Special Issue: Ceasefires

Journal of Conflict Resolution

2022, Vol. 0(0) 1–29
Costs and Cover: Explaining © The Author(s) 2022

the Onset of Ceasefires in Civil Article reuse guidelines:

sagepub.com/journals-permissions
Conflict DOI: 10.1177/00220027221129195
journals.sagepub.com/home/jcr

Govinda Clayton1 , Håvard Mokleiv Nygård2 ,

Siri A. Rustad1 , and Håvard Strand3 

Abstract
Ceasefires are common in civil conflict. Yet we have surprisingly little comparative
analysis of why and under what conditions they occur. A ceasefire provides
temporary relief from the costs of conflict, but also generates its own costs.
Building on this logic, we argue that conflict parties are more likely to accept the
costs associated with a ceasefire when the conflict costs are greater, in particular,
when: violence is intense; there are higher levels of ‘collateral damage’; and the
parties lack international support. Second, we contend that ceasefires are also more
likely in those periods in which the audience costs associated with entering into an
arrangement are lower, specifically, when the parties have some form of ‘political
cover’, such as during mediation. We find support for both arguments in an analysis
of a new dataset capturing all ceasefire in civil conflict from 1989-2020, using a
series of dyad fixed effect models.

Keywords
conflict management, civil wars, conflict resolution, dyadic conflict, internal armed
conflict, mediation, ceasefires

1
Center for Security Studies, ETH Zurich, Switzerland
2
Peace Research Institute Oslo, Norway
3
University of Oslo, Norway; Peace Research Institute Oslo, Norway

Corresponding Author:
Håvard Strand, Political Science, University of Oslo, Postboks 1097, Blindern, Oslo 1097, Norway.
Email: [email protected]
2 Journal of Conflict Resolution 0(0)

Introduction
Ceasefires are arrangements between conflict parties to suspend violent hostilities from
a specific point in time (Clayton et al. 2022a). They are a common feature in civil
conflict. Between 1989 and 2020, 2202 ceasefires were declared globally, and 21% of
all conflict years produced at least one ceasefire arrangement (Clayton et al. 2022b). In
2020 alone, new ceasefire arrangements arose in conflicts from Colombia, to Sudan,
Myanmar and Yemen. Why do conflict parties enter into a ceasefire? And under what
conditions are ceasefires more likely to occur?
A rich body of case and practitioner literature illustrates how ceasefires support
peacemaking (Åkebo 2020; Chounet-Cambas 2011a, 8; Mahieu 2007, 209; Smith
1995, 155-160); and war making (Crocker, Hampson, and Aall 1999; Smith 1995; Toft
2010) (for a summary of the literature see, (Clayton et al., 2022a). Recent studies offer
useful frameworks to consider the functions that different types of ceasefire can play at
different points in a bargaining process (Clayton and Sticher 2021; Sticher, and
Vukovic 2021), and how ceasefire outcomes should be considered (Clayton,
Nathan, and Wiehler 2021).
Despite the emerging body of research on ceasefires, the prior deficit of suitable data
means that we lack a systematic, global assessment of the conflict factors and con-
ditions that are more likely give rise to ceasefires. The prior lack of comparable data
also means that we only have limited knowledge about the extent to which insights
gained in a particular case can be generalized to the broader population of civil
conflicts.
Understanding the role that ceasefires play in shaping conflict dynamics, and the
contribution they make in peace processes, first requires a foundational understanding
of the conditions under which different forms of ceasefire arise (i.e. the data generating
process). Moreover, research on patterns of civil violence and conflict management
processes would also both benefit from a clearer understanding of where and when
ceasefires are more likely to appear.
To address this lacuna, we consider the strategic calculation that conflict parties face
when considering a ceasefire during civil war. Most simply, a (bilateral) ceasefire (if
honored by both sides) provides temporary relief from the costs of conflict.1 But
entering into a ceasefire also carries costs, such as potentially allowing an opponent an
opportunity to rearm and regroup, and triggering audience costs from those that remain
supportive of the violent struggle. States also risk legitimizing the non-state group (and
any gains that they have made), while non-state groups must in effect consent, albeit
temporarily, to the status quo in which the state maintains its position of ascendency.
Building on this logic, we develop two arguments. First, we argue that conflict
parties are more likely to accept the costs of ceasefire when the costs associated with
conflict are greater, in particular, when: conflict violence is more intense; there are
higher levels of ‘collateral civilian damage’; the parties lack international support; and
the conflict continues for long periods. Second, we contend that ceasefires are also more
likely in those periods in which the audience costs associated with entering into an
Clayton et al. 3

arrangement are lower, specifically, when the parties have some form of ‘political
cover’ (see, Allee and Huth 2006; Beardsley 2010). We focus on three forms of cover:
mediation; religious holidays; and the 2020 United Nations call for a Global Ceasefire
to fight the Covid-19 pandemic.
Utilizing the new civil conflict CeaseFire data (CF), we provide the first systematic
global analysis of ceasefire declarations from 1989-2020. In the analysis we focus on all
non-definitive ceasefire (i.e. non-permanent ceasefire) that match conflict dyads
recorded by the UCDP Armed Conflict Dataset.2 We estimate a series of conflict-dyad
fixed effects models that allow us to offer new insights on the specific conflict dynamics
associated with ceasefires, and thus when ceasefires are more likely.
In what follows, we first discuss the strategic calculation that faces conflict parties
considering a ceasefire, we then discuss conflict costs and political cover, setting out our
argument for how these factors shape the ceasefire process. We then present our re-
search design and finally our results.

The benefits and costs of ceasefires in civil conflict

Ceasefires provide conflict parties numerous benefits. For those seeking to manage or
resolve their dispute, ceasefires can: allow access to humanitarian aid (e.g. Aary 1995);
suspend violence (without addressing the incompatibility) (Hanson 2020); contain
violence (Chounet-Cambas 2011b); signal peaceful intent (Bara and Clayton 2022;
Clayton and Sticher 2021); demonstrate command and control (Åkebo 2020; Höglund
2011); increase civil society participation (Pinaud 2020); and create an environment
more conducive for negotiations (Smith 1995).
Ceasefires can also serve so-called ‘devious intentions’, meaning conflict parties
commit to an arrangement for reasons other than seeking peace (c.f. Richmond 1998).
The military benefits ceasefires can provide include: buying time to resupply and
regroup forces (Chounet-Cambas 2011b, 7-8; Mahieu 2007, 210-211); providing a
psychological break to increase the morale of the troops, or domestic constituents
(Mahieu 2007, 210); and consolidating territorial control (Sosnowski 2020; Woods
2011).
It is challenging for conflict parties to determine whether their opponent desires a
ceasefire for peaceful or devious reasons ex ante, as motives may be mixed, shift over
time, and the parties have clear incentives to conceal any non-peaceful motivations. As
such, while both sides may welcome many of the benefits accrued from a ceasefire, not
least the mitigation of conflict costs, they are also likely to be wary of affording benefits
that might enhance the military position of their opponent.
Audience costs are another obstacle for conflict parties seeking the benefits of a
ceasefire. Ceasefire negotiations can also often resemble the structure of Putnam’s
(1988) two-level game, whereby the negotiations between leaders at one level are
shaped by how both leaders anticipate the reaction of the domestic constituents. Any
such ‘cooperative’ behavior might be seen as a concession or signal of weakness,
generating discontent amongst supporters (and creating opportunities for outbidding
4 Journal of Conflict Resolution 0(0)

from political opponents). Constituents often need to consent to major policy changes,
such as making peace with the enemy, and often decide on the political future of the
leaders (Debs and Goemans 2010). Intuitively, we might expect that domestic pop-
ulations would be supportive of cooperative behaviors that move the parties towards
peace, in particular ceasefires that promise temporary respite from violence (Brutger
and Kertzer 2018; Kertzer and Brutger 2016). Yet domestic constituents often adopt
relatively “hawkish” policy preferences (Allee and Huth 2006, 222). This creates
significant audience costs if leaders attempt to deviate from their conflictual posture
(Kertzer and Brutger 2016). On many occasions, leaders intentionally generate such
audience costs in an attempt to strengthen their bargaining position, for example by
promising never to make a concession, or enter into dialogue with the other side (e.g.
“we don’t talk to terrorists”). (Fearon 1994). Yet if military victory proves untenable,
and they fail to extract sufficient concessions from their opponent, it creates a ‘bar-
gaining bind’ in that even though both parties might desire a ceasefire, the anticipated
audience costs prevent the parties from taking the necessary cooperative actions
(Fearon 1997).
The costs and benefits discussed so far broadly apply to all conflict actors. In
addition, there are some strategic calculations that differ between state and non-state
actors. For the non-state group, the most significant ceasefire cost is to accept the status
quo where the state remain superior. In intra-state conflict, the non-state actor seeks a
significant shift in the distribution of power within the state (Gleditsch et al. 2002).
Violence is the means through which the group achieves their leverage over the state,
and thus in principle, they are likely to be resistant to a ceasefire (in the absence of
sufficient concessions or military progress) that might release the hard-fought pressure
on the state (Mahieu 2007). Similarly, the state, as the power-holder, often favour a
ceasefire if it helps to sustain the favourable status quo.
Yet non-state groups can also gain recognition and legitimacy from a ceasefire.
During conflict the state usually dismiss non-state challengers as criminals or terrorists,
undeserving of a political voice. A ceasefire can provide non-state groups with le-
gitimacy, potentially increasing access to international aid and third-party support. In
some cases, these non-state benefits provide strong incentives for the state to resist a
ceasefire.3 Beyond concerns relating to the immediate opponent, the state often fear that
a ceasefire agreements might inspire other groups to mobilize, fight harder or longer for
similar concessions (though see, Bara and Clayton 2022). This mechanism has been
found for peace agreements (Bormann and Savun 2018; Walter 2009), but is also likely
to apply to ceasefires that temporarily legitimize any gains that the non-state group has
achieved on the battlefield.4
A ceasefire is thus always a strategic decision for conflict parties, whereby the
potential benefits of entering an arrangement must be weighed against the possible
costs. Many of the factors that shape this consideration are effectively unobservable, as
they are highly subjective and depend on the conflict parties’ reading of the conflict
situation. However, the key factors are those that shape the costs of conflict. When the
costs of conflict increase, conflict parties are likely to be more willing to take on the
Clayton et al. 5

costs associated with a ceasefire. We also know from prior research several of the
observable conditions that shape the costs of conflict. Similarly, some observable shifts
in the external context are also likely to impact the strategic calculations underlying the
adoption of a ceasefire. In particular, some conditions are likely to provide political
cover, meaning actors have some means of saving face which reduces the costs as-
sociated with entering into an arrangement. In the following discussion we set out how
conflict costs and political cover impact conflict parties decisions surrounding
ceasefires.

Conflict costs
Civil conflict requires significant financial, material, and human resources (Collier and
Hoeffler 2004; Gates et al. 2012). Sustaining violence for prolonged periods requires
maintaining an armed force of sufficient size, strength and military readiness, as well as
ensuring the force maintains the necessary basic supplies and war-making assets
(Balcells and Kalyvas 2014; Collier, Hoeffler, and Söderbom 2004; Cunningham,
Gleditsch, and Salehyan 2009). Conflict also complicates the provision of public
services, not least the maintenance of security (for states and insurgents, see e.g. Arjona
(2017)), making it more challenging to retain constituent support (Kalyvas 2000).
Conflict can also reduce extractive capacity and incoming revenue, placing resources
under yet further pressure (Costalli, Moretti, and Pischedda 2017; Koubi 2005;
Humphreys 2005). Conflict costs depend on technologies of rebellion adopted by the
non-state group (see, Balcells and Kalyvas 2014), but regardless of the form that
violence takes, conflict is always costly.5
A ceasefire can temporarily reduce many of these conflict costs.6 If a ceasefire is
agreed (and honored) by all conflict parties, there follows a relatively violence-free
period that should reduce many of the costs. The durability of a ceasefire varies greatly
according to its design and the function it is intended to perform (Clayton and Sticher
2021), and almost all ceasefires suffer some violations (Bara, Clayton, and Rustad
2021). Yet on average, ceasefires tend to produce at least a few weeks of relative calm
on the battlefield (Clayton and Sticher 2021). Thus, the immediate benefit afforded by
all ceasefires (that are honored) is a temporary relief from the costs of conflict.
The patterns of violence vary greatly across conflict, with significant differences in
the configuration of repertoires, targeting, frequency, and technique (Gutiérrez-Sanı́n
and Wood 2017). We focus specifically on changes in the frequency of deadly forms of
violence, as well as the presence or absence of external support. We acknowledge that
lethal violence is not always a good proxy for other forms of violence, and that our
approach captures but a few dimensions of conflict costs. Moreover, certain costs are
distributed asymmetrically, as on the battlefield gains for one side are often reflected in
loses for the other, meaning not all periods are equally costly for all parties. For
example, when the state disproportionally suffers the costs of conflict we would expect
the non-state group to be more resistant to a ceasefire. Yet all else being equal, periods
of intense violence impose costs on both parties, and can often be tough for many
6 Journal of Conflict Resolution 0(0)

belligerents to maintain, thus we expect that increases in factors associated with greater
conflict costs are likely to increase the probability of a ceasefire. Given the scarcity of
systematic research that explores the determinants of ceasefires, this seems a logical
point of departure.

‘Battlefield’ violence
The frequency and intensity with which the parties engage strongly shapes the costs of
the conflict. In particular, the number of casualties, material damage and drains on
supplies are critical dimensions. Variation in the technologies of rebellion likely ac-
count for much of the variation in costs across conflicts (Balcells and Kalyvas 2014),
and may well impact the propensity of the participants to engage in a ceasefire. Yet we
are more concerned with changes in the patterns of violence within conflicts. Here, we
expect that increases in the frequency of violence within a conflict increases the costs
for all involved and should therefore also increase the likelihood of a ceasefire.7 For
example, in South Sudan, a ceasefire was signed in July 2018 following the loss of four
hundred people, the most violent period in more than a year (Sundberg and Melander
2013). An imperfect, but relatively established proxy for the frequency of conflict
violence is the number of conflict party fatalities (see, Lacina 2006). From this dis-
cussion we derive our first hypothesis.

Hypothesis 1: Ceasefire are more likely in conflict months with higher combatant
fatalities.

Civilians are also often caught up in civil wars. It is common for conflict violence to
produce ‘collateral damage’, that is the “incidental loss of civilian life, injury to ci-
vilians and damage to civilian objects…caused by an attack on a lawful target”
(International Committee of the Red Cross 2021). ‘Collateral damage’ can increase
conflict costs in at least three ways.
First, collateral civilian damage can turn a domestic population against the armed
actor(s) deemed responsible (Condra and Shapiro 2012; Lyall, Blair, and Imai 2013;
Schutte 2016), reducing the supply of information and weakening an actor’s war
fighting effort (Kalyvas 2000; Kalyvas and Kocher 2009).
Second, there is pressure from international actors, in particular liberal democratic
states (Johns and Davies 2019), to avoid civilian casualties (though see, Salehyan,
Siroky, and Wood 2014). Conflict parties are then more likely to come under inter-
national pressure, potentially increasing conflict costs, in periods with increased
collateral damage.
Finally, while collateral damage is unintentional, it is not random, or evenly dis-
tributed across a dispute. The likelihood of civilians being inadvertently caught up in
the violence is correlated with the patterns of violence (Cronin 2013). Consequently, an
increased number of unintentional civilian fatalities is likely to be representative of shift
in the pattern of violence. Higher unintentional civilian casualties are more likely
Clayton et al. 7

during periods of particularly aggressive military operations, which, as we set out

above, are likely to increase conflict costs.
Of course, the impact of collateral damage is likely to depend on the theatre of
operations (e.g. urban or rural), the local political environment, and the actor seen to be
responsible for the violence (Condra et al. 2010; Condra and Shapiro 2012). If, for
example, one actor is seen to be solely responsible for the violence this might
strengthen civilian support for the other side. However, all else being equal, we expect
that periods with higher collateral damage are costlier for conflict parties, and thus more
likely to see a ceasefire.

Hypothesis 2: Ceasefire are more likely in conflict months with higher ‘collateral’
civilian fatalities.

Civilian targeting by non-state groups

Civilian casualties are not always unintentional. Over the last 30 years almost one
million civilians were killed in intentional attacks by conflict parties (Pettersson and
Öberg 2020). This form of violence goes by many names, including one-sided violence,
terrorism and civilian targeting. A burgeoning body of research explains the causes of
this form of violence (for an excellent review, see, Balcells and Stanton 2021). Tar-
geting civilians offers non-state groups an opportunity to generate revenue, increase
recruitment and punish unsupportive populations, while at the same time undermining
state governance structures (Humphreys and Weinstein 2006). This places additional
demands on state resources, who need to provide some form of protection to vulnerable
communities, and respond to possible downstream effects, such as mass displacement.
Prior research provides some evidence to suggest that civilian violence has the
‘power to hurt’, and can be effective at pushing the state to concede to political ne-
gotiations and concessions (Thomas 2014; Wood and Kathman 2014). Yet in the case of
one-sided non-state violence, the additional costs imposed on the state (and its con-
stituents) are likely to be counterbalanced by increases in the costs associated with a
ceasefire. Indeed, the state would likely suffer greater audience costs if it were to
respond to an increase in civilian atrocities with a commitment to suspend hostilities.
Since civilian loyalties often shift according the patterns of violence (Kalyvas 2000),
when civilians are targeted by non-state groups, they are often punished through the
withdrawal of vital support (Lyall, Blair, and Imai 2013; Pechenkina, Bausch, and
Skinner 2019), or the active collaboration with the opposing force (Clayton and
Thomson 2016; Schutte 2016). The public backlash against non-state groups has
previously been shown to undermine attempts to achieve political goals through
negotiation (Abrahms 2012; Fortna 2015).
We expect that this effect will be particularly pronounced with regards to ceasefires,
as unlike other concessions (e.g. starting negotiations), that can be sold to the public as
part of a mixed strategy of engagement (e.g. talking while fighting), a ceasefire commits
to a suspension of military operations, which is likely to be a difficult pill to swallow for
8 Journal of Conflict Resolution 0(0)

communities who have recently suffered at the hands of the non-state group. We
therefore expect that periods in which non-state groups increase one sided violence
should be less likely to produce a ceasefire.

Hypothesis 3: Ceasefire are less likely in conflict months with higher non-state
armed group violence against civilians.

Conflict duration
The conflict costs, and thus probability of a ceasefire, are likely to vary across the
duration of a dispute. Prior research has shown that the initial onset of violence reveals
important information about the capabilities and resolve of the conflict parties, the costs
associated with violence, and the likelihood of victory/defeat (Fearon 1995; Wagner
2000). This can help resolve problems of asymmetric information that often stand in the
way of an agreement, and so increase the likelihood of a ceasefire in the early phases of
a conflict. Yet after this initial bump in the availability of information, it is subsequently
likely to require longer periods of fighting for conflict parities to update their
knowledge and preferences (see, Sticher and Vukovic 2021). As such, ceasefire should
be more likely in the earlier periods of a dispute and then decline over time.

Hypothesis 4a: Ceasefires are likely in the initial months of a conflict and then
decline over time.

However, as the duration of a conflict increases, it often becomes clear that neither
actor is likely to prevail, as both sides have insufficient strength to overcome their
opponent, but sufficient strength and resolve to continue the conflict indefinitely. In this
case, even if the parties can endure the immediate costs that a conflict produces, the
anticipated costs of an indefinite conflict can shift the decision calculus between
fighting and a ceasefire. Duration provides a good proxy for the longer-term costs of
fighting, as a long history of conflict should lead conflict parties to expect more cu-
mulative costs in the future. Thus as the duration of a conflict increases, so should the
likelihood of a ceasefire.

Hypothesis 4b: Ceasefires are less likely in the early phases of a dispute and then
increase over time.

We therefore expect a non-linear relationship whereby a ceasefire is more likely in

the initial months, following this the likelihood of a ceasefire is likely to decline until
sufficient time has passed for a mutual recognition of the likely future cumulative costs
and unlikelihood of military victory to become clear. This is akin to mediation, which
has been shown to be more likely to be successful ‘early’ and ‘late’ in the duration of a
conflict (Greig 2001; Greig and Regan 2008; Regan and Stam 2000).
Clayton et al. 9

External support
The capacity of conflict parties to sustain costly violence can be shaped by the support
that they receive from other actors. Interested states commonly intervene in support of
one side or the other. When a third party provides military support, either through
troops, supplies or military technology, this can significantly shift the balance of power
within a conflict (Wood, Kathman, and Gent 2012). The effect that this has on the
likelihood of a ceasefire is likely to vary depending on the actor receiving the support.
The state is usually the stronger party, holding a significant military advantage
(Cunningham, Gleditsch, and Salehyan 2009). The state is only likely to concede to a
ceasefire when the non-state group impose sufficient costs. Previous research has
shown that non-state groups are more likely to force concessions from the state, in-
cluding ceasefires, when their capacity more closely approximates (or on rare occasions
exceeds) that of the state (Clayton 2013, 2016; Gent 2011; Hultquist 2013).
The introduction of external support for the state is likely to increase the power
asymmetry, allowing the state to sustain costly conflict for longer periods, counter-
balancing the pressure that rebels impose. In contrast, support for the non-state group is
likely to strengthen their military capacity, and push the parties closer to parity, which in
turn will increase the cost of the conflict. External support on behalf of the non-state
group is often particularly important, as rebels tend to have fewer resources, less
training and are poorer equipped (Cunningham 2010), and thus benefit enormously
from the external support. The support provided by other states can also free up the
groups from resource acquisition activities that take up time and resources, creating
more capacity that can be redirected at the war effort (Salehyan, Siroky, and Wood
2014).8 This increases the capacity of the non-state group to impose costs upon the
state, this has previously been shown to increase the likelihood of negotiated outcomes
(Balch-Lindsay, Enterline, and Joyce 2008; Cunningham, Gleditsch, and Salehyan
2009; Gent 2008), which we also expect to extend to ceasefires.

Hypothesis 5a: Ceasefire are less likely in conflict months with external support for
the state.
Hypothesis 5b: Ceasefire are more likely in conflict months with external support for
the non-state group.

Political cover
The prior discussion sets out how increased conflict costs raise the likelihood of a
ceasefire. But the likelihood of a ceasefire is also determined by the costs that entering
into an agreement imposes on the conflict parties. When ceasefires are ‘cheaper’, we
should expect conflict parties to be more likely to consider their adoption. To this end,
we are again not here concerned with the broader purpose that might underlie the
adoption of a ceasefire. As we discuss above, prior research has shown that peaceful
and devious intentions can underlie the adoption of a ceasefire. Instead, we are here
10 Journal of Conflict Resolution 0(0)

concerned with how moments of political cover that can arise during a dispute impact
the conflict party’s propensity to enter into a ceasefire arrangement.
Political cover is a means through which a conflict party can minimize the domestic
audience costs associated with (in our case) a ceasefire (Allee and Huth 2006;
Beardsley 2010; Druckman 1977). Recall that often conflict parties are reluctant to
enter a ceasefire due to the expected audience costs (often resulting from previous
hardline statements) that offset any potential benefits. Political cover is a means to ‘save
face’ when conceding to a suspension of hostilities with the enemy. In their seminal
work, Allee and Huth (2006) show that when states anticipate audience costs associated
with making voluntary negotiated concessions, they often seek political cover. Spe-
cifically, states find political cover in delegating responsibility for necessary con-
cessions to an international legal body, which makes it easier for leaders to justify the
desired concessions. This is more likely when states expect to suffer significant au-
dience costs from concessions, specifically in disputes that are highly salient for
domestic constituents, and when states are accountable to domestic political opposition
(Allee and Huth 2006; also see, Gent and Shannon 2010).

Mediation
Building on this work, Beardsley (2010) extends his analysis to international crises, and
how mediators can provide political cover. Mediators are likely to be less effective than
legal rulings, as the outcome is participant controlled and non-binding, and thus not so
easily shouldered on the third party (Beardsley 2010; Gent and Shannon 2010).
Nevertheless, Beardsley finds that mediation is also used by conflict parties as a form of
political cover. He finds that mediation is more likely when there is a greater threat of
domestic audience costs for cooperative behavior, and when the conflict parties have a
higher propensity to make concessions.
The political cover afforded by mediation should also extend to ceasefires. Conflict
parties seeking a suspension of hostilities are likely to suffer lower costs if the ceasefire
occurs in the context of mediation. In this case, the costs of conceding to a ceasefire can
be mitigated by the mediator, who bears some responsibility for the act (Beardsley
2010). In some cases, conflict parties may turn to mediation seeking cover for a
ceasefire. In other cases, the mediator may help the parties to agree a ceasefire that was
otherwise not possible, for example by enhancing the flow of communication, im-
proving the design of the process, and providing positive and negative inducements
(Beardsley et al. 2006). In this case, the political cover provided by mediation, rather
than motivate the onset of the process, can make a ceasefire more likely. For example,
during the negotiations between the Government of El Salvador and the FMLN, the
non-state group were unwilling to consent to a ceasefire absent sufficient progress in the
political negotiations. Alvaro De Soto, the UN Mediator, eventually managed to broker
a deal by framing the agreement as a concession and gesture of good will to the United
Nations Secretary General, not the government, thereby providing political cover that
was successful in producing a ceasefire agreement (de Soto and Frazier 2021).
Clayton et al. 11

Hypothesis 6: Ceasefire are more likely during a mediation process

United Nations call for a global ceasefire

Political cover can also arise spontaneously in response to events beyond the parties’
control. For example, natural disasters provide a ‘common enemy’ that might reduce
the audience costs associated with a ceasefire. In this case, a ceasefire is not seen as a
concession to an enemy, but a necessary response to a unique threat posed to con-
stituents. In most cases, this type of political cover is likely to be geographically and
temporally limited. However, the covid-19 pandemic offers a rare opportunity to
explore this effect on a global level.
On the 23 March 2020, as the threat posed by the pandemic became evident,
Guterres (2020), the United Nations Secretary General appealed for a global ceasefire to
“put armed conflict on lockdown and focus together on the true fight of our lives.” This
call was subsequently supported by almost all states, major religious figures and non-
governmental organisations (Ceasefire tracker 2020). This created a unique period of
political cover for any armed groups seeking a ceasefire for humanitarian, military or
political reasons. The threat posed by the pandemic also created incentives for hu-
manitarian ceasefires, and in itself provided a form of cover for conflict parties. But
Guterres’s call provided a unique a clear justification that should have reduced the costs
associated with initiating a ceasefire in the following period. For example, the Phil-
ippine Communist Party agreed to a 3 week ceasefire the day after the call, while Sudan
rebel groups responded by extending the already ongoing ceasefire agreement.

Hypothesis 7: Ceasefire are more likely following the UN SG call for a global
ceasefire

Religious holidays
Finally, periods of shared significance can provide political cover for a ceasefire. When
the state and non-state group both agree the significance of a certain period, this creates
cover for cooperative actions. The first ever ceasefire between the Colombian gov-
ernment and the ELN (in 2017-2018), for instance, was directly linked to the visit of
Pope Francis to Colombia. A common period of shared significance are religious
holidays. If both parties share the same religion, a ceasefire can be justified in relation to
the religious need, rather than as a concession to an opponent. Prior research has shown
that violence on religious holidays can often be disapproved of by civilians (Reese,
Ruby, and Pape 2017), which should reduce audience costs associated with a ceasefire.
Indeed, previous findings suggesting that violence is often reduced during religious
holidays, may be in part a result of the political cover that this provides for a ceasefire.

Hypothesis 8: Ceasefire are more likely on major religious holidays

12 Journal of Conflict Resolution 0(0)

Research design
Data structure
We test our hypotheses on all dyads in intra-state conflicts between 1989-2020 as
defined by the Uppsala Conflict Data Program (UCDP) (Gleditsch et al. 2002;
Pettersson, Högbladh, and Öberg 2019).9 We use a dyad-month structure – each
observation is a month from a particular conflict dyad. Note that a country may have
several on-going conflicts at any given point in time, and each of the conflicts may have
several dyads. A conflict-dyad enters our data on the month of the first fatality assigned
by the UCDP Georeferenced Event Dataset (GED) (Sundberg and Melander 2013), and
remains in our data each month until there are at least 2 months with no recoded
fatalities. Importantly we exclude all dyad months following a ceasefire in which there
are no recorded fatalities, assuming that the ceasefire is in effect and that the dyad as
such is not at risk of a new ceasefire.
We choose the dyad-month as we are interested in when ceasefire occur, and which
conditions correlate with their emergence. The dyad-month is practically the most
disaggregated temporal unit available to analyze conflict event data.10 While many
conflict events are precisely dated, a significant portion of the UCDP GED is not. A
dyad-day dataset would therefore be very vulnerable to measurement error, whereas a
dyad-year model would fail to pick up the dynamic element at the core of our study.
The dyad-month data structure produces 19,513 observations spread across
387 distinct dyads in 109 conflicts, with a total of 809 dyad-months (4.2%) with
ceasefires. Most dyads never see any ceasefire (234). Of the 153 conflict dyads that do
see a ceasefire, the number of ceasefires varies greatly, between 1 (54 instances) and 61
(the conflict between Philippines and CPP11) months with ceasefires.

Dependent variable
Our dependent variable is ceasefire onset. For this we draw on the new Civil Conflict
CeaseFire (CF) dataset (Clayton et al. 2022b). The CF dataset is the most compre-
hensive collection of ceasefire data currently available. In total, the CF dataset captures
2202 ceasefires in 109 civil conflicts between 1989 and 2020. The CF dataset adopts a
broad definition of a ceasefire as ‘an arrangement by or between conflict parties to
suspend violent hostilities from a specific point in time’. This definition captures the full
range of related security arrangements which conflict parties might use to temporarily
suspend or terminate hostilities, including arrangements labelled as truces, cessation of
hostilities, armistices, and preliminary ceasefire agreements (Clayton et al. 2019;
Clayton and Sticher 2021; PILPG 2013). In this way, the CF dataset considers ceasefire
an umbrella term capturing all arrangements that meet the above definition.12
We only include ceasefires that match one or more dyads recorded by UCDP. In
some cases, one ceasefire can include several dyads, such as the multilateral nationwide
ceasefire in Myanmar in 2018. Examples of ceasefires excluded from this analysis are
Clayton et al. 13

those solely between non-state groups. Furthermore, we exclude ceasefires that are
continuations of previously agreements.13 We also exclude definitive ceasefire
agreements i.e. permanent ceasefire that enter into effect with a peace agreement, based
on the assumption that they are born from a different data generating process than
temporary ceasefires. We discuss and test this assumption in our robustness checks.
Importantly, we focus on the declaration of a ceasefire from a specific point of time,
regardless of whether violence does indeed decline from the predetermined time or if
the ceasefire was in fact implemented. Thus, a ceasefire does not have to produce a
break in hostilities in order for us to treat it as a ceasefire.

Independent variables
From the theoretical discussion above we derived a series of hypothesis that either
relate to conflict costs or political cover. In the following we discuss the operation-
alization of the hypotheses in more detail.
Combatant fatalities are calculated using the total non-civilian battle deaths for each
month using the UCDP GED dataset (Sundberg and Melander 2013). This includes the
sum of state fatalities, non-state armed group fatalities, and fatalities to which the group
cannot be identified (i.e. in the GED Data deaths_unknown). The variable is log-
transformed using the formula ln(x+1) to avoid log of 0. See Appendix B for a
justification.
Collateral civilian fatalities are captured using the number of civilian deaths that
result from violence between the state and non-state group, but where there was no clear
intent on the part of either organized actor to target the civilians.14 Over the past
30 years the share of civilian deaths has been around 20-25% of total deaths. This is also
taken from the GED data, and log-transformed using the formula ln(x+1) to avoid log
of 0.
One-sided non-state armed group violence is captured by assigning all one-sided
events (i.e. intentional killings of civilians) in UCDP GED associated with a rebel
organization for a specific country and month to the dyad associated with this rebel
group.15
Duration is measured as the cumulative number of active months a dyad has seen up
until each dyad-month observation. We log-transform this variable and include first and
second polynomials to catch our theorized effect of an early high probability of
ceasefires, followed by a quick drop and a gradual increase.
External support is captured using two dummy variables. The first is coded 1 when a
foreign government supports the local government, and the second is coded 1 if the
opposition is supported. This is taken from the UCDP/PRIO Armed Conflict data,
which also indicates the support of other actors (i.e. side_a_2nd, and side_b_2nd
variables).
Mediation is identified using information from the Civil War Mediation dataset
(DeRouen, Bercovitch, and Pospieszna 2011) measured annually at the conflict level.16
14 Journal of Conflict Resolution 0(0)

UN call for a global ceasefire is captured with a dummy variable coded 1 for March
2021, the month in which the United Nations Secretary General announced the call.
This was also the month in which covid-19 was declared a pandemic. We expect that the
political cover to be highest in the immediate aftermath of the call. In robustness checks
we also explore other specifications of this variable.
Religious holidays are captured using the most significant Christian and Islamic
religious holiday. For majority Christian countries, we code December = 1. For
majority Islamic countries, we code Eid al-Fitr as 1 in the appropriate month. This is
based on data from Pew Research (2015). See Appendix D for a discussion of al-
ternative holidays.

Modelling approach
We analyse our data using fixed effects panel OLS regression models with robust
standard errors clustered on the dyad (Abadie et al. 2017; Angrist and Pischke 2008,
221).17 The dependent variable is dichotomous and highly skewed. The panel can be
viewed as a discrete duration design, where the theoretical observation is the active
conflict-dyad period, which is then divided into sub-units (months) to capture conflict
dynamics.
A fixed effect set up is well suited to identify the circumstances that trigger a
ceasefire. It allows us to evaluate how changes in the key covariates within a dyad
influence the likelihood of a ceasefire, while effectively controlling for many potential
omitted variables, such as economic development, political incompatibility, or conflict
contestation. This is the most suitable means for identifying which factors changing
during a conflict correlate with ceasefire onsets, helping to explain when a ceasefire is
more likely to occur.
A disadvantage with this approach is that it makes every invariant country-, conflict-
or dyad-specific variable redundant. This is not a problem in our case, as each of our key
independent variables are dynamic and do shift during conflict. Fixed effects do not
deal well with omitted factors that can change rapidly during the conflict, nor does fixed
effects ensure that these sub-units meet the i.i.d. criterion. We therefore run robustness
checks without fixed effects to further assess the validity of our results, these are
reported in Appendix F.
A key question separating the linear probability model and a logit model is whether
we believe the effects are multiplicative or additive. We discuss this in greater detail in
Appedix F, where we show that both a pooled logit and a conditional logit estimator are
mostly consistent but less conservative than our OLS estimates.
We control for proximity to previous ceasefire and use cluster-robust standard errors
to counter heteroscedasticity or non-identical distribution of residuals (Abadie et al.
2017; Cameron and Miller 2015). This is consistent with King and Roberts (2015)
critique of robust standard errors as the model misspecification in this case stems
directly from the data structure. Taken together with the fixed effects models, we
believe that this is the most conservative modelling choice available. The temporal
Clayton et al. 15

control is based on the decay function transformation of time (Raknerud and Hegre
1997) with a half-life parameter of 1 month. We run alternative approaches for both
standard errors and temporal control reported in Appendix E and G.

Analyses and results

Table 1 presents the results of a series of fixed effects panel OLS regression models.
Model 1 estimates only cost-related variables, Models 2 and 3 estimates political cover-
related variables and Models 4 and 5 estimates the two sets jointly. Unfortunately, the
mediation variable is missing from 2014 and onwards, and the Covid-19-related
variable is only positive in 2020. The two can therefore not be estimated in the same
model, which necessitates Models 2 and 3, and Models 4 and 5.
Hypothesis 1, that combatant fatalities are positively associated with ceasefires, is
strongly supported. In models 1, 4 and 5 the measure of combatant fatalities shows the
expected positive result, which is significant in all specifications. Substantively, for
every doubling of casualties, the probability of a ceasefire increases by 1pp. An increase
from 25 Battle Related Deaths to 1000 BRD leads to an increase of almost 4pp per
month and 36pp per year. Figure 118 illustrates the substantive effect in in the range
minimum (0) to 95th percentile (at 1096). Minor conflicts have a low likelihood of
ceasefires in general, but the marginal effect of an additional fatality is quite large for
minor conflicts. Increasing the number of fatalities from 1 to 150 per month increases
the likelihood for ceasefire by 4.9pp. Increasing from 150 to 1000, in contrast, adds
only 2.1pp to this likelihood.19 Hypothesis one is supported.
The second hypothesis stated a similar expectation regarding civilian collateral
fatalities. Again, we find a positive and significant result across all models. These
fatalities have a slightly larger marginal effect, but they are less frequent. Hypothesis
2 is also supported. Figure 2 is similar to Figure 1 in that the X axis range is minimum to
95th percentile.20 We see a very similar effect, albeit at a different scale. The marginal
effect of an additional civilian fatality is much higher, but the prevalence of these is also
much lower. Moving from 0 to 25 civilian fatalities increase the likelihood of a ceasefire
by about 3.8pp, while changing from 25 to 50 increases the likelihood by about
0.8pp. Hypothesis 2 is then supported. More costly conflict, whether measured through
combatant or unintentional civilian fatalities, increases the likelihood of a ceasefire.
The third hypothesis focuses on rebel violence that intentionally targets civilians.
We expect this violence to increase the barriers to ceasefires. Across models 1, 4 and
5 we find the hypothesized negative effect, which is significant according to con-
ventional levels. However, our argument here is more specific in that we suggested that
this form of violence would reduce the likelihood of the state being willing to enter into
an agreement with the non-state group, but should not impact the propensity of the non-
state group to enter into an arrangement. To test this, we divide ceasefires into uni-
lateral and bi/multi-lateral arrangements and estimate our fully specified model. We
present the results in Table 2. As expected, one-sided violence by the non-state group
has a consistent and significant negative effect on the likelihood of bi-lateral
 16

Table 1. Regression Results From Full Sample.

Model 1 Model 2 Model 3 Model 4 Model 5

Combatant fatalities (log) 0.011***(6.28) 0.011***(6.24) 0.011***(5.62)

Collateral Civ.Fatalities (log) 0.012***(4.43) 0.012***(4.52) 0.012***(4.50)
NSA one-sided Violence (log) 0.003 ( 1.46) 0.003 ( 1.48) 0.003 ( 1.97)
Internat.supp.Opposition 0.103**(2.70) 0.103**(2.73) 0.159***(3.41)
Internat.Supp.Govt 0.016 ( 1.48) 0.015 ( 1.39) 0.016 ( 1.36)
Duration (log) 0.001 (0.41) 0.001 (0.28) 0.000 ( 0.07)
Duration (log) # duration (log) 0.004***(3.82) 0.004***(3.78) 0.004***(3.21)
Recent CF (log) 0.071***(3.34) 0.093***(3.83) 0.079**(3.31) 0.071***(3.38) 0.056**(2.84)
Christmas 0.058 + (1.68) 0.068 (1.63) 0.059 + (1.69) 0.067 (1.62)
Eid al–Fitr 0.000 ( 0.03) 0.004 (0.29) 0.002 (0.24) 0.007 (0.58)
UNSG global CF call 0.061*(1.98) 0.068*(2.15)
Mediation 0.023**(2.65) 0.021**(2.61)
Constant 0.015*( 2.23) 0.033***(19.23) 0.029***(7.04) 0.017*( 2.52) 0.023**( 2.65)
N 19,513 19,513 14,220 19,513 14,220
CFs 809 809 643 809 643

Note: +p < 0.1, *p < 0.01, **p < 0.001, ***p < 0.0001.
Journal of Conflict Resolution 0(0)
Clayton et al. 17

Figure 1. Effect of combatant fatalities on monthly probability of ceasefire.

Figure 2. Effect of collateral civilian fatalities on monthly probability of ceasefire.

 18

Table 2. Regression Results from Bivariate and Univariate Sub-Samples.

Model 6 Model 7 Model 8 Model 9

Combatant fatalities (log) 0.007***(4.41) 0.007***(3.85) 0.005***(5.19) 0.006***(5.39)

Civ. Collateral Fatalities (log) 0.008***(3.71) 0.007***(3.58) 0.008***(4.18) 0.009***(4.03)
NSA one–sided.Violence (log) 0.004**( 3.28) 0.004**( 2.90) 0.000 ( 0.05) 0.001 ( 0.49)
Internat.supp.Opposition 0.102**(2.62) 0.167**(3.24) 0.022 (1.33) 0.025 (1.23)
Internat.Supp.Govt 0.005 ( 0.88)) 0.008 ( 0.92) 0.008 ( 0.84) 0.007 ( 0.92)
Duration (log) 0.001 ( 0.29) 0.002 ( 0.63) 0.002 (0.91) 0.002 (0.64)
Duration (log) # duration (log) 0.002*(2.40) 0.002+(1.92) 0.003***(3.54) 0.003***(3.22)
Christmas 0.007 (1.24) 0.008 (1.21) 0.053 (1.50) 0.060 (1.44)
Eid al–Fitr 0.007 ( 1.54) 0.006 ( 1.06) 0.009 (1.07) 0.015 (1.20)
UNSG global CF call 0.020 (1.15) 0.048+ (1.73)
Recent CF 0.046*(2.14) 0.046+(1.91) 0.067** (2.86) 0.042+ (1.93)
Mediation 0.015*(2.43) 0.008 (1.61)
Constant 0.012*( 2.07) 0.016*( 2.18) 0.010*( 2.54) 0.013**( 2.63)
N 19,513 14,220 19,513 14,220
CFs 409 317 455 369

Note: +p < 0.1, *p < 0.01, **p < 0.001, ***p < 0.0001.
Journal of Conflict Resolution 0(0)
Clayton et al. 19

Figure 3. Effect of dyad duration on monthly probability of ceasefire.

agreements with the state, but no effect on unilateral arrangements. Substantively, an

increase from 0 to 25 fatalities reduces the likelihood of a bi-lateral ceasefire by
1.3pp. Hypothesis 3 is then also supported.
Hypothesis 4a and 4b posits that ceasefires are more likely in the initial phase of a
conflict, before dropping down to a low level and gradually increasing from that level.
We find support for both these hypothesis. This is best understood with reference to
Figure 3. Initially the likelihood of a ceasefire is at its peak at 8.4% this then declines
rapidly to 4% after 9 months. The turning point seems to be around 29 months, at which
point the likelihood of a ceasefire is only 3.4%. The likelihood of a ceasefire then
increases again, but slowly. The monthly probability reached 5% at 18 years and 6% at
30 years. Both hypothesis 4a and 4b are then supported.
The fifth hypotheses set out our claim that external support for the non-state armed
group will increase the likelihood of a ceasefire by reducing the power asymmetry in
civil conflict, whereas external support on the government side is expected to have the
opposite effect. External support for the non-state group does indeed increase the
likelihood of a ceasefire in the range of 10-15pp per month, supporting Hypothesis 5a.
As a further test of this argument, we turn to Table 2. If external support does indeed
increase the relative capacity of the non-state group, and thus their capacity to extract
concessions from the state, then we would expect to see an increased likelihood in bi/
multi-lateral ceasefires that also involve the state. That is indeed what we find, offering
further support for Hypothesis 5a. At the same time, support for the rebel side is a rare
phenomenon, so we interpret this result with some caution.
20 Journal of Conflict Resolution 0(0)

We do not however find support for the government-based hypothesis. As expected,

the variable shows the negative sign, but is not significant in any model. Prior research
suggests that support for the government may be less influential on the relative dis-
tribution of power than support for the non-state group, which might in part explain this
result.
Next, we turn to the analysis of political cover. First, we explore if mediation impacts
the likelihood of ceasefire. Recall that mediation reduces the audience costs associated
with a ceasefire, as the outcome can be partially attributed to the third party. This should
be particularly the case with regards to bilateral ceasefires, as in the presence of a
mediator agreements with an opponent can more easily be ‘sold’ as a concession to the
process or mediator, rather than to the other side. We find support for this argument, the
mediation variable is positive and significant in all models, though the effect is rel-
atively modest at 2pp per month.21 As expected, the impact of mediation is more
important for bi-lateral agreements.22
Turning to the UN call for a global ceasefire, we find that in March 2020, the month
in which the initiative was announced the probably of ceasefire increased by around
6pp. It does therefore seem that the global call provided sufficient cover to encourage
conflict parties to enter into a ceasefire. In the Appendix L we run additional analysis to
determine if the effect can be seen in the months preceding the call, which might point
to a more general response to the emerging threat of the pandemic, or in the following
months, when the global community rallied in support of the initiative and the crisis
escalated. We find that only in March 2020 does the probability of a ceasefire increase
significantly. As the threat of the pandemic, and political pressure behind the global
call, arguably increased in the following months. That we only find an effect in March
points to the important role of political cover. All conflict parties who saw the political
cover as sufficiently significant as to shift their decision calculus appear to have signed
a ceasefire in March, meaning that by April, there were far fewer conflict parties whose
reluctance to enter into a ceasefire could be solved by cover. Unsurprisingly, given the
sudden nature of the UNSG call’s, the effect is limited to unilateral ceasefire. These
arrangements can be announced more quickly, taking immediate advantage of the cover
and placing the other side under pressure to reciprocate.
The final hypothesis asserts that religious holidays can be used as a political cover.
The evidence is mixed. Christmas has a fairly strong coefficient around 6pp, which is
par with the Security Council global call (but obviously happens every year), but the
Christmas effect is only statistically significant at the 10% level and does not hold up
when we look separately at unilateral and bilateral arrangements. The most plausible
interpretation is that political cover is relevant for both unilateral and bilateral
ceasefires, and that context dictates which type is chosen in specific situations. We find
no effect from Eid al-Fitr. We therefore find only limited evidence that religious
holidays offer political cover.
Clayton et al. 21

Robustness
To assess the robustness of our results we estimate some additional models. We provide
full detail of all the robustness checks in the appendix.

Conclusions
Ceasefires are common during civil conflict. Yet we still know very little about why and
under what conditions ceasefires more generally occur. To that end, we have situated
and conceptualized ceasefires as part of a broader understanding of the conflict process.
We find that conflict dynamics are strongly associated with the likelihood of ceasefires.
More intense fighting increases the likelihood of a ceasefire, whereas one-sided rebel
violence towards civilians has a negative effect, in particular on bilateral agreements.
International support for the rebels increases the likelihood of a bilateral ceasefire,
while support for the government side is inconclusive. Ceasefires are more likely early
in a dispute, and then become less likely for some time before eventually increasing
again.
We find partial support for the political cover hypotheses. Mediation does increase
the likelihood of ceasefires; future research should attempt to do more to distinguish
between the different mechanisms through which mediation might lead to a ceasefire to
tease out the independent effect of political cover. The UNSG call for a global ceasefire
in March 2020 did lead to a significant jump when controlling for conflict dynamics,
though a number of studies have subsequently shown that these arrangements had little
effect (Gowan 2020). Christmas is associated with more ceasefires, but not Eid al-Fitr.
Future studies could explore how the characteristics of societies, including religion and
ideology, shape the ceasefire process.
This study is only a first step towards building robust comparative knowledge on the
causes of ceasefires. We must delve deeper into the mechanisms and improve our
understanding of why and when belligerents declare ceasefires. In this, we also need
research that focuses on different forms of ceasefire (e.g. non-state ceasefire though see,
Duursma 2022; Lundgren, Svensson and Karakus 2022), and consider a wider range of
context factors (e.g. see, Braithwaite and Butcher 2022). We also need research designs
that get more squarely at the causal impact of various features of conflicts on ceasefires.
It is also not yet clear how differences across conflicts, including the type of warfare,
characteristics of non-state groups, and ideologies, shape the ceasefire process, and if
arrangements that emerge in different contexts produce heterogenous effects. De-
veloping this knowledge is a necessary step in understanding the various functions that
ceasefires play during intra-state conflict.

Acknowledgements
The authors thank the editor of JCR and anonymous reviewers.
22 Journal of Conflict Resolution 0(0)

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship,
and/or publication of this article: This project was made possible by generous funding from the
Folke Bernadotte Academy [DNR-17-00427], the Swedish Research Council [2018-01520], the
Centre for Security Studies at ETH Zürich, and the Norwegian Ministry of Foreign Affairs
[Conflict Trends QZA-21/0177].

ORCID iDs
Govinda Clayton  https://orcid.org/0000-0002-1426-0669
Håvard Mokleiv Nygård  https://orcid.org/0000-0002-8909-6146
Siri A. Rustad  https://orcid.org/0000-0001-9653-5488
Håvard Strand  https://orcid.org/0000-0002-1348-4900

Supplemental Material
Supplemental material for this article is available online. See www.ceasefireproject.org.

Notes
1. Ceasefires can of course also serve important peacemaking functions.
2. We exclude definitive ceasefires, i.e. permanent ceasefires that enter into effect with peace
agreements, as it is likely that these ceasefires arise through different causal processes than
temporary ceasefires, and thus do not fit with our theoretical arguments.
3. On rare occasions the state might also have an interest in using a ceasefire to (re-)gain
international legitimacy, for example when they have previously been associated with
atrocious forms of violence.
4. States can also generate a reputation for cooperation by agreeing and honoring ceasefires
(see, Bara and Clayton 2022).
5. For example, even irregular conflict as a significant economic impact (for example see,
Dorsett 2013).
6. Definitive ceasefire can also (in effect) terminate a violent conflict, but these are not the focus
of this study.
7. The distribution of battlefield casualties among the parties is probably very important to
understand the dynamic. Unfortunately not even the best data available allows us to separate
casualties between parties, and we therefore refrain from theorizing.
8. This has also been shown to increase the likelihood of violence against civilians by in-
creasing the likelihood of opportunist recruits, though this does vary according to the
preferences of the patron state, and the number of patrons (Salehyan, Siroky, and Wood,
2014).
Clayton et al. 23

9. We exclude interstate and extra-systemic conflict, as well as non-state conflicts and one-sided
violence.
10. The actor-level is conceivable, but hypotheses 1 and 3 relate to the interaction of parties in a
conflict and cannot be meaningfully measured at the actor level.
11. See https://ucdp.uu.se/statebased/411 for more details on this conflict.
12. In Appendix C we run several robustness tests applying several other criterions, such as
limiting the stated purpose to humanitarian or peace processes; and excluding bi- and uni-
lateral declarations.
13. In this case we still include the initial ceasefire declaration, but do not include the yearly
renewals which the CF dataset considers a part of the initial ceasefire.
14. This interpretation of collateral civilian damage was confirmed in personal correspondence
with members of the UCDP GED project.
15. Under the UCDP conflict definition, state-based conflicts and one-sided conflicts are two
different categories, and as such UCDP do not code one-sided violence within a civil war.
However, as a specific rebel organization can only participate in one conflict in a given
country, it is possible for us to attribute the violence of a group to one particular dyad. It is
possible for a non-state organization to be involved in more than one country, and therefore
in more than one conflict at a given time. IS was involved in 16 conflicts in 2020.
16. This was extended by Bara and Clayton (2022)
17. All models were estimated using STATA 16.
18. All figures are calculated based on Model 4.
19. Conflict parties might of course exacerbate violence in order to pressure the opponent into a
ceasefire. Yet even in this case, it is still the increased costs born from violence that increased
the parties willingness to accept a ceasefire.
20. The vast difference between the distribution of collateral civilian fatalities and combatant
fatalities is probably somewhat artificial as the substantial number of fatalities coded as
unknown in the UCDP GED dataset are classified as combatant in our analysis.
21. It is also possible that this results from problems with the mediation indicator, which is at the
yearly level.
22. It is possible that mediation occurs within a conflict due to increased international pressure,
and that this same pressure also produces a ceasefire. However, even in this case, the political
cover that mediation provides is likely to be influential. We reserve unpicking the different
mechanisms through which political cover might shape the relationship between mediation
and ceasefires for future work.

References
Aary, Vaughn (1995) “Concluding Hostilities: Humanitarian Provisions in Cease-Fire Agree-
ments.” Military Law Review 148: 186-273.
Abadie, Alberto, Susan Athey, Guido W Imbens, and Jeffrey Wooldridge (2017) “When Should
You Adjust Standard Errors for Clustering?” w24003. National Bureau of Economic
Research (https://www.nber.org/papers/w24003).
24 Journal of Conflict Resolution 0(0)

Abrahms, Max (2012) “The Political Effectiveness of Terrorism Revisited.” Comparative Po-
litical Studies 45 (3): 366-393.
Åkebo, Malin (2020) “Ceasefire Rationales: A Comparative Study of Ceasefires in the Moro and
Communist Conflicts in the Philippines.” International Peacekeeping 28 (2): 1-27.
Allee, Todd L, and Paul K Huth (2006) “Legitimizing Dispute Settlement: International Legal
Rulings as Domestic Political Cover.” American Political Science Review 100 (2): 219-234.
Angrist, Joshua D, and Jorn-Steffen Pischke (2008) Mostly Harmless Econometrics: An Em-
piricist’s Companion. Princeton: Princeton University Press.
Arjona, Ana (2017) Rebelocracy. Social Order in the Colombian Civil War. Cambridge:
Cambridge University Press.
Balcells, Laia, and Stathis N Kalyvas (2014) “Does Warfare Matter? Severity, Duration, and
Outcomes of Civil Wars.” Journal of Conflict Resolution 58 (8): 1390-1418.
Balcells, Laia, and Jessica A Stanton (2021) “Violence Against Civilians During Armed Conflict:
Moving Beyond the Macro- and Micro-Level Divide.” Annual Review of Political Science
24 (1): 45-69.
Balch-Lindsay, Dylan, Andrew J Enterline, and Kyle A Joyce (2008) “Third-Party Intervention
and the Civil War Process.” Journal of Peace Research 45 (3): 345-363.
Bara, Corinne, and Govinda Clayton (2022) “Your Reputation Precedes You: Ceasefires,
Credibility and Bargaining During Civil Conflict.” Journal of Conflict Resolution forth-
coming. DOI: 10.1177/00220027221126725.
Bara, Corinne, Govinda Clayton, and Siri Aas Rustad (2021) “Understanding Ceasefires.”
International Peacekeeping 28 (3): 329-340.
Beardsley, Kyle C (2010) “Pain, Pressure and Political Cover: Explaining Mediation Incidence.”
Journal of Peace Research 47 (4): 395-406.
Beardsley, Kyle C, David M Quinn, Bidisha Biswas, and Jonathan Wilkenfeld (2006) “Mediation
Style and Crisis Outcomes.” Journal of Conflict Resolution 50 (1): 58-86.
Bormann, Nils-Christian, and Burcu Savun (2018) “Reputation, Concessions, and Territorial
Civil War: Do Ethnic Dominoes Fall, or Don’t They?” Journal of Peace Research 55 (5):
671-686.
Braithwaite, Jessica Maves, and Charles Butcher (2022) “Muddying the Waters: The Anatomy of
Resistance Campaigns and the Failure of Ceasefires in Civil Wars.” Journal of Conflict
Resolution Forthcoming.
Brutger, Ryan, and Joshua D Kertzer (2018) “A Dispositional Theory of Reputation Costs.”
International Organization 72 (3): 693-724.
Cameron, A Colin, and Douglas L Miller (2015) “A Practitioner’s Guide to Cluster-Robust
Inference.” Journal of Human Resources 50 (2): 317-372.
Ceasefire tracker (2020) Ceasefires in a time of covid-19. https://pax.peaceagreements.org/static/
covid19ceasefires/
Chounet-Cambas, Luc (2011a) Negotiating Ceasefires: Dilemmas & Options for Mediators.
Geneva: The Centre for Humanitarian Dialogue.
Chounet-Cambas, Luc (2011b) Negotiating ceasefires. Mediation Practice Series Centre for
Humanitarian Dialogue: Geneva.
Clayton et al. 25

Clayton, Govinda (2013) “Relative Rebel Strength and the Onset and Outcome of Civil War
Mediation.” Journal of Peace Research 50 (5): 609-622.
Clayton, Govinda (2016) “Oil, Relative Strength and Civil War Mediation.” Cooperation and
Conflict 51 (3): 325-344.
Clayton, Govinda, Simon Mason, Valerie Sticher, and Claudia Wiehler (2019) “Ceasefires in
Intra-state Peace Processes.” CSS Analysis in Security Policy (252). https://css.ethz.ch/
content/dam/ethz/special-interest/gess/cis/center-for-securities-studies/pdfs/
CSSAnalyse252-EN.pdf
Clayton, Govinda, Laurie Nathan, and Claudia Wiehler (2021) “Ceasefire Success: A conceptual
framework.” International Peacekeeping 28 (3): 341-365.
Clayton, Govinda, Håvard Mokleiv Nygård, Siri Aas Rustad, and Håvard Strand (2022a)
“Ceasefires in civil wars: A research agenda (Introduction to Special Section of the Journal
of Conflict Resolution).” Journal of Conflict Resolution forthcoming. DOI: 10.1177/
00220027221128300.
Clayton, Govinda, Håvard Mokleiv Nygård, Håvard Strand, Siri Aas Rustad, Claudia Wiehler,
Tora Sagård, Peder Landsverk, Reidun Ryland, Valerie Sticher, Emma Wink, and Corrine
Bara (2022b) “Introducing the Civil Conflict Ceasefire Dataset.” Journal of Conflict
Resolution forthcoming. DOI: 10.1177/00220027221129183.
Clayton, Govinda, and Valerie Sticher (2021) “The Logic of Ceasefires in Civil War.” Inter-
national Studies Quarterly 65 (3): 633-646.
Clayton, Govinda, and Andrew Thomson (2016) “Civilianizing Civil Conflict: Civilian Defense
Militias and the Logic of Violence in Intrastate Conflict.” International Studies Quarterly 60
(3): 499-510.
Collier, Paul, and Anke Hoeffler (2004) “Greed and Grievance in Civil War.” Oxford Economic
Papers 56 (4): 563-595.
Collier, Paul, Anke Hoeffler, and Måns Söderbom (2004) “On the Duration of Civil War.”
Journal of Peace Research 41 (3): 253-273.
Condra, Luke N, Joseph H Felter, Radha K Iyengar, and Jacob N Shapiro (2010) The Effect of
Civilian Casualties in Afghanistan and Iraq. w16152. National Bureau of Economic
Research. https://www.nber.org/papers/w16152
Condra, Luke N, and Jacob N Shapiro (2012) “Who Takes the Blame? The Strategic Effects of
Collateral Damage.” American Journal of Political Science 56 (1): 167-187.
Costalli, Stefano, Luigi Moretti, and Costantino Pischedda (2017) “The Economic Costs of Civil
War: Synthetic Counterfactual Evidence and the Effects of Ethnic Fractionalization.”
Journal of Peace Research 54 (1): 80-98.
Crocker, Chester A, Fen Osler Hampson, and Pamela R Aall (1999) Herding Cats: Multiparty
Mediation in a Complex World. Washington, DC, US: United States Institute of Peace Press.
Cronin, Bruce (2013) “Reckless Endangerment Warfare: Civilian Casualties and the Collateral
Damage Exception in International Humanitarian Law.” Journal of Peace Research 50 (2):
175-187.
Cunningham, David E (2010) “Blocking Resolution: How External States can Prolong Civil
Wars.” Journal of Peace Research 47 (2): 115-127.
26 Journal of Conflict Resolution 0(0)

Cunningham, David E, Kristian Skrede Gleditsch, and Idean Salehyan (2009) “It Takes Two: A
Dyadic Analysis of Civil War Duration and Outcome.” Journal of Conflict Resolution 53
(4): 570-597.
de Soto, Alvaro, and Owen Frazier (2021) “The Ultimate Bargaining Chip: Ceasefires in the El
Salvador Peace Negotiations.” ETH Working Paper.
Debs, Alexandre, and HE Goemans (2010) “Regime Type, the Fate of Leaders, and War.”
American Political Science Review 104(3): 430-445.
DeRouen, Karl, Bercovitch Jacob, and Paulina Pospieszna (2011) “Introducing the Civil Wars
Mediation (CWM) Dataset.” Journal of Peace Research 48 (5): 663-672.
Dorsett, Richard (2013) “The Effect of the Troubles on GDP in Northern Ireland.” European
Journal of Political Economy 29 (C): 119-133.
Druckman, Daniel (1977) Negotiations, Social-Psychological Perspectives. Beverly Hills: Sage
Publications.
Duursma, Allard (2022) “Peacekeeping, Mediation, and the Conclusion of Ceasefires in Non-
State Conflicts.” Journal of Conflict Resolution Forthcoming.
Fearon, James D (1994) “Domestic Political Audiences and the Escalation of International
Disputes.” American Political Science Review 88 (3): 577-592.
Fearon, James D (1995) “Rationalist Explanations for War.” International Organization 49 (3):
379-414.
Fearon, James D (1997) “Signaling Foreign Policy Interests: Tying Hands versus Sinking Costs.”
Journal of Conflict Resolution 41 (1): 68-90.
Fortna, Virginia Page (2015) “Do Terrorists Win? Rebels’ Use of Terrorism and Civil War
Outcomes.” International Organization 69 (3): 519-556.
Gates, Scott, Håvard Hegre, Håvard Mokleiv Nygård, and Håvard Strand (2012) “Development
Consequences of Armed Conflict.” World Development 40 (9): 1713-1722.
Gent, Stephen E (2008) “Going in When It Counts: Military Intervention and the Outcome of
Civil Conflicts.” International Studies Quarterly 52 (4): 713-735.
Gent, Stephen E (2011) “Relative Rebel Strength and Power Sharing in Intrastate Conflicts.”
International Interactions 37 (2): 215-228.
Gent, Stephen E, and Megan Shannon (2010) “The Effectiveness of International Arbitration and
Adjudication: Getting Into a Bind.” The Journal of Politics 72(2): 366-380.
Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and Håvard
Strand (2002) “Armed Conflict 1946-2001: A New Dataset.” Journal of Peace Research 39
(5): 615-637.
Gowan, Richard (2020) “What’s Happened to the UN Secretary-General’s COVID-19 Ceasefire
Call?” International Crisis Group. https://www.crisisgroup.org/global/whats-happened-un-
secretary-generals-covid-19-ceasefire-call
Greig, J Michael (2001) “Moments of Opportunity: Recognizing Conditions of Ripeness for
International Mediation between Enduring Rivals.” Journal of Conflict Resolution 45 (6):
691-718.
Greig, J Michael, and Patrick M Regan (2008) “When Do They Say Yes? An Analysis of the
Willingness to Offer and Accept Mediation in Civil Wars.” International Studies Quarterly
52 (4): 759-781.
Clayton et al. 27

Guterres, Antonio (2020) “UN Secretary-General Calls for Global Ceasefire to Focus on Ending
the COVID-19 Pandemic.” https://www.un.org/en/academic-impact/un-secretary-general-
calls-global-ceasefire-focus-ending-covid-19-pandemic
Gutiérrez-Sanı́n, Francisco, and Elisabeth Jean Wood (2017) “What Should We Mean by “Pattern
of Political Violence”? Repertoire, Targeting, Frequency, and Technique.” Perspectives on
Politics 15 (1): 20-41.
Hanson, Kolby (2020) “Live and Let Live: Explaining Long-term Truces in Separatist Conflicts.”
International Peacekeeping 28 (3): 1-23.
Höglund, Kristine (2011) “Tactics in Negotiations between States and Extremists: The Role of
Cease-Fires and Counterterrorist Measures.” In Engaging Extremists: Trade-Offs, Timing,
and Diplomacy. Washington DC: United States Institute of Peace.
Hultquist, Philip (2013) “Power Parity and Peace? The Role of Relative Power in Civil War
Settlement.” Journal of Peace Research 50 (5): 623-634.
Humphreys, Macartan (2005) “Natural Resources, Conflict, and Conflict Resolution: Uncovering
the Mechanisms.” Journal of Conflict Resolution 49 (4): 508-537.
Humphreys, Macartan, and Jeremy M Weinstein (2006) “Handling and Manhandling Civilians in
Civil War.” American Political Science Review 100 (3): 429-447.
International Committee of the Red Cross (2021) “Rule 15. Principle of Precautions in Attack,
International Humanitarian Law.” https://ihl-databases.icrc.org/customary-ihl/eng/docs/v1_
rul_rule15
Johns, Robert, and Graeme AM Davies (2019) “Civilian Casualties and Public Support for
Military Action: Experimental Evidence.” Journal of Conflict Resolution 63 (1): 251-281.
Kalyvas, Stathis N (2000) Logic of Violence in Civil Wars. New York: New York University.
Kalyvas, Stathis N, and Matthew Adam Kocher (2009) “The Dynamics of Violence in Vietnam:
An Analysis of the Hamlet Evaluation System (HES).” Journal of Peace Research 46 (3):
335-355.
Kertzer, Joshua D, and Ryan Brutger (2016) “Decomposing Audience Costs: Bringing the
Audience Back Into Audience Cost Theory.” American Journal of Political Science 60 (1):
234-249.
King, Gary, and Margaret Roberts (2015) “How Robust Standard Errors Expose Methodological
Problems They Do Not Fix, and What to Do About It.” Political Analysis 23 (2): 159-179.
Koubi, Vally (2005) “War and Economic Performance.” Journal of Peace Research 42 (1):
67-82.
Lacina, Bethany (2006) “Explaining the Severity of Civil Wars.” Journal of Conflict Resolution
50 (2): 276-289.
Lundgren, Magnus, Isak Svensson, and Dogukan Cansin Karakus (2022) “Local ceasefires and
de-escalation: Evidence from the Syrian Civil War.” Journal of Conflict Resolution
Forthcoming.
Lyall, Jason, Graeme Blair, and Kosuke Imai (2013) “Explaining Support for Combatants during
Wartime: A Survey Experiment in Afghanistan.” American Political Science Review 107
(04): 679-705.
Mahieu, Sylvie (2007) “When Should Mediators Interrupt a Civil War? The Best Timing for a
Ceasefire.” International Negotiation 12 (2): 207-228.
28 Journal of Conflict Resolution 0(0)

Pechenkina, Anna O, Andrew W Bausch, and Kiron K Skinner (2019) “How do Civilians
Attribute Blame for State Indiscriminate Violence?” Journal of Peace Research 56 (4):
545-558.
Pettersson, Therése, Stina Högbladh, and Magnus Öberg (2019) “Organized Violence, 1989–
2018 and Peace Agreements.” Journal of Peace Research 56 (4): 589-603.
Pettersson, Therése, and Magnus Öberg (2020) “Organized Violence, 1989–2019.” Journal of
Peace Research 57 (4): 597-613.
PILPG (2013) The Ceasefire Drafter’s Handbook. Public International Law & Policy Group. https://
static1.squarespace.com/static/5900b58e1b631bffa367167e/t/5b730a224fa51ab1083c22bb/
1534265892577/PILPG+Ceasefire+Drafter%27s+Handbook+%28Including%2BTemplate%
2BCeasefire%2BAgreement%29-2.pdf
Pinaud, Margaux (2020) “Home-grown peace: civil society roles in ceasefire monitoring.”
International Peacekeeping 28(3): 1-26.
Putnam, Robert D (1988) “Diplomacy and Domestic Politics: The Logic of Two-Level Games.”
International Organization 42 (3): 427-460.
Raknerud, Arvid, and Håvard Hegre (1997) “The Hazard of War: Reassessing the Evidence for
the Democratic Peace.” Journal of Peace Research 34 (4): 385-404.
Reese, Michael J, Keven G Ruby, and Robert A Pape (2017) “Days of Action or Restraint? How
the Islamic Calendar Impacts Violence.” American Political Science Review 111 (3):
439-459.
Regan, Patrick M, and Allan C Stam (2000) “In the Nick of Time: Conflict Management,
Mediation Timing, and the Duration of Interstate Disputes.” International Studies Quarterly
44 (2): 239-260.
Richmond, Oliver (1998) “Devious Objectives and the Disputants’ View of International
Mediation: A Theoretical Framework.” Journal of Peace Research 35 (6): 707-722.
Salehyan, Idean, David Siroky, and Reed M Wood (2014) “External Rebel Sponsorship and
Civilian Abuse: A Principal-Agent Analysis of Wartime Atrocities.” International Orga-
nization 68 (03): 633-661.
Schutte, Sebastian (2016) “Violence and Civilian Loyalties: Evidence from Afghanistan.”
Journal of Conflict Resolution 61: 1595-1625.
Smith (1995) Stopping Wars: Defining The Obstacles To Cease-Fire. Boulder, CO: Westview
Press.
Sosnowski, Marika (2020) “Ceasefires as Violent State-Building: Local Truce and Reconciliation
Agreements in the Syrian Civil War.” Conflict, Security & Development 20 (2): 273-292.
Sticher, Valerie, and Sinisa Vukovic (2021) “Bargaining in Intrastate Conflicts: The Shifting Role
of Ceasefires”. Journal of Peace Research 58 (6): 1284-1299.
Sundberg, Ralph, and Erik Melander (2013) “Introducing the UCDP Georeferenced Event
Dataset.” Journal of Peace Research 50 (4): 523-532.
Thomas, Jakana (2014) “Rewarding Bad Behavior: How Governments Respond to Terrorism in
Civil War.” American Journal of Political Science 58 (4): 804-818.
Toft, Monica Duffy (2010) “Ending Civil Wars: A Case for Rebel Victory?” International
Security 34 (4): 7-36.
Clayton et al. 29

Wagner, R Harrison (2000) “Bargaining and War.” American Journal of Political Science 44 (3):
469-484.
Walter, Barbara F (2009) Reputation and Civil War: Why Separatist Conflicts Are So Violent. 1st
ed. Cambridge: Cambridge University Press.
Wood, Reed M, and Jacob D Kathman (2014) “Too Much of a Bad Thing? Civilian Victimization
and Bargaining in Civil War.” British Journal of Political Science 44 (3): 685-706.
Wood, Reed M, Jacob D Kathman, and Stephen E Gent (2012) “Armed Intervention and Civilian
Victimization in Intrastate Conflicts.” Journal of Peace Research 49 (5): 647-660.
Woods, Kevin (2011) “Ceasefire Capitalism: Military–Private Partnerships, Resource Conces-
sions and Military–State Building in the Burma–China Borderlands.” The Journal of
Peasant Studies 38 (4): 747-770.