Seeds of peace? Land reform and civil war recurrence following negotiated settlements

Abstract

Land reform has been depicted by some as an effective element of counterinsurgency strategy in nations experiencing peasant-based civil conflict. While some studies have argued that land reform reduces civilian support for insurgency, other research has demonstrated that these reforms are often undermined by brutal state repression. The study of land reform has also been driven largely by qualitative case study research, which has limited what we know about the cross-national efficacy of these reforms. This study contributes to the current literature by looking at the efficacy of land reform as part of the post-civil war peace process. Specifically, we examine whether land reform provisions included in comprehensive peace agreements reduce the risk of renewed civil war. Measuring the risk of civil war recurrence in all comprehensive peace agreements from 1989–2012, we find that the inclusion of land reform provisions in the post-war peace process substantially reduces the risk of renewed fighting.

Keywords

Civil war recurrence land reform peace agreements

No social group is more conservative than a landowning peasantry, and none is more revolutionary than a peasantry that owns too little land or pays too high a rental. (Samuel Huntington, Political Order in Changing Societies, p. 375)

Huntington’s quote alerts us to the fact that a number of studies have examined the relationship between inequitable distribution of land ownership and land tenure patterns and the onset of rural insurgency (Boix, 2008; Moore et al., 1996; Muller and Seligson, 1987; Paige, 1975). Huntington preceded the quote above by prescribing land reform as a strategy to inoculate rural populations against the temptation to revolt:

[A] government can … significantly affect the conditions in the countryside so as to reduce the propensity of peasants to revolt. While reforms may be the catalyst of revolution in the cities, they may be a substitute for revolution in the countryside. … Where the conditions of land ownership are equitable and provide a viable living for the peasant, revolution is unlikely. Where they are inequitable and where the peasant lives in poverty and suffering, revolution is likely, if not inevitable.

Rural insurgencies fought over land issues often become intractable. Rebels can find safe havens in rural areas, where they can avoid annihilation by the government’s armed forces, even if they cannot defeat the government and seize power themselves. They can cultivate ties with local peasant populations by promising land reform when they take power. The overwhelming majority of rebel fighters are themselves drawn from the peasant population as well, giving the rebels kinship, community, class, and other local ties that translate into local knowledge that government forces often lack and cannot easily cultivate, given this type of state’s reliance on repression as a strategy to preempt peasant unrest prior to the outbreak of civil war.

Boix (2008) argues that in a nation with an economy that is heavily based on agriculture and where land ownership and land tenure arrangements are highly inequitable, the land-owning class is likely to resist calls for economic reform (i.e. land reform) or political reform (i.e. democratization) because their primary asset—land—is not mobile: they cannot move their assets to another nation if a democratic regime decides to tax their wealth at a rate that land owners find unacceptable. Hence, Boix argues, the land-owning class is willing to fight a civil war rather than allow land redistribution or democratization. He then goes on to specify the conditions that determine whether the rebels win or the government wins the civil war.

What Boix does not consider is the outcome that has become the modal outcome of civil wars in the post-Cold War era: negotiated settlement. What we address in this article is not what makes a negotiated settlement more likely than a military victory by either side. Instead, we focus on the aftermath of conflict termination through negotiated settlement: the duration of the peace established by a peace agreement. In civil wars that end in negotiated settlements, does the inclusion of land reform provisions in the peace agreement reduce the risk of relapse into renewed conflict?

Existing research has shown that once a nation brings a civil war to an end, that nation has a high probability of relapsing into renewed civil war at a later date. In Breaking the Conflict Trap, Paul Collier and his colleagues noted that about half of the nations that experience civil war do eventually relapse into renewed conflict within a few years after the original war ends (Collier et al., 2003: 83). Licklider’s (1995) seminal work on civil war recurrence cautions us that there are strong reasons to expect that the peace established by a negotiated settlement is more fragile than the peace established by a decisive victory, regardless of whether it is the government or the rebels that prevail. Empirical evidence on this proposition is mixed, especially for the post-Cold War era. While Licklider (1995) and Toft (2010) found support for the notion that military victories produce a more durable peace, Walter (2004) and Kreutz (2010) found differences in the risk of peace failure following military victories versus negotiated settlements. Mason et al. (2011: 184–185) found that military victories produce a more durable peace than negotiated settlements, but that the difference declines over time: after about 14 years the risk of peace failure is lower following a negotiated settlement than a military victory. And, while Quinn et al. (2007: 184–188) found some evidence that military victories produce a more durable peace than negotiated settlements, they did find that peace agreements supported by UN peacekeeping forces are less likely to experience civil war recurrence than conflicts ending in a government victory. In general, this body of research has centered on identifying factors—attributes of the nation, characteristics of the now-ended conflict itself, and features of the post-conflict political, economic, and social environment—that are associated with an increased or decreased risk of peace failure. Our focus in this article is on the question of whether including provisions for land reform in a peace agreement reduces the risk of civil war recurrence. In civil wars where land inequality was implicated as a causal factor in civil war onset, we would expect the resolution or at least the moderation of those land issues to be a matter that, first, would be included in the terms of the peace agreement and, second, would contribute to a more durable peace by diminishing if not removing from the post-war environment one source of grievances that motivated peasant support for the rebels in the first place. Including land reform provisions in the peace agreement represents a costly signal on the part of the government, demonstrating its commitment to the peace process by agreeing to reforms that are damaging to its primary constituency: large land owners. In this manner, land reform provisions alleviate to some degree credible commitment problems that make rebels reluctant to sign on to a peace agreement.

We analyze the pacifying effects of land reform provisions by, first, presenting a theoretical framework on how land issues and land reform affect the risk of peace failure following a negotiated settlement to a civil war. We then test propositions derived from this theory with data on peace spells following civil wars that ended in negotiated settlements. Data on those peace spells are derived from the UCDP Conflict Termination Dataset (Kreutz, 2010), and data on the terms of peace agreements are derived from the Peace Accords Matrix Implementation Dataset (Joshi and Darby, 2013; Joshi et al., 2015). We present a series of hazard models with peace failure as the dependent variable and a conventional set of variables that previous research has shown to be related to the risk of peace failure. We add to that set of predictor variables measures of whether or not land reform provisions were included in the peace agreement and whether or not those terms were implemented. We find that the inclusion of land reform provisions in a peace agreement does indeed reduce the risk of peace failure. While the implementation of land reform provisions does not have a statistically significant effect on the risk of peace failure, the implementation of a broader range of accommodative reforms does have a pacifying effect. Thus, including land reform provisions does appear to serve as an effective costly signal of the government’s preference for sustaining the peace rather than returning to war, and the implementation of more accommodative provisions in the peace agreement, regardless of whether land reform provisions are one of these, reinforces rebels’ perception of the credibility of the government’s commitment to the peace process. The results of our analysis suggest that policymakers should focus not only on crafting peace agreements that promote security, but also address the underlying grievances that spurred the conflict to start in the first place.

Land tenure and civil war

In nations where agriculture constitutes a large share of the economy, and a substantial share of the labor force is engaged in agriculture, conflict can arise over issues of inequality in the distribution of land ownership and the land tenure patterns associated with such inequality. It is not inequality per se that drives peasant grievances, but the patterns of land tenure that define how much land peasant households have access to and what terms of trade and distribution of the risks of agriculture prevail between landed patron (the land-owning class) and peasant clients (the cultivating class). High levels of inequality in land ownership combine with the lack of occupational alternatives to agriculture to reduce most peasant households to a state of economic marginality. The risk of agriculture—that is, those events in weather, climate and markets over which peasants have no control but which can dramatically affect the size and value of their crops—confront peasant households with the constant risk of subsistence crisis. This risk compels them to seek the patronage of landed elites who can provide them with insurance against the risks of a subsistence crisis in the form of access to land and other benefits that amount to a ‘subsistence floor’ (Scott, 1976). Those guarantees include some mix of benefits such as rent delays and emergency food in bad crop years, emergency health care, seasonal wage labor work on the patron’s lands, production credit, seeds, and the use of equipment. In exchange for use of a plot of land and subsistence guarantees, peasant cultivators are expected to provide the landed patron with some mix of rent, crop shares, free labor, and other services when the patron demands them. They are also expected to support the patron politically by deferring to his instructions on matters such as whom to vote for in elections (Mason, 2004: 60–63; Scott, 1976; see also Joshi and David Mason, 2007: 398–399).

Given the grave consequences of a substance crisis, peasants are often willing to submit to highly inequitable terms of trade with a landowner, terms that virtually guarantee their perpetual poverty and dependence on the patron, because each client is far more dependent on the benefits provided by the landed patron than the patron is on the goods and services that any one client can provide him. The loss of a patron’s subsistence guarantees could be disastrous for the cultivator’s household. Without access to land, they would be cast into the pool of landless laborers, devoid of any subsistence guarantees and exposed to the uncertainties of the markets for land, labor, and food. By contrast, the defection of one client from a landed patron’s domain has little if any effect on that patron’s well-being because the defector can easily be replaced from the same pool of landless laborers.

The terms of trade and distribution of risks that prevail between landed patrons and peasant clients define the extent of peasant economic dependence on and political subordination to the landed patron. These terms vary according to whether the peasant operates as a smallholder, tenant, resident worker on a large estate, or landless wage laborer (Mason, 2004: 63–69; see also Paige, 1975). Smallholders are not directly dependent on landed patrons for access to land. However, their marginal economic status usually compels them to seek assistance from landed patrons for credit, seed, tools and equipment, and seasonal wage labor. All that stands between the smallholder and landlessness is one bad crop that compels him to sell his land or seek loans from the local patron under terms dictated by that patron. Resident workers on large estates are compensated not in wages but in housing and use of a household subsistence plot, as well as other forms of benefits that amount to a fairly secure subsistence floor. However, these ties of dependency bind them to the estate owner in a form of near serfdom. Tenancy arrangements can be based on fixed rents in cash or in kind, or on crop shares. The terms of the tenancy vary depending on the size of the landless population: the larger the landless segment of the rural population, the more unfavorable the terms of rental and tenancy arrangements will be because a large landless pool makes it easier to replace any given tenant who objects to the rental terms. Landless laborers exist outside of the network of patron–client exchanges. Their income is determined by the supply and demand for agricultural labor (see Joshi and David Mason, 2008; Mason, 2004: 63–69; see also Paige, 1975: 18–40).

Paige (1975) and others (e.g. Mason, 1986a, 2004) have theorized that the willingness of peasant cultivators to participate in dissident collective action, including armed rebellion, will vary according to the land tenure patterns under which they labor. Those embedded in land tenure patterns that leave them the most dependent on the patronage of the landed elite are less likely to participate in a rural insurgency or any sort of dissident collective action for fear of being evicted from their land and losing the patronage benefits that are their insurance against subsistence crisis. These include the various tenancy and sharecropping arrangements, as well as resident laborers on large commercial estates. Those who operate in land tenure categories marked by weaker ties of dependence on landed elites are more willing and able to participate in such movements. These categories include smallholders as well as landless laborers. Joshi and David Mason (2008) found that the higher the proportion of the population farming under the most dependent land tenure arrangements, the higher the voter turnout but the lower the vote for pro-reform parties (Joshi and David Mason, 2008, 2011). This is indicative of peasant subordination to the political preferences of their landed patron.

Boix (2008) presents a theoretical framework to account for the conditions under which civil war is likely to break out in a nation where agriculture constitutes a major share of the economy and a large share of the population depends on the agricultural economy for their well-being. He presents a model to explain how civil war becomes more likely in nations where land is the predominant asset in the economy and land ownership is highly unequal. In these circumstances the small land-owning class will prefer an authoritarian regime to a democratic one because under democracy, the poor will be able to elect a government that taxes the wealthy in order to provide public goods to the poor. The wealthy cannot credibly threaten to move their assets out of the country to avoid higher taxes: land is immobile. Therefore, the elite will resist—with repression when required—any social movement among the poor that is intended to redistribute land and/or to raise taxes on the wealthy in order to enhance the economic well-being of the poor through the provision of public goods. The landed class will act to maintain an authoritarian state that represses movements of the poor, but only so long as the costs of repression are low.

Although Boix does not discuss factors that increase or decrease the costs of repression, we can expect that those costs would be lower in agrarian economies where larger shares of the peasant population labor under dependent forms of land tenure, such as sharecropping and rental arrangements or as resident laborers on large estates. Under these land tenure patterns, the landed elite can preempt dissident movements among peasants by binding them in networks of clientelist dependency, whereby peasant political support (or at least quiescence) is preserved by providing them with subsistence guarantees. Where clientelist networks are more extensive and stable and incorporate a larger share of the rural population, the repression costs of preserving the dominance of the landed class against dissident movements should be relatively low. Where these networks are weak and larger shares of the peasant population exist outside of clientelist networks—that is, the share of the population that is either landless or smallholders—repression costs should be higher, because the landed elite cannot rely on ties of clientelist dependency to maintain peasant subordination without the overt use of force.

Inequality in land ownership is a condition that evolves over time. It is, like most structural features of a nation’s political economy, an attribute that changes only at a glacial pace. Hence, inequality in land ownership can account for which nations are at risk of the sort of peasant-based conflict explained in Boix’s theory, but it cannot explain when such conflict will occur. Some studies have argued that changes in agricultural production from cultivation for subsistence and for local markets to production for export markets exacerbates land inequality by creating incentives for large landowners to displace subsistence cultivators from the land and shift that land to the production of export crops, livestock, and other products that require fewer workers per unit of land. The result is that smallholders, renters, and sharecroppers are displaced from the land and cast into an ever-growing pool of landless laborers. These shifts are exacerbated by population growth, which increases land pressures generally and further swells the ranks of the landless population that is outside of the patron–client networks of dependency by which landlords control peasants and prevent the emergence of peasant collective action (Mason, 1986b, 2004: 74–79). As those manipulable ties of dependency erode, elites’ ability to control peasant behavior through patronage erodes, and the landed elite must devote more resources to overt repression. As the landed class escalates its use of repression, the incentives for peasants to engage in dissident collective action aimed at reform increase as well. Boix argues that those who control the land will resist efforts at land reform and democratization, but at some point increasing application of repressive violence against peasants will increase the risk of their shifting strategy from nonviolent forms of collective action to insurgent violence (Mason and Krane, 1989). Albertus and Kaplan (2013: 201) argue that these changes in the rural political economy can lower the opportunity costs for joining a rebellion and increase peasants’ estimate of the potential gains from supporting an insurgency that promises land reform (Joshi and David Mason, 2008; Mason, 1986a).

Civil war, land inequality, and peace agreements

Once a land-based civil war is underway, it can end in a rebel victory that results in the redistribution of land ownership (e.g. the Maoists’ victory in China, Castro’s victory in Cuba, the Sandinista victory in Nicaragua), or it can end in a government victory that preserves (or restores) pre-war patterns of land ownership. If neither the government nor the rebels can prevail over their rival, the war can end in a negotiated agreement (e.g. El Salvador, the Contra War in Nicaragua, the Maoist insurgency in Nepal).

Where such conflicts end in a negotiated settlement, land reform provisions are likely to be included in the terms of the peace agreement. Typically, land reform programs involve the redistribution of land ownership from large land owners to peasant farmers with some formula for compensating those whose lands are redistributed, with the state providing financing for the transactions. For instance, the Salvadoran land reform program called for in the Chapultepec Agreement resulted in the transfer of 103,300 hectares of farmland (about 10 percent of the nation’s total) to former combatants on both sides and to civilian supporters of the FMLN. These lands included properties owned by the state, properties that exceeded in size the 1983 constitutional limit of 245 hectares, and some properties in ‘conflictive zones’ that were offered for sale voluntarily. Land owners were compensated at market price for their land as determined by the 1980 land reform, with recipients paying off their mortgage over 30 years after a 4-year grace period (de Bremond, 2007: 1538).

As noted earlier, most if not all provisions for land reform in comprehensive peace agreements (CPAs) provide for compensation to the land owners whose land is redistributed. The conflict itself increases their incentives to sell off their land under the terms of the peace agreement. The costs of conflict they and their allies in the state have to absorb increase with the duration of the conflict. Moreover, the war itself serves to disrupt agricultural production, land tenure patterns (i.e. the distribution of peasant cultivators among different forms of production), and even de facto land ownership patterns. People are displaced from the land by war. Physical separation ‘changes, terminates, or puts on hold’ pre-war claims to land ownership and production (Unruh, 2003: 357). When rebels are able to seize control of territory in the countryside, they often expel land owners and redistribute land among local peasants as a way to break the power of the land-owning class and consolidate the support of peasant populations in the region. Sendero Luminoso insurgents did this in the southern sierra region of Peru (McClintock, 1984: 63; Mason, 1998: 222).Maoist insurgents followed this strategy in Nepal (Joshi and David Mason, 2010: 989). Thus, the inequality in land ownership that motivated the rebellion in the first place is altered by the war itself, and these changes make it more likely that the landed class will accede to land reform in the peace agreement.

When the war ends in a negotiated settlement, the de facto redistribution of land versus the pre-war legal claims to the same land becomes a primary source of contention to be resolved in the peace process. Unruh (2003: 357–358) argues that the extent to which the pre-war pattern of land ownership and land use remain intact at war’s end will vary depending on the duration of the conflict: the longer the war lasts, the more the pre-war landownership patterns will be disrupted, and the more resistant to a return to the status quo ante will be those who gained access to land as a result of ownership displacement during the war. Likewise, landowners who have had production disrupted by the conflict, especially those whose lands are in conflict zones, should be more willing to submit their lands for sale under the terms of the land reform rather than undertake the task of rebuilding production capacity disrupted by conflict.

Does the inclusion of land reform provisions in a peace agreement affect the risk of peace failure following a negotiated settlement to end a civil war? And how critical is the implementation of those provisions (as opposed to their mere inclusion in the agreement) to the durability of the peace? There are reasons to expect that the inclusion of such provisions in the peace agreement and their implementation should reduce the risk of a relapse into armed conflict. Redistributing land to rebels and their civilian supporters serves to relieve a major grievance that motivated their participation in the rebellion in the first place. For those who receive land under the terms of the program, land reform also raises the opportunity costs of returning to conflict as opposed to abiding by the peace agreement: under the land reform provisions of the peace agreement, a segment of the rebel’s support base would gain tangible assets. A resumption of armed conflict could put their claims to land in jeopardy, given the uncertainties surrounding the outcome of a renewed conflict. Should they find themselves on the losing side of a renewed war, the land reform provisions of the original peace agreement could be revoked or amended to their disadvantage. For instance, following the 1996 Abidjan Peace Agreement in Sierra Leone, land reforms were a conspicuous part of the settlement. After fighting renewed, though, these provisions were eventually dropped from the Lomé Peace Agreement that settled the dispute. The failure of the Abidjan agreement may explain, in part, the decision of the Kamajor militias, who depended heavily on the land, to engage in armed resistance against the RUF and AFRC operating out of Freetown. Moreover, the resumption of armed conflict would put beneficiaries’ newly acquired farm production at risk of the destructive and disruptive effects of war and put them at risk of being displaced from their land if it lies in a conflict zone. Thus, land reform beneficiaries have incentives to withhold their support from a movement to return to war.

H₁: the inclusion of land reform provisions in the peace agreement reduces the risk of peace failure in the aftermath of civil war.

The rebel organization’s broader goal is to manage a peace process that enables them to transition from an armed insurgent organization into a viable political party. The primary goal the FMLN sought from the peace negotiations in El Salvador was the establishment of a post-war environment that would enable them to make the transition from an armed insurgent organization to a competitive political party. This goal would mean that their priority was securing provisions that would dismantle the repressive machinery of the state, reform the Salvadoran military, provide security guarantees for FMLN fighters during disarming and demobilization, and codify political guarantees for the FMLN’s participation in the political process. In return, the FMLN agreed to a more modest land reform program that would include adequate compensation for land owners whose land was appropriated for redistribution (de Bremond, 2007: 15423).

Even though land issues are among the most salient grievances that drove the resort to violence in the first place, the peace agreement and its implementation tend to prioritize other provisions that are of more immediate and direct relevance to preventing a relapse into armed conflict: dismantling a repressive police and military apparatus and disarming, demobilizing and reintegrating rebel armed forces. These measures are critical to dismantling the capacity to resume armed conflict to the extent that the disarming, demobilizing, and reintegration (DDR) of rebel fighters is implemented. Likewise, provisions to reform, purge, and restructure the state security apparatus are the other side of the DDR coin, reducing the capacity of the state to visit repressive violence on supporters and the rebel organization.

Land reform provisions are intended to reduce the incentive of rebels to resume armed conflict. Land reform remains a critical element of the peace agreement for both practical and symbolic reasons. Practically, land reform can ameliorate to some extent the inequality in land ownership that fueled violence in the first place. In most cases, however, it is unlikely that land reform provisions of a peace agreement will redistribute enough land to enough households to accommodate most of those who supported the rebels because of land issues. In most agrarian economies, there simply is not enough land to provide every peasant household with enough land to be economically viable as farmers. However, Albertus (2015: 4–5) cites studies showing that land reform enhances the economic well-being even of non-beneficiaries in rural areas by stimulating economic growth and increasing demand for rural labor, creating upward pressure on wages and thereby stimulating growth in both farm and non-farm income. Thus, even peasants who do not receive land have less incentive to support or participate in a resumption of armed conflict. Albertus and Kaplan (2013) found that where widespread redistribution of land was concentrated in areas that had experienced higher levels of rebel activity, the level of rebel activity declined substantially after land reform. Where land redistribution was less extensive (due to landowner resistance), the level of rebel activity actually increased after land reform. This suggests that, whatever the limitations of land reform provisions in the peace agreement, if redistribution is concentrated in areas with high levels of rebel activity during the war, land reform should reduce the risk of peace failure in the aftermath of war.

More generally, this is an implementation question: if land reform provisions are included in the peace agreement, this represents a concession to the rebels and their supporters. But if it is not effectively implemented, those provisions amount to little more than ‘cheap talk’, and the risk of a resumption of civil violence should increase as implementation lags. In El Salvador disputes about the implementation of land reform in 1992 led the government to halt the restructuring of its security forces and the FMLN to halt the disarming and demobilizing of its armed forces. Until the implementation issues could be resolved, the peace was in jeopardy (de Bremond, 2007: 1544).

H₂: the implementation of land reform provisions should reduce the risk of civil war recurrence.

We turn now to the research design used to test these propositions.

Research design

To examine the effects of land reform on the durability of post-civil war peace, we investigate how the inclusion of land reform provisions in negotiated settlements shapes the risk of civil war recurrence following the establishment of a CPA. Our sample of CPAs is drawn from Joshi et al.’s (2015) Peace Accords Matrix Implementation Dataset (PAM_ID). As noted by Joshi et al. (2015), the PAM_ID includes detailed information on the types of provisions included in peace agreements from 1989 through 2013 as well as the degree to which each provision is implemented by the post-war government. In all, the PAM_ID includes detailed information on 34 different peace agreements with up to 51 different types of provisions in each. The PAM_ID also includes specific information on when and if the civil war resumes. The PAM_ID dataset also offers clear advantages over the UCDP Peace Agreement dataset (Harbom et al., 2006) for our project. Though the UCDP Peace Agreement dataset includes a wider temporal domain (starting in 1975), it fails to capture the annual implementation of peace agreement provisions. As one of our primary hypotheses focuses on the efficacy of land reform implementation, the PAM_ID dataset allows us to control for the implementation of other key peace agreement provisions (thereby ensuring our results are not spurious).¹ Using coding criteria similar to the UCDP Armed Conflict Dataset (UCDP/ACD; Kreutz, 2010), civil war recurrence is observed when fighting between the government and rebels generates 25 or more battlefield-related deaths in a given year. As with previous studies on civil war recurrence, we use a binary indicator for whether fighting resumes in a given peace year (Kreutz, 2010; Mason et al., 2011; Walter, 2004).

In order to isolate the effects of land reform provisions on the risk of civil war recurrence, we replicate models of civil war recurrence used by previous studies that employ the PAM_ID (Joshi and Quinn, 2015; Joshi et al., 2015, 2017). As with all the previous studies employing the PAM_ID, we use a Weibull hazard model. Unlike logistic regression models, hazard (or duration) models account for the role of time in estimating the risk of an event occurring. Using an accelerated failure time (AFT) estimation, the Weibull model accounts for how variables of interest increase or decrease the expected time until a failure event occurs. In other words, variables included in the models either increase or decrease the time until a civil war is likely to recur following the establishment of a CPA. Given that the Weibull model assumes that the baseline hazard of an event occurring increases or decreases with time, past research has cautioned against the use of this estimation technique when there is not an assumption as to how the risk of a failure event should change over time (Cunningham et al., 2009; Thyne, 2015). To ensure that our results are not driven by our model of choice, we also estimate a Cox proportional hazard model as a robustness check.²

Our primary independent variable, Land Reform, is a binary indicator of whether the peace agreement under observation includes a provision that addresses land reform issues. Land reform provisions in peace agreements generally address issues surrounding land ownership, access to farming land, and/ or agricultural support services such as technical assistance and access to credit. For instance, El Salvador’s 1992 Chapultepec Peace Agreement included provisions that dealt specifically with land tenure grievances, land transfers, and equity for farmers. Similarly, Sudan’s 2005 Comprehensive Peace Agreement included provisions that targeted longstanding issues around land use and ownership within the South. Niger’s 1995 peace agreement between the government and the ORA (Organization of the Armed Resistance) offered technical assistance for aggrieved farmers. Because variables identifying land reform provisions are not included in either the PAM_ID or the UCDP Peace Agreement Dataset, we consulted the primary documents included in the PAM_ID as well as UCDP Peace Agreement Dataset to identify which peace agreements included some form of land reform. With regard to the PAM_ID, many land reform provisions have been identified either as decentralization, self-determination, development, or natural resource management provisions. On the other hand, UCDP has excelled at mapping key provisions related to military, political, economic, and territorial reforms (as well as post-war justice provisions), but focuses less on key indices related to natural resource management (such as land reform provisions). In total, 38% of the CPAs in the PAM_ID include land reform provisions. To measure land reform implementation, we rely on Peace Accords Matrix implementation histories³ as well LexisNexis Newswire searches as to the progress of the implementation of each provision. From these sources, we generated a binary variable for whether land reform provisions reached at least partial implementation. For partial implementation, we investigated whether the process of redistributing land is underway and mechanisms for determining land ownership/land use have been established (though all claims have not been resolved).

One benefit to using the PAM_ID is that it enables us to compare the effect of land reform provisions with other key peace agreements provisions that previous studies have shown to reduce the risk of renewed fighting between both signatories and non-signatories to peace accords. The PAM_ID codes the annual implementation of each peace agreement provision. Each provision is coded on a 0–3 ordinal variable, with a zero indicating that the provision is absent from the peace agreement and a three indicating that the provision is fully implemented. Our models include a control for the implementation of security sector reforms and accommodative reforms (Joshi et al., 2015), as well as the total implementation of peace agreements (Joshi and Quinn, 2015). Joshi et al. (2015) identify eight security sector reforms included in CPAs: disarmament of rebels, demobilization of armed forces, reintegration or dissidents, military reform, police reform, establishment of a ceasefire, dissolution of government paramilitary organizations, and withdrawal of armed forces. Joshi et al. (2015) argue that these processes are often interrelated, suggesting that an aggregate measure more accurately captures the success of each peace agreement’s implementation. Following the procedure of Joshi and Quinn (2015), we generate an aggregate implementation rate (ranging from 0 to 1) of security sector reforms (SSR Implementation) as well as a control for the number of security sector reforms. Similarly, Joshi et al. (2015) identify the effects of accommodative reforms as a mechanism of addressing combatant commitment problems. Joshi et al. (2015) identify accommodative reforms as amnesty provisions, prisoner transfers, and the establishment of a transitional political power-sharing agreement. Following the procedure employed in Joshi and Quinn (2015), we generate an annual implementation rate (ranging from 0 to 1) of accommodative reforms (Accommodation Implementation). We also include a control for the total number of accommodative provisions in the peace agreement. Finally, we generate a control for the annual implementation of all peace agreement provisions included in each CPA (Joshi and Quinn, 2015).

We include a number of controls that account for other factors that increase or decrease the risk of civil war recurrence. For instance, given that previous research has demonstrated that the intensity of the previous war often affects the likelihood that renewed fighting will take place (Kreutz, 2010; Mason et al., 2011; Quinn et al., 2007; Walter, 2004), we include a control for civil war intensity as measured by the number of battlefield deaths (in thousands) during the previous civil war. Past research has suggested that the duration of the now-ended civil war also provides valuable information revealing consideration about the prospects of an actor’s resumption of armed conflict producing a better outcome than abiding by the existing peace agreement (Mason et al., 2011; Quinn et al., 2007; Walter, 2004). With that in mind, we include a control for the duration of the previous civil war. We also include controls for whether the previous civil war was fought over territorial or government incompatibilities, as well as whether UN peacekeepers were deployed to enforce the CPA. Finally, we include controls for inequality as measured by the infant mortality rate (Walter, 2004) and the relative degree of democracy within the post-civil war state as measured by the revised Polity II indicator from the Polity IV Database (Marshall et al., 2011).

Results

Our initial set of results suggests strong support for the proposition that land reform provisions increase the durability of post-civil war peace. The models found in Table 1 estimate the risk of civil war recurrence between signatories of the CPA. All models are estimated using the AFT form. AFT models estimate whether variables increase or decrease the time until an event occurs. For our purposes, we use the AFT models to estimate whether peace agreement provisions increase the amount of time until a civil war recurs (ensuring a more durable post-civil war peace). Models 1–5 suggest that the inclusion of land reform provisions in CPAs substantially increases the time until a civil war recurs. Model 1 estimates the effects of including a land reform provision in the CPA without controlling for the effects of implementing other key peace agreement provisions. The results of Model 1 suggest that the inclusion of land reform provisions significantly increases the time until fighting renews between signatories of a CPA. The effects of land reform provisions appear to increase once we control for the implementation of security sector reforms that are a part of the CPA (Model 2). In calculating the marginal effects, there is a 62.8% increase in the duration of peace once we account for the security sector reforms as compared with the effects in Model 1.⁴

Table 1.

Civil war recurrence between CPA signatories (AFT Weibull Hazard Model).

	Model 1	Model 2	Model 3	Model 4	Model 1
Land reform provisions	2.789**	3.776***	1.704*	4.351**	4.349**
Land reform provisions	(1.401)	(0.509)	(1.008)	(2.033)	(2.043)
Land reform implementation					3.232
Land reform implementation					(3.260)
Security sector implementation		0.211***
Security sector implementation		(0.0342)
Security sector provisions		−0.316***
Security sector provisions		(0.0766)
Accommodation implementation			0.0307***
Accommodation implementation			(0.0109)
Accommodation provisions			0.0516
Accommodation provisions			(1.142)
Total PA implementation				0.0655**	0.0655**
Total PA implementation				(0.0272)	(0.0273)
Conflict intensity	0.00111	−0.000133	−0.000674	−0.00107	−0.00107
	(0.00178)	(0.000710)	(0.000896)	(0.00102)	(0.00102)
War duration	−0.0116***	−0.00329	−0.00173	0.00179	0.00179
	(0.00358)	(0.00596)	(0.00273)	(0.00789)	(0.00789)
UN peacekeeping	0.0181	0.0274***	0.0106	0.0256**	0.0256**
	(0.0193)	(0.00530)	(0.0126)	(0.0109)	(0.0109)
Infant mortality rate	−0.0434**	−0.0320***	−0.0219*	−0.0259**	−0.0259**
Infant mortality rate	(0.0215)	(0.00825)	(0.0123)	(0.0107)	(0.0107)
Territory conflict	0.551	0.986***	0.185	1.413	1.412
	(1.099)	(0.330)	(1.094)	(1.079)	(1.081)
Lagged polity II	0.0953	0.0232	0.0859	0.0335	0.0335
	(0.0762)	(0.0259)	(0.0675)	(0.0782)	(0.0782)
Constant	6.375***	3.466***	1.886	−0.833	−0.832
	(2.174)	(1.299)	(2.166)	(2.743)	(2.747)
Observations	219	219	183	218	218
Wald Chi squared	57.79***	935.05***	114.63***	215.97***	410.23***
P	1.01	1.13	1.46	2.52	2.52

Robust standard errors in parentheses.

***

p<0.01, **p<0.05, *p<0.1.

The results appear to weaken, however, when we account for the impact of Accommodation Reforms, such as Transitional Political Power-Sharing and Amnesty, on the risk of civil war recurrence (Model 3). While statistically significant, the relationship between land reform and the risk of civil war recurrence becomes weaker when we control for the impact of the implementation of reforms designed to assuage the credible commitment concerns of combatants. The results, though, appear to be much stronger when we account for the total amount of peace agreement implementation (Model 4). As noted by Joshi and Quinn (2015), the continued process of peace agreement implementation, regardless of the provisions, is often the most significant factor in generating durable peace following the establishment of a CPA. The results of Model 4 would suggest that, even after accounting for the total degree of peace agreement implementation, the effect of including land reform provisions has a significant and independent positive effect on the duration of peace. Figure 1 includes the coefficient plots for each model.

Figure 1.

Coefficient Plots for Table 1.

To illustrate the effects of land reform provisions, we plotted the calculated cumulative hazard of the risk of civil war recurrence with and without land reform. As demonstrated in Figure 2, the absence of land reform provisions in the final CPA stimulates a precipitous increase in the risk of civil war recurrence between signatories of the agreement. Specifically, the risk of renewed fighting continues to increase steadily with the passage of time. On the other hand, the inclusion of land reform provisions ensures that the risk of renewed fighting remains low over time. For instance, after accounting for the effects of Accommodation Reforms, the inclusion of a land reform provision decreases the risk of civil war recurrence by 447% (or exp^1.7 – 1 = 4.47).

Figure 2.

Cumulative hazard of civil war recurrence following a CPA.

Interestingly, our variable Land Reform Implementation appears to have no statistical relationship with the risk of civil war recurrence when estimated in conjunction with the inclusion of land reform provisions as well as the total degree of peace agreement implementation. These results may be driven, in part, by the fact that the process of implementing land reform provisions is often both lengthy and costly. Post-war land commissions within Nepal, South Sudan, and South Africa have been routinely criticized for slowly adjudicating land issues. New claims often emerge in South Africa that the land commission must add to a growing list to investigate. Equally, in South Sudan, the land commission’s efforts have been slowed in part by chronic corruption that has looted many of the assets reserved for administering claims. It would appear, at the moment, that the promise of land reform may actually be more attractive than the difficult process of establishing land reform.

Selection effects

There is a risk that our results may be driven by selection effects. Unlike the establishment of security sector or accommodation reforms, land reform is tied intimately to specific grievances within the conflict. It may be unlikely that reforms needed to mollify the concerns of rural dissidents should have equal weight in more urban conflicts such as the longstanding insurgency in Northern Ireland, or in ethnically based civil wars where the grievances are based in identity conflicts. Rather, land reforms are specific concessions made by the central government to resolve grievances among rural populations. This would suggest that land reforms may be distributed unevenly within our universe of cases, possibly biasing our results.

In order to ensure that the effects of land reform implementation are not driven by selection effects, the second stage of our analysis uses a seemingly unrelated recursive bivariate probit regression (Gartner, 2011; Greene, 2003; Hartzell and Hoddie, 2015). While selection effects are often modeled using a Heckman probit regression, that technique would be inappropriate for our needs. Specifically, Heckman probit models are appropriate for modeling necessary conditions as part of the selection stage. For instance, in order to estimate when mediation is accepted in civil war, one must first predict whether mediation has been offered by third parties (Greig and Regan, 2008). The likelihood of civil war recurrence, though, is not predicated on the inclusion of land reform provisions. By only examining the survival of CPAs with land reform provisions, we would exclude the universe of cases where civil war renewed and land reform was not part of the peace process. Using a bivariate probit model, we can compare the risk of civil war recurrence among cases that were identified in the selection stage as well as cases that were not identified. Using the bivariate probit model, we also account for the effects of the selection stage within the outcome stage, ensuring that the effects of land reform provisions are not driven by underlying selection effects (Gartner, 2011; Greene, 2003).

As part of the selection stage, we use four key variables to predict the inclusion of land reforms in the peace agreement: duration of the previous conflict, end-of-war democracy, the number of peace agreement provisions, and agricultural economy. As land reforms represent costly concessions, we assume that land reforms are offered following lengthy civil wars. Since land reform is directed largely at disaffected rural citizens, land reform provisions should be more likely to be included in peace agreements in economies that are primarily based around agriculture. Our variable, agricultural economy, measures agriculture as a share of the post-civil war state’s gross domestic product. This variable is drawn from the World Bank Indicators website. Equally, we also assume that land reform is likely to be offered if the government negotiating the settlement is democratic. Unlike more autocratic rulers who rely on a small cadre of elites, democratic leaders are forced to solicit votes from a majority of the citizenry. Democratic leaders are therefore likely to agree to land reform as part of any settlement as they can campaign on agrarian reforms in order to solicit votes. To address the exclusion restriction, we include a variable measuring the total number of provisions within the peace agreement. As each provision represents a different issue area within the peace agreement, we assume that the greater the number of issues covered, the more likely it is that land reforms will be part of the peace process. On the other hand, there is little evidence that peace agreements with numerous provisions are more capable of reducing the risk of civil war recurrence than peace agreements with fewer provisions. As we contend in our theory, it is the resolution of contentious issues (such as land-based grievance) that we expect will reduce the risk of renewed fighting. Equally, past research has shown that it is the process of implementing multiple provisions that reduces the risk of renewed fighting (Joshi and Quinn, 2015).

Model 1 in Table 2 examines the relationship between the inclusion of land reform provisions and our variables of interest included in the selection stage. Using a probit model, the results demonstrate that agricultural economies, end-of-war democracies, and the total number of peace agreement provisions significantly increase the likelihood of the inclusion of land reform provisions in a peace agreement. Equally, the results of Model 2 demonstrate that these same variables do a poor job predicting the likelihood of renewed fighting between the combatants. The results of Table 2 suggest that, once selection effects have been accounted for, land reform provisions substantially reduce the risk of civil war recurrence between CPA signatories. As the selection stage suggests, both agricultural economy and end-of-war democracy are predictive of land reform provisions being included in the final peace agreement. Once these selection effects have been accounted for, the inclusion of land reform provisions appears to significantly reduce the likelihood that signatories to the peace agreement will engage in renewed violence. It should also be noted that the Rho statistic is not significant, suggesting that the outcome equation is not tied to the selection equation (suggesting little in the way of selection effects). This provides us with greater confidence that the effectiveness of land reform provisions is not tied to underlying selection effects.

Table 2.

Civil war recurrence between CPA signatories (recursive bivariate probit).

	Model 1Land reformprovision(probit)	Model 2Civil warrecurrence(probit)	Model 3Selection stage(bivariate probit model)	Model 3Outcome stage(bivariate probit model)
Land reform				−7.085***
				(1.332)
Conflict intensity				−0.000308
				(0.000545)
War duration			0.00377*	0.00982**
			(0.00203)	(0.00399)
UN peacekeeping				−0.00961
				(0.0113)
Infant mortality rate				0.0142
Infant mortality rate				(0.0131)
Territory conflict				−0.0622
				(0.853)
Lagged polity II				−0.117
				(0.0940)
_Spline 1		0.207*		0.0868
		(0.112)		(0.0856)
_Spline 2		−0.295		−0.186
		(0.205)		(0.163)
_Spline 3		0.277		0.202
		(0.229)		(0.168)
Total number of provisions	0.0423***	−0.0236	0.0125
Total number of provisions	(0.0113)	(0.0298)	(0.0367)
Agricultural economy	0.0174***	−0.0186	0.0288*
Agricultural economy	(0.00589)	(0.0142)	(0.0170)
End-of-war democracy	0.991***	−0.516	1.051**
End-of-war democracy	(0.179)	(0.473)	(0.515)
Constant	−1.785***	−0.337	−2.040**	−3.440*
	(0.286)	(0.590)	(0.821)	(1.771)
Observations	253	253	234	234
Log-likelihood	−144.12431	−27.361369		−139.33607
Rho=0				0.123

Standard errors in parentheses.

***

p<0.01, **p<0.05, *p<0.1.

Conclusion

The results suggest that the inclusion of land reform provisions significantly reduces the risk of civil war recurrence following the establishment of a negotiated settlement. As we argue in the theory, the inclusion of land reform provisions both addresses longstanding grievances held by dissidents and increases in the opportunity costs associated with future rebellion. The results remain robust while accounting for the implementation of salient provisions, such as security sector reforms or accommodative provisions, as well as the total degree of peace agreement implementation. Equally, our results remain robust when we account for possible selection effects on when and where land reform provisions are used as part of the peace process. In short, we feel that the use of land reform provisions appears to clearly mollify the grievances of signatories to negotiated settlements.

Issues surrounding land use and land ownership still represent a threat to stability in many fragile states. For instance, land ownership and use was a major issue in the Casamance insurgency in Senegal. The Movement of the Democratic Forces of the Casamance (MFDC) has engaged in years of fighting to redress longstanding grievances over the state’s appropriation of land and the mismanaged environmental degradation of farmland. The failure of the two parties to negotiate clear land reforms as part of their 2004 agreement has contributed to recent renewed violence and infighting within the MFDC. The international community has a clear interest in helping parties identify and establish remedies to protracted civil wars.

With that in mind, though, the results also underscore clear limitations of land reforms in reducing the risk of civil war recurrence. For instance, the results also underscore that the implementation of land reforms appear to have no independent effect on the risk of renewed fighting by signatories of the peace agreement. As noted earlier, this result may be driven by the long and complicated process of implementing these costly provisions. Further research should examine what factors act as a barrier to land reform implementation. While elites may agree to land reforms as a precondition to peace, they may have little interest in actually disrupting the status quo. Rather, vested interests may attempt to slow or halt the process of implementation so as to ensure that meaningful change never occurs.

An additional shortcoming of our analysis is that these findings only demonstrate the effectiveness of land reforms following a negotiated settlement. Land reforms included within CPAs should inevitably reflect compromises in the goals of dissidents as well as the concessions granted by land owners. What remains unclear is how effective unilateral land reforms offered by a victorious incumbent government or a new revolutionary government are at reducing the risk of renewed fighting. For instance, it stands to reason that victorious rebel groups have an incentive to offer generous land reforms as they dismantle the vestiges of the previous regime. Future research should explore the nature of civil war outcomes in shaping the scope and effectiveness of land reform in the post-war environment.

Footnotes

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Notes

Author biographies

Eric Keels is a Post-Doctoral Research Fellow in Global Security at the Howard H. Baker Center for Public Policy. His research focuses on civil wars, conflict management, democracy, and African politics.

T David Mason is Regents Professor of Political Science and the Johnie Christian Family Professor of Peace Studies at the University of North Texas. He has published widely on land reform, civil war, and the durability of peace after intrastate conflicts.

References

Albertus

(2015) Autocracy and Redistribution: The Politics of Land Reform. New York: Cambridge University Press.

Albertus

Kaplan

(2013) Land reform as a counterinsurgency policy: Evidence from Colombia. Journal of Conflict Resolution 57(2): 198–231.

Boix

(2008) Economic roots of civil wars and revolutions in the contemporary world. World Politics 60(3): 390–437.

Collier

Elliott

Hegre

et al . (2003) Breaking the Conflict Trap: Civil War and Development Policy. Washington, DC: Word Bank.

Cunningham

Gleditsch

Salehyan

(2009) It takes two: A dyadic analysis of civil war duration and outcome. Journal of Conflict Resolution 53(4): 570–597.

de Bremond

(2007) The politics of peace and resettlement through El Salvador’s Land Transfer Programme: Caught between the state and the market. Third World Quarterly 28(8): 1537–1556.

Gartner

(2011) Signs of trouble: Regional organization mediation and civil war agreement durability. Journal of Politics 73(2): 380–390.

Greene

(2003) Econometric Analysis (5th edn). Upper Saddle River, NJ: Prentice Hall Publishers.

Greig

Regan

(2008) When do they say yes? An analysis of the willingness to offer and accept mediation in civil wars. International Studies Quarterly 52(4): 759–781.

10.

Harbom

Högbladh

Wallensteen

(2006) Armed conflict and peace agreements. Journal of Peace Research 43(5): 617–631.

11.

Hartzell

Hoddie

(2015) The art of the possible: Power-sharing and post-war democracy. World Politics 67(1): 37–71.

12.

Joshi

Darby

(2013) Introducing the Peace Accords Matrix (PAM): A database of comprehensive peace agreements and their implementation, 1989–2007. Peacebuilding 1(2): 256–274.

13.

Joshi

David Mason

(2007) Land tenure, democracy, and insurgency in Nepal: Peasant support for insurgency versus democracy. Asian Survey 47(3): 393–414.

14.

Joshi

David Mason

(2008) Between democracy and revolution: Peasant support for insurgency versus democracy in Nepal. Journal of Peace Research 45(6): 765–782.

15.

Joshi

David Mason

(2010) Land tenure, democracy, and patterns of violence during the Maoist Insurgency in Nepal, 1996-2005. Social Science Quarterly 91: 984–1006.

16.

Joshi

David Mason

(2011) Peasants, patrons, and parties: The tension between clientelism and democracy in Nepal. International Studies Quarterly 55(1): 151–175.

17.

Joshi

Quinn

(2015) Implementing the peace: The aggregate implementation of comprehensive peace agreements and peace duration after intrastate armed conflict. British Journal of Political Science 47: 869–892.

18.

Joshi

Melander

Quinn

(2017) Sequencing the peace: How the order of peace agreement implementation can reduce the destabilizing effects of post-accord elections. Journal of Conflict Resolution 61(1): 4–28.

19.

Joshi

Quinn

Regan

(2015) Annualized implementation data on comprehensive intrastate peace accords, 1989–2012. Journal of Peace Research 52(4): 551–562.

20.

Kreutz

(2010) How and when armed conflicts end: Introducing the UCDP Conflict Termination Dataset. Journal of Peace Research 47(2): 243–250.

21.

Licklider

(1995) The consequences of negotiated settlements in civil wars, 1945-1993. The American Political Science Review 89(3): 681–690.

22.

McClintock

(1984) Why peasants rebel: The case of Peru’s Sendero Luminoso. World Politics 37: 48–84.

23.

Marshall

Jaggers

Gurr

(2011) Polity IV Project: Political Regime Characteristics and Transitions, 1800–2010. College Park, MD : Center for International Development and Conflict Management, University of Maryland.

24.

Mason

(1986a) Land reform and the breakdown of clientelist politics in El Salvador. Comparative Political Studies 19: 487–516.

25.

Mason

(1986b) Population growth and the ‘Struggle for the Hearts and Minds’ of the rural population in Central America. In: Saunders

(ed.) Population Growth in Latin America and U.S. National Security. New York: Allen & Unwin, pp. 183–222.

26.

Mason

(1998) Take two acres and call me in the morning: Is land reform a remedy for peasant unrest? Journal of Politics 60: 199–230.

27.

Mason

(2004) Caught in the Crossfire: Revolution, Repression, and the Rational Peasant. Boulder, CO: Rowman & Littlefield.

28.

Mason

Krane

(1989) The political economy of death squads. International Studies Quarterly 33(2): 175–198.

29.

Mason

Gurses

Patrick

et al . (2011) When civil wars recur: Conditions for durable peace after civil wars. International Studies Perspectives 12: 171–189.

30.

Moore

Lindström

O’Regan

(1996) Land reform, political violence, and the economic inequality-political conflict nexus: A longitudinal analysis. International Interactions 21: 335–363.

31.

Muller

Seligson

(1987) Inequality and insurgency. American Political Science Review 81: 425–451.

32.

Paige

(1975) Agrarian Revolution: Social Movements and Export Agriculture in the Underdeveloped World. New York: Free Press.

33.

Quinn

David Mason

Gurses

(2007) Sustaining the peace: Determinants of civil war recurrence. International Interactions 33(2): 167–193.

34.

Scott

(1976) The Moral Economy of the Peasant: Rebellion and Subsistence in Southeast Asia. New Haven, CT: Yale University Press.

35.

Thyne

(2015) The impact of Coups d’Etat on civil war duration. Conflict Management and Peace Science. Epub ahead of print 25 March. DOI: 10.1177/0738894215570431.

36.

Toft

(2010) Ending civil wars: A case for Rebel Victory. International Security 34(4): 7–36.

37.

Unruh

(2003) Land tenure and legal pluralism in the peace process. Peace and Change 28(3): 352–379.

38.

Walter

(2004) Does conflict beget conflict? Explaining recurring civil war. Journal of Peace Research 41(3): 371–388.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB