Evaluating the scope and intensity of the conflict trap
| DOI | 10.1177/0022343316684917 |
| Author | Håvard Hegre,Håvard Mokleiv Nygård,Ranveig Flaten Ræder |
| Published date | 01 March 2017 |
| Date | 01 March 2017 |
| Subject Matter | Research Articles |
Evaluating the scope and intensity
of the conflict trap: A dynamic
simulation approach
Håvard Hegre
Department of Peace and Conflict Research, Uppsala University
& Peace Research Institute Oslo (PRIO)
Håvard Mokleiv Nygård
Peace Research Institute Oslo (PRIO)
Ranveig Flaten Ræder
Office of the Auditor General of Norway
Abstract
Several studies show that internal armed conflict breeds conflict by exacerbating conditions that increase the chances
of war breaking out again. Empirically, this ‘conflict trap’ works through four pathways: conflicts increase the
likelihood of continuation, recurrence, escalation, and diffusion of conflict. Past empirical studies have underesti-
mated the scope and intensity of the conflict trap since they consider the impact of conflict only through one of these
pathways and rarely across sufficiently longtime periods. This article shows that simulation and forecasting techniques
are useful and indeed necessary to quantify the total, aggregated effect of the conflict trap, over long time periods and
across countries. We develop a country-year statistical model that allows estimating the probability of no conflict,
minor conflict, and major conflict, and the probabilities of transition between these states. A set of variables denoting
the immediate and more distant conflict history of the country are used as endogenous predictors in the simulated
forecasts. Another set of variables shown to be robustly associated with armed conflict are treated as exogenous
predictors. We show that the conflict trap is even more severe than earlier studies have indicated. For instance, if a
large low-income country with no previous conflicts is simulated to have two to three years of conflict over the 2015–
18 period, we find that it will havenine more years of conflict over the 2019–40 period than if peace holds up to 2018.
Conversely, if a large low-income country that has had major conflict with more than 1,000 battle-related deaths in
several of the past ten years succeeds in containing violence to minorconflict over the 2015–18 period, we find that it
will experience five fewer years of conflict in the subsequent 20 years than if violence continues unabated.
Keywords
conflict diffusion, conflict escalation, conflict recurrence, conflict trap, forecasting, simulation
Introduction
Internal armed conflict remains a major societal problem
in many parts of the world. Over the past couple of
decades, these conflicts disproportionately occur in a
group of less than 50 countries that transition in and
out of conflict – the ‘bottom billion’ population of the
world (Collier, 2007). Structural factors that tend to
facilitate conflict – poverty, poor governance, and non-
participation in the modern, global economy – are clus-
tered. This convergence, however, is intensified by the
‘conflict trap’ (Collier et al., 2003).
Corresponding author:
havard.hegre@pcr.uu.se
Journal of Peace Research
2017, Vol. 54(2) 243–261
ªThe Author(s) 2017
Reprints and permission:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0022343316684917
journals.sagepub.com/home/jpr
The presence of a conflict trap is well established in
the literature,
1
butweknowlittleaboutitssizeand
severity. In this article we show that the conflict trap is
more intense than suggested by previous studies. If a
country similarto Tanzania experiences two to three years
of conflict over the 2015–18 period, we find that it can
expect nine additional years of conflict over the subse-
quent 20 years compared to a scenario where peace con-
tinues up to 2018. Conversely, if a country like Nigeria
succeeds in de-escalating its conflicts to less than 1,000
battle-related deaths in three out of four years over the
2015–18 period, the number of years with major conflict
over the subsequent two decades will be reducedby 50%.
Quantifying the substantive effect of a phenomenon
like the conflict trap is an essential complement to the
focus on statistical significance in previous studies. Asses-
sing the intensity of the conflict trap is demanding, how-
ever, since the ‘trapping effects’ extend far beyond the
country year for which we have direct statistical estimates
– they involve a set of factors active over a significant
period of time, and work through several channels. We
identify four empirical pathways for the conflict trap: an
onset of a conflict increases the risk of conflict in the next
year (pathway 1: continuation); it increases the risk of
future conflict in the same country (2: recurrence); it
(often) leads to intensified fighting that subsequently is
harder to end (3: escalation); and it often spreads to
neighboring countries (4: diffusion). Thanks to recent
analyses disaggregated to the country-year level we now
have fairly precise estimates for annual effects along each
of these pathways. However, these studies fail to take
into account the combined effect of them as well as their
long-run implications. A comprehensive analysis of the
conflict trap cannot be done by simply discussing the
estimates from statistical models conducted at a disag-
gregated level of analysis, but requires the use of simula-
tions based on the disaggregated statistical estimates.
Such simulations are common in forecasting applications
and also used in some of the articles in this special issue
(e.g. Witmer et al., 2017)
We utilize the forecasting/simulation framework
developed in Hegre et al. (2013) to assess the extent to
which the outbreak of conflict affects the risk of future
conflict in the same country, its neighborhood, and
regionally. Since the conflict trap is related to onset,
termination, and recurrence of conflict, we study the
incidence of conflict. In our specification, we model how
the risk of conflict depends on the conflict history of the
country itself as well as its neighborhood, a set of risk
factors established as robustly related to conflict, and
time-invariant country effects.
To assess the intensity of the conflict trap, we conduct
a set of statistical ‘experiments’ within our simulations.
For instance, we investigate the model’s implications for
the future risk of conflict in the group of previously
peaceful low-income countries if one of them experi-
ences conflict in the 2015–18 period. These experiments
allow us to report a measure akin to an ‘average treat-
ment effect’ of conflict, such as the one for Tanzania
referred to above. The estimated magnitude of the con-
flict trap indicated by our results for low-income coun-
tries is considerable and larger than shown in previous
studies. This estimate of the substantive effect has impor-
tant implications for a policymaker that considers taking
costly steps to prevent the escalation to armed conflict in
a country in crisis. In deciding how to respond to a crisis,
it makes a difference for a policymaker that a successful
intervention that results in the prevention of armed con-
flict in the short run also helps avoid armed conflict for
most of the subsequent two decades. In comparison, if
the detrimental effect of an onset of fighting is contained
to the next two years, intervention might be seen as
carrying higher than expected costs and risks. As such,
our results and estimation methodology can support
decisions that allocate limited resources to conflict pre-
vention and peacekeeping operations.
The article reviews the literature and discusses the
different pathways through which a conflict trap oper-
ates, looks at some descriptive statistics related to the
conflict trap, describes and discusses the statistical model
and the simulation technique, performs out-of-sample
evaluation of a set of candidate models for the simula-
tion, reports the simulation results, and analyzes the
intensity of the conflict trap.
Why conflict traps
Multiple explanations for the presence of the conflict
trap have been established. The literature points to both
societal and economic impacts of war and to how both of
these impacts are transmitted to neighboring countries
through diffusion effects.
Societal impacts
Wartime transformation of social actors, structures,
norms, and practices have long-lasting effects on society
1
See, for instance, Lichbach & Gurr (1981); Quinn, Mason &
Gurses (2007); Dahl & Høyland (2012); Hegre & Nygård (2015).
Collier, Hoeffler & So
¨derbom (2008) estimate the risk of conflict
reversal to be around 40% during the first post-conflict decade.
244 journal of PEACE RESEARCH 54(2)
To continue reading
Request your trial