Approaching Resilience for Displaced People in the C ongo
Charles A Jones, PhD
Senior Modeler, Climate Interactive
63 Eastland Rd
Jamaica Plain MA 02130
+1 781 691 5417
cjones@ climateinteractive.org
“UK Aid from the British People”
This project was funded by the Department for International Development
Abstract
A multi-organization consortium is engaged in a project to improve resilience among people
affected by multiple displacement in the Democratic Republic of the Congo. Funded by the UK
Department for Intemational Development (DFID), the project includes humanitarian,
peacebuilding, and data collection organizations. System dynamics modeling is an important
innovation of this three year project, along with field research and policy advocacy; all leading to
pilot interventions on the village scale. While the research phase is ongoing, the SD approach has
already yielded insights that build on humanitarian and development mental models, such as
short term versus long term trade-offs, and how conditions before displacement and in potential
host villages can affect the recovery of the displaced community.
Keywords: Internal displacement; development; humanitarian aid; resilience;
Conditions in the Eastem Democratic Republic of the Congo (DRC) are dire. The latest report
from the UN Security Council-appointed Group of Experts (Group of Experts 2015) catalogs a
host of problems. Numerous Congolese and foreign armed groups continue to operate,
contributing to instability. Civilian populations suffer from violence, intimidation, enslavement,
and expropriation of assets by armed groups, causing frequent displacement as people seek
safety elsewhere. Parts of the economy and resource wealth of the DRC have been diverted by
armed groups and elements within the government. Neither Congolese military (FARDC) nor
United Nations forces (MUNESCO) have succeeded in establishing security or rule of law.
Over ten years of intensive efforts by UN agencies and intemational non-governmental
organizations (NGOs) have saved lives but have made little difference in the long-term situation
of people in the DRC (White 2014). The acute needs have often overwhelmed planning for the
future, leading to calls for taking a long-term development approach even during humanitarian
aid (DFID 2013, Rudolf 2014, Mosel & Levin 2014). This paper reports on work in progress
with that approach.
Given the frequency and prevalence of displacement in the Eastern Congo, many approaches
have centered on resilience in the face of disaster rather than the prevention of disaster.
Resilience in general refers to avoiding or recovering from harm; for the purposes of this project
we define it as the ability of a community to withstand or recover from damage caused by
displacement (or other shocks), and return to a state as good or better than before. In a sense,
people of the Kivus have demonstrated remarkable resilience, in that a majority withstand the
harms of multiple displacement; but neither recovery nor improvement have been common in the
long term. There is clearly room for improvement in the approach of international aid
organizations.
One problem has been poor understanding of the nature of resilience; attempts to measure or
quantify have failed (Levin 2014). We propose that the trouble is that resilience is not itself a
variable or stock that can be built up, but a descriptive feature of system behavior. Because of
this resilience is only observed in context, when the community is recovering or not. Models
overcome part of this limitation, by allowing users to simulate things that cannot be measured.
System dynamics was selected as the modeling paradigm because the system in which
community resilience occurs is characterized by the accumulation of assets, flows of production
and consumption, and feedbacks between them.
The UK Department for International Development has funded this project, led by the
Norwegian Refugee Council (NRC). Climate Interactive is providing modeling and analysis for
the project. Our other partners are International Alert, and the Intemal Displacement Monitoring
Centre (IDMC). The three-year, $9m project includes ongoing research; pilot programs to
increase resilience while testing emerging theories of resilience; and advocacy, to use these
findings to influence future humanitarian responses to displacement.
Methodology
System dynamics modeling in this project supports the analysis of qualitative and quantitative
data, theory building, and communication. The immediate aim is to inform decisions about pilot
projects, which will include interventions in areas where people are affected by multiple
displacement in an effort to improve their resilience. As the projects operate over the following
18 months, observed results will further inform theory building, allowing us to refine the model.
The simulation tool is intended to support both decision making for action on resilience, and for
communication and outreach. We plan to use simulation to communicate our findings
convincingly, in an attempt to improve the overall humanitarian response to displacement.
Information to build the model began with a narrative, describing the general events and effects
associated with displacement, assembled with our consortium members with knowledge of the
system. We created a first iteration of the model, focused on the causes of displacement and
2
movement after displacement. The first model was essentially a straw man — known to be
“wrong” (more so than the way all models are wrong) but useful for highlighting knowledge
gaps. On our first trip to DRC, we showed the simulation to our consortium, in a workshop for
Congolese aid workers, and to some of the many intemational humanitarian staff stationed there.
By showing a model that was very simple but still had plausible behavior, we were able to elicit
more detail. Experts were led to explain things they thought were obvious. Within our
consortium, we showed how the model changed from day to day, familiarizing them with system
dynamics methods. When showing the model was impossible or inappropriate, we conducted
semi-structured interviews and focus groups. We held focus groups with intemally displaced
people in two camps, each separated into male and female groups in respect of cultural norms. In
the end we had collected system information from people with various areas of expertise, levels
of aggregation, and time-scales of action.
As the important sectors to model were revealed, the consortium decided on research questions,
methods, and instruments. The main field data come from surveys, with versions for displaced,
host families, and returnees; and qualitative focus groups with the same sub-samples. The data
collection was contracted to Congolese partners, to account for language and cultural sensitivity.
Data was collected mostly in Kiswahili, with some exceptions in other local languages or French
as appropriate. Data analysis produced notes and summaries in French and English.
New information was incorporated into model structure, which was checked with partners and
other experts. The model has developed in an iterative process and will continue to change over
the course of the project.
The Community Resilience Model
The model developed for the project represents the conditions in a community (in general, a
village) over the course of 20 years. Different versions of the model interface have been built for
use by or presentation to different audiences. The version provided for this paper allows for up to
five episodes of displacement, with policies for emergency and programmed assistance and help
from a host community.. In the logic of the model, the village is the people rather than the
location. When displacement occurs the model follows the people, and simulates their conditions
even if they are separated, rather than representing a place and tracking flows of people in and
out.
Figure 1 shows the main sectors of the model. People use resources to produce the things they
want and need. They can apply that production either to goods that relate to their quality of life,
or back into resources for future production. The availability of assets then feeds back to affect
people, their conditions such as health and social cohesion. Production depends on both the
quality and quantity of labor and resources. Decisions to invest in resources depend in part on the
conditions in the community.
PEOPLE
Popuiaton [em |
—_—
Displacement
Household
Goods
ASSETS
Production
Figure 1: Resilience Model Roadmap
Displacement affects people both directly, and by taking away their possessions and cutting them
off from their resources. The most important problem cited in most displacements is being cut
off from access to land, this being a predominantly agricultural economy. Land tenure laws in
the DRC are complex; there is no private ownership of land, and rights to use are the subject of
several overlapping bodies of law, so disputes are common. Even after the end of displacement it
can take time to restore access to land.
“Markets” in this model means the ability to earn from non-agricultural work. As a supplement
to farming, and as sometimes the only income during displacement, people engage in wage labor
and small trading. When displacement occurs wages are depressed and product markets are
disrupted to the disadvantage of the small trader. Even within the depressed wage market,
displaced people tend to be paid less than their host community and neighbors. Theft and
prostitution are also reported by displaced people as means of livelihood.
“Capital” includes all possessions that help people to earn a living — seeds, tools, livestock, long-
lived plants, bicycles, etc. This is a low tech economy, so capital is less significant than in a
western economy — zero capital does not imply zero production, for example — but the it is one
means of production that responds dynamically to people’s decisions. One of the most important
recovery dynamics passes through capital, as shown in Figure 2. Investing in capital competes
with basic needs. Especially when access to land is disrupted, a shortage in production can lead
to a vicious cycle of declining production. People must sacrifice some quality of life in the
present in order to recover in the long run.
Spending Decisions
Fod =<
Housing — Education
etc
Investing
Production 9) Capital ¥
+ Capital: Seek Access to
Sons. other income
R
Education:
Weak, Slow
Access to
Land
Figure 2: Capital and Education Loops
Education acts in a similar manner, under the assumption that higher education can yield greater
production for the community, but it acts over much longer time scales. And, in this economy,
there are fairly limited opportunities for higher earning power, so the loop is weaker than the
capital loop. Never the less, people continue to place a high priority in keeping children in school
when possible, despite high school fees and challenging access. In essence, people’s priorities on
education must not be based on purely economic considerations. Education should be considered
a quality of life issue as well as an economic one.
Other dynamics display the short-versus-long term tradeoff. Most prominently, environmental
degradation and land quality show a tradeoff between coping mechanisms and sustainability. The
need for immediate consumption can pressure people to use land more intensely, by skipping
crop rotation and fallow cycles, not using terracing, or planting and harvesting more frequently
than supported by the growing season. As shown in Figure 3, these shortcuts can increase
production in this environment, but they may deplete the soil or increase erosion. Over a few
years, agricultural production falls below the level before over-intensive practices, further
encouraging bad practices.
Food
Housing
etc ‘
Pressure to
4B) Produce More
B
ort Term
‘oping
Production
Land use
Intensity
Sh
Ce
4R) Long Term
Access to Degradation
Land
Land Quality
Figure 3: Land Use Intensity and Degradation
Material well-being is central to the feedback loops in this model, so production is an important
sector. The humanitarian community uses terms like livelihood, which in this model would
include a complex of variables in the production and consumption sectors. Production is
calculated in a function similar to Cobb-Douglas — it is multiplicative for the appropriate factors
of production, but as mentioned contains a correction to account for the fact that in this low-tech
economy, capital is not necessary for there to be some production. The equation for agricultural
production is:
( Labor )
*
Base Labor
yCapital
yLand yLabor
Land
Base and)
Capital + Min
(Fase Capital + iim)
Base Production * (
Non-agricultural production is similar, but with the Markets variable in place of land. Production
accumulates in the Stores stock, from which Consumption draws to be allocated between
Household Goods, Investment / Capital, Current Spending (or Immediate), Education and other
costs. The basic stock and flow is shown in Figure 4
Production
Smooth Time
Minimum
Consumption Time
Lagged P roduction <———.
Initial Stores
+
Q
. Expected
Production
Consumption Consumption
+
Y
@ Stores Galys +
Desired Total
ive)
Stores Losses
‘Consumption
Figure 4: Production and Consumption
For parsimony, consumption in the model is split into categories that have the same dynamics or
effects on other variables, as shown in Figure 5. Production can also be saved, accumulating in
the Stores stock. The level of Household Goods, and the flow rate of Immediate Consumption,
together determine Quality of Life. Quality of Life is both an important output metric, anda
determinant of other variables such as Health and Social Cohesion.
y
3° For now ”
or ng” neha
er? Food ““m, “©
sf Shelter Dy,
Oe Water “On
Clothing
Fuel 6
po a
8 2
a :
Tools . &
ry Seeds Education
© Livestock Health Care
For the future
Figure 5: Types of Expenditures
Decisions in the model are based on simple floating goals: people act to close the gap between
current levels and accustomed levels, where the accustomed levels slowly adjust to current. The
time constants are such that people get used to improved conditions more quickly than they
accept degraded conditions.
With the ability to cause displacement in the simulated environment, the resilience of the
community is seen through the behavior over time of its variables. One can look at the
qualitative behavior and end point of indicator variables: Quality of Life (QoL), Total
Production, Total Assets. We also calculate the ratio of some variables six months after the start
of recovery compared to their peak, pre-displacement values. Resilience is again, not a variable
but a description of how the system responds to a shock.
A model text file written in Vensim (DRC Community Resilience SDS 2.mdl), and associated
files accompanies this paper.
Model Behavior
The behavior of key model variables to a simple displacement that takes away half of the
community’s possessions and access to land is shown in Figure 6. When displaced, people
immediately lose assets. They are also cut off from their land, and markets are disrupted, so their
production fallf immediately as well. In this case, people still have some access to land, so
during the displacement their QoL rises slowly even during the displacement. A fter the
displacement is ended, the community slowly regains access to their land, allowing production —
and therefore assets and QoL — to recover. There is a small bump in QoL as people have enough
to save some stores. None of these variables ever quite reaches its starting point, because people
have become used to a lower level of capital.
Events Production (indexed)
Figure 6: Response to a One-year, 50% Loss Displacement
In amore severe displacement, where land access is only 5% of the pre-displacement case — not
only does production fall farther than before, but conditions continue to degrade over the course
of displacement. See Figure 7.
Events Production (indexed)
3 150
200 Tz ft
50
| °
a 0 = 7 a5 i 5
Total Assets (indexed) Quality of Life
Figure 7: A Displacement with 50% Loss of Assets and 95% Loss of Land Access
With multiple displacements, both the immediate and long-term conditions are worse. As shown
in figure 8, even after the recovery has progressed for a couple years, and has nearly recovered in
Production, some assets are still below sustainable levels; QoL drops further than before and the
eventual recovery point is even lower.
Events Production (indexed)
Total Assets (indexed) Quality of Life
Figure 8: Two One-year, 50% Loss Displacements
Finally, the model allows for different types of aid as well as different conditions of
displacement. Figure 9 zooms in on the QoL graph with three different aid treatments on the
same displacement event. The no-aid case is shown for comparison: in this event QoL falls,
remains steady throughout displacement, then rises with increasing access to land and build-up
of assets. A cash distribution has an immediate and obvious change in quality of life, but almost
no long term effect — in neither case does QoL reach its pre-displacement level in the new steady
state. In fact, increasing the amount of cash cannot overcome that dynamic. In contrast, a
distribution of capital has very little effect on QoL immediately — just a small impact as presence
of free capital frees up production to be assigned to household and immediate needs. But the
capital aid does have a long term effect, causing QoL to exceed the pre-displacement level, an
outcome not possible with any amount of cash. It is a graphic demonstration of the tradeoffs
shown in Figure 2 above.
Quality of Life
2.25
0
00 05 10 15 20
Total Quality of Life : demo no aid
Total Quality of Life : demo cash
Total Quality of Life : demo capital
Figure 9: Cash Aid versus Capital Improvement
Results, Insights, and Next Steps
The modeling for this project is designed to be iterative. An important part of the cycle is to
show the simulation to Congolese and interational experts in situ, to test for both model validity
and for its usefulness for outreach and education. Showing the simulation has already yielded
insights as people react to model behavior and compare it to their own experience. Some of the
structure in the current version was elicited precisely because of experts’ reaction to the concrete
example of prior versions. Thus many of our insights are in the form of new phenomena to be
investigated through field research, modeling, and further feedback.
The capital recovery dynamic is the most important determinant of resilient behavior in the
model. While we investigate whether capital is the only or most important mechanism, it is clear
that some kind of immediate versus future tradeoff guides the recovery dynamics.
Figure 2 would seem to imply that people just need to learn to set priority on capital if they wany
to recover or improve their conditions, but that is not the whole story. Figure 10 expresses the
same tradeoff in a different way, highlighting the limits on people to he able to invest. If
immediate needs are not met, then health and the related ability to work suffer, lowering
production even further. Shifting priorities either way can lead to a spiral of vicious cycles,
dragging down the conditions and production in the community.
10
Health
Housing ete
Investing
é
ees ‘Ne
Production AR) Capital +
Capital: Stock =
aN Strong Ability
to Work
t2)
Heath: Rapid
Figure 10: Health and Investment, Tradeoff and Spirals
Even given adequate resources to invest in the future while not sacrificing current needs, people
might not be able to make it work. Both the rationality of and ability to apply available resources
to the future are affected by conditions that are themselves affected by displacement. It makes no
sense to invest in capital if one expects to lose it in a future displacement next year. There is both
a rational expectation and an emotional aspiration component to the decision to forgo present
consumption for future gain, both of which could be expected to fall with repeated
displacements.
Ability &
Hope
Social
Conditions
Production
Capital:
Strong
Access to
other income
+h
Education:
Displacement Weak, Slow
Figure 11: Social Interactions with Investing and Education
There are also mechanisms within the community that enable setting aside for the future. Each
household might have a little extra; the community as a whole might have enough to invest for
the community’s well-being; but if that savings cannot be brought to bear on the problems at
hand it does not matter. Community credit and savings schemes are built on trust, which can be
damaged when the displaced community is split up among different sites.
11
In exploring the validity of the capital recovery dynamics, several ideas came out that warrant
further elicitation. The network of factors is illustrated in Figure 12. Even if ore than basic needs
are met, the ability to recover is dependent upon: external factors, like security and access;
economic factors, like markets; personal factors, like skills and perception. Many of these are
dependent on social phenomena — the psychosocial impact of being repeatedly uprooted. Social
cohesion can cut across many of these factors by affecting credit, cooperation, expectation, and
hope.
Physical Resources Capacity to Set Aside
Money, Tools, Time Credit, Cooperation, Skills, Markets
A
Ability to Improve
=
Conditions to Set Aside Decision to Set Aside
Access, Security, ce Hope, P
Figure 12: Factors in the A bility to Recover or Improve through Investment
wr
It is also an important insight of the research that host communities affect displaced
communities. Most of the time in the Eastern DRC, displaced people are staying in the homes of
a host family. Even when they are not, sites of displacement are near or in an existing
community, whose resources affect and are affected by the presence of displaced people. In
addition to social cohesion within communities, relations between communities are important.
One leverage point to improve resilience might be to support regions so that if any community
suffers a disaster, other communities are more able and more willing to help.
Thus the research program for the next iteration of model development has a deeper focus on
how social factors affect material well-being, and how those social factors are themselves
affected. The next iteration n the model structure will include two communities and the transfer
of people, goods, and good will (or disputes) between them. Pilot projects are expected to start
this calendar year, yielding more insight as our theories of resilience are tested in action.
References
DFID 2013. Building Resilience in DRC: Linking the Humanitarian and Development Sectors.
London: Department for International Development.
Group of Experts 2015. Final report of the Group of Experts on the Democratic Republic of the
Congo. UN Security Council Report number S/2015/19 (D. Fahey, V. Anya, F. Bemard, Z.
Bahalal, G. Mbaye, E. Serralta).
Levin, Simon 2014. Assessing resilience: why quantification misses the point. Humanitarian
Policy Group report. London: Overseas Development Institute.
Mosel, Irina & Simon Levin 2014. Remaking the case for linking relief, rehabilitation and
development. Humanitarian Policy Group report. London: Overseas Development Institute.
Rudolf, Markus 2014. Assessing the humanitarian response to chronic crisis in North Kivu.
White, Stacey 2014. Now What? The International Response to Internal Displacement in the
Democratic Republic of the Congo. Washington DC: Brookings-LSE.
13
