Abstract
Security of property rights, or tenure,1 is central to the economics of development and is recognized as important for the adoption and implementation of sustainable forest management. Despite the potential importance of tenure security, this paper shows that there is a great deal of variation in how security is defined and measured, which may be responsible for difficulties that empirical studies have had in linking tenure security with economic behavior. The objective of this paper is to clarify concepts of tenure security that will hopefully allow for improvements in empirical studies. (JEL Q15)
I. INTRODUCTION
Security of property rights, or tenure,1 is central to the economics of development. For example, Besley (1995) states that the evolution of property rights and their effect on investment are “central issues” in the political economy of development, and that specification and protection of rights are important preconditions for economic growth. Tenure security is recognized as important for the adoption and implementation of sustainable forest management (Owubah et al. 2001), which becomes especially important when considering the high rates of tropical deforestation in the developing world in recent decades (Deacon 1999).
Despite the potential importance of tenure security, this paper will show that there is a great deal of variation in how security is defined and measured, which may be responsible for difficulties that empirical studies have had in linking tenure security with economic behavior. The objective of this paper is to clarify concepts of tenure security that will hopefully allow for improvements in empirical studies.
II. A MODEL OF TENURE SECURITY
As will be discussed below, there are a great variety of definitions regarding tenure security. As a basis for deciding on a preferred definition, we begin with the proposition that the concept of tenure security has largely evolved in response to efforts to clarify investment incentives of property holders (e.g., Bohn and Deacon 2000). In a general sense, authors have been concerned that insecure tenure to a resource reduces incentives to invest by reducing the benefits that the resource can yield. As such, tenure security has evolved to be a potentially important determinant in explaining the economic behavior of property holders. It is within this context that we consider concepts of tenure security in the model developed below.
We begin with the concept of property rights. Because property rights have been defined as benefit streams (e.g., Bromley 1991), they may be interpreted as the utility received by a property right holder of a resource over time that arises from practices such as extraction, consumption, or habitation. Thus, property rights may be thought of as arising from the use of resources within the context of social rules such as laws, regulations, and norms (e.g., Demsetz 1967). These social rules specify the conditions of holding tenure to a resource. Therefore the utility derived from a property right will be influenced by the resource asset in question, and the social setting shown by 
[1]
where UR is the present value of utility derived from resource R, or the total value of a given resource property right to the holder over some time horizon T; δt is a discount factor; Rt is resource R at time t, which is a representation of the physical state of the resource including its mass, age structure, location, and other physical features; Ct is a vector of social rules at time t that may constrain or accommodate2 securing the benefit stream that the property right provides; uRt(·) is utility derived from resource R at time t, which is a function of Rt and Ct.3 As described further below, the condition of Rt and Ct in t + 1 can be influenced by choices made in t.
In order to address concepts of tenure security, we introduce stochastic elements to the property right model in equation [1]. Framing property rights in a stochastic framework requires introducing the expected value of utility, E(UR). Introducing an expected value changes equation [1] as follows:
[2]The expected utility of the property right is now shown to be influenced by the starting conditions of the resource (R0) and social rule determinants (C0), as well as a stochastic series of their values over time, r(Rt), and c(Ct), where r(·) = xtβ + εrt and c(·) = ztα + εct.4
These expressions show that the arguments of these functions contain predictable elements, contained in the vectors xtβ and ztα, and random components (εrt, εct). The predictable elements represent features of the resource or social rules that are known and/or influenced by the tenure holder. The random components, though not predictable, may be disclosed to the property holder over time.
To illustrate these expressions, consider investments that influence future benefits that may be derived from the resource.5 Investments in resource management could influence elements in xt, or investments in changing social rules, through lobbying efforts or contributing or assisting with political campaigns, could affect future resource benefits by influencing elements in zt. Because the efficacy of these investments may be uncertain, they are represented as stochastic elements. There may also be stochastic effects that are beyond the control of the tenure holder, which would also be captured by εrt and εct. For example, weather events such as rainfall, or a lack thereof, could be important in influencing r(·). Moreover, unanticipated changes in government policies at national levels (such as decentralization) could change c(·).
Following the specification of equation [2], we can investigate a number of concepts associated with tenure security. Because our purpose is to conceptualize tenure security in a way that informs economic behavior, we proceed by investigating influences of various determinants on the benefit streams that property rights confer. In the following discussions we denote secure and insecure tenures as giving rise to differentiated expected utilities (i.e., respectively, E(URS+) and E(UR S−)).
To begin, we consider the relationship between the more general concept of stability of property rights and tenure security. In equation [2], a stable property right (i.e., a benefit stream), ceteris paribus, could be associated with R0 = Rt and C0 = Ct. A stable flow of resources would represent sustained yield concepts of equal periodic harvests, while a stable vector of rules would represent institutional stability. But neither resources flows nor rules are likely to be stable empirically. Ecological, climate, and economic variability are likely to cause benefits from resource harvests to vary over time, sometimes increasing, sometimes decreasing. Therefore, if one were to equate tenure security with the stability of the property right benefit stream influenced by resources and rules (i.e., E(URS+) iff R0 = Rt, C0 = Ct), then we are unlikely to ever find examples of secure tenures. Instead, empirical concepts of tenure security tend to focus more on the vector of rules.
Tenure security could be defined by looking at changes in expected benefit streams while holding the resource constant and considering different sets of rules. The literature review below discloses that this approach to conceptualizing tenure security is a common practice. For this definition, a secure, versus an insecure tenure would have values of C0 and c(·) such that E(URS+)>E(URS–). But such a definition ignores the differentiated concepts of “substance” and “assurance” put forth by Sjaastad and Bromley (2000), which may be represented, respectively, by the two social rule determinants (C0 and c(·)). Sjaastad and Bromley (2000) assert that it is only the assurance of rights, and not the substance, that truly determines tenure security. As such, two rights with identical substance could differ in assurance and, therefore, levels of security. Similarly, a highly valuable right could be less secure than a less-valuable right. For example, one set of rules could involve very high but entirely stable levels of income taxes, while another set could involve low but variable tax rates. Utility could be greater from the latter even though security is greater in the former.6 Accordingly, the concept of tenure security is best restricted to considering changes in rules, and the starting conditions, C0, are not needed to define tenure security. That is, although both C0 and c(·) are determinants of property right values, c(·) alone represents the assurance aspects, which encompass tenure security.
Following the discussion above, we define tenure security by holding the resource variables and C0 constant, while letting factors in c(·) vary. That is, a secure versus an insecure tenure could be defined as having the following conditions:
[3]
[4]Thus, tenure security arises from expectations that changes in rules over time will increase (or not decrease) utility, while tenure insecurity arises from expectations that changes in rules over time will decrease utility.7
Further to our discussion of r(·) and c(·) above, the expected value of the property right, E(UR), may change in response to a number of influences exerted by the property right holder. For example, an investment, IR, in natural resource productivity could change the value of the property right by increasing the value of the resource. If we assume that r(·) = ƒ(IR) and ∂r(·)/∂IR > 0 and ∂E(UR)/∂r(·)>0, then ∂E(UR)/∂IR>0.8 Similarly, the holder of a property right could invest in attempting to change the social rules governing the benefit stream. Such investments, IC, could be efforts to change social rules (e.g., investing to register land or lobbying to reduce property right constraints) or investments in the resource, IR, which could also change social rules (e.g., planting trees to strengthen the claim to land). Therefore, if we assume that c(·) = ƒ(IR, IC) and ∂c(·)/∂IR,C>0 and ∂E(UR)/∂c(·)>0, then ∂E(UR)/∂IR,C>0.9
In the above discussion, we have defined tenure security as an expectation occurring at a given point in time. But, the ability to influence tenure security with investments highlights the prospect that tenure security is likely to change over time. For example, learning over time could disclose whether returns to lobbying government are successful or not, thereby influencing future expectations of utility. As such, the security of tenures is likely to change as property rights and their granting environments evolve.
Thus far, our definition of tenure security has rested on the assumption that the content, C0, and assurance, c(·), aspects of property rights are independent variables. But as will become evident in the literature review below, some authors imply that they may be correlated. For example, some authors interpret insecure tenures as being synonymous with short durations of property rights. Part of this interpretation could be based on the belief that both aspects of property rights, tenure insecurity and shortened duration, are believed to reduce investment incentives; the first through reduced expectations in benefit streams, and the second because of a curtailed time horizon within which to capture benefits. But these two effects can be considered independently as, respectively, assurance and content. That is, following the definition of tenure security above, we could have secure and short tenures, or insecure and long tenures. In the literature review below, the more common case regarding the relationship between duration and security arises when it is assumed that shorter tenures tend to be more insecure, perhaps because they must be renewed more often than long tenures, and because renewals are positively correlated with changes to rules (i.e., ∂c(·)). But note from equations [3] and [4] that ∂E(UR)/ ∂c(·) may be positive or negative. Therefore, an additional assumption regarding the positive correlation between reduced duration and insecurity is that renewals give rise to rule changes that reduce benefit streams (i.e., ∂E(UR)/ ∂c(·)<0). In sum, in order for shorter duration tenures to be correlated with insecure tenures, it must be assumed that (1) shorter duration tenures are renewable, (2) renewal results in increased probabilities of changes to current rules. and (3) these changes in rules decrease the value of the property right to the holder.
An important aspect of the depiction of tenure security developed above is that it may be deployed on a resource by resource basis while considering the numerous conditions relevant to a given resource. Though much of the tenure security literature refers to “land tenure,” the framework above recognizes that rights are frequently granted separately to land-based resources, each of which may yield its own benefit stream. For example water and timber resources frequently have different owners and differing sets of social conditions, implying that each could be considered, analytically, as a separate benefit stream (i.e., E(UR)). Moreover, as will become evident below, much of the tenure security literature recognizes only a small subset of the potential vector of relevant rules (i.e., Ct) that could potentially condition benefit streams.
III. A LITERATURE REVIEW ON TENURE SECURITY
The literature on tenure security covers a number of different areas. In the review below, we start with definitions and measures of tenure security. Next we look at empirical studies in the literature that have attempted to link tenure security to two types of economic behavior: investments and deforestation.
Definitions and Measures of Tenure Security
Table 1 summarizes definitions and measures of tenure security that have been used in the literature. The table discloses a great variety of concepts associated with tenure security. But notably absent in most studies is a basic definition of tenure security that can be used to attempt to link security to economic behavior. Instead, authors tend to define tenure security in relation to specific conditions that are found, and believed to be important, at their study sites. As such it is difficult to identify common analytical threads that link these studies. But if we employ Sjaastad and Bromley’s (2000) concepts of content and assurance developed in the conceptual model above, we can potentially make some sense of the variability of definitions of tenure security found in the literature.
Definitions and Measures of Tenure Security Used in Previous Literature
There are a number of studies that present definitions based on assurance concepts of tenure security. These studies refer to uncertainty of rights (e.g., Sjaastad and Bromley 2000; Robinson 2005; Gavian and Fafchamps 1996; Li, Rozelle, and Brandt 1998; Owubah et al. 2001; Smith 2004); probability or perceived probability of losing all or part of rights held, including eviction and expropriation (e.g., Besley 1995; Bohn and Deacon 2000; Carter and Olinto 2003; Hayes, Roth, and Zepeda. 1997; Holden and Yohannes 2002; Jacoby, Li, and Rozelle 2002; Otsuka et al. 2001; Place and Otsuka 2000, 2001, 2002; Sjaastad and Bromley 1997); expected time of residence before eviction (e.g., Cattaneo 2001); uncertainty over changes in government policy (e.g., Feder et al. 1992); probability of extension or renewal (e.g., Nautiyal and Rawat 1986); freedom from imposition from outside (e.g., Brasselle, Gaspart, and Platteau 2002); and the expected impact of changes to various individual attributes of a tenure (e.g., Luckert 1991). There are also papers that define security in terms of the substance of rights rather than the assurance of those rights. Examples of definitions describing the substance of rights include duration of rights (e.g., Zhang and Pearse 1996, 1997; Brasselle, Gaspart, and Platteau 2002; Gavian and Fafchamps 1996; Godoy, Kirby, and Wilkie 2001), legal title to land (e.g., Feder and Onchan 1987), renewability of rights (e.g., Zhang and Pearse 1996; 1997), and the right to sell or transfer land (e.g., Gavian and Fafchamps 1996).
The diversity of definitions of tenure security has, not surprisingly, led authors to adopt a great variety of measures. Table 1 summarizes the measures of tenure security used in the reviewed literature. Measures that describe the assurance of rights include the probability of eviction or expropriation (e.g., Carter and Olinto 2003; Jacoby, Li, and Rozelle 2002; Sjaastad and Bromley 1997; Mendelsohn 1994; Holden and Yohannes 2002; Li, Rozelle, and Brandt 1998; Feder et al. 1992), expected time until eviction (e.g., Cattaneo 2001), probability of extension or renewal (e.g., Costello and Kaffine 2008; Nautiyal and Rawat 1986), conflict with abutters or owners of adjacent lands (e.g., Godoy et al. 1998; Godoy, Kirby, and Wilkie 2001), political stability (e.g., Bohn and Deacon 2000; Deacon 1994, 1999), and perceptions of good governance (e.g., Kaufmann, Kraay, and Mastruzzi 2003). Measures of security that describe the substance of rights include the degree to which rights are transferable (e.g., Hayes, Roth, and Zepeda 1997; Brasselle, Gaspart, and Platteau 2002; Kabubo-Mariara et al. 2006; Place and Hazell 1993), legal land title (e.g., Feder and Onchan 1987; Smith 2004; Owubah et al. 2001; Besley 1995), type of land tenure or method of acquisition (e.g., Gavian and Fafchamps 1996; Kabubo-Mariara 2007; Kabubo-Mariara et al. 2006; Otsuka et al. 2001; Place and Otsuka 2000, 2001, 87(2) Arnot, Luckert, and Boxall: What Is Tenure Security? 303 2002; Besley 1995), range of use rights (Brasselle, Gaspart, and Platteau 2002); duration of possession (e.g., Li, Rozelle, and Brandt 1998; Besley 1995; Zhang and Pearse 1996, 1997), previous litigation (e.g., Besley 1995), renewability of tenure (e.g., Zhang and Pearse 1996, 1997), and obligation to share financial returns with government (e.g., Zhang and Pearse 1996, 1997).
Table 1 also discloses frequent differences between the way that tenure security is defined and measured. In many cases, the variance between definitions and measures of security probably occurs because information is not available for the preferred measures. Some authors (e.g., Place and Otsuka 2001; Li, Rozelle, and Brandt 1998) discuss such information problems as justification for the use of tenure content measures that are assumed to be proxies of assurance aspects of tenure security. For example, Li, Rozelle, and Brandt (1998) discuss the fact that their use of length of time holding a plot as a measure of tenure security may be problematic and counter this by adding a dummy variable for the perception that the plot will be taken back at the end of the current crop year. Similarly, Place and Otsuka (2001) state that it would have been ideal to have information on explicit measures of security but that the necessary data were unavailable, and therefore, they use method of acquiring land as a measure for security.
Most authors (e.g., Brasselle, Gaspart, and Platteau 2002; Hayes, Roth, and Zepeda 1997; Owubah et al. 2001; Zhang and Pearse 1996, 1997), however, do not justify their use of security measures that differ from their provided definitions. For example, Owubah et al. (2001) define tenure security as confidence in rights, but then proceed to use the capability to legally register land in the owner’s name as a measure of security. Presumably, these authors assume strong correlations between their definitions and the proxies they use to measure security.
Although definitions and measures of tenures security vary in Table 1, there are some common approaches and assumptions that have been employed. These include the common use of two content measures of property rights as proxies for tenure security: legal title and duration of tenure.
While holding legal title to land may be positively correlated with security of tenure, a number of papers suggest that legal title does not guarantee secure rights. For example, Place and Otsuka (2001) use legal title as a measure of security but state that it would be better to have information on explicit measures of security. The key issue is whether legal title is sufficiently correlated with tenure security such that it is a good proxy. Brasselle, Gaspart, and Platteau (2002) state that tenure security can be obtained only when land is registered and protected by a legal title. But Razzaz (1993) states that “legality of tenure is not necessarily a precondition for security of tenure.” This statement is supported by studies in Thailand by Feder and Onchan (1987) and Feder, Onchan, and Chalamwong (1988), who show that illegal squatters on public lands in Thailand actually have a relatively secure tenure, since they face low lifetime eviction probabilities. In this situation, the Thai squatters have relatively secure tenure despite their lack of legal, governmentendorsed title. Place and Otsuka (2000) find a similar pattern in Uganda on lands under mailo10 ownership. On these lands, only the landowner can obtain legal title to the land, but tenants on the land have very strong rights including protection from eviction (Place and Otsuka 2000). Such results are also evident looking across countries. Deacon’s (1994, 1999) research supports the theory that legal title is not always correlated with security of tenure. If the title is formal, legal, and state enforced, but the government is unstable, then the property rights associated with this title may not be very secure.
A number of studies use duration of tenure as a measure of security (e.g., Zhang and Pearse 1996, 1997; Besley 1995; Godoy, Kirby, and Wilkie 2001). But holding a right for a long period is not necessarily correlated with the probability that the right will change to make the landholder better or worse off. Along these lines, Smith (2004) suggests that a landowner with long duration rights but poor assurance will have different incentives than a landowner with assured but shorttermed rights. For the case of fisheries concessions, Costello and Kaffine (2008) conclude that if the probability of renewal is sufficiently high, then any tenure length is sufficient to incent resource stewardship, while if the probability of renewal is sufficiently low, then no tenure length can incent stewardship. The distinction between duration and tenure security has also been explicitly recognized in the empirical specifications of regressions that authors have chosen. For example, Place and Otsuka (2002) and Li, Rozelle, and Brandt (1998) include duration and separate security measures in their models.
Another notable feature of the definitions and measures of tenure security in Table 1 are evident in the breadth, or lack thereof, of the resources and rules that the studies attempt to address. The vast majority of the studies are broad with respect to the benefits conferred under the concept of “land tenure,” as opposed to identifying specific benefit streams associated with specific resources. Conversely, the breadth of the rules associated with any one benefit stream is frequently defined and/or measured as one small part of the rules that condition benefit streams. The approaches range from depicting tenure security with a single variable (the case for most studies) to the extreme case where Luckert (1991) considers the security of 25 rules associated with a forest tenure in British Columbia.
Impacts of Security on Investments
Most of the papers reviewed assumed that increased security of property rights will lead to increased investment in land (e.g., Besley 1995). Security of property rights is hypothesized to increase investment through two mechanisms. The first is through increased assurance (e.g., low probability of expropriation) that the investors will be able to reap the benefits of their investments (e.g., Besley 1995). The second mechanism is through increased access to funds that can be used for investing (e.g., Feder and Onchan 1987). The latter is hypothesized to occur because having secure, legal title allows occupants to use the land as collateral to obtain institutional credit. However, following the discussion above, it could be argued that access to capital represents content rather than assurance, since the literature discusses legal title to be important as collateral (e.g., Cattaneo 2001; Smith 2004).
Though most papers deal with single-country settings, Bohn and Deacon (2000) investigate cross-country data using a theoretical model of ownership risk. Results indicate that the effects of insecurity vary depending on whether the resource in question requires large amounts of capital investment for extraction. That is, security increases investment in the extraction of resources that require high capital investment (i.e., oil) but decreases investment in the extraction of resources that require low capital investment (i.e., forests). In contrast to this cross-country approach, most analyses concentrate on a particular developed or developing country setting.
There are few papers discussing tenure security and investment in developed country settings, and most of the papers found discuss security in the context of Canadian forest tenures. Nautiyal and Rawat (1986) use a theoretical model to show that tenure security, defined as the probability of a forest tenure being renewed for an additional term, would lead to increased capital investment in firms’ timber processing plants. Luckert and Haley (1990) use differences in perceived security for holders of two tenure types in British Columbia, as measured by Luckert (1988), as a possible reason for the differences in investment in silviculture for two forest tenure types. Zhang and Pearse (1996, 1997) also examine, respectively, the effects of tenure security on investment in silviculture and on the occurrence of not satisfactorily restocked (NSR) lands in British Columbia. These two articles conclude that tenure security is positively correlated with silvicultural investment and negatively correlated with occurrence of NSR land. It should be noted, however, that the variables used in data analysis by Zhang and Pearse (1996, 1997) are the types of tenure, not measures of the tenure holders’ assurance of their rights.
In contrast to the relative shortage of studies examining the effects of tenure security in developed nations, there is a large body of literature discussing this issue in developing countries. Sjaastad and Bromley (1997) develop a theoretical model to show that tenure insecurity in Sub-Saharan Africa leads to suboptimal incentives for investment, but no empirical results are presented. There are, however, a large number of empirical studies examining the relationship between tenure security and investment in land. The results of these studies are summarized in Table 2. In Africa, Besley (1995) (Ghana), Place and Otsuka (2002) (Uganda), Smith (2004) (Zambia), Kabubo-Mariara (2007) (Kenya), Kabubo-Mariara et al. (2006) (Kenya), Gavian and Fafchamps (1996) (Niger), and Hayes, Roth, and Zepeda (1997) (Gambia) all found positive relationships between tenure security and investment. Similar results were found in Thailand by Feder and Onchan (1987); in Paraguay by Carter and Olinto (2003); and in China by Li, Rozelle, and Brandt (1998), and Jacoby, Li, and Rozelle (2002). But in contrast to these papers, Place and Hazell (1993) conclude that tenure security had only a positive impact on some (mostly long-term) improvements in some of the regions studied in Sub-Saharan Africa.
Empirical Results on the Impacts of Tenure Security on Investment in Developing Countries
However, as we discussed before, many of these results should be interpreted with caution given the variables used to measure security in analyses. Of the papers reviewed that concluded positive impacts of tenure security on investment in developing countries, only Jacoby, Li, and Rozelle (2002) used a measure of the assurance of landholders’ rights (estimated risk of expropriation). They found that the reduced risk of expropriation increased use of organic fertilizers. Results reported by Place and Otsuka (2002), Carter and Olinto (2003), and Gavian and Fafchamps (1996) were for differences in agricultural investment on land held under different tenure types. Feder and Onchan (1987) and Smith (2004) found positive influences of legal title on fixed investments and land improvements, respectively. Hayes, Roth, and Zepeda (1997) concluded that increased transferability of rights increased the propensity to make fixed investments and also increased the probability of finding trees on the tenure holder’s land. Kabubo-Mariara (2007) and Kabubo-Mariara et al. (2006) found that private ownership of land and increased rights of transfer increased soil and water conservation. The mixed results found by Place and Hazell (1993) were found using complete transfer rights as a measure of security.
In contrast to this part of the literature that concludes that increased security leads to increased investment, there are studies that find no relationship and/or opposite results. For example, Otsuka et al. (2001) found a statistically insignificant effect of tenure type on tree planting and tree crop yields in Sumatra, and Hayes, Roth, and Zepeda (1997) found that the most transferable rights category had an insignificant influence on medium-term improvements, while the second most transferable category had a significant negative effect on medium-term improvements. Similarly, Feder et al. (1992) found that the perceived likelihood of contract disruption in the short term and perceived likelihood of retaining the same land in the long term had only a significant positive effect on crop-related investment and housing investment in one of four regions studied in China. Holden and Yohannes (2002) found that the perception of tenure security had no significant impact on purchase of farm inputs and planting of perennials. Brasselle, Gaspart, and Platteau (2002) concluded that there was no significant correlation between investment and the range of use and transfer rights held by farmers in Burkina Faso. Again, of the above papers, only those by Feder et al. (1992) and Holden and Yohannes (2002) used measures of security that account for the assurance of rights rather than the substance of rights.
Yet another problem with these empirical studies arises when considering the direction of causality regarding tenure security and investments. It is suggested in many papers that there may be a reciprocal causality between tenure security and investment (e.g., Robinson 2005; Sjaastad and Bromley 1997; Besley 1995; Godoy et al. 1998; Godoy, Kirby, and Wilkie 2001; Otsuka et al. 2001; Place and Otsuka 2002; Southgate, Sierra, and Brown 1991; Brasselle, Gaspart, and Platteau 2002; Place and Hazell 1993). If this reverse causality does exist, then it becomes reasonable to assume that tenure insecurity can cause increased investment if investments allow individuals to obtain more secure rights to the land. Otsuka et al. (2001), for example, found that in Sumatra, relatively strong transfer rights are granted to people who clear communal forests and to people who plant trees. Similarly, in Uganda, where evicted tenants must be compensated for lost investments, high-valued investments reduce the likelihood of eviction, thus increasing security of tenure (Place and Otsuka 2002). Robinson (2005) illustrates an example in Karnataka, India, where some farmers chose to invest in permanent improvements with the belief that it would reduce the likelihood they would be evicted by government, while other farmers avoided any kind of permanent investment because of the risk of eviction. Brasselle, Gaspart, and Platteau (2002) initially found that security (measured by the breadth of use and transfer rights) significantly influenced investment. But, once they allowed for endogeneity between security and investment, they found that security had no significant impact on investment, while investment had a significant positive impact on security. It should be noted, however, that tenure security may not always be a function of investment. For example, Hayes, Roth, and Zepeda (1997) defend their decision to model security and investment as exogenous by stating that in Gambia, tenure security is dependent on the method of acquiring land and is, thus, dependent on factors beyond the control of the individual.
Although many of the papers reviewed discuss the possibility that this reverse causality may exist and discuss its implications for data analysis, very few take this endogeneity into account in their modeling. The only paper reviewed that carried out empirical analysis allowing for endogeneity of security and investment was that of Brasselle, Gaspart, and Platteau (2002).
Impacts of Security on Deforestation
Whereas impacts of tenure security on investments have been investigated in both developing and developed countries, the impact of tenure security on deforestation is discussed in a developing country context. Most of the papers reviewed assumed that increased security of property rights will lead to decreased deforestation (e.g., Cattaneo 2001). Table 3 provides a summary of conclusions from reviewed papers regarding the effects of tenure security on deforestation. In a cross section of countries, Deacon (1994) and Bohn and Deacon (2000) found a positive relationship between political instability and deforestation, and Deacon (1999) found a similar relationship in an examination of historical records of the past 3,000 years. These conclusions are also supported by more regionally focused studies such as that of Cattaneo (2001), who found that expected time of residence until eviction and rate of deforestation are negatively correlated in Brazil, and Southgate, Sierra, and Brown (1991), who found a negative relationship between adjudicated (i.e., legally recognized) agricultural land and deforestation.
Conclusions on the Impacts of Tenure Security on Deforestation
There are also studies, however, with conclusions that do not support these results. Godoy et al. (1998) found contrary results in Bolivia, where conflict with ranchers significantly increased deforestation, while conflict with loggers and smallholders had no significant impact on deforestation. Conflict with all types of abutters (ranchers, loggers, smallholders, and an oil firm) was found to be jointly significant. In a later paper, Godoy, Kirby, and Wilkie (2001) found no significant effects of conflict with abutters (adjacent land users) or duration of residence on deforestation in Bolivia. Owubah et al. (2001) found that capability to register land did not significantly affect farmers’ conservation decisions for natural forests or establishment of tree plantations in Ghana. In Malawi, Place and Otsuka (2001) found that method of acquiring land did significantly affect deforestation in some regions. However, the effects were positive in some regions and negative in others. In Sumatra, Otsuka et al. (2001) concluded that the type of land ownership had no significant impact on deforestation.
There are several reasons that may explain why the results in these papers appear to be inconsistent. First, as with the earlier discussion of investment, differences in results for the impact of tenure security on deforestation may be explained by differences in how tenure security is measured. The only papers discussed in the previous paragraph that used measures of security that reflect the assurance of a landholder’s rights are those by Bohn and Deacon (2000), Cattaneo (2001), Deacon (1994, 1999), Godoy et al. (1998), and Godoy, Kirby, and Wilkie (2001). All of the papers that use measures of the assurance of rights found negative correlations between security and deforestation, except for that by Godoy, Kirby, and Wilkie (2001), who found no significant effect.
Measures of tenure security based on the substance of rights include duration of residence (Godoy, Kirby, and Wilkie 2001) (in conjunction with conflict), type of land tenure (Otsuka et al. 2001), capability to legally register land (Owubah et al. 2001), method of acquiring land (Place and Otsuka 2001), and adjudicated agricultural land (Southgate, Sierra, and Brown 1991). Of those papers using measures of the substance of rights, all found insignificant impacts of their measures on deforestation, except for that by Place and Otsuka (2001), who found positive correlations in some regions and negative ones in others.
A second explanation for the disparity in results is differences in how investment is defined in relation to the activity of deforestation. If one accepts the hypothesis that increased security of property rights leads to increased investment, then this could cause increased or decreased deforestation. This occurs because in some situations both planting trees and clearing forests can be viewed as forms of investment in land (Otsuka et al. 2001). Planting trees on land can be seen as a long-term commitment to the productivity of the land (Sjaastad and Bromley 1997), while clearing forestland can also be seen as an investment in land, especially if the land is cleared for agricultural use or to claim the land (Godoy, Kirby, and Wilkie 2001). Angelsen (2007) discusses this issue and develops a theoretical model showing that tenure insecurity has a direct negative effect on the net present value of cleared forest land, reflecting the risk of losing land and a positive indirect effect on the net present value of such land through increased rent. Angelsen (2007) concludes that it is impossible to determine the net effects of insecurity on a theoretical basis, corroborating the discussion above that suggests that tenure insecurity can lead to increased investment.
IV. SUMMARY AND CONCLUSIONS
The literature on tenure security has had difficulty in establishing empirical links to economic behavior regarding investments and deforestation. Our review of the literature discloses a number of potential reasons behind this problem. First, the definition of tenure security is not clear. In most cases, authors do not specify a definition of tenure security, and in the few cases where definitions are offered, a wide range of concepts is used. If the objective of defining tenure security is to further our understanding of economic behavior, we argue, following Sjaastad and Bromley (2000), that content and assurance aspects of tenures should be distinguished. It is the assurance aspects of tenures, and their impacts on expected benefit streams, that relate to concepts of tenure security. But the fact that the literature frequently uses both content and assurance aspects in attempting to address tenure security points to a second potential area of confusion: the measures of content and assurance may be correlated.
The potential correlation of content and assurance aspects of tenures is both a blessing and a curse. Given the difficulties associated with obtaining data on assurance, most empirical studies have adopted content measures as proxies (e.g., duration of tenures and land titles). Therefore, potential correlations between content and assurance have allowed empirical studies to proceed despite the absence of assurance data. But the lack of clarity regarding definitions of tenure security has caused the literature to fail to explicitly recognize potential problems with settling for less than ideal measures of tenure security. In particular, evidence has emerged that suggests that the content proxies may, in some cases, be poor measures of tenure security, thereby leading to the misspecification of empirical estimations.
A third problem arises from the potential endogeneity of tenure security and economic behavior. Though numerous authors have acknowledged the potential for investments and deforestation to impact tenure security, very few studies have addressed these issues in their empirical models. The result, again, is the potential for misspecification of empirical estimations.
A fourth problem arises from difficulties associated with modeling the breadth of resources and rules associated with tenure security issues. Despite the fact that lands frequently produce multiple types of benefit streams, each with its own sets of rules, research has frequently failed to distinguish between these potential differences in resource tenures. Moreover, the complexity associated with the breadth of rules, which we find in practice, is generally proxied by one or just a handful of assurance and/or content variables. Again, without careful consideration regarding issues associated with choices of proxies, the potential for misspecification is rife.
Connecting concepts of tenure security to economic behavior is a vexing problem. Any individual study site is plagued with numerous natural and social characteristics that make identifying impacts of tenure security on economic behavior a Herculean task. But facing such complexity will require researchers to start from a clear conceptual foundation about what tenure security is, and how it can be measured. Only then will we be able to assess the problems and promise of using alternative proxies to investigate tenure security. Until we clarify these beginnings, we are unlikely to progress toward understanding the importance and specific impacts of tenure security on economic behavior.
Acknowledgments
Funding for this work was provided by the Sustainable Forest Management Network. Thanks to the reviewers for their comments, including Philippe Marcoul, Sandeep Mohapatra, and James Rude.
Footnotes
The authors are, respectively, graduate student; professor; and professor, Department of Rural Economy, University of Alberta, Edmonton, Canada.
↵1 Following common use in the literature, we use the terms property rights and tenure interchangeably. Specific definitions are presented below.
↵2 Although the economic literature on property rights frequently considers conditions of holding property rights as constraints (i.e., as “attenuations,” see Pejovich 1990), these conditions may sometimes also include subsidies that increase the value of property rights. For ease of exposition, in the discussion that follows we follow the convention that considers rules as constraints.
↵3 Note that Rt and Ct are not likely to be independent, as rules are thought to evolve in response to characteristics of resources (e.g., Pearse 1988). This lack of independence, however, does not influence the definitions of security that follow.
↵4 The dynamic stochastic equations could assume a number of different forms. A more precise specification would include defining states of nature and probabilities associated with each, and characterizations of utility associated with the various outcomes. However, we maintain a general form here, as the precise specification does not influence the development of the concepts of security that follow.
↵5 We introduce investments more explicitly into our formulation below.
↵6 We are grateful to a reviewer for this example.
↵7 Note that the expected utility that arises from changes in rules will depend on the risk preferences of the property right holder. To the extent that property holders are risk averse, the mere presence of variability will decrease utility, while expected variability regarding negative changes will amplify reductions in utility and tenure insecurity.
↵8 More specifically, if expected utility and investments are expressed in common units (i.e., U.S. dollars), then ∂E(UR)/∂IR>1, or the investment would not be undertaken.
↵9 As was the case above, if expected utility and investments are expressed in common units (i.e., U.S. dollars), then we can expand the conditions for investment to include conditions and resources; if ∂E(UR)/∂IR,C>1, the investment will be undertaken.
↵10 Mailo ownership is the major form of land tenure in the Buganda region of central Uganda. Mailo lands are lands that were given by colonialists to “notables and elites,” beginning in the early 1900s (Place and Otsuka 2000).
