Violence, Death, and Displacement in Darfur 1
A System Dynamics A pproach to Analyzing Violence, Death, and
Displacement in Darfur
Jeremy B. Sato
Ph.D. Student, Electrical and Systems Engineering
Washington University in St. Louis
2 Hobbs Mill Court
St. Charles, Missouri 63303, USA
Phone: 636-922-9921
jbs2@ wustl.edu
Bonnie J. Stansen
MSW Student, George Warren Brown School of Social Work
Washington University in St. Louis
8032 Venetian Drive
Clayton, Missouri 63105, USA
Phone: 314-283-8004
bstansen@ wustl.edu
This paper presents a System Dynamics approach to analyzing the violence and widespread
death and population displacement in Darfur, Sudan, as observed since 2003. We lay a founda-
tion for using simulation to investigate the underlying structure and effects of violence; model
analysis indicates that the dynamics may be driven more by the population at risk than by ag-
gressor intent. This model can aid in future policy analysis and establishes a foundation for us-
ing System Dynamics to understand the structure and pattern of genocide. We present several
challenges to analyzing the Darfur crisis including observability, information delays, and the
choice of metrics. Finally, we discuss modeling results and options for intervention and propose
several policy questions and areas for future research.
Keywords: Darfur, Sudan, Violence, Refugees, Population Displacement, Policy, Genocide
Our research focuses on the dynamics of the mass killing and displacement of the people
of Darfur, in the Westem region of Sudan, since 2003.
In Darfur, violence has led to an estimated 450,000 deaths and 2.5 million displacements,
including refugees fleeing to Chad and internally displaced persons (IDPs) in settlement camps
within Darfur. The affected populations are facing widespread disease and starvation, and many
humanitarian organizations have been unable to serve in the region due to the ongoing violence,
lack of security, and lack of access. In many cases, the Sudanese government limits the form of
international aid. Additionally, since the peace agreement in May, the under-staffed and under-
funded African Union (AU) Forces have been unable to stabilize the violence and IDP crisis.
Sudan refuses to allow additional help from United Nations (UN) troops.
The observed patterns of violence - mass killings, destruction of land, physical abuse,
and forced displacement - that have evolved since the conflict began in February 2003 demon-
strate the dynamic nature of this problem; it has gone through periods of acceleration and decel-
eration, indicating dynamic behavior. We note many citations from literature on the causal rela-
Violence, Death, and Displacement in Darfur 2
tionships between the actions of resistance groups, militias, and their effect on humanitarian aid.
Policy is influenced by international awareness of the crisis and actions carried out affect the re-
sistance and militia groups, as well as the Sudanese government. These causal relationships cre-
ate a highly complex and dynamic environment where the coupling of problems demands careful
evaluation of policy. This paper presents a method for explicitly analyzing the intemal dynamics
of the Darfur crisis. In addition, it presents an analytical tool that helps us develop an endoge-
nous explanation for the crisis, serves as a method for deciding the case of genocide, and acts as
a test-bench for simulating the effectiveness of future policy options.
Background of Darfur
Geography
Darfur is the largest region of Sudan, situated in the West, and is about the size of France.
Darfur borders Chad, the Central African Republic (CAR), and Libya. Only 15 percent (six mil-
lion) of Sudan’s population lives in Darfur. Sudan is the largest country in Africa and is ap-
proximately one quarter the size of the United States. Water, a critical resource in Sudan and
throughout Africa, is channeled by the Nile River and its tributaries. This covers an area of
about 2.5 million square kilometers with a population of nearly 40 million, primarily living in
urban centers. Sudan’s political stability is of great international interest because of the number
of countries it borders: Egypt in the North; Eritrea, Ethiopia, and the Red Sea in the East;
Uganda, Kenya, and the Democratic Republic of Congo in the South; and Chad, the CAR, and
Libya in the West. Sudan is rich in petroleum and has small reserves of other minerals such as
iron ore, copper, zinc, tungsten, gold, and silver (World Bank, 2005).
Ethnography
For political strategies and expediency, the Darfurian identities have been polarized into
“Arab” and “African”. The Goverment of Sudan (GoS) has exploited these categorizations for
its own benefit (De Waal, 2004b). Despite this categorization, however, it is rarely possible to
tell, on the basis of skin color, to which group a person of Darfur belongs. Understanding this
subtlety is essential to analyzing the crisis, since preconceived notions shape the way the dynam-
ics are studied, beginning with selection of variables. There are between 40 and 90 ethnic groups
in Darfur (Flint & De Waal, 2005). Most groups have been in the area for centuries and all are
Muslims of one sect or another, mainly Sufi or Madhi. Until the Darfur conflict, the different
groups mixed freely and intermingled, uniting in marriage and family groups, and forming pas-
toral, farming groups and semi-nomadic herding groups. These nomadic groups occupied well-
defined areas and followed established courses (De Waal, 2005).
Political/Military Conflict
During the last few decades, several factors contributed to the violence in Darfur. Wide-
spread use of the land and environmental factors have resulted in desertification. The Sahara de-
sert is gradually expanding into the Sahel region, shrinking water supplies and reducing available
land for herding and even less arable land for farming. In the wake of the drought and famine of
1984-1985, a series of local conflicts erupted. Though the people of Darfur already lived in a
resource-poor region, the herding groups were set against the farmers in a struggle for diminish-
ing resources. No government intervention occurred to alleviate the conflict, and as a result,
both sides armed themselves.
Violence, Death, and Displacement in Darfur 3
The GoS marginalized the region of Darfur, ignoring the basic infrastructure, health and
educational needs of its people. Additionally, Darfur was not included in the negotiations and
subsequent peace agreement that ended Sudan’s long civil war in the South. “Cognizant of the
fact they had been left out of the peace process whose results would not help them, and frustrated
by government inaction on past land conflicts and persistent complaints for being marginalized
by the GoS, including loss of their jobs to Arabs, sections of non-Arab Darfurians launched at-
tacks on several army positions” (Nasong’o &Murunga, 2005). The GoS, fearing that an armed
rebellion would inspire resistance groups in other parts of the country, began the campaign
against Darfur. The Sudanese military began aerial bombings and the GoS armed a paramilitary
group called the Junjaweed. The Junjaweed were needed because the majority of the armed
forces were from Darfur (Nasong’o & Murunga, 2005); the capital city, Khartoum, feared that
the forces would refuse to fight family and friends if they were sent to Darfur. Encouraging the
ideology of Arab supremacy (Arabization) established among some of the Bedouin groups by
Libyan President Muammar Gaddafi during the 1980s, the GoS recruited and armed the Jun-
jaweed, who “stole, burned, mutilated, killed and raped - subjecting tiny communities to uni-
maginable horrors” (De Waal, 2004). The Junjaweed, often with full army support, eradicated
whole villages and destroyed wells and fields so that they could not be used again.
The government policy was not just one of counterinsurgency but to destroy a way of
life. The Junjaweed and the Sudanese military either stole or destroyed all food and animals.
Anything of value was either removed or burned. Villages, including schools, clinics and mar-
kets were decimated. Wells were polluted - often with dead bodies - and fields were bumed and
salted. With over two million people displaced, the GoS then began blocking humanitarian aid
efforts. The Junjaweed blockaded displacement camps and prevented delivery of aid. In Khar-
toum, supplies would not be released to agencies. On the few roads that exist in Darfur, road-
blocks and hijackings occurred regularly.
Since February 2003, an estimated 450,000 people have died and over two million have
been displaced. What is being labeled as a conflict between “Arabs” and “Africans” to the inter-
national community is an attempt to obfuscate by the central Sudanese government. The GoS
has attempted “to suppress a rebellion motivated by a demand for basic human rights and re-
gional autonomy” (Howard-Hassman). The Junjaweed, together with the Sudanese military,
have continued to systematically commit acts of mass murder, rape, torture and mutilation in
Darfur (State Department, 2004).
Genocide
The United Nation’s (UN) Convention on Genocide of 1948 (signed by over 100 nations in
1951) defines genocide as a punishable crime. Article Two of the Convention defines genocide
as “any of the following acts committed with the intent to destroy, in whole or in part, a national,
ethnic, racial or religious group, as such:
e Killing members of the group
Causing serious bodily or mental harm to members of the group
Deliberately inflicting on the group conditions of life calculated to bring about its physi-
cal destruction in whole or in part
e Imposing measures intended to prevent births within the group
¢ Forcibly transferring children of the group to another group (United Nations, 1948)
Violence, Death, and Displacement in Darfur 4
Article Three of the Convention continues on to specify that in addition to the act of genocide
itself, conspiracy, direct and public incitement, and the attempt to commit genocide, as well as
complicity in genocide are equally punishable acts. Beyond punishing the perpetrators of such a
crime, the convention also imposes the general duty on the signatories “to prevent and punish”
those who commit genocide.
Model Structure
Modeling Goals
Our analysis is directed towards the UN and intemational community in service to the
people of Darfur. The UN has a vested interest in the Darfur crisis because of their ability and
obligation to respond to the crisis, as the Sudanese government has been largely ineffectual. In
addition, a UN recognition of genocide leads to additional obligations for UN members. There-
fore, the objective of our modeling and analysis is to provide insight for the UN and international
community on the effectiveness of alternative policies.
In this paper we focus on the goal of restoring peace to Darfur by ending and preventing
future violence-related deaths (both direct and indirect), and displacement (both refugees and
IDPs). This goal is measured by tracking the deceleration, stabilization, and eventual decrease in
displacements and deaths due to disease, starvation, and violence. Future works will address sub-
sequent goals: insuring that all the people of Darfur are geographically, socially and economi-
cally reintegrated into Darfur society; and, opening the political space necessary for reconcilia-
tion and creating opportunities for interchange with the UN, AU, and neighboring countries (e.g.
Chad and the Central A frican Republic).
Objective
Our objective is to present a System Dynamics approach to analyzing the crisis in Darfur
from February 2003 to the present (A pril 2007). We describe a simulation model and key causal
relationships and feedback structures to analyze the trends of displacement and deaths due to vio-
lence. We explore the following seven questions:
What are the primary dynamics that might drive excess mortality and displacement?
What are the current dominant feedback loops?
What are possible effects on neighboring countries?
What are the linkages between structure and behavior in cases of genocide?
What are the opportunities for intervention?
How is policy effectiveness measured?
How is model success determined?
SF SER GOS
Model Formulation
Our research focuses on observed events, phenomena, and time-series data from February
2003 to present. We focus on the following main stock variables which form the basis for future
policy and action (see A ppendix A for the complete list of included and excluded variables):
e Non-displaced people of Darfur
e Displaced people of Darfur in transit
e Displaced people of Darfur in settlement camps
e Deaths in Darfur due to direct violence
Violence, Death, and Displacement in Darfur 5
e Deaths in Darfur due indirectly to violence and displacement (starvation and disease) and
lack of aid
e Militias and Sudanese military
e Resistance groups
Exogenous model variables consist of the fractional rates, time delays, and constants that govern
the flow processes between stocks.
Figure 1 illustrates how the people of Darfur transition through various phases. Initially,
all people of Darfur begin at home. Conflict then causes people to join resistance groups and
militias, causes displacement, and results in death due to violence. Because people in transit are
more susceptible to attacks, we disaggregate the process of displacement into two stocks: “in:
transit” and “at displacement camps”. This also allows us to capture the dynamics of multiple-
displacement. Death can occur from all states. Militias, military, and resistance groups grow
due to both voluntary and involuntary recruitment. Our simulation model is shown in Appendix
B and model equations are documented in A ppendix C.
Militias /
Military
Sudanese
recruitment
recruitment
People of Darfur
displacement arrival
Excess
At Home In Camp
a | Deaths
H 5 mortality
reintegration multiple fron
displacement
& violence
F
Resistance LE"
Groups
Observed Flows i>
Potential Flows —>
Figure 1. Simplified Stock and Flow Diagram for Genocide and Displacement in Darfur
The model of the Darfur crisis consists of six key components:
1. Militias and Military
2. Resistance Groups
3. The Process of Killing
4. The Process of Displacement
Violence, Death, and Displacement in Darfur 6
5. Displacement Camps and Multiple Displacement
6. Disease, Starvation and Aid
Militias and Military
Militia and military violence are abstracted as a single concept since both groups inflict
violence against resistance groups and the people of Darfur. Sources describe coordinated at-
tacks between the GoS and the Junjaweed where the Government attacks first with bombs, and
the Junjaweed follow with attacks by horseback and camelback (Patrick, 2005). One factor that
increases the number of militia and military aggressors is the process of involuntarily recruiting
children into armed groups (Office of UN Deputy, 2006).
Militia and military growth is modeled as a Bass Diffusion process, shown in Figure 2.
The first important assumption is that aggressors recruit new aggressors, both voluntarily and
involuntarily. Societal pressures contribute to voluntary recruitment (diffusion element), and as
noted above, there is evidence of involuntary recruitment of children (forcing function element).
This structure gives rise to S-Shaped growth of militia/military membership, as seen in Figure 3,
where growth accelerates initially as potential members are plentiful, and then decelerates to
equilibrium as the stock of potential members drain. Figure 4 shows the predicted behavior using
the model’s current parameters and initial conditions, where initial aggressors and a high invol-
untary recruitment rate dominate the Aggressor’s goal-seeking feedback loop throughout the
time horizon.
Fractional Rate of "
Involuntary Fractional Rate of
Recruitment Voluntary Recruitment
a5 VK a cy Joining Militias, Military =
SS Resnaed Pherae eonel Hal of Voluntary Jeena *
Potential - otential A ggressors * A ggressors
Agaressots J jolning Milltas, Aggressors (Potential A ggressors + A ggressors) +
analy Fractional Rate of Involuntary Recruitment *
tial Potential A ggressors
- 1S>
Figure 2. Structure of Militia and Military Growth
Aggressors
200,000
150,000
i 100,000
50,000
A
0 12 23 35 46
Time (Month)
Violence, Death, and Displacement in Darfur 7
Figure 3. Aggressors vs. Time, S-Shaped Growth
Aggressors
40,000
32,500
25,000
Aggressor
17,500
10,000
0 12 23 35 46
Time (Month)
Aggressors : Current
Figure 4. Aggressors vs. Time, Goal Seeking Growth
Resistance Groups
Since resistance groups in Darfur took up arms in February 2003, the Sudanese govern-
ment has recruited and supplied militia forces in a sweeping campaign of attack against the civil-
ian population of Darfur (Leaning, 2004). Boys affected by the violence become ripe for re-
cruitment into these resistance groups (Reeves, 2006).
The growth of resistance group involvement is modeled as a first-order process as shown
in Figure 5, govemed by a fractional recruitment rate. This is a balancing feedback loop where
the number of new recruits is a function of the potential new recruit population. The growth
shown in Figure 6 is slightly damped due to additional draining of the stock “Darfurians At
Home” from death and displacement. The recruitment process can be expanded into an SI or
Bass Diffusion process, similar to the previous structure for militia membership growth.
Fractional
Recruitment Rate
-
Darfurians in +/+ @& | pdarturiansat | |!
Resistance Groups Home
Recruitment .
Recruitment = Darfurians at Home *
ation
Fractional Recruitment Rate
Figure 5. Structure of Resistance Group Growth
Violence, Death, and Displacement in Darfur 8
Darfurians in Resistance Groups
Datsion
s
s
0 12 23 35 46
‘Time (Month)
Dacisans in Resistance Gmups + Cument
Figure 6. Resistance Groups vs. Time, First-Order Process
The Process of Killing
The militias target civilians sharing the same ethnicity as the resistance groups in Darfur
in an effort to stem alleged support for enemy groups (Office of UN Deputy, 2006). Militias also
directly target young men and boys (Patrick, 2005). Groups that have been attacked (sedentary
farming communities) self-identify as not being aligned with groups that support militia groups
whereas self-identified Arab-communities are relatively untouched by the Junjaweed and gov-
emment forces (Patrick, 2005). The GoS has restricted access to Darfur over the past three
years, making it difficult for UN teams to estimate excess death due to violence. In 2005, how-
ever, the death rate was estimated at roughly 10,000 per month (Udombana, 2005).
The killing and displacement processes are identical in structure (compare Figures 7 and
9). Both depend on the number of militia or military members and the frequency at which they
attack the villages, as well as the size of the susceptible population. Both have the basic structure
necessary for S-shaped growth. The first part of the process is dominated by a reinforcing loop
due to the growth of aggressors and the large population at risk. The process then shifts loop
dominance to a balancing behavior due to the draining of the susceptible population due to dis-
placement and killing.
Violence, Death, and Displacement in Darfur 9
a B
Darfurians at ome_D_
Recruitment
Population
arfur>
Fixed Flow .
Threshold H stopulation of -
Killing Rate in ~ Village
Villages per Mortality
Aggressor
<Aggressors>
Village Mortality = A ggressors *
Excess Deaths in Killing Rate in Villages per Aggressor *
Villages MIN(1, Darfurians at Home /
(Fixed Flow Threshold H * Population of Darfur) )
Figure 7. Structure of Killing
Total Excess Deaths
600,000
450,000
5
300,000
fs ge
150,000 a
pa
0 =
0 2 23 35 16
Time (Month)
Total Excess Deaths : Current
Total Excess Deaths : Excess Deaths from Data
Figure 8. Total Excess Deaths vs. Time: Model Behavior and Reference Mode Data (Bureau of
Intelligence and Research, 2005)
The time-series data shown in Figure 8 starts at the beginning of the conflict and ends
two years into the conflict. The data represents excess deaths in Darfur beyond typical mortality
rates for the region. The initial population of Darfur in February 2003 was six million. As of
2005, 180,000 to 350,000 have been killed, 300,000 fled to Chad, and there were two million
IDPs in Darfur, with greater than three million who were conflict-affected (about half the popu-
lation of Darfur) (Patrick, 2005).
Currently, the model does not reflect AU forces. Another aspect excluded is the violence
within the Southern Liberation Army (SLA), which adds to the abuses suffered by civilians (Of-
fice of UN Deputy, 2006).
Violence, Death, and Displacement in Darfur 10
The Process of Displacement
Intensified fighting and insecurity has led to large population displacements, even when
the planting season encourages a retum to the farmlands (Office of UN Deputy, 2006). Large
movements of populations are recorded after military and militias clash with SLA forces (Office
of UN Deputy, 2006). Furthermore, an increase in abuse and violations against civilians has
been compounded by their lack of protection (Office of UN Deputy, 2006). Villages are looted
and bumed, resources are stolen, wells are poisoned, land and fields are destroyed, women are
raped, and populations are displaced into inhospitable IDP camps (Patrick, 2005).
The process of displacement uses the same structure presented previously for killing, as
indicated when comparing Figures 7 and 9. Displacement moves populations from home to in-
transit.
Displacement Rate
<Aggressors> Per Aggressor
ia Fixed Flow <Popul
| Threshold D Dar
Kot] ——
+ | - rt
Displacement Rate = A ggressors *
Displacement Rate per A ggressor*
MIN(1, Darfurians at Home /
(Fixed Flow Threshold D * Population of Darfur)
/
Darfurians at
Home
Figure 9. Structure of Displacement Process
Darfurians at Home
8M
65M
5M ——
}
35M —
2M
0 12 23 35 46
‘Time (Month)
Darfurians at Home : Current,
Figure 10. Darfurians at Home vs. Time
The model behavior in Figure 10 is consistent with current projections of IDPs and refugees.
Violence, Death, and Displacement in Darfur 11
Darfurians In Transit
40,000
30,000 an ——.
i | ee
20,000
° !
10,000
0
0 12 2 35 16
Time (Month)
Darfurians In Transit : Curent
Figure 11. Darfurians In Transit vs. Time
We explored modeling killing and displacement as a constant process, a first-order proc-
ess, or a constant process that transitions into first-order. Systematic acts of violence driven pri-
marily by aggressor size and capability may produce a constant rate of killing and displacement.
If violent acts are more random and, in addition to aggressor capability, are also driven by the
available population at risk, a first-order process is appropriate. A population that is drained at a
constant rate may eventually shift to a first-order flow once the susceptible population reaches
some fraction. In order to evaluate these processes, our model allows us to change structure
where killing and displacement are constant through some fraction of the initial population, and
then shifts into a first-order process.
Displacement Camps and Multiple Displacement
IDPs continue to be targeted by the militias due to their extreme vulnerability, and many
have been displaced multiple times (Office of UN Deputy, 2006). In addition, distributions of
food and non-food items have been low due to insecurity, lack of funding, and inaccessibility
which also forces large numbers of IDPs to flee the camps. (Office of UN Deputy, 2006) It is
difficult to estimate the number of IDPs because headcounts are not conducted every month, re-
sulting in a time lapse between displacement and registration. This, together with continuous
movements which are not easily tracked, results in significant delays before IDPs appear in sta-
tistics (Office of UN Deputy, 2006).
We model camp arrivals as a first-order process with an average travel time, as shown in
Figure 12, while also noting that people in transit are susceptible to attacks. The process of mod-
eling multiple-displacement is similar to the process for displacement. In our current model,
multiple displacement is caused by external violence. We have not yet analyzed factors of over-
capacity, lack of aid, and intra-IDP fighting (Office of UN Deputy, 2006). We estimate that the
average time to reach displacement camp is two weeks, although it is difficult to estimate this
parameter accurately due to the lack of data and variability.
Violence, Death, and Displacement in Darfur 12
Re-Displacement
<Aggressors> — Rate per Aggressor
Fixed Flow
Threshold R
<Population of
Darfur>
us
Darfurians In Leaving paeenet
Transit Camps Camps
Camps ~~ avg Time to Amriving at Camps = Darfurians In Transit /
Reach Camp Avg Time to Reach Camp
Figure 12. Structure of IDPs and Multiple-Displacement.
Figure 13 displays time-series data for the number of IDPs starting one year into the con-
flict through mid-2006. The model structures are not adequate for capturing the rise, overshoot,
and settling behavior, while still satisfying reasonable death estimates in the same simulation
run. The model reproduces the rise and fall with certain parameter settings; however, it is not
possible with the current structure to simultaneously track both IDP and death reference modes.
Darfurians in Displacement Camps
4M
\
0 12 23 35 46
‘Time (Month)
amps vs. Time: Model Behavior and Reference Mode
Data (Office of UN Deputy, 2006).
Disease, Starvation, and Aid
Many challenges prevent the adequate administering of aid to IDPs, resulting in deaths
due to disease and starvation. Humanitarian access in Darfur is affected by the degree of insecu-
rity, denial of access, and the continued harassment of humanitarian organizations and workers
(Office of UN Deputy, 2006). In addition, intra-SLA fighting causes humanitarian organizations
to withdraw (Office of UN Deputy, 2006). Funding shortages result in ration cuts (Office of UN
Deputy, 2006). It is difficult to find consistent, comprehensive estimates of mortality that distin-
guish disease- from starvation-related deaths.
Starvation affects members of a homogenous population independently and equally, and
at a very high level of abstraction is modeled as a first-order process, as shown in Figure 14.
Disease however has the added element of diffusion through contagious infected populations and
requires an SI or SIR structure at minimum. Although we did not implement this structure, it is a
natural extension to our model. Figure 15 shows the simulated trajectory of excess deaths at
camps.
Violence, Death, and Displacement in Darfur 13
+ 8
g Darfurians in
6 Displacement
Camps
Fixed Flow Fractional
Threshold C Starvation Rate
+
Killing Rate at — Displacement
Camps Per Aggressor Camp Mortality
_ . Displacement Camp Mortality = A ggressors *
“Aggressons> Killing Rate at Camps Per Aggressor *
MIN(1, Darfurians in Displacement Camps /
Excess Deaths at (Fixed Flow Threshold C * Population of Darfur) ) +
Displacement Camp Fractional Starvation Rate *
Darfurians in Displacement Camps
Figure 14. Structure of Displacement Camp Mortality
Excess Deaths at Displacement Camp
400,000
300,000
200,000
100,000
0
0 12 23 35 46
‘Time (Month)
ues Dh Displcnet Camp: Cumet
Figure 15. Excess Deaths at Displacement Camp vs. Time
Displaced population is not a useful metric for evaluating policy effectiveness. An increase in
displaced persons obviously is not desirable when the client's goal is to decrease displacement.
On the other hand, a decrease in the displaced population does not necessarily indicate
successful re-integration into society; it could indicate an increase in mortality or that more ID Ps
are in transit.
Validation and Testing
Structural Assessment
All stocks represent human populations, and therefore never go negative. In our model
this is assured by using first-order material delays or hybrid first-order, constant flow processes
which never drain stocks below zero. Stocks are only drained into other stocks or ultimately
Violence, Death, and Displacement in Darfur 14
death. The stocks represent a collectively exhaustive, mutually exclusive set of states under our
level of disaggregation. Limits to growth for each of the stocks is captured by the fact that
stocks receive no more than the total of the stocks that flow into them. That is, there are no open
sources or sinks in our stock flow diagram. The sum of all the “people of Darfur” stocks (includ-
ing death), therefore, is equal to the initial number of people of Darfur at home (six million).
This verifies flow conservation.
Behavior Reproduction
We compared simulation results against time-series data for death and displacement;
however, because the majority of data collected is inconsistent or incomplete, we have only
rough estimates. Behavior reproduction tests demonstrate that with the basic structures em-
ployed, we match qualitative trends; however, systematic bias remains indicating that missing
feedback loops are not yet captured.
Model Limitations
The model does not include economic or political factors, policy, or the dynamics of AU
Forces. After calibrating to past behavior, the next step is to incorporate observation feedback
that informs or influences policy. We are currently researching the dynamics of conflict between
militia and military members and resistance groups. Our current model focuses more on the ef-
fects of violence than the cause.
Model Suitability
This model is suitable for evaluating the effects of systematic or random acts of violence
against the people of Darfur and for understanding flow processes between states. We use the
model to test dynamic hypotheses derived from reports and literature on the crisis, and whether
they are capable of producing observed behavior. The model is a basic stock-flow structure, ex-
pandable to incorporate and test feedback links that affect death and displacement. The model
can be expanded to incorporate policy levers based on observation of the crisis.
Results
In response to the seven questions we explored throughout our analysis, we have summa-
rized our results and insights based on model exploration, testing, and sensitivity analysis.
1. The dynamics of mortality and displacement are driven primarily by two feedback loops:
the reinforcing loop that increases violence (growth in militia/military population and
violent acts), and the balancing loop of the susceptible population at risk. Our model cur-
rently does not model the dynamics of AU troop protection, or military and humanitarian
resource depletion, which may also contribute to the dynamics.
2. The dominant feedback loop currently driving displacement and mortality is the balanc-
ing loop associated with the depletion of the population at risk, as opposed to the rein-
forcing loop of increasing militia and military population. This is why even in the ab-
sence of AU or peacekeeping forces, the model predicts a deceleration in the number of
deaths and displacements. When compared to observed data, we also see that currently
the number deaths and new IDPs are decelerating or holding constant, despite the ineffec-
Violence, Death, and Displacement in Darfur 15
tiveness of AU presence. The depletion of the stock of people at risk is slowing down the
displacement and death rates, making it difficult to judge the individual impact of current
efforts and future intervention based on death and displacement statistics alone.
3. One possible effect on Sudan’s neighboring countries is that an abundance of new people
at risk (a fresh stock) could again accelerate the current crisis, if left unabated.
4. Important linkages between structure and behavior in the case of genocide may be found
in the flow process of displacement and killing. A systematic process may be more likely
to produce a constant flow of death and displacement, not as sensitive to the susceptible
population at risk until a large portion of the population has been removed. More random
acts of violence would resemble a first-order flow response. In our model for Darfur, we
can test for systematic or random violence by changing the point where displacement and
killing shifts from a constant flow to a first-order process. Initial simulation runs are in-
conclusive due to incomplete reference mode data sets.
5. Options for intervention fall into three categories: reducing vulnerability of the popula-
tion at risk (e.g. deploying additional peacekeeping forces); reducing the means of violent
incidents (e.g. affecting military resource supply or military/militia recruitment), and re-
ducing will for violent incidents (e.g. peace treaties, additional resources, political or eco-
nomic incentives).
6. Policy effectiveness can be measured in part by tracking additional displacements and
excess deaths due to disease, starvation, and violence. As we have shown, however, we
must also measure re-integration and gain more insight into what happens to people dur-
ing transit and re-displacement.
7. In order to use the model to design effective policy, we must first show that it is consis-
tent with data on displacement and deaths provided by the UN (UNICEF, OCHA and
WHO), USAID, CARE, and Doctors Without Borders. Then we must identify possible
lever points for intervention. Model success requires identifying high-leverage opportu-
nities to introduce balancing loops that would both stabilize the crisis and lead to an envi-
ronment conducive to reintegration efforts. Success of the model also requires the ability
to gain insight on why certain policies would be ineffective.
Implications and Next Steps
Even with the UN’s definition of genocide, detection in Darfur is complex. Our approach
is to study the conflict in Darfur as a system and investigate the dynamics. System Dynamics al-
lows a unique approach to investigate, compare, and contrast patterns of violence that arise from
different roots. Genocidal acts intentionally target national, ethnic, racial, or religious groups in
whole or in part. This creates a causal-loop structure distinct from other violent political actions
that can be tested through simulation. Policy designed to address genocide must consider the
effects of mass death and displacement on future generations. The loss of nearly half a million
people and the displacement of half the population of Darfur will have profound effects on rein-
tegration and future economic and social policy.
Violence, Death, and Displacement in Darfur 16
In modeling the crisis in Darfur and designing effective policy, two challenges arise.
First, there is a lack of data for people in transit, and it is difficult to obtain accurate excess mor-
tality rates. Also, the significant delay time between camp arrival and camp registration leads to
misestimates of the current crisis, making it difficult to assess the effectiveness and conse-
quences of policy.
Our model has focused on the social and military domains, and the effects of violence.
Future work will expand the baseline model to incorporate political, economic, religious, and
geographic factors that influence conflict and violence. In developing the conflict model, we
will consider the effects of military and humanitarian resource availability, accessibility, peace-
keeping efforts, and AU troop protection.
Our goal is to use the model to evaluate alternative policies and their effectiveness, pro-
jecting several years into the future, with the goal of answering the following policy-related
questions:
e Can/should the UN declare genocide in Sudan (Patrick, 2005)? If so, what actions fol-
low?
e Should the international community impose targeted sanctions against Sudan and Suda-
nese nationals accused of war crimes in Darfur (BBC News, 2006)? What type of sanc-
tions (economic, travel) (Patrick, 2005)?
e What is the most effective deployment of peacekeeping troops (who, where, when, and
how many)?
e Should a protection force be commissioned, in addition to the AU, to counter all aspects
of violence with a full mandate to protect civilians at risk (Patrick, 2005)?
e What is an adequate mission deployment and how do we define its mandate (Pendergast,
2007)?
e What is the most effective use of aid workers (who, where, when, how many, and what
type of aid)?
e Should new or revised peace agreements be enacted? How (Kristof, 2006)?
To answer these questions, several aspects of the crisis entail further investigation and
data mining. The dynamics of voluntary recruitment into both resistance forces and militias re-
quire research and validation. Additional information is needed to calibrate the fractional re-
cruitment rates of children, the initial number of potential militia and military members, and the
initial number of actual militia and military members in February 2003. Also, the time from
when boys are affected by violence and when they are susceptible to recruitment needs to be ana-
lyzed. Further work is necessary to study the dynamics of how AU forces affect resistance
groups, militias, and the GoS, and the protection they provide to IDPs and humanitarian aid
workers. More research is needed to study displacement rate and its relationship to military, mili-
tia, and at-risk population size. Another area of concern is the anti-D PA sentiment in IDP camps:
this has led to attacks on AU personnel and assets, forcing the AU to reduce its presence in a
number of IDP camps, and subsequently increasing camp vulnerability to outside attack (Office
of UN Deputy, 2006). Lastly, we seek current data that tracks population movements, death at
displacement camps and the phenomena of multiple displacement.
References
British Broadcasting Company News. 2006. Q&A: Sudan's Darfur conflict. British Broadcasting
Company, http://news.bbc.co.uk/2/hi/africa/3496731.stm (accessed 9/13/2006)
De Waal, A. 2005. Briefing: Darfur, Sudan: Prospects for peace. African Affairs, 104,
no.414:127-135.
De Waal, A. 2004a. Counter-insurgency on the cheap. London Review of Books, no.15 (August
5):
De Waal, A. 2004b. Who are the Darfurians? Arab and A frican identities, violence and external
engagement. J ustice Africa.
Edkins, J. 2005. Mass starvations and limitations of famine theorizing. International Develop-
ment Society Bulletin, 33, no.4: 12-18.
---Eighth report of the Secretary General to the Security Council. 2006. Darfur: Whose responsi-
bility to protect? no.(October 26). Human Rights: New Y ork.
Flint, J. & De Waal, A. 2005. Darfur: A short history of a long war. Zed Books: London and
New Y ork.
Howard-Hassman, R.E. 2005. Genocide and state-induced famine: global ethics and westem re-
sponsibility for mass atrocities in A frica. Perspectives on Global Development and Tech-
nology, 4, no.3-4: 487-516.
Kristof, N. 2006. China and Sudan, blood and oil. The New York Times, National edition, April
23, Sec. 1:15.
Kristof, N. 2006. When genocide worsens. The New York Times, National edition, July 9,
Sec1:13.
Leaning, J. 2004. Diagnosing genocide - the case for Darfur. New England J ournal of Medicine,
351, no.8:735-739.
Nasong’o, S. W. & Murunga, G.R., 2005. Lack of consensus on constitutive fundamentals:
Roots of the Sudanese civil war and prospects for settlement. African and Asian Studies,
4, 1-2: 51-82.
Office of UN Deputy Special Representative of the UN Secretary-General for Sudan, UN Resi-
dent and Humanitarian Coordinator 2006, July 1. Darfur Humanitarian Profile No. 24.
New Y ork: United Nations.
Patrick, E. 2005. Intent to destroy: the genocidal impact of forced migration in Darfur, Sudan.
Journal of Refugee Studies, 18, 411 - 427.
Pendergast, J. 2007. The answer to Darfur: How to resolve the world’s hottest war. Enough: The
Project to Abolish Genocide and Mass Atrocities. International Crisis Group
Reeves, E. 2006. Darfur, eastern Chad face unconstrained human destruction. Sudan Tribune,
November 27.
Reeves, E. 2006. The future of Darfur: Next causality, The New Republic. May 15, 234: 18-20.
The Res Publica, 2006. The Darfur resistance: who they are, what they want, and what they need
to help create a new Sudan. Retrieved from http://www.therespublica.com. (accessed Oc-
tober 10, 2006).
Timiraos, N. 2006,. Sudan - divestment activists get act together. The Wall Street J ournal Na-
tional Edition, July 19, Sec. C1.
Udombana, N. J. 2005. When neutrality is a sin: the Darfur crisis and the crisis of humanitarian
intervention in Sudan. Human Rights Quarterly, 27: 1149-1199.
---United Nations, 1948. Convention on the Prevention and Punishment of Genocide, Created on
August 16, 1994 / Last edited on January 27, 1997.
http: //www.hrweb.org/leqal/qenocide.html (accessed on 12/16/2006).
---The World Bank, 2006. Sudan-Country Brief. Retrieved from
http://web.worldbank.org/W B SITE/EX TERNA L/COUNTRIES/A FRICAEXT/SUDANE
. (accessed on11/10/2006).
APPENDIX A: Simulation Model Boundary Chart
Endogenous Variables | Exogenous Variables Excluded Variables
Affects Model Xx x
Behavior
Behavior
Affected By x
Model
Stocks: e Initial Conditions for |e Weather, Seasons,
People of Darfur ... All Stocks and Acts of Nature
e In Home Villages e Domestic Issues in
¢ In Resistance Auxiliary Variables: Neighboring Coun-
Groups Fractional Rates: tries
e In Transit from/to e Resistance Recruit- e Media Observation
Displacement Camps ment e Economic Variables
e In Displacement
Camps
e Dead (Excess Deaths
Due to Violence,
Disease, Starvation)
¢ Potential Militia or
Military Members
(Potential A ggres-
sors)
e Militia or Military
Members (A ggres-
sors)
Flows:
e Recruitment into Re-
sistance Groups
¢ Joining Militias,
Military
e Mortality (People of
Darfur)
e Displacement
e Arriving at Camps
e Leaving Camps
e Militia Involuntary
Recruitment
e Militia Voluntary Re-
cruitment
e Killing Rate Against
Resistance Groups
(Per Aggressor)
e Killing Rate in Vil-
lages (Per A ggressor)
e Killing Rate in Transit
(Per Aggressor)
e Killing Rate at Camps
(Per Aggressor)
e Displacement Rate
(Per Aggressor)
e Re-Displacement Rate
(Per Aggressor)
e Starvation/Disease
Death Rate
e Average Time to
Reach Camp
(e.g. Oil Revenues
and Distributions,
Economic Sanctions,
Foreign Investment,
Divestment)
Peace Treaties
Politics Between
UN, Sudan, and US
Military Supplies
AU Forces
UN Forces
Births
Non-Violence Re-
lated Deaths
APPENDIX B: Simulation Model Stock Flow Diagram
Finetionel Rate of Fractional Rate of
Involu .
prolate Voluntary Recruitment
<Total Excess Deaths
Potential ‘1 rr
joining Militias, Aggressors
Aggressors J Military
<lnitial Potential <iitial
Aggressors> Aggressors> ,
Displacement Rate a a fe Diniseanet
PA rt ggressors: ate per Aggressor
anngesties oie i Fixed Flow
es
Fractional Fixed Flow <Population of Threshold R
Recruitment Rate +| Threshold D Darfur>
+
|
Fat e : Darfurians in
Dafuransin | Darfurians at Home| >| Daria Ga Displacement
Resistance Groups REGAIN Di = - Camps
<itia Dafifiansin itil Population |
Resistance Gfloups> ofDarfur> } ‘Avg Time to
| Reach Camp
AB) | <Populationof |) 4B) |
' f Fixed Flow | <Population of Darfur>
Sep Tae fat |
Fixed Flow Fractional
Killing Rate in Fixed Flow Threshold C. c Starvation Rate
: Villages per - Threshold T
Rebel Displacement
5 f ‘Aggressor Wvi : Killing Rate at n
Aggressor Killing + ortality Village lortality in + a) Camp Mortalit
Rate Against + Mortality Killing Rate in ransit Camps Per Aggressor“ pMoreny
Rebels <A ggressors>- Transit per Aggressor Aguesas>
Excess Rebel Excess Deaths in Excess Deaths in Excess Deaths at
Deaths Villages Transit Displacement Camp |
APPENDIX C: Simulation Model Equations
(01) Aggressor Killing Rate Against Rebels=
0.001
Units: Darfurian/(Month*Aggressor) [0,0.05,0.001]
Exogenous Parameter: Rebels killed per month per aggressor (by aggres-
sors).
(02) Aggressors= INTEG (
"Joining Militias, Military",
Initial Aggressors)
Units: Aggressor [0,?]
Stock: Aggressors represent the militia and military population.
(03) Arriving at Camps=
Darfurians In Transit/Avg Time to Reach Camp
Units: Darfurian/Month [0,?]
Flow: First-order flow from Darfurians in transit to displacement camps,
dictated by average time to reach camps.
(04) Avg Time to Reach Camp=
0.5
Units: Month [0,12]
Exogenous Parameter: Average time for Darfurians to reach camp once dis-
placed, or average time in transit during multiple-displacement.
(05) Darfurians at Home= INTEG (
-Displacement Rate-Recruitment-Village Mortality,
Initial Population of Darfur)
Units: Darfurian [0,?]
Stock: People of Darfur all originate at home, and leave “home” via re-
cruitment into rebel groups, death, and displacement.
(06) Darfurians in Displacement Camps= INTEG (
Arriving at Camps-Displacement Camp Mortality-Leaving Camps,
9)
Units: Darfurian [0,?]
Stock: Population of displacement camps.
(07) Darfurians in Resistance Groups= INTEG (
+Recruitment-Rebel Mortality,
Initial Darfurians in Resistance Groups)
Units: Darfurian [0,?]
Stock: Population of rebels.
(08) Darfurians In Transit= INTEG (
+Displacement Ratet+Leaving Camps-Arriving at Camps-Mortality in Tran-
Sit,
0)
Units: Darfurian [0,?]
Stock: People of Darfur in a state of transition between home and camp.
(09) Displacement Camp Mortality=
Aggressors*Killing Rate at Camps Per Aggressor*MIN(1, Darfurians in
Displacement Camps
/(Fixed Flow Threshold C*Population of Darfur
) )+Fractional Starvation Rate*Darfurians in Displacement Camps
Units: Darfurian/Month
Flow: Camp mortality includes direct killing and disease/starvation. Di-
rect killing follows either a first-order flow based on the aggressor
and the camp population, or a constant flow based on the aggressor
population alone, or a hybrid of the two. A hybrid is dictated by the
“Fixed Flow Threshold” ranging from 0 to 1 which represents the frac-
tion of the initial camp population where the mortality shifts from
constant to first-order process. Starvation is a first order flow dic-
tated by the fractional starvation rate of the camp population.
(10) Displacement Rate=
Aggressors*Displacement Rate per Aggressor*MIN(1,Darfurians at
Home/ (Fixed Flow Threshold D
*Population of Darfur) )
Units: Darfurian/Month
Flow: Displacement moves population from home to in transit. The process
follows either a first-order flow based on the aggressor and the sus-
ceptible population at home, or a constant flow based on the aggressor
population alone, or a hybrid of the two. A hybrid is dictated by the
“Fixed Flow Threshold” ranging from 0 to 1 which represents the frac-
tion of the initial home population where displacement shifts from a
constant to first-order process.
Acts of violence include looting and burning villages, stealing re-
sources, poisoning wells, destroying land and fields, raping women, di-
rectly killing young men and boys, and forcing displacement into inhos-
pitable IDP camps (Patrick, 2005).
(11) Displacement Rate per Aggressor=
2
Units: Darfurian/(Month*Aggressor) [0,?,0.001]
Exogenous Parameter: People of Darfur displaced per month per aggressor
(by aggressors).
(12) Excess Deaths at Displacement Camp= INTEG (
Displacement Camp Mortality,
6)
Units: Darfurian [0,?]
Stock: Counts total excess deaths in displacement camps.
(13) Excess Deaths in Transit= INTEG (
Mortality in Transit,
0)
Units: Darfurian [0,?]
Stock: Counts total excess deaths of those who are in transit.
(14) Excess Deaths in Villages= INTEG (
Village Mortality,
®)
Units: Darfurian [0,?]
Stock: Counts total excess deaths of people of Darfur at home, in vil-
lages.
(15) Excess Rebel Deaths= INTEG (
Rebel Mortality,
®)
Units: Darfurian [0,?]
Stock: Counts total excess deaths of rebels.
(16) FINAL TIME = 46
Units: Month
The final time for the simulation.
(17) Fixed Flow Threshold C=
al
Units: Dmnl [0.01,1]
(18) Fixed Flow Threshold D=
Units: Dmnl [0.01,1]
(19) Fixed Flow Threshold H=
unite: Dmnl [0.01,1]
(20) Fixed Flow Threshold R=
Units: Dmnl [0.01,1]
(21) Fixed Flow Threshold T=
tintts: Dmnl [0.01,1]
(22) Fractional Rate of Involuntary Recruitment=
units: 1/Month [0,0.1,0.01)
Exogenous Parameter: Fraction of potential aggressors that are involuntar-
ily recruited each month.
(23) Fractional Rate of Voluntary Recruitment=
0.7624
Units: 1/Month [0,?]
Exogenous Parameter: Fraction of potential aggressors that are voluntarily
recruited each month.
(24) Fractional Recruitment Rate=
0.0001
Units: 1/Month [0,0.001,0.0001]
Exogenous Parameter: Fraction of people of Darfur at home that are re-
cruited into resistance groups each month.
(25) Fractional Starvation Rate=
0.005
Units: 1/Month [0,0.1,0.005]
Exogenous Parameter: Fraction of people in displacement camps that starve
each month.
(26) Initial Aggressors=
20000
Units: Aggressor [0,?]
(27) Initial Darfurians in Resistance Groups=
20000
Units: Darfurian [0,?]
(28) Initial Population of Darfur=
6e+006
Units: Darfurian [0,?]
(29) Initial Potential Aggressors=
15000
Units: Aggressor [0,?]
(30) INITIAL TIME = 0
Units: Month
The initial time for the simulation.
(31) “Joining Militias, Military"=
Fractional Rate of Voluntary Recruitment*Potential Aggres-
sors*Aggressors/
(Potential Aggressors + Aggressors) + Fractional Rate of Involuntary Re-
cruitment
*Potential Aggressors
Units: Aggressor/Month [0,?]
Flow: Moves people from potential aggressors to aggressors. Process fol-
lows a Bass-diffusion process where two forces contribute: voluntary
joining and involuntary joining.
(32) Killing Rate at Camps Per Aggressor=
0.24
Units: Darfurian/(Month*Aggressor) [0,0.5,0.01]
Exogenous Parameter: Number of people of Darfur in camps that are killed
per aggressor (by aggressors), per month.
(33) Killing Rate in Transit per Aggressor=
0.001
Units: Darfurian/(Month*Aggressor) [0,0.05,0.001]
Exogenous Parameter: Number of people of Darfur in transit that are killed
per aggressor (by aggressors), per month.
(34) Killing Rate in Villages per Aggressor=
0.08
Units: Darfurian/(Month*Aggressor) [0,0.5,0.01]
Exogenous Parameter: Number of people of Darfur at home in villages that
are killed per aggressor (by aggressors), per month.
(35) Leaving Camps=
Aggressors*"Re-Displacement Rate per Aggressor"*MIN(1, Darfurians in
Displacement Camps
/(Fixed Flow Threshold R*Population of Darfur
))
Units: Darfurian/Month
Flow: In this model, people leave camps due to aggressor violence. The
process follows either a first-order flow based on the aggressor and
the camp population, or a constant flow based on the aggressor popula-
tion alone, or a hybrid of the two. A hybrid is dictated by the “Fixed
Flow Threshold” ranging from 0 to 1 which represents the fraction of
the initial camp population where the process shifts from a constant to
first-order process.
(36) Mortality in Transit=
Aggressors*Killing Rate in Transit per Aggressor*MIN(1, Darfurians In
Transit
/(Fixed Flow Threshold T*Population of Darfur
))
Units: Darfurian/Month
Flow: Mortality is due to direct killing (aggressor violence). The proc-
ess follows either a first-order flow based on the aggressor and the
susceptible population, or a constant flow based on the aggressor popu-
lation alone, or a hybrid of the two. A hybrid is dictated by the
“Fixed Flow Threshold” ranging from 0 to 1 which represents the frac-
tion of the initial susceptible population where mortality shifts from
a constant to first-order process.
(37) Population of Darfur=
Darfurians at Home+Darfurians in Displacement Camps+Darfurians in Re-
sistance Groups
+Darfurians In Transit
Units: Darfurian [0,?]
Auxiliary Variable: Sum does not include militia/military population.
(38) Potential Aggressors= INTEG (
-"Joining Militias, Military",
Initial Potential Aggressors)
Units: Aggressor [0,?]
Stock: Population of people who can potentially be recruited into militias
or the military.
(39) "Re-Displacement Rate per Aggressor"=
0
Units: Darfurian/(Month*Aggressor) [0,?]
Exogenous Parameter: Number of people of Darfur in camps that are re-
displaced per aggressor (due to aggressor violence), per month.
(40) Rebel Mortality=
Aggressors*Aggressor Killing Rate Against Rebels*Darfurians in Resis-
tance Groups
/Population of Darfur
Units: Darfurian/Month
Flow: Mortality is due to direct killing by aggressors. The process fol-
lows either a first-order flow based on the aggressor and the rebel
population, or a constant flow based on the aggressor population alone,
or a hybrid of the two. A hybrid is dictated by the “Fixed Flow
Threshold” ranging from @ to 1 which represents the fraction of the re-
bel population where mortality shifts from a constant to first-order
process.
(41) Recruitment=
Darfurians at Home*Fractional Recruitment Rate
Units: Darfurian/Month [0,?]
Flow: Moves population of Darfur at home to resistance groups. A first-
order process determined by a fractional recruitment rate.
(42) SAVEPER =
TIME STEP
Units: Month [0,?]
The frequency with which output is stored.
(43) TIME STEP = 0.0078125
Units: Month [0,7]
The time step for the simulation.
(44) Total Excess Deaths=
Excess Deaths in Villages + Excess Deaths in Transit + Excess Deaths at
Displacement Camp
+Excess Rebel Deaths
Units: Darfurian [0,?]
(45) Village Mortality=
Aggressors*Killing Rate in Villages per Aggressor*MIN(1, Darfurians at
Home
/(Fixed Flow Threshold H*Population of Darfur
))
Units: Darfurian/Month
Flow: Mortality is due to aggressor violence (direct killing). The
process follows either a first-order flow based on the aggressor popu-
lation and the population at home, or a constant flow based on the ag-
gressor population alone, or a hybrid of the two. A hybrid is dictated
by the “Fixed Flow Threshold” ranging from 0 to 1 which represents the
fraction of the initial home population where mortality shifts froma
constant to first-order process.
Acts of violence include looting and burning villages, stealing re-
sources, poisoning wells, destroying land and fields, raping women, di-
rectly killing young men and boys, and forcing displacement into inhos-
pitable IDP camps (Patrick, 2005).
