“T RUTH, J USTICE AND P EACE ”
A quantitative analysis of the impact of Truth and Reconciliation Commissions on conflict recurrence
Elsa Solvin
Dept. of Peace and Conflict Research Uppsala University
Bachelor’s thesis January 2021 Word Count: 9337 Supervisor: Susanne Schaftenaar
ii
Abstract
Time has passed since truth and reconciliation commissions (TRCs) started to grow in popularity. The general patterns of their effect on the post-conflict societies are still unexplored. The main claim is that establishing a TRC will reduce the likelihood of conflict recurrence compared to other forms of transitional justice as TRCs are especially well equipped to mitigate reasons for conflict recurrence. This paper uses quantitative methods with the PCJ dataset and the UCDP dyadic dataset to analyse the trends of different types of transitional justice between the years 1946-2006. The logistic regression showed a negative relationship between establishment of TRCs and conflict recurrence. The main implication of these findings is that there are general patterns of TRCs having an effect, which need to be further researched.
iii
Table of content
1. Introduction ... 1
2. Theoretical framework ... 2
2.1. Previous research ... 2
2.2. Conceptualisation of dependent and independent variables... 6
2.3. Theoretical argument/Causal mechanism ... 7
3. Research design ... 9
3.1. Method ... 9
3.2. Independent variable ... 9
3.3. Dependent variable ... 10
3.4. Control variables ... 11
3.5. Causal criteria ... 13
3.6. Notes on sources ... 13
4. Results and analysis ... 13
4.1. Descriptive statistics ... 13
4.2. Regression analysis ... 15
4.3. Discussion ... 17
5. Conclusion ... 20
5.1. Avenues for future research ... 21
6. Bibliography ... 22
Appendix ... 26
iv
List of figures and tables
Table 1. Causal mechanism ... 8
Table 2. Descriptive statistics ... 14
Table 3. Logistic regression of conflict recurrence likelihood ... 15
Table 4. Logistic regression conflict recurrence likelihood (any TJ)... 26
Table 5. Robustness test conflict recurrence likelihood ... 22
1
1. Introduction
“The world exists on three things: truth, justice, and peace” is a Hebrew proverb, but are these related? After termination of armed conflict, there is always a possibility of violence erupting again and recurrence of that conflict. Research has found that reasons for conflict emergence and recurrence tend to relate to some aspects of a country which are affected and can be amplified by armed conflict. Some of these are grievances and/or greed (Collier and Hoeffler, 2004; Buhaug, Cederman and Gleditsch, 2014), commitment problems (Svensson, 2007), information asymmetries (Öberg, 2002) and poor governance (Hegre and Nygård, 2015). The peace following termination usually has some form of reconciliatory effort, through different kinds of transitional justice (TJ). Starting at the end of the 20th century one type of TJ, Truth and Reconciliation Commissions (TRCs), has grown in popularity for dealing with the aftermath of conflict at the local level (Wiebelhaus-Brahm, 2010, p. 477). This has largely been parallel to the increase in intrastate conflict, where TRCs are also most applicable. This is due to the necessity of the different parties of the conflict to live within the same borders after the conflict is over. The main goal of TRCs tend to be creating more reconciled attitudes between groups, while promoting truth-telling and communal healing (Begley, 2020, p. 1). In the long-term the consequences of TRCs are not as clear.
Given this, this paper aims to answer the question ‘what is the long-term impact of
establishing a truth and reconciliation commission in transitional justice projects on conflict recurrence?’. So far, research has done little quantitative analysis of the long-term impact of these commissions as they are still relatively new, the main research has been case based such as Indonesia and Timor Leste (Stahn, 2001; Roosa, 2007), South Africa (Bozzoli, 1998;
Gibson and Louis, 2004) and Sierra Leone (Shaw, 2007; Millar, 2012; Cilliers, Dube and Siddiqi, 2016). As TRCs work differently and are located in different places it is difficult to measure overall success of the commissions (Kubicek and Walker, 2020, p. 296). This paper will therefore use the definitions provided by Dancy, Kim and Wiebelhaus-Brahm (2010) to include the findings in the wider field of TRC research. There is also little research on TRCs compared to other forms of TJ. I argue that the establishment of a TRC in the post-conflict context addresses all of these possible reasons for armed conflict leading to a significant reduction in the likelihood of more violence. I found support for my theory using quantitative analysis. The main finding was that for the conflicts between 1946 and 2006 the probability of recurrence was significantly reduced when a TRC was established. From about 28% chance
2 without a TRC compared to about 20% chance if a TRC was established. The findings hold when controlling for the type of termination the conflict had, the intensity of the conflict and the GDP of country. Even though the findings were not statistically significant, the hope to still contribute to the general understanding of the long-term effects of truth and reconciliation commissions.
2. Theoretical framework
2.1. Previous research
There are several reasons for armed conflict emergence and recurrence. Similarly, there are mitigating reasons and institutions to prevent armed conflict. Two of the main reasons for armed conflict argued by researchers are greed and grievances. These are often put in contrast, although there is support for both having a causal relationship with armed conflict (Collier and Hoeffler, 2004). Grievances are relative deprivations of a person or a group of people. These can lead to armed conflict as the relatively deprived feel like it is the only option to improve their situation. Most of the literature has focused on individual grievances, which compared to greed has a smaller impact. Buhaug, Cederman and Gleditsch (2014), on the other hand, analysed the effects of group relative deprivation, namely horizontal
inequality. Horizontal inequality exists where there are inequalities between groups as defined by culture (Langer, 2005, pp. 26–27). Different cultures define groups in different ways, these identities also differ in importance for different people, meaning the horizontal inequality can have different effects on different people. Buhaug, Cederman and Gleditsch (2014, p. 429), found that even though there was less support for individual grievances causing armed
conflict, there was significant support for horizontal inequality increasing the risk especially if the group was ethnically defined. Horizontal inequality is also likely to help with the
collective actions problem, i.e. the problem of not cooperating to reach a goal but instead
“free-riding” on others and achieving the same goal without having taken any risks (Willer, 2009, p. 23).
In contrast to the argument of grievances is the argument that greed is what causes armed conflict. Greed relates to armed conflict as there is possibility for profit. Described as the opportunities in civil war, Collier and Hoeffler (2004, pp. 565, 568). Greed in this case is, for example, to benefit from natural resources, or finances from either a foreign government or ethnic diaspora. Natural resources have in themselves also been causally linked to an
3 increased likelihood of intra state armed conflict (Ross, 2004, p. 61). Greed can have a
particularly large impact on countries that are less economically developed. This is largely due to the reason that there tend not to be as many opportunities for individuals to earn money. Resulting in them turning to a rebel leader in order to earn money to support themselves (Humphreys and Weinstein, 2008, p. 451).
This relates to the realist theory of armed conflict relating to cost-benefit calculation and signalling theory. Fearon (1995, pp. 386–387) argues that there is a bargaining range, in which there is an agreement where both parties would be better off than they would be if they wasted resources on conflict. Coming to an agreement instead of engaging in armed conflict would therefore be mutually beneficial. Signalling theory states that this range only can be found when there is complete information to and from each side of the conflict. The parties therefore want to signal strength through different tactics. This gives them a better position in negotiating, leading to the other side wanting to signal strength, and thus, conflict escalate (Öberg, 2002, p. 28). Consequently, if both parties are given information on the other, both sides will have no reason for further armed conflict as they both know the other’s relative strength.
A clear trend is that most current and recent armed conflict is intrastate (Themnér and Wallensteen, 2011, p. 526), meaning they take place within the borders of a state. This can have many different reasons. In the UCDP, these are coded as being either based on territory or government (UCDP, no date). Conflict has different reasons for emerging in the interstate context, compared to interstate. As this paper focuses on TRC processes and they are used almost exclusively on the intrastate context (there are exceptions, such as Indonesia and Timor Leste) (Wiebelhaus-Brahm, 2010), combined with it being the most common form of conflict, means that this paper will focus on intrastate conflict. Apart from most conflicts being intrastate a large number of modern conflicts tend to be continuations of old conflicts.
All conflicts between 2003 and 2010 were continuations of previous conflicts (Walter, 2010, p. 1), these types of conflicts are the focus in this paper.
A mitigating effect of intrastate armed conflict recurrence is a to have a state with democratic institutions and good governance, especially conflicts that arise from grievances. (Hegre and Nygård, 2015) Good governance can come from both formal and informal institutions (Hegre
4 and Nygård, 2015, p. 987). TJ processes tend to strengthen both formal and informal
institutions. This is done by the enforcement of rule of law as well as inclusion of local rituals and rites (Bozzoli, 1998, p. 168). After an intrastate conflict, there is need for reconciliation as well as the building of strong institutions. These political institutions are key in avoiding what has been called the ‘conflict trap’, where a country is more likely to experience armed conflict if they have in the past (Walter, 2010, p. 3). This is in stark contrast interstate conflict where there is no need for the parties involved to achieve an agreement where they can live in the same country after the conflict is over.
There are different ways for a conflict to end. It could end in a one-sided victory, or in a bargained solution which are both common ways. Different forms of transitional processes are used and needed in these different instances. A common problem stemming from bargained solutions are commitment problems. These occur when a side in a conflict believes that the other side will not commit to the terms and therefore will not commit themselves, or the actual reneging of a side after the agreement. These occur for a variety of reasons and particularly in the intrastate context (Walter, 1997, pp. 137–139). Svensson (2007, p. 179) argues that this is mainly a concern for governments in intrastate conflicts. In a conflict situation they still have a chance to keep the power they had before the conflict, in a
bargained agreement, however, they stand to lose some of that power. They can, after having gained some power through the agreement, leverage that power in order to achieve more. In intrastate conflicts there is also a need for both sides to disarm following the termination of conflict as the parties should be able to coexist following the agreement.
After the termination of conflict there is need for some sort of process to ensure that the fighting stops these are usually some sort of transitional justice. When most people think about TJ, it relates to trials and the International Criminal Court. These are only a few examples of TJ, however. Domestic or international trials could create accountability for deeds done during the conflict and set norms for the post-conflict society. There are other processes that fall under the same category, however. Another way is to provide amnesties in order to ensure a full surrender from the losing party. A third form of TJ is to give some sort of reparations to affected members of a community or state for any losses suffered during the conflicts. A fourth is to establish a TRC. These are usually carried out in combination with other reconciliatory efforts. The purpose of TJ processes is usually to aid in the reconciliation
5 process (Binningsbø et al., 2012a, pp. 733–736). They can also be used as a ‘carrot and stick’
in order to terminate a conflict (Stanley, 2001, p. 531). For that purpose, amnesties can be useful. This could include leaders being offer certain amnesties in order to get the leaders to surrender or agree to negotiate. If there is more risk in failing and being punished leaders might instead risk continued conflict to possibly gain some protection once fighting is over (Binningsbø et al., 2012a, pp. 733–736).
This process could possibly have some drawbacks, however. From the field of psychology there have been findings of TRCs being harmful to individual wellbeing. Cilliers, Drube and Siddiqi (2016, pp. 792–793) looked at the effects of the TRC in Sierra Leone. They analysed survey data from different villages in Sierra Leone. They found that even though it led to reconciled attitudes, however, several of the respondents were found come out of the process with worse trauma and psychological problems such as PTSD. Sierra Leone was specifically looked into by Millar (2012, pp. 722–723) as well, who found that the reconciled attitudes were not due to the TRC, but rather the attitudes and culture of the people in Sierra Leone as well as the conflict dynamics. He argued that the conflict in Sierra Leone was not due to increased cleavages in society and therefore the TRC served little purpose. Drawing from both findings it is possible that TRCs might do more harm than good in some circumstances.
It is reasonable to believe that coming to terms with what has happened in a conflict is easier if the people involved believed that they were fighting for a just cause and was fighting an enemy. Finding out the truth in these circumstances might lead to more reconciled attitudes.
Success and failure of TRCs are also difficult to measure this partly due to the different definitions of TRCs, but also due to the different mandates and aims of different TRCs (Dancy, Kim and Wiebelhaus-Brahm, 2010, p. 3). There are ways to mitigate the risks of mental health issues, but it demands a support system. In order to avoid the risk, it is therefore necessary to implement training, making the commissions even more complex. There is therefore need for research on the benefits to ensure that the establishments of them is definitely beneficial.
Another case-based example of TRC is research of South Africa. Studies have found that there is some connection of the TRC and reconciled attitudes. Gibson and Louis (2004) explored the effects of knowing and accepting of the truth of the past (as established by the commission) on reconciled attitudes among South Africans. Their main finding was that the
6 more extreme groups on both sides of the spectrum were less likely to be accepting of the truth, in this case the negative and large impact of the apartheid system. Most of these groups agreed with the overall ideas behind the system, the problems lying rather in the
implementation. They also found that the acceptance of the truth had a greater impact on reconciliation for non-black South Africans compared to the black population (Gibson and Louis, 2004, p. 215). Meaning it is possible that victimised groups could have less reconciled attitudes if they are aware of the crimes committed against their group. From this it is possible to draw the conclusion that as groups are less accepting of the truth, they are also less likely to hold reconciled attitudes. Further, as one can argue that extreme groups are more likely to engage in violence, recurrence of conflict is more likely when these are not reconciled.
This paper aims to fill the gap of the long-term implications of the creation of TRCs. The main goal of all TJ processes should be to avoid more violent conflict. They should therefore strive to reduce the likelihood of conflict recurrence. This paper aims to see if TRCs are the most effective at reaching this goal. This will be done by quantitative analysis of all TJ processes between 1946 and 2006.
2.2. Conceptualisation of dependent and independent variables
The dependent variable in this study will be conflict recurrence. This refers to the parties involved in armed conflict returning to armed conflict after having terminated for some reason. This concept is sometimes called conflict relapse, this concept, in some cases includes other conflicts arising after termination of conflict, not necessarily related to the first. Instead, these conflicts arise from conditions left by the conflict. Usually, the state of the economy and lack of opportunities (Walter, 2010). To specify that this paper specifically refer to the same parties returning to conflict, the word conflict recurrence will be used.
The independent variable of this paper will be establishment of Truth and Reconciliation Commissions as part of the transitional justice process. Definitions of TRCs tend to vary.
Most definitions state that they need to be a temporary investigative body. The question of who and what the commission are specifically supposed to investigate is more disputed. As the parties involved in a conflict tend to be complicated and both state and non-state actors a general definition of the parties involved may be difficult. What types of abuses the
commission is supposed to investigate is also complicated. Some look at human rights abuses
7 broadly, while others look at specific actions taken during conflict (Dancy, Kim and
Wiebelhaus-Brahm, 2010; Kubicek and Walker, 2020). A focus of most TRCs is also to improve human rights, which they tend to do to some extent, in combination with other forms of TJ (Olsen et al., 2010, p. 475).
Most research into TRCs are single case or small-N studies. This is due to the unique nature of TRCs that tend to be focused on individual countries’ needs and characteristics. There are some things these different commissions tend to have in common, and that serve a similar purpose. Analysis of these common aspects will be the aim of this study. This will be done to see the benefits of the general ideas of TRCs and their implications. The similarities were addressed by Dancy, Kim and Wiebelhaus-Brahm (2010, pp. 8–14). They argue that there are 5 characteristics that all TRCs have in common. They based this on several other definitions to come up with criteria that could be used by all research in order to do comparative
research. These include that a commission needs to be temporary. This means that it should be used as a part of the transitional process, rather than an institution in end of itself. The
commission also needs to be sanctioned by the state, this is so that the transitional power of the state can be measured. They also argue that the aim of the commission should be to publish a paper of some sort. Even if that paper is not published in the end. The commission will also have needed to start operating. The commission should also investigate abuses over time.
As mentioned previously, most modern conflicts are intrastate. TRCs also tend to be held for parties that exist in the same country. This is partly as some aspects of TRCs are designed to build strong institutions, which are not as necessary for people who live across boarders as for people that have to live in the same country after the termination of. Apart from institutions there is also a larger need for groups that live in the same country to be reconciled as they need to share physical space (Nadler et al., 2008, pp. 15–16). This means that this study will be largely focused on intrastate conflicts.
2.3. Theoretical argument and causal mechanism
Finally, the paper argues that there is a link between TRCs and conflict recurrence though the following causal mechanism:
8 Based on the literature in the field of conflict recurrence there are some factors that stand out.
The main findings in the previous research are that these are grievances, greed, poor
governance, commitment problems, and information asymmetries. According to the findings or TRCs, TRCs should be particularly well equipped at addressing these causes. Grievances can be reduced by the ability of victims of crimes to be heard in order to achieve communal healing (Pruitt and Kim, 2004, p. 219; Cilliers, Dube and Siddiqi, 2016, pp. 792–793) . Risk coming from greed can be reduced, partly by possible trials, but also by increasing costs when perpetrators are held accountable for their actions (Stanley, 2001, p. 531). One of the main components of TRCs are to come to a consensus of what happen during conflict and expose any human rights abuses (Olsen et al., 2010, p. 475). TRCs also tend to strengthen formal and informal institutions (Bozzoli, 1998, p. 168). When the truth has been established, there should no longer be any information asymmetries, meaning bargained solutions are a more viable option when the parties are aware of the stakes and the other parties’ strength.
Table 1. Causal chain
It could be argued that as TRCs can cause harm to participants, that TRCs could create grievances and therefore not reduce tensions caused by grievances. I would argue however, that the harm done by TRCs as found by research are on an individual basis. This is due to the fact that there is little evidence to say that individual grievances are enough to lead to armed conflict, rather, it is group grievances caused by horizontal inequalities (Langer, 2005, p. 43).
These individual grievances would therefore not serve to make conflict recurrence more likely. The types of problems these individuals have do not seem to be similar to the types of grievances that leads to armed conflict.
Establishment of Truth and Reconciliaiton
Commission
- Reduced grievences - Incentives relating to greed reduced
- Credible commitment facilitated
- Strengthened institutions and accountability - Removal of information asymmetries
Reduced likelihood of
conflict
recurrence
9 The hypothesis, thus, is:
Establishing a Truth and Reconciliation Commission in the transitional justice process will reduce the likelihood of conflict recurrence long-term.
3. Research design
3.1. Method
The aim of this paper is to see the effects of the establishment of a TRC in a TJ process to see if there is a decreased likelihood for conflict recurrence if there is a commission. As suggested by the theoretical background, this should be the case. In order to find this correlation
quantitative methods will be used. I will run a logistic regression, to see if there is a decreased chance for conflict recurrence. To ensure that there are no other latent variables, some control variables will be used. These are discussed below. In quantitative methods there is a risk of not having fully valid definitions and operationalisations of the concepts. In order to avoid this, I will clearly define both definitions and operationalisations so that future researchers may use the same or different definitions to ensure the robustness of my findings. In order to better display and interpret data there will be a formula for predicted probabilities of conflict recurrence. As all variables used are categorical or dichotomous the values used for the calculation will be the mode of each variable. There will also be robustness tests run these will use the same definitions and method as the main findings, apart from the allotted amount of time within which the conflict will have needed to recur.
3.2. Independent variable
The independent variable of this study will be the presence or absence of TRCs in TJ
processes. As mentioned in the review one of the main reasons for the lack of comprehensive research on the topic of TRCs is the lack of a general understanding of what a TRC is. Some commissions call themselves TRC without fulfilling what most people see as the criteria for TRCs. In order to combat the issue of having different definitions for TRC in different research papers Dancy, Kim and Wiebelhaus-Brahm (2010) set a list of criteria. These are developed from several other researchers’ definitions and establishes a clear list of
requirements. These are “1. The mechanism was a newly established and temporary
commission. 2. It was officially sanctioned by the state. 3. Its mandate includes investigative powers. 4. It actually began operating. 5. It investigated a pattern of abuses to personal integrity rights that occurred over a period of time, some of which were perpetrated by state
10 actors.” (Dancy, Kim and Wiebelhaus-Brahm, 2010, p. 7). For the purposes of including this study in the broader narrative, this paper will use this same operationalisation.
Transitional justice are the processes that happens post conflict in order to deal with the consequences of armed conflict. These can include many different types of processes. They tend to be some combination of trials, truth commissions, reparations, amnesties, purges, and exiles (Binningsbø et al., 2012a, p. 732). For the purposes of this study any post-conflict context that has had any of these processes will be considered to have had TJ.
In order to measure the independent variable this paper will be using the Post Conflict Justice (PCJ) dataset (Binningsbø et al., 2012b). It is a systematically collected dataset of all instance of a period of peace after the termination of conflict (Binningsbø et al., 2012a, p. 733). These include instances of trials, truth commissions, reparations, amnesties. In the dataset they are only coded as present or absent (Binningsbø et al., 2012c, pp. 6–7). These types of processes are dynamic and can have several different facets and characteristics, which could have different types of effects on the outcome. Capturing complex concepts is difficult in quantitative analysis, however, and as creating a new dataset is beyond the scope of this paper, this paper will also use presence and absence of these as well. This might prove an issue with validity, as these narrow definitions may not fully cover the concepts. I argue that the main aim of this paper will still be able to be fulfilled with these slightly unnuanced definitions of these conflicts, as it aims to see general patterns between conflict recurrence and type of TJ.
3.3. Dependent variable
The dependent variable for this paper will be conflict recurrence.
In order to establish the long-term effects of the TRC this paper will be addressing if there up until the year 2019 has been a recurrence of conflict. In order to avoid newer TRCs, of which you cannot yet see long-term effects, this paper will only look at TRCs completed before 2006. Conflict recurrence will be defined as appearing in the UCDP as an active conflict, using the dyadic dataset (Harbom, Melander and Wallensteen, 2008; Pettersson and Öberg, 2020). This means that the dyad will have to be the same. In order for a conflict to be shown as active, there needs to be at least 25 battle deaths in a calendar year (UCDP, no date). There
11 are issues of reliability in using this dataset. This is due to the possibility of the parties of a dyad changing their names after termination of conflict. In the UCDP, this will mean that it is coded as a new conflict. The reliability will be assured through manual correcting where applicable. The data on conflict recurrence will be transferred through manual coding. This means that there could be some risk of random error by the coder, especially as the coder is not experienced in coding. As coder reliability is always a risk, and is not systematic in any way, it should not lead to major inconsistencies in the findings, if someone else were to use the dataset, however, there might be a need to verify the dataset.
For the purposes of this study “long-term” will be defined as within 14 years. This is so that the maximum number of conflicts from the PCJ dataset. The latest conflict in the dataset terminated in 2006. As this paper was written in the end of 2020 it is possible for all conflicts to recur in the time span. In the future even longer times may be evaluated as the more TRCs are put in place and there is longer time for the conflicts to recur. There is need for this type of study to be made again when more time has passed so that more cases can be included, to increase generalisability, or for longer time to have passed, to increase validity. For the robustness tests it will be 20 and 5 years, respectively.
3.4. Control variables
Intensity of a conflict has been shown to have an increase of the likelihood of conflict recurrence. This has been theorised to a need for revenge when costs were high (Quinn, Mason and Gurses, 2007, p. 185). In the UCDP a conflict is described as a war if there are over 1 000 battle related deaths in a calendar year (UCDP, no date). For the purposes of this study this measure will be used to control for intensity. If there has at any point been a year with over 1 000 battle related deaths it will be viewed as an intense conflict. There could be some problems with validity using this definition. Depending on the size of the country 1 000 deaths can be more or less intense. It could also be argued that there are other ways of
defining the intensity of a conflicts, such as sexual violence and civilian deaths. This could make these definitions seem less valid. Since this is a quantitative study however, it is impossible to fully capture complex concepts. This paper therefore argue that the measure is valid enough to be used in this study.
12 There are also different effects of the types of termination the armed conflict came to.
Binningsbø et al. (2012a, p. 738) refer to victories, bargained solutions and others. In victories there may be a complete change of leadership for example, whereas in a bargained solution the main structure of main remain similar to before the conflict. They also found that the different types of termination tended to have different TJ processes. Different terminations may also have different effects on the post conflict society. To ensure that these aspects are considered, it will be controlled for in the analysis. This measure is considered valid in terms of the comparison between bargained solutions and victories. The “other” category comes from the conflict having no or low activity (Binningsbø et al., 2012c, p. 7). This is defined as the conflict reaching below the 25 battle related deaths per year threshold and is therefore viewed as inactive (Kreutz, 2010). As the termination is controlled for in the regression it should not make the results less valid or reliable.
Another strong correlation that has been found is the relationship between the economy of a country and its likelihood for conflict and conflict recurrence. This can be used as a proxy for grievances, even though it is debated as high GDP does not necessarily relate to the economic status of individuals in that country. Even though the research as to why GDP tend to affect the likelihood of conflict is indecisive there is a correlation between GDP and the likelihood for recurrence of armed conflict (Quinn, Mason and Gurses, 2007, pp. 168–169). In order to control for this correlation this paper will control for GDP. In order to avoid difficulties of inflation making the numbers impossible to compare, it will use the Gapminder (no date) dataset, which the Penn World Table (PWT) 9.1 dataset (Feenstra, 2019) adjusted for PPP in United States dollars in 2011, meaning the numbers should be comparable across time and space. The dataset reports the GDP with adjustment for PPP for the area of the current official state boarders. This might become an issue as the dataset from Binningsbø et al. (2012c, p. 6) uses the states boarders as they were when the peace period started. This means that countries which have changed names of had a change in boarders will be coded in manually. And as previously stated there is risk for random error when coding. As it is random, it should not make a systemic difference to the results. For all observations, the GDP will be coded for the year of conflict termination.
13 3.5. Causal criteria
To verify the causal relationship in this study it will attempt to qualify according to Kellstedt and Whitten (2018, p. 60) by overcoming their four hurdles. These are to find a plausible causal mechanism, ensuring the causal direction i.e., that X leads to Y, establishing
covariation and controlling for confounding variables. The causal mechanism as explained above is deemed by the author to be plausible. There is a link between the mediating variables and the independent and dependent variable. For the second criteria, time order is important, this means to ensure that X happens before Y. In this case that the TRC is established before the evaluation of whether or not the conflict recurred. This is clearly fulfilled as the evaluation exclusively looks at the time post-conflict. Covariation will be explored by using regression on the statistical data. As for confounding variables, these could still exist, even though some are controlled for. These are discussed further in the discussion part of this paper.
3.6. Notes on sources
Finally, a note on the sources used in the analysis. The UCDP uses different methods to assure that their numbers of battle related deaths are correct. They also include the range of estimates, with what they deem to be the lowest estimate possible based on their findings.
They also include the highest number they find in a range from highest to lowest estimate.
The number used in this analysis will be their best estimate. Which is the one they believe to be the most plausible number of deaths. The main reason for using the UCDP is that the PCJ dataset is based on same system. As the PCJ dataset is the only one of its kind it was
necessary to find other datasets that was compatible. Gapminder as well as the PWT have sourced how they set the values of the GDP for different countries, meaning their findings can be validated, making the dataset reliable.
4. Results and analysis
4.1. Descriptive statistics
In order to understand the data descriptive statistics were created. As the values in the first table are not as easy to interpret, a frequency table was created for dichotomous and categorical variables.
14 Table 2. Descriptive statistics
Statistic N Mean St.Dev. Min Pctl(25) Median Pctl(75) Max
TRC 173 0.052 0.2 0 0 0 0 1
Recurrence within14
173 0.15 0.4 0 0 0 0 1
Civil war status
173 0.44 0.5 0 0 0 1 1
Termination 173 1.8 0.8 1 1 2 3 3
GDP 173 5698 9598.3 247 1454 2605 6033 80944
The descriptive statistics show most conflicts that involved a TJ process did not recur within 14 years, only about 15% did. This also includes some instances of the same conflict recurring several times. As can also be seen TRCs were not common up until 2006 this will have some effect on the generalisability of the findings. Most armed conflicts during the period did not escalate into wars, but rather stated below the threshold of 1000 battle related deaths per year during any year the conflict was active. The proportion were about 56% armed conflicts and 46% that escalated into a war. Most termination falls under the category “Other”, which was the way about 46% of conflicts ended. On the other hand, “Victory” and “Bargained” was the termination of about 28% and 27%, respectively. The mean GDP as adjusted for PPP was 5698.387 rounded to 5689. The median, in contrast, was 2605, this indicates that the data is skewed, which can also be seen in the summary. A better measure of tendency in this case is therefore the median, which will be seen as the average. This means that for the predicted probability will use the conditions that the conflict did not escalate to war status and ended with an “Other” solution as this is the average conflict in the dataset as well as the average GDP.
Statistic N no yes
TRC 173 169 9
Percentage 100% 95% 5%
Recurrence within 14 years 173 147 26
Percentage 100% 85% 15%
Civil war status 173 97 76
Percentage 100% 56% 44%
N Victory Bargained Other
Type of termination 173 48 46 79
Percentage 100% 27.7% 26.6% 45.7%
15 4.2. Regression analysis
In order to examine my hypothesis, I will have to do two things. First, I will analyse whether or not I can reject my null hypothesis. The null hypothesis is that the establishment of a TRC will have no effect on the likelihood of conflict recurrence. This becomes problematic as the findings are not statistically significant at any level. This could mean that the findings cannot be generalised to all conflicts. One could argue, however, that as this sample includes all instances of transitional justice up from 1946 up until when the dataset was created, means that there is no need to generalise around that time period. The question is whether it can be generalised to conflicts later than 2006. This holds true even when you control for both intensity of the conflict and the reason for termination of the conflict. There is no possibility to statistically reject the null hypothesis fully.
Table 3. Logistic regression of conflict recurrence likelihood
Dependent variable:
Conflict recurrence
(model1) (model2) (model3)
TRC -0.041 -0.085 -0.085
(0.123) (0.128) (0.128)
terminationOther 0.008 0.009
(0.074) (0.075) terminationVictory -0.188*** -0.188***
(0.069) (0.069)
civilwar -0.097* -0.098*
(0.056) (0.057)
GDP -5.172*10-7
(2.806*10-6)
Constant 0.152*** 0.281*** 0.284***
(0.028) (0.063) (0.065)
Observations 173 173 173
Log Likelihood -68.398 -61.986 -61.969 Akaike Inf. Crit. 140.795 133.972 135.937 Note: *p<0.1; **p<0.05; ***p<0.01
SE in parentheses
16 The victory condition of termination seems to also decrease the likelihood of recurrence of conflict. This could be due to violent being registered as conflicts in this dataset. It is possible that the correlation instead coup d'état is between coup d'état and lack of conflict recurrence.
This could be an interesting area for future research, to control for coup d'état and/or look at the relationship it has to conflict recurrence. As the model showed some relationship between the type of termination and conflict recurrence it was useful to have as a control in this study.
The findings of the impact of TRC should therefore hold regardless of the type of termination the conflict had. It should also be noted that the termination condition was so called dummy variables, which means that the results are compared to the omitted category, in this case bargained solutions. This means that the significant decrease in likelihood of recurrence after a victory is when it is compared to the likelihood after a bargained solution.
As expected, higher GDP tended to reduce the likelihood of conflict recurrence. This follows the consensus the countries with more economic wealth tend to not be as likely to have conflict. It did, however, not significantly change the effect of TRCs when used as a control or have a significant impact on the likelihood of conflict.
Another surprising discovery was the impact of the intensity of the conflict. In this sample conflicts that crossed the threshold of 1000 battle related deaths in a calendar year were less likely to re-emerge compared to conflicts that did not. The findings were also statistically significant at the 90% confidence level. As this test only included cases were there was some form of transitional justice, that could be the reason for the direction of the coefficient. It is possible that there is something relating to TJ processes that makes intensity have a negative correlation to conflict recurrence. One could argue that if there is a process which draws attention to, and punishes human rights violations and high levels of lethality, that the knowledge of the atrocities of armed conflict would make people less prone to engage in it again. These findings would therefore not necessarily contradict the general conclusion of intensity of conflict leads to an increased likelihood of conflict recurrence.
To understand these findings one can run a predicted probability of conflict recurrence in a given country. This is done by taking the values of an average country and changing the values of the independent variable. The average values were determined by the most appropriate measures of tendency for each variable. This gives a probability of 0.2918 or
17 about 29% for conflict recurrence in a country where there is a TJ process and no TRC,
whereas in a county with a TRC the probability is 0.2068 or about 21%. This means that the probability of conflict recurrence is reduced by almost a third by including a TRC. This is a significant difference. Although as stated previously the findings relating to TRCs are not statistically significant. The claim that TRC reduce the likelihood by this amount on average in all conflicts could therefore be misleading. What can be said is that in all countries with a TJ process between the years 1946 and 2006, the likelihood of conflict recurrence was significantly lower for those that had a TRC.
In order to see if the impact of TRCs could be viewed as general there was also a test to see the impact of any type of TJ process compared to countries with no form of TJ (for regression table, see appendix, table 4). The results were not only statistically significant, using predicted probabilities showed that the effect was significant as well. A conflict had a probability of 0.4770 or about 48% chance of recurring if there was to process of TJ while one with a process had a probability of 0.2781 or about 28%.
In order to validate some of these results, a robustness test was run (for regression table, see appendix, table 5). This was done using the same variables as the main model apart from the time allowed for the conflict to recur was 20 and 5 years, respectively. The findings from those regressions showed the same trends as within 14 years. This supports the hypothesis by addressing some concerns of validity with the definition of long-term.
4.3. Discussion
The results show that for the armed conflicts in the analysis, there was a clear reduction in the number of conflicts that recurred if that country established a TRC for the conflicts between 1946 and 2006 as compared to other forms of TJ. This aligns with other previous research in conflict recurrence and TRCs.
I would argue that since this analysis covers all relevant conflicts that could have recurred within 14 years, generalisability is therefore not a problem, the findings apply to all relevant cases. In order to combat this to some extent a robustness test was performed using the same variables. This included fewer cases, making it less generalisable than the main analysis. This analysis also gave the result that TRCs had a negative relationship with the recurrence of
18 conflict. In future, researchers will have more data to analyse as more time will have passed.
Making them more generalisable and increasing the validity of those studies.
Since the long-term effects of TJ processes in general has not been fully researched, this analysis also ran a regression of the effects of having any TJ process at all. And it might be worth noting that any TJ process had an impact on the reduction of the likelihood of conflict recurrence. This is likely due to there being an effort of reconciliation in the country and should not be surprising. The test mas mainly run to see the difference any type of TJ makes in order to make the impact of TRCs more concrete. This might be an area for future research as well as the specific impacts of TRCs.
The findings in this paper does to some extent fulfil the causal criteria from Kellstedt and Whitten (2018, p. 60). The causal mechanism is still viewed as plausible, although they have not been tested in this study, there are theoretical reasons to believe that those are the
mediating variables for the reduced likelihood of recurrence of conflict could be due to addressing grievances, commitment problems and the information gap as well as increasing costs for armed conflict. The causal direction can be confirmed as the recurrence or lack thereof is measured within 14 years of conflict termination, with the first observation being after the termination of conflict and establishment of, or beginnings of the establishments of, TRCs. Covariation can be established in the data observed, although does not hold statistical significance, meaning this covariation could be dubious. As for confounding variables, there are some issues. There is always a risk for unknown variables to affect both the dependent and independent variable. For future research inclusion of more possible confounders is therefore recommended. Limited by the scope of this study, the main confounding variables that were found were controlled for.
Consequently, there is a possibility that there is some alternative explanation. One of these could be the types of conflicts that leads to the establishment of TRCs, and the states where one has the opportunity to. In order to establish a TRC there needs to be intention in the country to move past the conflict. This could be a reason for the reduced numbers of conflict recurrence in those countries. In order to combat this future research could attempt to do a small-N comparative study between countries where a TRC was established and where it was not. Then using those findings as control variables in future quantitative analysis. There is
19 also need for the resources to establish a TRC. This was somewhat controlled for as GDP could be an indicator of these resources.
Another possible reason for the different success rates of TRC could be related to the internal factors. Different groups in society have different impacts on the adoptions of TRCs as well as their power. Civil society organisations need a strong base of international support in order to work both from grassroots as well as apply pressure from above. The reason, therefore, some truth commissions are more successful than others are due to a strong civil society (Zvobgo, 2020, pp. 1–3). There is need for greater study of the process of international support for the civil society. Analysis of this is beyond the scope of this specific paper, however, and is left to future researchers.
As for the general discussion of TRCs, this paper claims that there is a comparative advantage to the establishment of them in the post-conflict context. This can be seen when controlled for other factors for conflict recurrence. The main counterclaim to this could be an unexplored confounding variable in the types of countries that choose to establish a TRC. It is possible that these countries, or their leaders, have a vested interest in a more stable country. They might also be willing to implement changes and listen and take what the commission suggest into consideration. This would mean that if the international community were to put pressure to force an establishment, they might not work as effectively. There are, however, other possible benefits of TRCs, not explored in this paper, which could still have an effect in countries that have less vested interest. Building on what Zvobgo (2020, pp. 1–3) claims, there are also benefits from the involvement of the international community on TRC success.
Research building on this paper could therefore have policy implications for the international community.
This paper gives no definitive answer in the discussion of the main reason for armed conflict between greed and grievances. One conclusion that could be drawn from these findings is that, at least for these conflicts specifically, there was a significant difference between conflicts which had a TRC as compared to other TRCs. This means that TRCs has a comparative advantage to the other types. TRCs should be better at addressing grievances than the others. One could argue that as this is the only cause TRCs should be significantly better at addressing are grievances. The other main reason for conflict recurrence that TRCs
20 likely perform better to avoid is related to governance. TRCs are temporary bodies and should not be viewed as the types of democratic institutions that have been research previously, but rather as a bridge from armed conflict. As this is not tested in this study no conclusions can be drawn from it, however. This study has shown that the reasons for conflict recurrence may be one that was hypothesised.
5. Conclusion
Through the use of statistical tools this paper aimed to see the effects on the likelihood of conflict recurrence when there was an established TRC. It is argued that there is a relationship between conflict not recurring and the establishment of TRC. Through the use of
multivariable regression while controlling for some aspects that might affect the success of TRCs specifically and conflict recurrence generally. The hypothesis was found to have some support in the findings. From this one could draw the conclusion that one should aim to always establish a TRC in a TJ processes. Due to the limitations of this study that may be to rash, however. The main conclusion to draw from this study is that there is a link between establishment of TRCs and the probability of conflict recurrence. How strong that connection is remains to be seen.
The main doubt of the conclusion that there is a connection between TRCs and conflict not recurring comes from the lack of control variables. One of these is the possible connection between countries that choose to establish a TRC that also causes the conflict to not recur in those counties. There are also possible other confounding variables that would benefit the paper if they were included. The reason for them not being included in this paper is partly due to the inexperience of the author with quantitative methods, making it more difficult to find and ensure reliability of datasets online. The other main reason is that this paper was written during the second wave of Covid-19 in 2020-2021. This means that isolation was strongly recommended, impacting the possibility to discuss with peers.
The implication of this is that there is a strong incentive for the international community to continue helping and applying pressure to ensure post-conflict societies to establish TRCs after a violent conflict. This seems to hold true for especially violent conflicts as well. This could have policy implications for the international community in post-conflict context after intrastate wars. As the intent of most peace processes is to use the opportunity to create more
21 democratic societies, the reduction of deaths due to armed conflict should greatly aim their goals in general.
In conclusion this paper establishes, with reservation for possible limitations, that there is a negative relationship between the establishment of TRC post conflict and the likelihood of conflict recurrence as compared to other forms of TJ. This claim holds when looking at conflict recurrence after 20 years and 5 years as well as 14 years, when controlling for some other factors.
5.1. Avenues for future research
It is possible that there are specific types of countries where TRCs are more applicable than others. It is also possible that one of these reasons could also lead to an increased or decreased likelihood of conflict recurrence. Future research should try to identify these characteristics in order to control for them in the future. This was beyond the scope of this paper, however.
22
6. Bibliography
Begley, T. B. C. (2020) ‘Strengthening the Right to Know through Truth and Reconciliation Commissions’, Human Rights Brief, 23(1), pp. 25–31.
Binningsbø, H. M., Loyle, C. E., Gates, S. and Elster, J. (2012a) ‘Armed conflict and post- conflict justice, 1946–2006: A dataset’, Journal of Peace Research, 49(5), pp. 731–740. doi:
10.1177/0022343312450886.
Binningsbø, H. M. , Loyle, C. E., Gates, S. and Elster, J. (2012b) ‘PCJ dataset’, Journal of Peace Research. SAGE Publications Ltd, 1 September. Available at: http://www.justice- data.com/pcj-dataset (Accessed: 10 November 2020).
Binningsbø, H. M. , Loyle, C. E., Gates, S. and Elster, J. (2012c) ‘Post-Conflict Justice Dataset Codebook’, Justice data, 1 September. Available at: http://www.justice-data.com/pcj- dataset/PCJ%20codebook%20-%20Binningsb%C3%B8%20et%20al%20JPR_49(5).pdf (Accessed: 10 November 2020).
Bozzoli, B. (1998) ‘Public ritual and private transition: the truth commission in Alexandra township, South Africa 1996’, African Studies, 57(2), pp. 167–195. doi:
10.1080/00020189808707894.
Buhaug, H., Cederman, L.-E. and Gleditsch, K. S. (2014) ‘Square Pegs in Round Holes:
Inequalities, Grievances, and Civil War’, International Studies Quarterly, 58(2), pp. 418–431.
doi: 10.1111/isqu.12068.
Cilliers, J., Dube, O. and Siddiqi, B. (2016) ‘Reconciling after civil conflict increases social capital but decreases individual well-being’, Science, 352(6287), pp. 787–794. doi:
10.1126/science.aad9682.
Collier, P. and Hoeffler, A. (2004) ‘Greed and grievance in civil war’, Oxford Economic Papers, 56(4), pp. 563–595. doi: 10.1093/oep/gpf064.
Dancy, G., Kim, H. and Wiebelhaus-Brahm, E. (2010) ‘The Turn to Truth: Trends in Truth Commission Experimentation’, Journal of Human Rights, 9(1), pp. 45–64. doi:
10.1080/14754830903530326.
23 Fearon, J. D. (1995) ‘Rationalist Explanations for War’, International Organization, 49(3), pp. 379–414.
Feenstra, R. C. I. (2019) ‘Penn World Table 9.1’, Groningen Growth and Development Centre. Available at: https://www.rug.nl/ggdc/productivity/pwt/pwt-releases/pwt9.1 (Accessed: 8 January 2021).
Gapminder (no date) ‘GD001’, Gapminder. Available at:
https://www.gapminder.org/data/documentation/gd001/ (Accessed: 28 December 2020).
Gibson, J. L. and Louis, S. (2004) ‘Does truth lead to reconciliation? Testing the causal assumptions of The South African Truth and Reconciliation process’, American Journal Of Political Science.
Harbom, L., Melander, E. and Wallensteen, P. (2008) ‘Dyadic Dimensions of Armed Conflict, 1946—2007’, Journal of Peace Research, 45(5), pp. 697–710. doi:
10.1177/0022343308094331.
Hegre, H. and Nygård, H. M. (2015) ‘Governance and Conflict Relapse’, Journal of Conflict Resolution, 59(6), pp. 984–1016. doi: 10.1177/0022002713520591.
Humphreys, M. and Weinstein, J. M. (2008) ‘Who Fights? The Determinants of Participation in Civil War’, American Journal of Political Science, 52(2), pp. 436–455. doi:
https://doi.org/10.1111/j.1540-5907.2008.00322.x.
Kellstedt, P. M. and Whitten, G. D. (2018) The fundamentals of political science research.
Third edition. New York: Cambridge University Press.
Kreutz, J. (2010) ‘How and When Armed Conflicts End: Introducing the UCDP Conflict Termination Dataset’, Journal of Peace Research, 47(2), p. 12.
Kubicek, P. and Walker, C. (2020) ‘Do Truth and Reconciliation Committees Improve
Human Rights? Evidence From Africa’, The Journal of the Middle East and Africa, 11(3), pp.
295–310. doi: 10.1080/21520844.2020.1808401.
Langer, A. (2005) ‘Horizontal Inequalities and Violent Group Mobilization in Côte d’Ivoire’, Oxford Development Studies, 33(1), pp. 25–45. doi: 10.1080/13600810500099634.
24 Millar, G. (2012) ‘“Our brothers who went to the bush”: Post-identity conflict and the
experience of reconciliation in Sierra Leone’, Journal of Peace Research, 49(5), pp. 717–729.
Nadler, A. Malloy, T. and Fisher, J. D. (2008) Social Psychology of Intergroup
Reconciliation: From Violent Conflict to Peaceful Co-Existence. Oxford University Press, USA.
Öberg, M. (2002) The onset of ethnic war as a bargaining process: testing a costly signaling model. Uppsala, Sweden: Dept. of Peace and Conflict Research, Uppsala University (Report, no. 65).
Olsen, T. D. et al. (2010) ‘When Truth Commissions Improve Human Rights’, International Journal of Transitional Justice, 4(3), pp. 457–476. doi: 10.1093/ijtj/ijq021.
Pettersson, T. and Öberg, M. (2020) ‘Organized violence, 1989–2019’, Journal of Peace Research, 57(4), pp. 597–613. doi: 10.1177/0022343320934986.
Pruitt, D. G. and Kim, S. H. (2004) Social conflict: escalation, stalemate, and settlement. 3rd edn. New York: McGraw-Hill.
Quinn, J., Mason, T. and Gurses, M. (2007) ‘Sustaining the Peace: Determinants of Civil War Recurrence’, International Interactions, 33, pp. 167–193. doi: 10.1080/03050620701277673.
Roosa, J. (2007) ‘How Does a Truth Commission Find out What the Truth Is? The Case of East Timor’s CAVR’, Pacific Affairs, 80(4), pp. 569–580.
Ross, M. L. (2004) ‘How Do Natural Resources Influence Civil War? Evidence from Thirteen Cases’, International Organization, 58(1), pp. 35–67.
Shaw, R. (2007) ‘Memory Frictions: Localizing the Truth and Reconciliation Commission in Sierra Leone’, International Journal of Transitional Justice, 1(2), pp. 183–207. doi:
10.1093/ijtj/ijm008.
Stahn, C. (2001) ‘Accommodating Individual Criminal Responsibility and National Reconciliation: The UN Truth Commission for East Timor’, American Journal of International Law, 95(4), pp. 952–966. doi: 10.2307/2674655.
25 Stanley, E. (2001) ‘Evaluating the Truth and Reconciliation Commission’, The Journal of Modern African Studies, 39(3), pp. 525–546.
Svensson, I. (2007) ‘Bargaining, Bias and Peace Brokers: How Rebels Commit to Peace’, Journal of Peace Research, 44(2), pp. 177–194. doi: 10.1177/0022343307075121.
Themnér, L. and Wallensteen, P. (2011) ‘Armed conflict, 1946–2010’, Journal of Peace Research, 48(4), pp. 525–536. doi: 10.1177/0022343311415302.
UCDP (no date) Definitions, Uppsala Conflict Data Program. Uppsala University, Sweden.
Available at: https://www.pcr.uu.se/research/ucdp/definitions/ (Accessed: 14 December 2020).
Walter, B. F. (1997) ‘The Critical Barrier to Civil War Settlement’, International Organization, 51(3), pp. 335–364.
Walter, B. F. (2010) ‘Conflict Relapse and the Sustainability of Post-Conflict Peace’, World Development Report background papers, 1(1), p. 41.
Wiebelhaus-Brahm, E. (2010) ‘Truth Commissions and Other Investigative Bodies’, in Bassiouni, M. C. (ed.) The pursuit of international criminal justice: a world study on
conflicts, victimization, and post-conflict justice. Antwerp; Portland: Intersentia, pp. 477–574.
Willer, R. (2009) ‘Groups Reward Individual Sacrifice: The Status Solution to the Collective Action Problem’, American Sociological Review, 74(1), pp. 23–43. doi:
10.1177/000312240907400102.
Zvobgo, K. (2020) ‘Demanding Truth: The Global Transitional Justice Network and the Creation of Truth Commissions’, International Studies Quarterly, 64(3), pp. 609–625. doi:
10.1093/isq/sqaa044.
26
Appendix
Table 4. Logistic regression conflict recurrence likelihood (any TJ)
Dependent variable:
Conflict recurrence
PCJ -0.199***
(0.049)
terminationOther 0.053
(0.061) terminationVictory -0.247***
(0.063)
civilwar -0.066
(0.047)
Constant 0.477***
(0.061)
Observations 326
Log Likelihood -171.752
Akaike Inf. Crit. 353.504
Note: *p<0.1; **p<0.05; ***p<0.01
27 Table 5. Robustness test conflict recurrence likelihood
Dependent variable:
Conflict recurrence
Within 5 years Within 20 years (model4) (model5)
TRC -0.150 -0.039
(0.112) (0.160)
terminationOther 0.021 0.036
(0.065) (0.082) terminationVictory -0.157*** -0.152**
(0.060) (0.075)
civilwar -0.084* -0.080
(0.049) (0.061)
GDP -0.00000 -0.00000
(0.00000) (0.00000)
Constant 0.227*** 0.267***
(0.057) (0.072)
Observations 173 162
Log Likelihood -38.257 -66.567
Akaike Inf. Crit. 88.515 145.134
Note: *p<0.1; **p<0.05; ***p<0.01 SE in parentheses
