars argue that customary tenures have been mischaracterized and misunderstood. Customary tenures are not anachronisms impeding economic progress but instead are dynamic institutional arrangements characterized by a mix of property rights (some private, some shared) which has adapted over time to meet community needs.[15] Such tenures can be inherently secure and conducive to economic growth - nothing less than "private property for the group."[16] The failure of many African societies to move toward individual tenures is thus an indicator of the inappropriateness of these tenures for the African social context and a challenge to claims of universality for the ETLR. The land policy implications are also clear:  forcible reform of land tenure using Western institutional models is not advisable. To improve tenure security governments should clearly define and enforce all property rights regimes, including indigenous institutions.[17]

Behind this debate on land tenure and land policy is an even more fundamental disagreement over human decision-making and rationality. Adherents of the ETLR view decision-making through the dominant theoretical model in economics, that of the "economic man."[18] As every Economics 101 student knows, economic man (and woman) is a self-interested, atomistic actor endowed with a set of preferences whose decision-making is expected to be "rational." Rational behavior is defined as utility-maximizing behavior, that is, behavior that makes the economic man better-off.[19] The evolution of land institutions described by the ETLR assumes the self-interested rationality of the economic man. Faced with scarcity and the difficulties of community-based tenures, a rational decision-maker will choose to hold land under individual tenure in order to maximize his/her utility.   

Scholars skeptical of the ETLR, such as mainstream institutional economists and economic sociologists, not surprisingly are also quite critical of the economic man. They question the concept of human nature central to this paradigm. Human beings are cast as rational actors driven by the need to maximize their utility, yet every day one sees evidence in actual behavior that humans can and do act "irrationally" in the economic sense.[20] One also sees instances when people knowingly act against their individual self interest.[21] They argue that there is a range of motives for human behavior and decision-making in addition to self-interest. Humans are motivated by principles such as altruism, cultural constructs such as tradition and nationalism, as well as by ignorance and irrationality.[22] A second critique relates to the under-socialized or atomistic nature of the individual. Economic man is presented as a creature born with a given set of preferences. He is depicted as an individual acting in isolation, deliberating solely upon his own welfare and acting purposively to maximize that welfare.[23] Mainstream institutional economists, most vocally, find this presentation problematic because it does not explain where preferences - the source of purposive action - come from and why they change. In neo-classical economic theory, preferences are "immanently conceived" and fixed.[24] Without an explanation of why people have certain preferences and why preferences may change, mainstream institutionalists assert that economics offers little meaningful explanation for purposeful action.[25]

In contrast, mainstream institutional economists and economic sociologists have clear notions about the source of purpose. They argue that human beings are social animals born into a society endowed with culture, beliefs, and institutions. Human preferences and hence purpose are determined in an interactive process whereby the cultural and social factors that dictate what is considered acceptable or unacceptable behavior interact with individual perspectives and motivations.[26] Granovetter calls this view of human behavior "embeddedness." Embeddedness represents a middle way between the atomized, undersocialized perspective of neo-classical economics and the oversocialized concept of human nature once characteristic of sociology.[27] "Actors," Granovetter argues, 

do not behave or decide as atoms outside a social context, nor do they adhere slavishly to a script written for them by the particular intersection of social categories that they happen to occupy. Their attempts at purposive action are instead embedded in concrete, ongoing systems of social relations.[28]

These on-going systems of social relations or institutions must be integrated into economic analysis in order to really understand purposive human behavior and predict economic outcomes.

These two debates - one over the evolution of property rights in Africa, the other an underlying disagreement on human decision-making - provide an interesting theoretical frame for examining the community decision-making that took place in the settlement upgrading initiative in Voi. Using the insights of ETLR, one would expect that given high levels of population growth, growing land scarcity, and the government's wholehearted embrace of private property and the market, any decision-maker offered the choice between individual or group leasehold would select individual title. The deviance - or irrationality - of the Voi community in selecting community-based property thus is striking. Why did their land tenure choice differ from other communities within Kenyan society? What factors do Voi community members cite as being weighed in their decision? In what terms do they describe their decision? Do they speak of the decision in economic terms? Are factors of community identity and belonging significant? What sort of expectations, moreover, do the residents of Tanzania-Bondeni have of their "common" property? Do these expectations differ from those of individuals opting for individual property rights? 

TANZANIA-BONDENI SETTLEMENT AND THE UPGRADING PROJECT

The settlement of Tanzania-Bondeni is situated approximately 1.5 kilometers from the commercial heart of Voi town, which is located on the main highway running from Mombasa to Nairobi. In 1989 just prior to the start of the upgrading project, the population of Voi town was estimated to be 13,202 people.[29] Using the growth rate of the prior decade (5.79%), in 1999, the time of fieldwork, the town was home to approximately 23,200 persons. Government statistics are not available for the settlement of Tanzania Bondeni, but Asienwa estimated its population at 2,993 persons.[30] Assuming a similar rate of growth, the settlement would have contained approximately 4,971 persons in 1999. The Tanzania Bondeni settlement covers approximately 22 hectares of land and is physically cut into two parts by a railroad track belonging to Kenya Railways. The larger built up area to the south of the track is Tanzania. This part of the settlement is bounded by the Voi river to the south and the Voi Sisal Estate to the west. The smaller built up area to the north of the rail line is Bondeni, which means "in the valley" in Swahili. Wedged between the embankment that supports the current rail line and the embankment for a disused rail line to the north, this neighborhood indeed can be seen as lying in a valley (please see Appendix 1 for a demographic profile of the settlement).

The need to conduct settlement upgrading in the settlement emerged through a local five-year planning process, known as a Local Authority Development Programme.  Upgrading projects, in brief, provide basic urban services to settlements (e.g., water, sanitation, roads, and paths). Such projects also generally seek to provide some level of land tenure security. Formal legalization (i.e., survey, titling, and registration) is one method of conferring security that has been used in these projects (Gulyani and Bassett, forthcoming).  In agreeing to assist the local authority with the task, planners at the Ministry of Local Government (MLG) and the GTZ Small Towns Development Project (STDP) had two main objectives.[31] The first objective was to improve the financial health of the local government by broadening its property tax base.[32] So long as settlements remain informal - illegal with land ownership undetermined and unregistered - local authorities are unable to assess these properties and add their values to the property tax rolls.[33] Secondly, planners saw the projects as an opportunity to test an upgrading approach that would be affordable and replicable by a local authority on its own. 

Prior to implementation, the planners working with the STDP reviewed the lessons learned in other upgrading projects in Kenya and elsewhere and formulated an approach that would avoid repeating past mistakes.[34] The upgrading approach was organized around certain key principles and guidelines that reflected these lessons learned - a key principle was providing affordable and sustainable tenure security to settlement residents. Tenure security, however, was seen as a challenge. There was a consensus that past upgrading projects had not succeeded in providing sustainable security of tenure.[35] Past Kenyan projects had issued long-term leasehold titles with five-year restrictions on sale, but over the course of time many low-income project beneficiaries eventually sold their leaseholds to other more affluent individuals.[36] 

Kenyan planners give two reasons for land sales.[37] The first is that settlement residents are not particularly interested in owning land in the city. They are only there for their working lives and they really want land in their rural home areas. As a result they "cash-out"­- treat the project as an unexpected windfall, obtain whatever money they can for their plot, and go to squat elsewhere. The second reason is more forgiving toward the slum resident:  sales occur due to financial distress brought about by occurrences like an illness or the death of a family member.  Residents want the asset, but as the poorest strata of society, they do not have the means to withstand bad times.

While there was agreement that past attempts to provide security of tenure to the urban poor had failed, there was little sense of how to provide affordable, sustainable land rights in the upgrading initiatives. Sanctioned government options were limited: leasehold or freehold title. The application of restrictions on sale on leasehold titles were intended to prevent squatters from "cashing out"­­­­­- these restrictions, however, are easily circumvented by informal transactions and have been shown to be ineffective elsewhere.[38] Restrictions on sale, moreover, did nothing to assist those residents wishing to retain land, but with no alternative disposable assets.

While the STDP was formulating its upgrading approach and deliberating upon security of tenure, the Ford Foundation sponsored a study to examine the viability of utilizing the American Community Land Trust Model as a model for delivering affordable shelter to the urban poor.[39] A year later the CLT model was picked up for consideration by the STDP project as a possible tenure form. A Community Land Trust, briefly, is a community-based, democratically controlled organization formed to hold and acquire land for the use of its members.[40] Its most defining characteristic is the splitting of ownership: individuals own houses while the CLT owns the land upon which the houses sit. CLTs are formed to hold land in perpetuity thereby removing it from the open market. The CLT, moreover, plays a role in the sale of houses - in most cases retaining the first right of refusal. The CLT model splits usufruct for two main reasons. First, the model recognizes that value in property comes from different sources.[41] The value of real estate is largely determined by public investments (e.g., infrastructure, schools) created by society as a whole. The model ensures that any increase in property value attributable to the land (e.g., location coupled with public investment) is captured by the community.[42] Secondly, group ownership of the land is tactical:  it helps buffer lower income residents from the effects of gentrification and rising land prices since the CLT removes land from the market, limits equity appreciation, and takes on the task of meeting property taxes. The CLT provides a social safety net to ensure that poor people get and retain better housing.

While the CLT model draws some of its inspiration from customary tenures in Africa, the model is highly formal and replaces or extinguishes other tenure institutions or property rights configurations.[43] CLTs are incorporated entities - in the US they are not-for-profit corporations with an elected Board of Directors. While membership is open to all interested persons, to obtain land trust land an individual (or would be land lessee) must become a member and is expected to participate in community planning and governance. The relationship between the trust and land lessees is outlined in written lease agreements. These agreements detail, for instance, rules regarding the sale of houses, payment of fees, and valuation of assets.[44] 

The Tanzania-Bondeni Tenure Decision

By late 1992 upgrading activities for the Voi settlement had made significant progress.  Members of a national level Project Promotion Committee (PPC) had visited the settlement and were ensuring support for upgrading in Nairobi. A local level Technical Task Force (TTF) composed of officers from the ministerial line agencies and representatives of the local authority had been constituted. Social planners had completed a "listening survey" intended to uncover the felt needs of the residents as well as to identify important opinion leaders.  Preliminary community mobilization had occurred: residents had indicated interest in participating in the upgrading exercise. Finally, community members had elected a Residents Committee (RC) to represent their interests in the upgrading exercise. This committee had been apprised of its responsibilities and powers and had a good understanding of the upgrading approach, including the issue of land tenure.

From the outset, the Tanzania-Bondeni Residents Committee demonstrated a strong ability to mobilize its community and gain support for upgrading. Committee members enthusiastically embraced their role - moving through the settlement daily to update residents on project activities and remind them of their obligations.[45] Payments such as survey fees accrued quickly. Other small projects flourished.[46] 

The Residents Committee also responded with great interest to the idea of the Community Land Trust model. The discussion of security of tenure and land loss by squatters resonated with the Voi leadership. In her report, the GTZ Social Planner described the response of the Voi leadership as being to a "large extent influenced by the flux [of] people who went to the project area with the intention of buying out the beneficiaries."[47] Disturbed by this development, the Tanzania-Bondeni RC indicated its interest in learning more about the land trust option. The RC's interest in turn sparked debate and lobbying within the community. According to Muchene: 

once the two types of tenure system [sic] were explained to them at the leaders workshop, they embarked on an awareness campaign to convince the people to accept the community land trust. Some individuals who had bought structures for speculative purposes also campaigned for the individual title. [48]

In November 1992, planners from the MLG and STDP held a final discussion on tenure and the various options for land ownership with the Voi Municipal Council and the members of the Residents Committee. This discussion was framed around a matrix illustrating the options for land ownership and community organization. The three options identified were individual leasehold title, individual title coupled with the formation of housing cooperatives, and group leasehold title with the formation of a community land trust. After this meeting, the RC held a series of six community meetings in which the three options were explained. At the end of each meeting, community members were asked to vote for the tenure form of their choice. Observers from the STDP attended each meeting to ensure that only bona fide structure owners were allowed to vote and to see to it that both individual and community tenure were fully explained with sufficient space given for questions and debate. The vote was overwhelmingly in favor of the CLT model: 239 structure owners for the CLT model versus 19 structure owners in favor of individual leasehold title.[49]

THE LAND TENURE DECISION: RESIDENTS' EXPLANATIONS

Methodology

To understand the rationale behind the Voi tenure decision two sets of fieldwork plus a brief follow up visit were conducted. The first field work was conducted in 1996 and consisted of interviews with project officers, government of Kenya officials, and members of the Tanzania-Bondeni residents committee as well as a field visit to the settlement.[50] The 1999 field research consisted of in-depth interviews with community residents, additional informant interviews, and archival research at the Kenya National Archives. Finally, the author made a brief visit to Kenya in 2003 to track progress in project implementation and gather documents related to land policy reforms.

The data reported in this paper primarily arises from fieldwork that took place in March-June 1999. During this period, the author (with the assistance of a research assistant) completed 50 in-depth interviews with Voi residents, all of whom were structure owners.  Structure owners were interviewed as they were the residents empowered to make the land tenure decision. Interviewees were selected using a modified "snowball" technique.[51] The sampling process began by looking for indicators of long-term residence, namely numbers painted on doors (an indicator that they lived in the settlement when an initial socio-economic survey was conducted by the STDP in 1990) or rusty roofs. After completing an interview, the structure owner was asked if he or she knew another structure owner who might be interested in speaking with the interviewers. This approach was modified, however, by a mapping process in which the approximate location of each respondent was mapped on the physical development plan for the settlement. Once a certain area was adequately represented, the interview process moved to another section of the settlement and began the process over again. Interviews were conducted in Swahili with notes taken by both the research assistant and the author. Interviews ranged from 45 minutes to two and a half hours. Interview notes were transcribed at the end of each working day and texts were reviewed by the author and research assistant to ensure accuracy. Content analytical procedures and pattern matching were used to identify key themes and narratives arising from the interviews.[52]

Eleven interviews were with structure owners living in Bondeni and the rest conducted with Tanzania owners. The disparity in representation between Bondeni and Tanzania is purposeful. Tanzania is the larger, more populous settlement area. Additionally, two discussions were held with members of the Managing Committee (nee Residents Committee) of the Tanzania-Bondeni Settlement Society. The first meeting was a briefing on the planned research; the second meeting focused on the committee's experience implementing the CLT model. Several opportunistic interviews occurred, including one with the former Youth Representative on the RC. 

A demographic profile of the respondents is provided as Appendix 2. As it shows, the majority of respondents were women. While a concerted effort was made to tap the views of more men, it was without much success. Several interviewees said that there were few men in the settlement; the former youth leader (a man) described the settlement as being full of "old people and women."[53] Respondents ranged in age from their mid-twenties to into their seventies. The exact the age of respondents was not asked, instead it was estimated from their physical appearance and narratives. The majority of respondents were between the ages of 31 to 50. They had lived in the settlement anywhere from 1 to 57 years. Most of the respondents were not community leaders. Four were RC members, four were in leadership positions in the local cooperative housing societies, and two were leaders in other committees, namely a youth and a women's group. Through the interview narrative (e.g., stories of migration) and through observation/listening, ethnic background was also identified.[54] The ethnic make-up of the respondents was quite uniform: of the 50 people interviewed 48 (or 96%) were Taitas. Two were from other areas in Kenya.

Findings from Residents Interviews

Questions were asked that directly related to the tenure decision. Structure owners were asked whether they could remember the tenure options offered to the community by the government. They were also asked to explain the process they used for choosing between these options and whether they had personally participated in the selection process. Finally, structure owners were asked to explain why they felt the community had selected group title and their opinion of the decision. Respondents were then asked to elucidate what they felt particular advantages and disadvantages were associated with both community-based and individual tenure.

Out of the fifty structure owners interviewed, 16 indicated that the settlement had voted to select its tenure form. They all knew that the selection was between individual title (kumiliki kibinafsi) and group ownership (kumiliki kijamii). The majority of respondents indicated that they could not recall that there were options in land tenure. Those who indicated that they didn't know there was a choice tended to be either younger settlement residents (who at the time of selection were generally in their teens) or they were newcomers to the settlement who had gotten land after the 1992 decision. The longer-term residents who did not remember the tenure choice indicated that they were away at the time (e.g., seriously ill, working elsewhere). Five respondents contested the idea that there really was a choice.  They said that the decision was made by the Residents Committee.[55] 

Respondents were also asked about the process used to select tenure for the upgrading project. A greater number of respondents were able to describe the series of community meetings held in November 1992 and the resultant vote than had been able to remember that the community had been given options in land tenure. Presumably, discussion of tenure options triggered something in their memories. Alternatively they may have been told by others, such as the Residents Committee, that the tenure form was selected through a vote by the majority of structure owners in a series of neighborhood meetings. Eighteen persons indicated they had gone to the community meetings and had actually voted (they were not asked how they voted). Of those who participated in the meetings, 15 were women and 3 were men. The participants fell into the following age groups: 20-30 years of age (2 participants), 30-40 (5 participants), 40-50 (3 participants), and older than 50 (8 participants).

When asked to explain why the community voted to hold land together respondents gave answers which roughly fall into four categories. These four categories of answers, each of which will be dealt with in turn, were: (1) I don't know or I refuse to speculate; (2) Because of poverty and a lack of economic ability; (3) In order to prevent loss of land; (4) Because the Residents Committee decided.[56] The term "roughly" is used to characterize responses because although the answers were grouped together reflecting the same central contention, in their narratives residents often forwarded slightly more nuanced explanations that distinguished their views from those of their neighbors.

The first group, indicating that they did not know why group tenure was selected, consisted of 16 people. Of these, six individuals completely refused to speculate on the reason for the selection. I attributed their answers to the politicized atmosphere of the settlement at the time of the research and the fact that the tenure regime was the subject of much debate and acrimony. The remaining respondents indicated that they did not know why group tenure was selected. People who said they did not know why group tenure was selected were generally (1) people who were not present at the tenure decision and therefore refused to hypothesize on the decision; (2) younger members of the settlement who had not participated in the settlement decision and had gotten plots as a result of inheritance or by being a child of a structure owner who was allocated more than one plot in the first allocation exercise; or (3) newcomers to the settlement who had gotten plots after the decision was made.

The most common explanation of the tenure decision dealt with the economic status of settlement residents and the cost of owning land. These respondents emphasized that settlement residents are poor. At the time of the decision, they felt that they did not have the economic ability to meet the costs associated with individual leasehold. They saw group tenure as being advantageous to their interests since it would enable them to get the land more inexpensively. Since it also entailed sharing the costs of keeping the land, group tenure would assist them to retain the land for themselves and their children. Group ownership also implied additional community help for the less able, a feature that these respondents said appealed to those voting for the tenure form. In short, as the following quotes illustrate, Voi residents felt they had insufficient economic ability to own land individually:

• "It fit our ability. Here in Tanzania-Bondeni we don't have the ability to pay for an individual title deed, therefore we choose group ownership."[57]
•   "I think they saw that they would be helped through group ownership, for instance to join together and to build houses collectively for people who lacked ability."[58]
• "It fit and once we were explained {about tenure}, we agreed to choose group ownership. Because we were told that with group ownership we will pay less money than for private ownership. It was because of this that we agreed."[59]

A second explanation for the tenure decision cited by respondents was the Voi residents selected to go with the CLT model in order to prevent community members from losing their land. To many of these respondents the coercive side of the CLT model that limits the rights of community members to dispose of land was paramount.  Settlement residents wanted to limit "cashing out" behavior.  As one respondent explained, residents feared that this unexpected gift of land would be squandered by their neighbors:  "Because there are people who can't be satisfied. They are greedy. And also they are not grateful. Therefore they are able to sell and go somewhere else. Therefore they embraced group ownership so that these people would not sell."[60]

Additionally, some respondents in this grouping emphasized that the CLT model was attractive since in addition to limiting their own rights it also limited the ability of outsiders to buy into the settlement. For these respondents, the defensive nature of the no-sale rule was also important. As informant explained, group tenure was selected: "Because the people of Voi love to sell land. Therefore they saw that group ownership will prevent sales and also it will prevent outsiders from stealing the land of Tanzania-Bondeni."[61]

The final set of explanations for the tenure decision averred that the Residents Committee played the decisive role in selecting group tenure. Answers here break down into roughly three categories. There were some respondents who indicated that the RC selected group tenure itself but did so out of goodwill because they felt was the best tenure form for the community.[62] In this interpretation, the RC was said to have learned from the teachings of the STDP in the various training events which were held as part of the upgrading process.  One resident noted: "We did not give our opinions about land ownership. But the committee explained to us. They decided to own land together because of this and that [various reasons]."[63] Secondly, there were three respondents who attributed the decision to the committee but were totally at a loss to explain the committee's rationale. They suggested we talk directly to the committee. Thirdly, a minority of respondents averred that the committee tricked the community into selecting group title for its own purposes. According to this version of the selection, the committee tricked people by telling them that the STDP would only build houses for settlement residents if they selected group title. As this has not happened, and as the committee members have received more than one plot, some of these respondents are convinced the RC wanted group title in order to remain powerful people in the local community.[64] As one respondent averred: "We were not given an opportunity to choose. They said with community ownership we will have houses built for us, but with individual ownership you will not get a house built."[65] Finally, one respondent in this grouping acknowledged that there was a vote on tenure with wide participation from the settlement residents, but he stressed that the vote was not really free and fair: "Yes I voted.  But they used an arm of the government, like the chief [to run the meeting]. They said it was group ownership and it became that."[66]

Pre-Project Institutions and Security of Tenure

Using these answers to reflect on institutional theory, the information appears inconclusive and somewhat contradictory. These answers appear to uphold the perspective that the decision to form the institution was primarily driven by self-interest: residents banded together in order to achieve something that they felt they could not achieve individually. Responses also give credence to a perspective that group tenures survive due to the machinations of elites who gain power from land allocation, that RC members manipulated the decision to select the community-based tenure regime in order to maintain their positions of power and prestige in the settlement.[67] The answers do not provide much evidence that land had an explicit role in determining community identity and social belonging. Nor do the answers provide an indication that pre-project institutions played a role in the decision.

A series of questions was asked regarding access to land and security of tenure prior to the upgrading initiatives. Responses provide data for understanding the above factors, including the pre-project institutional arrangements in the settlement. These answers also provide information that enables a picture of settlement relations at the time of the tenure decision to be pieced together. Three significant themes emerged in this discussion. These were: (1) the perceived degree of tenure insecurity in the settlement and residents' fear of demolition by the municipal council; (2) the level of economic ability in the settlement and residents' perceptions that their land was vulnerable to more powerful outsiders; and (3) community interdependence and the degree of trust that existed in the community at the time of the land tenure decision.

Access to Land/Security of Tenure

Respondents were asked to explain to us how they got access to land in the settlement in the first place. All fifty respondents answered these questions. With the exception of one respondent who indicated that he just cleared bush and built, and five others who indicated that they inherited land from their parents, respondents all got permission to settle in the area. There were, in short, procedures for gaining access right from the beginning of the settlement. Two of the older residents obtained permission to settle from the original owners: one woman relocated there with her mother when the District Commissioner outlawed cattle in town. A second woman worked for the sisal estate and erected a house on their land. The most common way of gaining access to land and permission to build, however, was to ask the wazee wa mtaa (neighborhood elders), a committee constituted by the area chief.[68] Two respondents indicated they got land by asking neighbors for a corner of "their" land. In the project period (post-1991), land access remained regulated. One respondent indicated that she was a tenant who got a plot as part of the project; another indicated he had asked the RC for a plot. Four respondents indicated that they bought a house from a departing resident; two of them purchased their house after the project had begun.[69]

After getting access to land, most respondents proceeded to build their own houses. Building houses was not as easy as gaining access to land. To build a house to completion required effort and forward planning. Several respondents indicated that although the wazee wa mtaa granted permission to settle often in exchange for a "small token," they also warned residents that the land was not their property ¾ as squatters they could face demolition at any time if the government wanted or needed the land. As one of the respondents described, despite the quasi-sanctioned mode of gaining land, it took determination to get a house built:

We built at night. Chief Ali did not want us [to build]. We begged the village elder for permission to build. But he said if you get in trouble, you have to defend yourself. We paid him a little money, like a token. I was caught by the municipal council. They took me to the chief and the District Officer. But I did not lose heart, I continued to build.[70] 

In the face of opposition by municipal council, the chief, and the District Officers, Tanzania-Bondeni residents banded together to help each other out in getting houses completed. As one informant related, the embattled squatters built at night since at that time the municipal askaris (enforcement officers) did not work.[71] Another tactic was building over the weekend. One respondent indicated that people would accrue the building materials over the course of the week and then in the weekend with the assistance of their neighbors, they would build.[72] By necessity residents had to build fast ¾ if the council found a house half built in the daytime they would demolish it.  As one resident of Bondeni described her experience:

I spoke with colleagues at the market and they told me that there was an elder who was able to give me a plot. I paid this man 270 shillings [c. 1984] and he told me that my house should be built in two days because the municipal council would demolish it [otherwise].  But I built in two days, therefore, they did not demolish my house.[73]

Once a house was built, there were some implicit rules that needed to be followed in order to avoid demolition. First, the municipal council would not destroy a house which was occupied. Settlers adopted a tactic of sitting outside their houses when the municipal askaris were passing through. According to one elderly resident, the most effective tactic was to have a woman with a baby and a suffuria (cooking pot) in front of the house ¾ in that case the house would not be touched.[74] The second rule related to money. Once a house was built, settlers needed to go to the municipal council and register. If they registered and paid a fee to the council, then they would not be harassed by the askaris.[75] According to one RC member, people did not want to pay the fee, but they had no choice.[76] The fee was categorized by the Municipal Council as a "conservancy fee" (a fee for rubbish collection), although council did not provide the service to the settlement. According to the former Voi Municipal Clerk, this peculiar situation arose as a result of the initial registration of squatters that took place under the Commissioner of Squatters in the 1970s.  Although the Commission was disbanded without solving the squatter problem, the council kept up the registration system. Fifteen of 50 respondents indicated that they paid this fee.

A series of questions was asked relating to tenure security. Given the situation above, one might expect that the settlers would feel relatively secure once they had successfully built a house and registered with the council. The perspective that there was relatively good tenure security is also reflected in the social survey conducted by Asienwa and her observations on tenants aspiring to purchase houses in Tanzania-Bondeni.[77] In responding to the questions on tenure security, a majority of the respondents indicated that they knew that they were settled on government land and that their residency was thus technically illegal. Twenty-one said they did not think they were breaking the law. A number of these respondents indicated that they felt that the sanction of the wazee wa mtaa and payment to the town council was sufficient. As one protested: "We were shown this opportunity [plot of land to settle] by the chief, and he is the government."[78]

Despite this, half of the respondents did indicate that there was a prevailing sense of tenure insecurity in the settlement prior to the advent of the upgrading project. Out of the people who answered the question, a majority indicated that they felt some level of insecurity. Fifteen said they felt a little or somewhat insecure. The main complaint of these respondents was that they were forced to "live temporary," that is accept a lower standard of living than their wealth would afford.[79] The construction of a temporary house was allowable, but erecting a permanent house would bring trouble. Although many residents had the resources to invest in better quality housing, they were constrained by fear of losing their investment. The main fear was demolition: respondents indicated that they were scared that their houses could be demolished at any time without warning. Despite the implicit rules listed above, the Voi Municipal Council was still seen as a potential threat to the settlement. Among the respondents, these quotes are illustrative of this level of tenure insecurity:

• "We were unable to build permanent houses because we were told that we were able to be driven out at any time."[80]
• "We were really scared to build permanent houses because we knew we were squatters, and at any time we could be moved."[81]
• "This situation disturbed me because I feared they were able to come and demolish at any time. I did not want to build a nice house. Even now [the houses of the] people of Mwakigali are being demolished."[82]

A smaller group of respondents said they felt very insecure. They were the most worried that their houses would be destroyed and that they would be evicted from the land. As a group these respondents tended to be older people and/or those with limited economic ability. One single mother expressed her anxiety over tenure insecurity by saying: "This situation disturbed me a lot. When I heard that I was breaking the law and that this land was not ours. [I feared] because my children don't have two [parents]. I did not know where I would take them and even if I die where will they go?"[83]

Not surprisingly, people whose homes had been demolished previously also indicated greater tenure insecurity. As one resident explained: "The county council really disturbed us.  We built and the council demolished it every time."[84]

Notably, all the RC members interviewed stated that they felt very insecure in their tenure status. As a group these individuals had all faced demolition ¾ three had experienced having their homes destroyed, the one who escaped was a male who as a former employee of a municipal council dared to chastise the askaris who threatened his house.[85] According to one RC member, the threat of demolition and displacement was always with Tanzania-Bondeni.[86] Bondeni was particularly at risk as it was dominated by railway's land and the corporation kept threatening to build a warehouse there.[87] The prospect of demolition only receded once the project began and houses were enumerated for legalization. 

Economic Ability

The second theme that arose in the interviews regarding the tenure decision was the perceived level of economic ability of Voi residents. Voi residents tend to perceive themselves as extremely poor people. To make this distinction in degree requires reliance on the use of language in the interviews. Significantly, an unusual Swahili word for describing poor people, wanyonge, occurred repeatedly in several of the Voi interviews.[88] For instance, as an explanation for the land tenure decision: "Because we were very poor people. We don't have money and it fit therefore. The tenure forms were explained to us all, we agreed to own together. Because unity is strength."[89]  Again, when a resident described how she learned about the group tenure decision and speculating on the reason it was selected: "One day in a meeting we were told that we had agreed to group ownership and that group ownership will be cheaper than individual ownership. Because we, the people of Voi, are very poor, that's why I think we agreed to own land together."[90] And finally, when discussing the possible use of the CLT model elsewhere: "It depends if the people want [to own land] like this. It can succeed in a place where the people are very poor and they are unable (lit: are defeated) to pay the high price required to own individually."[91]

In Swahili, the common term for a poor person is mtu maskini, and indeed that word was used in most of the Voi interviews. The word wanyonge is not frequently heard and indicates a more severe state of poverty and powerlessness. The Standard Swahili-English Dictionary translates the word mnyonge (singular of wanyonge) as "a humble, abject, low, debased person."[92] Bakhressa's Kamusi ya Maana na Matumizi  defines the word as mtu maskini asiyekuwa na kitu, "a poor person who has nothing".[93] My research assistant translated it as "weakling."

For Voi residents, the implication of extreme poverty was not simply that they lacked the ability to pay for land and housing. Being a myonge also meant that you lacked political power and patronage networks. As the allocation of land is one of the most politically manipulated processes in modern day Kenya, Voi residents were understandably concerned that the gift of this project ¾ the land ¾ might be lost to outsiders with greater political power and connections.

In speaking with residents about their opinion of the tenure decision and the respective advantages and disadvantages of the two tenure forms, the fear of land loss was repeatedly emphasized. Respondents stressed that they feared that they or their neighbors could be cheated out of their land by more powerful or savvy people. In discussing this threat, residents used a variety of Swahili verbs, including kupokonya (to take away by force, rob, or plunder); kunyakua, the causative form of to grab or snatch); kudhulumu (to treat unjustly, defraud, oppress); and kunyang'anya (to take by force, steal or rob).[94]

Respondents were split, however, on whether group ownership would do a better job protecting their land against predatory outsiders than individual title deeds. Not surprisingly, those who were pleased with the CLT decision felt that protection from political manipulation was one of the main advantages of group title. As one respondent explained her perspective on the two tenure forms: "If you unite you cannot be robbed of the land. The whole family can get land. There are no disadvantages. With private ownership, someone can be given [land] by powerful people and he could rob you of your plot."[95]

The reasoning of some other respondents was the complete reverse. Although he shared the concern for land loss, one respondent felt that group ownership placed him at greater risk for land loss since his individual rights were subordinated to the decisions of the group:  "The society can decide to steal this land, or that it should sell [my] place here."[96] A second respondent pointed out that one title deed made them all vulnerable since if the society is cheated by someone, everyone could become landless.[97] "Also if a person wants to defraud us if he uses this single title deed he defrauds us [all]."[98] Finally, echoing the sentiments of many of those who were dissatisfied with the group title decision, a third respondent stressed that the land would only really be protected from predation if she had the title deed herself:

I prefer individual ownership. There is no disadvantage to private ownership. The advantage: I am able to get a loan from someplace and build my house. My children are able to know this plot is my private property and they won't get complications here as was the case previously. No one can rob my children of this land because they have permission [to stay].[99]

Community Interdependence and Trust

The third theme that arose in the interviews that is crucial to an understanding of the tenure decision was the level of community interdependence and unity that existed in the settlement in late 1991. Prior to legalization, as has been noted above in the discussion on security of tenure, settlement residents had relied upon each other to accomplish their individual goals. They had worked together to build their homes and they had united to defend their homes against threats of demolition from the local administration. They had created a number of self-help institutions, such as youth groups, savings societies, and women's groups to advance their economic status.[100] They had worked together to build a nursery school for their pre-primary age children. Now in the course of the upgrading project residents were being offered the choice between continuing with this unity by holding the land as a group or to stand on their own by selecting individual leasehold. That the residents should choose to hold land together is not exactly surprising since they knew from experience that they could defend their land together, but many were unsure of how they might fare on their own.

This is not to suggest that the decision to hold land together was predestined, that because they had been together they wished to continue together. Rather, the importance of this shared history is that it created a key variable ¾ trust ¾ that made the decision to hold land collectively a possibility. Granovetter reminds us that most decisions are not made in the social void of the Prisoner's Dilemma. Decisions are embedded ¾ they are made in the context of concrete personal relations. Personal relations in turn are an important variable to consider here because these relations provide the basis for trust or distrust in any society. People learn whether they can trust each other by interaction; decisions on how to structure future relations are based upon past interactions.

This insight is particularly important for understanding the Voi decision. In the case of Voi, when dealing with land issues residents had strong expectations of good behavior (or no "malfeasance" to use Granovetter's term) from their fellow residents and their leadership. They had united as a community to protect their land against outsiders. They had established a network of relationships that would assist them in time of need. Voi residents felt they knew and could predict the behavior of their neighbors. In short, they felt they could trust their neighbors.

To return to the evidence of the interviews, the existence of trust in Voi is more implied in the discussions than expressly stated. Part of this is the result of the interviewing approach ¾ questions were open ended so as to allow residents to speak in their own words. There were several places in the interview process where the answers of residents provide insight on community characteristics and evidence of the existence of trust and its relationship to the tenure decision. The most significant evidence was provided in discussions around the replicability of the CLT model in other settings.

In thinking about where the CLT might be successfully tried in the future, respondents not surprisingly tended to reflect back on the experience of the Tanzania-Bondeni project and used what they knew about their settlement to give advice about its future use. Some of the answers were admittedly prosaic: "It can succeed if the residents of the place are able to get along with GTZ."[101] Other answers, however, gave indications of key settlement characteristics that point toward the existence of trust. In addition to being poor, three factors were most frequently mentioned: residents must be united, they must be able to cooperate with each other, and they must care for each other's welfare.[102] Some expressions of this include:

• "It depends on the people who live in the place and also it requires a place [in which] people cooperate well." 
• "It is able to succeed very well in every place. People who are able to care for each other [lit. love] In a place of love everything is possible."
•   "Us here in Tanzania Bondeni we know each other. That is why it is able to succeed. It is necessary that people themselves understand each other."
• "The people should live [together] with unity and cooperation."[103]

Although the terms are not the same as Granovetter's, taken together these analyses point toward the existence of trust in the community. The basis of cooperation, unity, and even love is fundamentally trust. To cooperate one must have expectations that your partner will fulfill his/her side of an agreement (or in economic terms believe that there will be no free-riding). To be unified one must believe, or trust, that all members of your group share the same goal (Which is why being a traitor is so mercilessly punished ¾ one has betrayed the trust of others). Finally "to love" (or in the context of this paper, to care for the social welfare of another) requires a significant level of understanding, empathy, and expectation of reciprocity.  This latter factor is only created through on-going interpersonal relations characterized by trust.

CONCLUSIONS

The interviews conducted for this study illuminate conflicting reasons for the selection of the CLT model as the land tenure form for the upgrading project. Most commonly, residents said that the CLT model was selected because it fit their economic ability ¾ the lower cost and shared obligations of the model were decisive. Others indicated that it was selected because it would prevent people from selling the land of Tanzania-Bondeni to outsiders. Its defensive nature, in short, was important. Finally, a number of respondents place the blame ¾ or credit ¾ at the feet of the Residents Committee. In this perspective, the CLT model was not really democratically selected¾it was the choice of the leadership of the settlement.  Some described their leadership as making this selection because it was in the best interests of the settlement, while others saw less benign impulses and felt the RC was attracted to the CLT model because it would maintain their importance within the settlement.

I have argued that the CLT model was selected because Tanzania-Bondeni residents perceived themselves as very poor and powerless. The settlement's history was marked by an adversarial relationship with the town council while the specter of legalization brought with it an influx of outsiders ready to purchase plots from would-be sellers. Selecting the CLT model was a sensible option for keeping land in the community and protecting the least able. Group tenure, however, was not a foregone conclusion. The ability to select group tenure was only an option because of the pre-existing institutional arrangements in the settlement¾the history of self-help and community cooperation¾that had fostered a sense of trust which is a necessary precursor to collective action.

At the outset I suggested that an understanding of the decision-making process of the residents of Tanzania-Bondeni would provide data for reflecting on the theoretical debate over land rights and African development, as well as helping to inform theoretical perspectives over the nature of human decision-making. To recap, the Voi land tenure decision appears to fly in the face of the ETLR, which would predict that in the face of increasing land scarcity and population growth, residents would select to hold land individually. By implication the decision also appears irrational-moving the economic decision maker away from, not toward, the most utility maximizing institution, namely individual ownership.

Evidence from the interviews indicates that Tanzania-Bondeni residents were clearly acting rationally. Residents selected to hold land in common because this institution was perceived as best serving their individual self-interest. Group tenure was advantageous: it would enable them to get the land more inexpensively and _­­with the provisions for community control and the prohibition on selling land, the CLT model would defend their land against outsiders and retain the land for themselves and their children. The difficulties of cooperation, which were acknowledged by many of those who supported group tenure, were offset by the benefits of assured land access, secure tenure, and promised social support. The rationality of the decision holds even if one believes that the Residents Committee manipulated the decision:  residents still thought the benefits of cooperation (i.e., land and a house) would outweigh the costs of having to cooperate with others.

An understanding of the Voi tenure decision lends credence to the arguments of mainstream institutionalists and economic sociologists that to understand human decision-making we need to understand the embedded context of the decision-maker. In particular, the decision challenges fundamental economic assumptions about human preferences and the nature of costs and benefits. As Bromley has stressed, benefits and costs are not universal - they are socially constructed.[104] They are largely determined by the institutional setting. Tanzania-Bondeni residents were situated in a society characterized by great economic and political inequalities in which they perceived themselves as "at risk" and vulnerable to outside manipulation. They also had a history of cooperation to protect themselves and their possessions from outside threats. Embedded in such a setting, residents saw cooperation as relatively low cost and beneficial, while going it alone - the presumed efficient and more rational decision - was deemed the riskier and more costly option.

Finally, the tenure decision in Voi also appears to undermine a critical dimension of the ETLR, namely the primacy of scarcity as the driver of institutional change. Tanzania-Bondeni residents were acutely aware of the scarcity of land in their country; they often expressed their thankfulness for the upgrading project and the opportunity to get land as a "gift from God." But scarcity did not drive them toward individualism. Rather the specter of scarcity and the fear that they might lose this precious resource lent support to a decision to hold land collectively. Using insights from the Voi experience one might conclude that the commons persist even in the face of scarcity because scarcity is a necessary - but not sufficient - condition for the emergence of private property rights. 

To apply these insights to the debate on African land tenure institutions and reform processes, the Voi experience supports those who argue for the inclusion of a range of tenure options in land policy reforms. Forcing all communities to adopt individual land tenure is not advisable for such an ecologically, ethnically, and economically diverse continent.  Simply providing tenure options, however, is an insufficient approach. Institutional variety should be coupled with procedural approaches for selecting land institutions and overseeing land allocation. Community members know their histories: they understand their internal dynamics, needs, and aspirations. To build on this knowledge, reforms must make provisions for transparent participatory community decision-making processes with sufficient safeguards for the interests of women, minorities, and other marginalized groups. Even with such policy changes, the commons will likely persist and individual tenures will likely expand in Africa - but with this approach they will do so at the behest of communities and not economic theoreticians. 

Endnotes: