Armed Groups and Militarized Elections.

Armed Groups and Militarized Elections
Paul Staniland
Assistant Professor of Political Science
University of Chicago
[email protected]
DRAFT
September 2014
Abstract: Non-state armed groups are often involved in electoral violence, but we know little
about the origins or fates of these groups. This article develops an interactive theory of relations
between governments and electoral armed groups. Governments assign different political roles to
armed groups that reflect their ideological position and electoral value. State strategies flow from
these political roles, but groups’ organizational capacity can allow them to resist government
efforts to control, destroy, or incorporate them. The interaction of regime political interests and
armed group autonomy leads to five distinct government-armed group trajectories, ranging from
incorporation to violent conflict. Comparative evidence from militarized elections in Karachi is
used to illustrate the validity of the concepts and assess the power of the theory, while also
exploring their limits and future research. The argument and findings improve our understanding
of the linkages between democracy, civil conflict, and state power.
Word Count: 11,653
Acknowledgments: Thanks to Mike Albertus, Laia Balcells, Ahsan Butt, Christopher Clary,
Jesse Driscoll, Kristin Fabbe, Laurent Gayer, Morgan Kaplan, Matthew Kocher, Peter Krause,
Austin Long, Aila Matanock, Asfandyar Mir, Vipin Narang, Dann Naseemullah, Irfan
Nooruddin, Dan Slater, Abbey Steele, Emmanuel Teitelbaum, Monica Toft, Steven Wilkinson,
Adam Ziegfeld, and participants in seminars at the Department of Political Science at the
National University of Singapore, R. Allan Bennett Lecture Series at the University of
Connecticut, and Security, Peace, and Conflict workshop at Duke University for helpful
comments on various versions of this paper. Excellent research assistance by Saalika Mela and
Jonathan Weatherwax is gratefully acknowledged.
2
In 1984, the Sikh Golden Temple in the Indian state of Punjab was stormed by the Indian
Army, Prime Minister Indira Gandhi was assassinated by her Sikh bodyguards, and an
insurgency erupted that took tens of thousands of lives over the next decade. At first glance, this
seems like just another anti-state rebellion in South Asia. But it actually emerged from a
conscious attempt to use violence to win elections by key figures in the ruling Indian National
Congress, who had propped up a radical Sikh preacher to splinter a powerful regional party. This
electoral strategy exploded out of the government’s control, at huge human cost.
Punjab is a particularly dramatic case of electoral militarization run amok, but it points to
an important gap in our understanding of electoral violence. Existing research assumes that
governments can control non-state armed actors, allowing them to carefully unleash and rein in
violence to win elections (Wilkinson 2004, Hafner-Burton et al. 2014). This article shows instead
that armed groups can defy, manipulate, or bargain with governments. Rather than assuming topdown control, it advances an interactive theory to explain the varying political relationships
between governments and electoral armed groups.
The diverse trajectories that these armed actors follow – from being co-opted into the
state to turning against the political system – are crucial to explaining patterns of democratization,
civil war, and international intervention. Militarized elections are common, ranging from
contemporary Pakistan to 1950s Burma to 19th century America. They pose huge challenges to
international organizations tasked with facilitating democratization, external interveners trying to
keep and build peace, and citizens and politicians caught in the path of violence. In Libya after
the overthrow of Muammar el-Qaddafi, the persistence of militias means that elected politicians
lead or ally with non-state armed groups, rather than using the state to straightforwardly foster or
crack down on violence. This is true beyond Libya. In Iraq, the Philippines, and Lebanon, for
3
instance, armed factions are deeply intertwined with “mainstream” electoral politics. We need
new concepts, theories, and evidence to make sense of these political dynamics.
This article explains the origins and trajectories of government-electoral armed group
relationships. Armed groups occupy different political roles as a result of their ideological
position and electoral value. These roles, in turn, shape how governments treat them. But groups
are not necessarily pawns of regimes. Instead, they can resist co-optation, repression, or control
when they have organizational power autonomous from government patronage. Organizationally
dependent groups are purged, incorporated into the ruling party or state, or controlled as proxies
of the regime, depending on the group’s political role. Autonomous armed groups, by contrast,
resist incorporation and control to become “normalized” actors mixing ballots and bullets or defy
suppression and emerge as resilient opponents of state power.
The article first introduces two new concepts – militarized elections and electoral armed
groups – and identifies why they matter for understanding democratization, civil conflict, and
international interventions. It then theorizes the relationships that emerge between governments
and electoral armed groups, showing how groups become resilient opponents, incorporated,
purged, normalized, or proxies. These trajectories are illustrated with examples from a wide
variety of cases. Third, comparative evidence from Karachi’s lethal electoral politics is used to
more systematically explore the theory’s explanatory power. This research design leverages
over-time and cross-group comparisons within a shared context, revealing strengths and limits of
the argument. The theory and evidence help us understand when and how coercion and
democratic practice are fused together, creating distinctive forms of armed politics in an era in
which elections have spread but violence persists.
4
Militarized Elections and Electoral Armed Groups
Violent elections occur with some regularity (Straus and Taylor 2012, Hyde and Marinov
2012). This article focuses on a subset of those elections – electoral environments in which at
least two contenders for power use lethal violence and its threat to try to coerce or displace
opponents, manipulate voting, and/or intimidate the press and other observers.1 I refer to these as
militarized elections. Two dynamics can encourage the emergence of militarized elections. First,
regimes that are forced into partial political liberalization, often by international pressures, can
sponsor non-state violence as a substitute for direct state repression (Reno 2011, Levitsky and
Way 2010, Bekoe 2012, Roessler 2005, Collier 2009). Second, militarized elections can emerge
in weakly institutionalized democracies, as politicians restrain coercion against electorally
valuable armed groups that represent a political party or important social constituency
(Wilkinson 2004, Brass 1997, Snyder 2000).2 Opposition parties can in turn acquire weapons,
build armed wings, or cultivate armed actors – such as street gangs and insurgents – to engage in
these “armed politics.”
Electoral armed groups are the non-state armed actors, linked to either governments or
oppositions, involved in militarized elections. They include militias, the armed wings of parties,
criminal networks, militant groups, insurgents, and student activist organizations. Government
leaders, the international community, and state security forces have received the overwhelming
1
I do not study armed groups that are not involved in trying to win elections, the government
forces involved in electoral conflict, elections in which only the state uses violence. These are
important areas, but beyond the scope of this article.
2
For more on electoral violence, see Varshney 2002, Roessler 2005, Dunning 2011, Flores and
Nooruddin 2012, Brancati and Snyder 2011.
5
bulk of attention in electoral violence research, but the trajectories of electoral armed groups
themselves need careful attention. They can influence four important areas of political life.
First, the activities of electoral armed groups can allow regimes to evade international
strategies for post-conflict democratization (Brancati and Snyder 2012, Nooruddin and Flores
2012, Radin 2012, Paris 2004) and election monitoring (Kelley 2012, Hyde 2011, von
Borzyskowski 2014). By claiming that they are not responsible for electoral violence, regimes
may limit their vulnerability to international sanctioning while still retaining power. International
intervention and democratization policies are often thwarted or manipulated by the violence
committed by electoral armed groups. In Iraq, Afghanistan, and Libya, for instance, politicianand party-linked paramilitaries and private armies have posed huge challenges for
democratization and stabilization missions.
Second, militarized elections can contribute to the onset of civil wars, through two
mechanisms. In the first pathway, electoral armed groups may turn against the state or be
targeted by it for destruction, leading to internal conflict. The case of the Indian Punjab
introduced above is a striking example of this pathway to civil conflict, while Reno (2011) finds
in Africa that “the appearance of armed factions associated with past and present governments,
conceived in part as instruments to bolster these governments, came to be the principal threat to
their security” (Reno 2011, 244). In the second pathway, even when under government control,
electoral armed groups can be tools of ethnic cleansing and civilian displacement, operating as
the shock troops of regimes when electoral polarization escalates into full-scale conflict (Snyder
2000). These groups add dangerous volatility to already unstable electoral competitions.
Third, the long-term fates of electoral armed groups can shape the structure and extent of
state power, even in the absence of civil conflict. If governments are able to incorporate or purge
6
these groups, regimes can centralize coercion. This is an important mechanism of state building.
Cooperation with independent armed groups, by contrast, can leave certain geographic areas and
functional domains under the de facto control of these non-state actors. Complex “topographies”
of state power (Boone 2003, Staniland 2012) emerge when state and non-state violence become
intertwined in this manner.
Fourth, the quality of democracy hinges on the use of violence in the electoral arena. If
electoral armed groups operate autonomously, they become power-brokers that control patronage
and political mobilization at the point of a gun. If these groups are controlled by the government,
they may be used as tools of social control that allow regimes to discipline and co-opt local
communities. In both cases, the existence of electoral armed groups can hollow out democratic
representation and institutional responsiveness, creating an “armed clientelism” (Eaton 2006).
Yet electoral militarization can also have positive consequences for democracy. Violent elections
sometimes provide a way of bringing armed factions into the political system by creating
incentives to engage with formal electoral processes, rather than opposing them from the outside
(Reno 2011). Such elections may even give voice to previously excluded social sectors,
facilitating democratization from below (Wood 2000). Electoral armed groups can bargain with
regimes for political concessions, creating an unorthodox and double-edged, but important,
pathway of electoral participation.
Theorizing Armed Group Trajectories
Despite these high stakes, existing research has not explored where electoral armed
groups come from or what trajectories they take over time. This oversight flows from two
assumptions: first, that non-state violence is manipulated by ruling politicians simply trying to
7
win elections and, second, that when violence has served its electoral purpose, politicians can
shut down armed groups (Wilkinson 2004, Hafner-Burton et al. 2014).
This section develops an interactive theory that offers two challenges to this top-down
conventional wisdom. First, the incentives of governments are not purely electoral. Ideology
conditions which groups are most likely be cooperated with or repressed, interacting with
electoral incentives to create a more powerful explanation of government strategy toward armed
groups. Second, state control over armed groups is a variable rather than a constant. In some
cases, including India’s Punjab in the 1980s, governments completely lose control of armed
groups, in others, such as contemporary Libya, ruling politicians have to cooperate with powerful
armed actors, and in yet others, electoral armed groups are pawns of the government.
Five distinct relationships can develop between governments and electoral armed groups:
groups become normalized, incorporated, purged, resilient opponents, or proxies. In two of these
trajectories, the group ceases to exist (incorporation and purging); in the other three, the group
continues to operate, but with varying levels of independence from and cooperation with the
state (normalized, proxy, and resilient opponent groups).
Political Sources of Government Strategy
Government assigned armed groups different political roles. Internal security apparatuses,
militaries, and politicians pay careful attention to the rhetoric and behavior of armed actors,
including both existing groups and new splinter cells, in order to understand if these groups pose
threats or could be possible allies. Political categorization is crucial for rendering the polity
legible (Foucault 1978, Scott 1998) and providing decision rules for policy. As I discuss in
greater detail below, these evaluations may change over time as initial categorization are updated
and new actors enter the scene.
8
Regimes use two important variables to assess an electoral armed group’s political role:
its electoral usefulness and ideological fit with the regime’s political project. Ideological fit and
electoral value can align but also diverge. The ideological worldviews of ruling governments
make sense of political threat. Elected officials and the security personnel they control draw on
their political goals and institutional preferences to perform this crucial interpretive task.
Building on Brass (1974) and Boudreau (2004), I argue that regimes have different perceptions
of threat and beliefs about the acceptable “rules” of political life. Governments assess whether
the current and past rhetoric and behavior of a group is seen to threaten or bolster the core
symbols and political support bases upon which a government rests. For instance, a regime may
be deeply averse to encouraging armed mobilization along religious lines, while perfectly willing
to cooperate with armed groups that mobilize linguistic cleavages (Ganguly and Brown 2003);
some governments are existentially threatened by a militant Left, but others see it as a potential
ally. Ideological position determines which groups a government will look to as allies and
enemies. Though ideology has been overlooked in work on state repression (Davenport 2007), it
is essential for explaining regime perceptions of the value of or threat posed by armed actors.
Government ideological projects are rooted in the goals of ruling parties, their
bureaucracies’ and security forces’ lessons from foundational historical experiences, and beliefs
about which political cleavages and symbolic appeals are politically dangerous. They are
reproduced within parties, militaries, and interior ministries, creating powerful continuities that
are primarily disrupted by regime change or a major change in government. Political beliefs
shape both the background “common sense” (Laitin 1986) and the specific rationales deployed in
deliberations and debates over how to deal with armed actors. Levitsky and Way (2012) show
that even governments like Mugabe’s in Zimbabwe, generally dismissed as non-ideological
9
kleptocracies, still draw on deep ideational legacies from anti-colonial struggles that provide a
shared understanding of political threat and non-material sources of regime cohesion.
An armed group’s political role is further shaped by whether it can deliver concrete
electoral benefits to the government (Wilkinson 2004, Lacina 2013). Politicians are forced to
carefully calculate their possible path to take or hold power, and so they consider whether armed
groups – within ideological constraints – can provide an advantage in voting and coalition
formation. The electoral value of an armed group varies by context, but includes the ability to
directly deliver votes, to act as a coalition partner at the national or regional level, and/or to use
plausibly “deniable” (Carey et al. 2013) violence that demobilizes the government’s electoral
opponents. Governments are most likely to collude with electoral armed groups that are
ideologically compatible and able to deliver electoral benefits (Acemoglu et al. 2013);
nevertheless, desperation may also induce them to cooperate with incompatible groups.
Governments try to shut down groups that are not electorally useful, but the preferred
strategy for achieving this goal depends on ideological fit. 3 Ideological allies are targeted for
incorporation, turning party militias into police, thugs into ruling party strongmen, or street
fighters into cogs in state patronage networks. Ideological opponents who cannot or will not help
a government win elections, by contrast, are targeted for violent elimination. These groups
cannot be easily accommodated because of political incompatibilities, and it is safest to try to
eliminate their capacity to harm the regime.
Armed Group Autonomy
3
Armed groups can deliver other benefits, such as operating as agents of internal control or as
cross-border proxies, that I do not theorize here. On cross-border dynamics, see Byman 2005. On
the use of pro-state militias, see Carey et al. 2013, Staniland 2012b.
10
Government political interests drive preferred state strategy, but governments are not
hegemonic. An armed group’s organizational autonomy – its ability to generate men, manpower,
and internal control independent of state support – conditions the effectiveness of these state
policies and can sometimes force strategic change. Electoral armed groups are not necessarily the
pawns of cunning elites; they can be powerful actors that push back against regime strategies.4
Autonomous groups emerge from preexisting social and organizational networks distinct
from government patronage and sponsorship (Reno 2011, Staniland 2014). Political parties,
student groups, religious networks, and other potential opposition social bases - what Wolf
(1999) refers to as “fields of leverage” - become the basis for armed electoral groups. This
process of “social appropriation” (McAdam, Tarrow, and Tilly 2001, 44) facilitates the creation
of organizational structures impermeable to state control. Autonomous groups may cooperate
with the government, but are not reliant on it. In Iraq, for instance, the Kurdish armed political
parties in the north have organizational roots independent of the central state; they bargain with
Baghdad rather than being controlled by it.
In contrast, groups that emerge with the active sponsorship and support of a government
are dependent on it for resources, manpower, and influence within the political system.5 The
youth groups linked to Bangladesh’s two main political parties are a classic example, being
intertwined with the parties’ patronage networks and control mechanisms (Ali 2010, 254, 281).
4
For a broader discussion of politicians and communal riots in India, see Varshney and Gubler
(2012) and Wilkinson (2013).
5
This echoes the distinction that Shefter (1994) draws between political parties that emerged
reliant on government patronage compared to those that mobilized outside of these patronage
resources.
11
Dependent groups are easier to control or eliminate than autonomous groups. Given the
difficulties of sustained, violent collective action, most electoral armed groups are likely to be
dependent.
Groups can shift from dependence to autonomy by building new networks and acquiring
new resources flows, but this requires luck and skill. Factions of the radical sectarian militia,
Lashkar-e-Jhangvi (LeJ), in Pakistan provide an example. Though initially patronized by
politicians and parts of the security services in the early and mid-1990s, some factions of LeJ
have now linked up with other groups in Pakistan and Afghanistan, creating new alliances that
have allowed them to turn against their previous sponsors (Nasr 2002, 101-108; Zahab and Roy
2004). The Pakistani security apparatus has had to politically categorize these actors as they have
emerged, leading to new relationships and state strategies.
The case of the LeJ also reveals, however, that the autonomous/dependent
conceptualization does not apply when an armed group is deeply divided and different factions
hold varying relationships with the government. In these situations, we need to study the
individual factions rather than the umbrella organization: the LeJ as an overall group is difficult
to categorize because of its fragmentation, but particular factions within it can be evaluated as
more autonomous or dependent.
Table 1 summarizes the argument. Each cell represents a different political relationship
between an armed group and the government. Within each cell the group’s organizational
autonomy/dependence determines which specific outcome is most likely. Regimes may control
dependent groups as armed proxies, incorporate them into the state itself, or efficiently purge
these groups. Government strategy determines the trajectories of dependent groups.
Organizationally autonomous groups, however, can use their power to push back against state
12
policies. Those that are ideologically compatible with or electorally useful to a government
become “normalized” as mainstream political players mixing ballots and bullets; governments
will either try to control them but fail, or recognize their autonomy and treat them as a business
partner. Autonomous groups that are ideologically incompatible and electorally unhelpful will be
targeted for suppression, but they can fight back and become resilient opponents of the
government. At the end of this section I identify important mechanisms driving changes in group
relationship with a government across and within the calls of Table 1.
Table 1. Electoral Armed Group Trajectories
(AG = Armed Group)
Electorally Useful?
Ideologically Yes
Compatible?
No
Yes
AG Autonomous: Normalized
AG Dependent: Proxy (Deep
Cooperation)
[A]
AG Autonomous: Normalized
AG Dependent: Proxy (Thin
Cooperation)
[C]
No
AG Autonomous: Normalized
AG Dependent: Incorporated
[B]
AG Autonomous: Resilient
Opponent
AG Dependent: Purged
[D]
Unitary and Divided Governments
This discussion above assumed a unitary government that is able to engage in coherent
strategies. This assumption does not always hold. In situations where there are multiple groups of
elected politicians who control different security forces, we can disaggregate “the government”
using my framework. For instance, in federal systems local police are often controlled by statelevel elected politicians, while the military and central security forces are controlled by national
politicians. When national and state political leaders have different electoral interests or
ideological goals, we should see distinct patterns of strategy toward electoral armed groups by
each set of security forces. Similar fractures can develop within factionalized state apparatuses.
13
In Northeast India, for instance, local politicians and police have sometimes been accused
of colluding with insurgent groups during electoral campaigns (cells A or C depending on the
case) even as security forces controlled by central politicians try to destroy the same groups (cell
D) (Bhaumik 2009). In southern Thailand during the Thaksin administration, the army and police
were linked to different civilian factions polarized around the political power of the monarchy
and traditional Bangkok elite (McCargo 2006, 9-10). Different parts of the state may act at crosspurposes because of the conflicting interests of political leaders, for the reasons that my theory
highlights. Different parts of the governing apparatus will assign different groups to different
roles in Table 1 and deal with them accordingly.
The interactive theory is much less helpful when security forces are not beholden to
elected officials, even if elections are ongoing. Politically powerful militaries, in particular, may
pursue their own political and organizational interests. In these contexts, my approach should
nevertheless explain the preferences and advocated policies of elected politicians, whether or not
their preferred policy is implemented. This creates a further observable implication of the
argument: if we see politicians disagreeing with security forces along the lines I predict (and data
on these policy positions is credible), it makes us more confident that the underlying political
logic I propose is operating among elected officials. With these caveats in mind, the rest of this
section unpacks each trajectory, and then considers shifts between them over time.
Ideologically Compatible and Electoral Useful Armed Groups [A]
Electorally useful dependent groups become proxies. If their ideological outlook aligns
with that of the government, it will use them as tools of the regime. These are relationships of de
facto control by the government, even if it denies responsibility for the group’s behavior. Proxies
in this context are often linked to the ruling party or leaders’ networks but officially distinct from
14
the state, including youth and student wings, party militias, criminals linked to the ruling regime,
and provocateurs. They are unleashed against opposition politicians, their supporters, and
dissidents around election time, but can be kept in reserve after an individual election has passed.
This trajectory helps regimes evade international monitoring, and is likely to substantially
degrade the quality of electoral representation and state accountability.
Proxies have attracted the most attention in work on electoral violence, generally as the
instigators of riots, intimidation, and local street clashes; they are the stereotypical
lumpenproletariat shock troops of a regime deployed to try to minimize its culpability for
violence (Levitsky and Way 2010, Reno 2011). In India, Brass (1997) and Wilkinson (2004)
argue that local networks of criminals and Hindu nationalist organizations are strategically
deployed and restrained by politicians. The “pocket armies” of Burma’s ruling party in the 1950s
operated as the local enforcers of politicians in the rural countryside (Callahan 2004). When
groups switch from this proxy trajectory, it is usually to become incorporated once they are no
longer electorally useful.
The other trajectory an armed group can follow in this political context (cell A) is
becoming normalized. Normalized groups mix “mainstream” politics with the continued threat
and use of violence. They participate in elections, fund-raising, and/or contentious politics in the
political system but retain coercive capacity. Unlike proxies, normalized groups are free agents
with their own recruiting networks, logistical and financial operations, and command structures
that governments cannot easily co-opt. Because of shared electoral and ideological interests,
cooperation between normalized groups and governments in this context is close and durable, but
more equal than in proxy relationships. The private armies of politicians in the Philippines are an
example (Sidel 1999), as are armed local powerbrokers in Nigeria who cooperate with national
15
politicians (Reno 2011). In Iraq after the American invasion, Shiiite political parties with armed
militias became normalized groups linked to governing coalitions. This trajectory can undermine
the influence of the “median voter” in favor of pacts and deals between the government and
electoral armed groups.
Electorally Useful but Ideologically Incompatible Groups [C]
Most electoral armed groups are likely to be found in cell A because strategically
calculating governments will try to cooperate with electorally useful groups sharing political
goals. But sometimes governments cooperate with electoral armed groups despite underlying
political incompatibility (cell C). This is most likely in a weakly institutionalized electoral
environment with a government desperately bidding to retain power.
Proxy relationships can emerge in these contexts as governments prop up dependent, but
ideologically divergent, armed actors to attack electoral rivals. Armed provocateurs and splinter
groups are used against the government’s electoral opponents in a shadowy world of
provocateurs, collusion with warlords, and other forms of deniable electoral violence. SubSaharan Africa after the Cold War has seen a number of cases (Collier 2009, Bates 2008, Reno
1998) in which governments patronized the enemies of their enemies. Governments and their
proxies in this context are strange bedfellows only able to engage in a fragile cooperation. These
relationships are likely to be linked to human rights abuses, assassination campaigns, and murky
and protracted violence that undermine state accountability and the quality of democracy.
Proxies of convenience will be targeted for repression by the government as soon as they stop
being electorally useful (cell D).
We can also see normalization when, despite differences in preferences, an autonomous
group is willing to cooperate with the government as part of a coalition or in a live-and-let-live
16
collusive deal. Bargaining and mutual deterrence emerge in these contexts. Examples include the
relationships between the non-Shiite parts of the Lebanese state and Hezbollah and between the
Kurdish Democratic Party and the central government in Baghdad. The KDP and Hezbollah are
powerful actors with deeply embedded organizational structures. Normalization in this context is
a transactional, often tense relationship, but the balance of power and interests can maintain
cooperation. Democratic representation may completely break down as direct armed bargaining
between the state and armed groups determines policy outcomes, creating “armed
consociationalism” (Gayer 2014, 207).
Ideologically Compatible but Electorally Unhelpful Groups (B)
Some armed groups are ideologically compatible with the government but not electorally
useful (cell B). Dependent groups in this context are incorporated, usually by being absorbed
into the state and/or ruling party. This is most likely to happen to groups that were previously
proxies after an electoral victory and/or in the face of severe criticism of the government for their
actions. Such groups are compatible with and dependent on the government, and can be
accommodated into the state relatively easily. This may be simply a matter of shifting them from
de facto to de jure arms of the state. Folding armed groups into the “mainstream” is a crucial
mechanism through which regimes centralize control over violence.
In the Philippines, Human Rights Watch (2012) reports that local political clans’ private
armies are often formally rebranded as state security forces, and put on the government payroll,
as a way of consolidating central power and avoiding criticism from outside observers. In the
early-1970s Calcutta, Kohli shows that the Indian National Congress absorbed street fighters,
criminals, and other armed actors into its party structures after previously using them as proxies
17
in the brutal battles of 1967-1972 (Kohli 1990). These elements became part of the Congress as
it ruled the state of West Bengal.
Autonomous groups in this context, however, continue to be normalized. The government
cannot incorporate them against their wishes, and since they do not pose a deep ideological threat
to the regime they can be tolerated as local power brokers or opposition parties.6 This means that
the armed politics found in other political contexts endures in this one as well, characterized by
opportunistic cooperation between state and armed group.
Electorally Unhelpful and Ideologically Incompatible Armed Groups (D)
When electoral armed groups pose a serious political threat to a government, conflict is
likely (cell D). This context is rare, as regimes will usually not back groups with radically
different goals, and armed groups will try to avoid a debilitating reliance on hostile governments.
Yet such relationships can emerge as ruling governments and their interests change and as armed
groups radicalize in response to internal competition or external repression.
Two mechanisms are likely to lead to this political context. The first is when a new
government comes to power with different electoral and ideological interests than its predecessor,
and is faced with electoral armed groups that were sponsored by the previous government.
Armed group political roles shift radically after this change in power, as does state strategy. The
second mechanism occurs when a politically incompatible group has been brought into the
electoral arena by a desperate or miscalculating regime (as in cell C) and then radicalizes. In both
of these circumstances, attempts at bargaining and compromise are challenging because of
6
I assume that autonomous groups generally prefer to maintain independence as a means of
advancing their goals and interests free of government control, but when this assumption does
not hold we will see incorporation in this context.
18
problems of trust and the difficulty of assessing relative power and armed group organizational
autonomy (Fearon 1995).
Dependent armed groups in this context are rapidly purged. A purge often happens after a
government uses a politically incompatible group as a lowest-common-denominator ally (as in
cell C), and then destroys it after it has served its purpose. These proxies of convenience cannot
be easily incorporated into the state structure because of their distinct political origins and goals,
but destroying them is straightforward. Purges also occur when a new government targets thugs
and strongmen who had been dependent on the old regime. When the Burmese military seized
power in 1962, for instance, it quickly eliminated the local armies of the APFPL politicians
(Callahan 2004). As with incorporation, however, successful purges assert regime power and
consolidate coercive capacity within the state.
Resilient opponent are groups that ideologically and electorally oppose the ruling
government. They are targeted for destruction because they pose an important political threat to
the government.7 Unlike purged groups, however, they have the organizational capacity to fight
back and resist suppression for a sustained period of time. Civil war can be avoided if the armed
group and state have some overlapping interests that provide the basis for a compromise, deescalation, and a return to normalization after a period of conflict. When radicalizing armed
groups or unaccommodating states (Goodwin 2001) are involved, however, a “militarized
elections” pathway to civil war onset can emerge. Electoral militarization may escape the
manipulation of even powerful political leaders, leading to a massive, often unexpected, violence
7
There may be situations in which an autonomous armed group deters a crackdown despite
being seen as a political threat, but these relationships are likely to be fragile.
19
exploding out of electoral militarization run amok, as in the Punjab case introduced in the
beginning of this article, and in some periods of Karachi’s politics described below.
Dynamics and Change in Group Trajectories
These state-group trajectories are not locked in stone. Groups can move from being
proxies to being normalized, or transform from normalization into resilient opposition; their
political roles can shift between and within the cells in Table 1. These dynamics involve
contingency, interaction, and agency, and no theory can accommodate all of the pathways of
change. Nevertheless, my framework helps to analyze interaction over time by identifying the
importance of ideological positioning, electoral interests, and organizational autonomy. It
provides a new analytical map to structures the likely pathways of change.
There are important constraints on change. If a group is successfully purged or
incorporated, it ends as an actor and cannot move between political roles. Conflict between a
government and a resilient opponent can forestall a return to normalization if trust breaks down
and a total war of annihilation develops. Normalized groups are hard to purge or incorporate, and
their endurance limits what governments can do to them. Groups with a long history of adhering
to a particular ideological line may not be able to credibly change their political position, at least
in the short run, and may remain deeply suspect in the eyes of governments. This means that
changes are not totally fluid; certain options can be off the table for historical reasons.
Within these constraints, several important mechanisms drive change. The most dramatic
shifts are between the cells in Table 1, creating new political relationships between regimes and
armed groups. New governments can come to power with different ideological projects and/or
electoral interests. The replacement of a leftist government by a right-wing coalition or the
collapse of an electoral authoritarian regime by a democratizing mass movement, for instance,
20
can cause a dramatic reassessment of enemies and allies. Major coalitional changes within
governments may have similar effects. Threat perceptions and electoral needs often change with
regime politics. Previous proxies can be targeted for purges and resilient opponents may become
normalized as political roles change.
The most dramatic shifts in trajectories should therefore occur after a major change in
government and/or regime, as new rulers assesses the armed landscape and armed groups probe
the rules of the game. A process of violent learning often ensues in these periods (Tajima 2014),
with groups and governments trying to measure their relative power and establish ideological
positions. This can trigger a temporary period of conflict followed by normalization if both sides
end up finding a mutually-agreeable equilibrium, or it can lead to escalating warfare.
Electoral armed groups themselves can radicalize and moderate their ideological
positions, both in and outside of these transitional periods. A powerful cause of uncompromising
or even radicalizing ideological positions is the existence of threat of intra-movement
competition and splintering (Pearlman 2011, Krause 2013/14). Adopting a rigid stance may be
necessary to avoid charges of selling out or being co-opted, heightening ideological tensions and
limiting opportunities for electoral cooperation. Some ideological positions may be unacceptable
to the group’s supporters. This mechanism can lead to sustained conflict and keep groups in
resilient opposition with a central government.
Cutting in the opposite direction, however, are causes of ideological moderation. Two in
particular stand out. First, electoral armed group groups may decide to cooperate with the
government after realizing that their goals are unattainable and government repression too much
to bear. War-weariness can force a group to operate within the government’s rules of the game as
long as that position does not fundamentally undermine the organization’s raison d’etre. This
21
change is limited by internal organizational dynamics and the political preferences of a group’s
social base (Kaplan 2014), so ideological positions may be quite “sticky” over time. Second,
groups can moderate their ideological position to reap the benefits of patronage, political power,
and protection that becoming a proxy or normalized actor confers (Reno 2011). Ideological
malleability in exchange for material benefits is especially common for groups born as
dependent proxies without a preexisting political agenda.
As discussed earlier, changes in organizational autonomy may occur as a result of regime
strategies and armed group innovation (Weinstein 2007, Staniland 2014). Shifts in this balance
of organizational power can affect state-armed group trajectories within the cells in Table 1. If
groups grow more or less dependent on regimes, their options and vulnerabilities change. These
various processes of interaction are endogenous and iterative, but I offer a structured set of
theoretical processes that can be empirically assessed and compared.
Elections and Violence in Karachi
Research Design
This section examines Karachi’s violent electoral politics to probe in greater detail where
the theory succeeds and fails. The research design leverages comparisons across groups and time
in Karachi to fill out the typology in Table 1, identifying both the origins of armed group
political roles and groups’ reactions to consequent government policies. These comparative cases
constitute a plausibility probe (Eckstein 1975) that illustrates and establishes greater confidence
in the explanatory power of the typological theory (Elman 2005), while also highlighting areas
for future research.
Karachi has advantages and limitations as a context for assessing my theory. Multiple
armed groups have been involved in Karachi’s national and provincial elections over the last
22
three decades. These groups have had different levels of organizational autonomy and dealt with
governments possessing different electoral incentives and ideological goals. This structure of
variation allows a careful examination of government-armed group trajectories that holds
constant many broader structural features of the polity. There is some available qualitative data
that allows us to process-trace specific interactions and decision-making processes,
complementing the comparisons of outcomes (George and Bennett 2005; Brady and Collier
2010).
There are also important limitations to this study. As shown below, the enduring political
power of Pakistan’s military means that internal security is not under the straightforward control
of elected officials. The complexity of Pakistan’s internal security arrangements can undermine
data quality and mask responsibility for policy outcomes decision-making processes: the Army is
a powerful actor that can influence the behavior of state-level police forces, as can politicians
and local criminal networks. Yet this also means that if there is substantial support for my
argument in Pakistan, similar dynamics are likely to operate in more “normal” electoral contexts.
The power of the Army explains why the theory fails in two important instances, but also shows
that that the evidence is not being cherry-picked to support the theory. As I discuss below,
Pakistan suggests the need for more systematic research on when and how militaries thwart
civilian politicians in electoral environments.
Empirical Overview
Karachi’s politics since the 1980s have been highly militarized. Political parties have
created armed wings, criminal gangs are allied with political actors, and the military has both
colluded with and tried to suppress key electoral armed groups in the sprawling metropolis.
Elections, policy-making, and coercion are entwined: “Violence in the city is frequently used as
23
a negotiating tactic to sway provincial- or federal-level decision-making” (Yusuf 2012, 5). I
compare the Muttahida Qaumi Party (MQM), its splinter the MQM-Haqiqi (MQM-H), and local
student and criminal groups that have become linked to the Pakistan People’s Party (PPP). These
are the main electoral non-state armed actors in Karachi; while other militant groups have a
presence in the city, including the Pakistani Taliban and Lashkar-e-Jhangvi, they are not as
explicitly involved in electoral competition.
Table 2 summarizes the theoretical expectations and empirical outcomes in this set of
comparisons. The MQM’s organizational autonomy has consistently been high, but its electoral
value and ideological compatibility with elected governments have varied over time. The MQM
has been normalized when it is electorally valuable to national political leaders and when it has
restrained itself from indiscriminate communal killing and/or hints at separatism. It was a
resilient opponent against the Benazir Bhutto PPP-led government from 1993-1996 because it
was not electorally useful to Bhutto and was locked in a brutal ethnic war with the PPP’s Sindhi
base. However, the military’s crackdown on the group in 1992 shows that my theory loses
explanatory power when coercion is not directed by politicians.
Karachi also provides examples of proxies, purging, and incorporation. During the early
1990s the MQM-Haqiqi broke away from the MQM. It became a proxy for the military that
fought the MQM and stood for elections, but in 2003 the MQM-H was purged by the army as
part of a deal between Pervez Musharraf and the MQM. My theory correctly predicts its role as a
proxy from 1992-2003, but the 2003 purge is a failure for the argument, which predicts
incorporation. The military’s independent power helps to explain this failure. Incorporation
occurred when the PPP has brought thugs, students, and gangs into its party organization. The
PPP’s relationship with another armed group, the People’s Aman Committee (PAC), provides a
24
mix of normalization and resilient opposition, as a PPP-led provincial government ended up first
cooperating and then attacking the PAC.
Table 2. Electoral Armed Groups in Karachi
(Bolded outcomes are those not correctly predicted by theory)
Groups
Ideological
Fit
Electoral
Value
Org.
Predicted
Autonomy Trajectories
Actual
Trajectories
MQM
Somewhat
incompatible
1992-97, but
largely
compatible
otherwise,
especially after
1997
Compatible
High 1988-89,
1990-93, 1997-8,
1999-present;
Low 1990, 19931996
High
Normalized (1988-93),
Resilient Opponent
(1990, 1993-96),
Normalized (1997present)
Normalized (1988-89,
1990-1992)
Resilient Opponent
(1990, 1992-1996)
Normalized (1997present)
Moderate, 19922003; Low 2003present
High since 1980s
Low
Proxy (1992-2003),
Incorporated (2003)
Proxy (1992-2003)
Purged (2003)
Variable,
though
mostly Low
Weak local groups
incorporated into
party; autonomous
groups become
normalized or resilient
opponents
Incorporation
dominant pattern;
PAC normalized and
then resilient
opponent
MQM-H
Local armed
gangs/student
groups
Compatible
The MQM as Normalized Electoral Armed Group, 1988-1992
In the 1980s, the mobilization of ethnic Mohajirs (Urdu-speaking migrants from prePartition India) transformed the MQM from a student organization into a major political party.
This process was accompanied by substantial violence, while the MQM’s dominance of the
Mohajir vote made it electorally valuable for governments trying to take and hold power in
Islamabad and the Sindh Provincial Assembly. Crucially, the MQM emerged in the 1980s with a
tight organizational structure rooted in student mobilization, dominated by leader Altaf Hussain,
that was not dependent on state patronage or resources. Cohen describes the MQM as “a tightly
organized political party structured along Leninist lines” (Cohen 2004, 216), Lieven argues that
it was “built on its highly disciplined and ruthless student organization” (Lieven 2011, 253), Haq
identifies its “strict party discipline” (Haq 1995, 997) and Inskeep observes that “MQM
25
politicians were fiercely disciplined” (Inskeep 2011, 172). This has been, and remains, an
organizationally autonomous armed group that cannot be easily controlled or destroyed by
Pakistan’s security forces.
In the wake of 1988’s democratization, the MQM became a valuable electoral partner for
both Nawaz Sharif’s Pakistan Muslim League (PML-N) and Benazir Bhutto’s Pakistan People’s
Party (PPP). In 1988, the MQM linked up with the PPP to support Bhutto’s first government,
which felt that it needed the MQM’s support in the Sindh Provincial Assembly to avoid a
polarizing urban-rural divide and to bolster the PPP’s factionalized party structure in the
province. In return, the MQM was allowed to operate with impunity in Karachi, despite deep
underlying disagreements about the distribution of political power between the MQM and the
PPP’s Sindhi ethnic base. The MQM was normalized as a mainstream actor (cell C in Table 1),
agreeing to the Karachi Accord of 1988 with the PPP. However, this “very brittle relationship”
(Haq 1995, 999) fell apart when the MQM clashed with Sindhi activists and PPP cadres. The
MQM, as a free agent rather than a proxy, abandoned Benazir in late 1989. PPP-controlled
security forces then tried to crack down on the MQM in the summer of 1990. It fought back and
became a resilient opponent (Shah 2014, 169), but Benazir was soon removed from power.
With Benazir’s government brought down, the MQM became part of the ruling coalition
at both the Sindh and national levels between 1990 and 1992. The new Prime Minister, Nawaz
Sharif, held back state coercion during the 1990-92 period, even as violence in Karachi
continued. The MQM was electorally useful and broadly ideologically acceptable to the PML-N
(cell A), since the MQM’s vicious battles with Sindhis and Pashtuns were far less threatening to
the Punjabi-based party than to the PPP. Sharif was particularly beholden to the MQM because it
26
was essential to a tenuous coalition in the Sindh Provincial Assembly intended to keep the PPP
out of power in the province (Abbas 1990b, 48).
Neither Sharif nor Bhutto could incorporate the MQM into the state or their national
parties even when cooperating with it. In this political context, the MQM’s “parallel rule was
tolerated only as long as the MQM kept to the ‘rules’ and had an influential protector in [Sindh]
Chief Minister Jam Sadiq” (Frotscher 2008, 217). The MQM was engaged in “armed clientelism”
that made it a crucial political player mixing ballots with bullets (Gayer 2012, 78). Frotscher
shows that during the MQM coalition periods of 1988-89 (with the PPP) and 1990-92 (with the
PML-N) “the state had lost its monopoly of power to the MQM in these districts” and “murders
or other crimes carried out in the name of the party were not investigated by the legal authorities”
(Frotscher 2008, 175). Electoral value and organizational autonomy gave the MQM the clout to
become a normalized armed party, as my theory predicts.
The MQM as a Resilient Opponent, 1992-1996
The period from 1992-1996 saw a crackdown, first by the Army acting on its own
initiative in 1992 and then by the second Benazir Bhutto government from 1993-1996. As noted
above, the Pakistani state has multiple nodes of power, including above all the military. The
army crackdown of 1992 is not explained by my argument, but the 1993-1996 period of Bhuttoled suppression is. In her second term, the MQM was not electorally valuable to Bhutto and both
its rhetorical and physical attacks on the PPP’s ethnic support base made it a deep political threat.
These changes in first military and then civilian political interests drove the MQM from
normalization to resilient opposition.
In 1992, the army went into Karachi as a part of Operation Clean-up because it felt that
the MQM was threatening military interests, in particular due to the MQM’s kidnapping of an
27
Army major in 1991 and its penetration of the police in the province (Nawaz 2008, Staniland
2010). The unelected army had no electoral use for the MQM and its activities were seen as an
affront to the military’s self-perception as the guardian of the state. Though the operation was
ostensibly a sweep across Sindh, the Army primarily focused on the MQM, engaging in major
operations in Karachi from mid-June 1992. My theory does not explain this 1992 crackdown:
Sharif still had electoral uses for the MQM, but the army was able to operate as it wished.
However, it is clear that the Sharif government was not in favor of the crackdown. Indeed,
by August 1992, Herald reported that “the civilian and the military leadership was now
completely at odds regarding the scope and execution of the army operation” which had been a
“unilateral” decision by the Army (Bakhtiar 1992, 26). Shah’s (2014, 172) recent assessment
agrees: the Army aggressively targeted the MQM “despite the government’s objections.” Gayer
notes that the Army cracked down “apparently without Nawaz Sharif’s knowledge, let alone
approval” (2014, 31), and Ahmed argues that “[Chief of Army Staff] Janjua and the corps
commander of Karachi, Lieutenant General Naseer Akhtar, wanted to cut the MQM down to size,
but Nawaz was not willing to do that as it was part of his ruling coalition at the Centre “(Ahmed
2013, 286).
Though the MQM becoming a resilient opponent versus the military in 1992 is not
predicted by my argument, it does explain the political preferences of the elected politicians, and
suggests that this shift would not have occurred if Sharif had controlled the army. The MQM was
able to absorb this repression without being purged; Jafri reports that military officers viewed the
MQM as “a highly organized unit with a centralized command structure, damaged as it may be”
(Jafri 1992, 30); Gayer (2014, 75) writes that Clean-up “brought the MQM to heel but did not
break it.”
28
In 1991-2, the splinter group MQM-Haqiqi (the “Real” MQM, MQM-H) broke from
Altaf Hussain’s MQM, as I discuss below, with military sponsorship and became a proxy used in
both violence and elections against the MQM (Haq 1995). Sharif was removed as Prime Minister
in 1993, and Benazir Bhutto returned to national power in Islamabad in 1993 without relying on
the MQM. There was no electoral incentive for her to do business with the MQM (Gayer 2014,
211), and the brutal ethnic politics of Sindh made the MQM an ideological threat to the
government’s preferred vision of Pakistani politics (cell D), as did rumors (some planted by the
military) that the MQM might push for a separate province – or even an independent country –
for urban Sindh.
Negotiations between the PPP and the MQM over a coalition in the Sindh provincial
assembly collapsed in early 1994, as “the utility of the MQM had gradually diminished to
nothing, at least from the point of view of votes, and it was then that the PPP leadership stopped
making overtures and became increasingly unwilling to talk” (Bakhtiar 1994, 31). Bhutto backed
the Army crackdown, which finished at the end of 1994, while extending additional repression
through Interior Ministry and provincial security forces through 1996. These forces also answer
to the Army, but in this case were heavily directed by Naseerullah Babar (a retired major-general
and Interior Minister under Bhutto) (Gayer 2014, 215). This 1993-1996 period of “open conflict”
(Gayer 2014, 100) under Bhutto is far better explained by my theory than the 1992 Army
operation.
The MQM Returns to Normalization, 1997-Present
Thousands died in this conflict, especially in 1995, but the MQM and the state did not
end up in a total war. The Mohajir constituency’s distributive demands could be met within the
confines of the political system, as long as secessionism was avoided and ethnic violence
29
calibrated. The MQM entered a “gray zone” (Auyero 2007) of ideological compatibility, with
elements of both overlap and resistance to the military and civilian leadership’s goals and
worldviews. This provided a basis for compromise that prevented a spiral into permanent conflict.
As a result, “periods of calm and of intense violence chequeuered the 1990s” (Khan 2010, 45).
The MQM returned to normalization when the second Sharif government came to power
in 1997 (cell A). The MQM explicitly toned down its rhetoric and tried to portray itself as an allPakistan national party, while the PML was not invested in Sindhi ethnic assertion. It became a
coalition partner of Nawaz Sharif’s second government in 1997 (Frotscher 2008, 255; Nawaz
2008, 454). Thereafter the “MQM systematized a form of violent government” (Gayer 2012, 83)
in Karachi. Since 1997 – other than a falling-out with Sharif’s government in 1998 after the
alleged MQM assassination of a prominent personality in Karachi – the MQM has been a
consistent ally of ruling governments at the national and provincial levels. Yusuf argues that the
level of national repression and cooperation by civilian leaders toward the MQM is driven by the
group’s electoral and coalitional importance (Yusuf 2012, 7).
Normalization endured under Pervez Musharraf’s electoral authoritarian regime (the first
election, though rigged, under his rule came in 2002) and the 2008-2013 PPP-led democratic
government. In exchange for stability and support, the “party was given a free hand to rule as it
saw fit in urban Sindh” (Frotscher 2008, 264) under Musharraf (Gayer 2012, 78-9). Despite its
bitter history with the MQM, once Asif Ali Zardari became president, the PPP made deals with
the party (Gazdar 2011, 17) because it continued to be a badly needed partner for stabilizing
Sindh. The PPP had learned the cost of attacking the MQM, and the MQM had a clearer sense of
the acceptable rules of the game. The same rough equilibrium has held since Sharif returned for
his third round in power after 2013, even as violence continues to stalk Karachi’s streets. Indeed,
30
“one of the fundamental features of the party is that it uses both violence and parliamentary
politics to pursue and assert its interests” (Frotscher 2008, 130).
The experience of 1992-1996 seems to have induced mutual learning: the MQM
identified the limits of its influence, and civilian politicians and the military we made very aware
of how resilient and disruptive the MQM could be. Karachi’s armed politics stabilized in the
2000s compared to the rapid shifts between cells in Table 1 that characterized the 1990s.
Governments and armed groups can adjust their perceptions and behaviors over time. This was a
dynamic process in which both ideological adjustment and electoral incentives played a central
role. The MQM has been able to operate within the Pakistani political system to remarkable
effect, contributing to the destabilization of several civilian governments (Gayer 2014, 31) and
amassing enormous political power from this leverage.
This trajectory is in sharp contrast to explicitly ethno-separatist armed groups, like those
in Baluchistan and (early 1970s) East Pakistan, that have been perceived by political elites as
fundamentally inimical to the idea of Pakistan (Cohen 2004, Fair 2014), or other groups in
Karachi discussed below that lack organizational autonomy. Governments can do business with
the MQM as long as it does not become explicitly separatist or uncontrollably communal, the
MQM can protect itself from incorporation, control, or destruction because of its organizational
structure, and it has therefore become an integral part of Karachi’s politics of “ordered disorder”
(Gayer 2012).
MQM-Haqiqi as Proxy and Purge Victim
The MQM-H split from the MQM in 1992 around the time of the Army crackdown. The
military provided it with guns and protection as it pursued counterinsurgency operations (Gayer
2012, 77-78; Haq 1995, 1001; Lieven 2011, 316). The MQM-H also stood for elections, though
31
without much success, making it an electoral armed group. I code it as moderately electorally
useful, since it put up candidates in an effort to displace the MQM as the electoral representative
of the Mohajirs. But it quickly lost discipline and it never won much electoral support, becoming
an unambiguous proxy of the military (Bakhtiar and Jafri 1994, 32). It was dependent on the
Army for survival, rather than having a well-established social or organizational base: Haq
argues that the Army’s manipulation “profoundly delegitimized the Haqiqi among the majority
of MQM supporters, who regarded the dissidents as traitors” (Haq 1995, 1001).
In 2003, the MQM and Pervez Musharraf cut a deal that solidified their alliance and the
MQM’s role as electoral power and armed enforcer in Karachi. The MQM-H posed no
ideological challenge to the government, but it was no longer valuable for either winning seats or
targeting the MQM, and had been in decline for years. The military removed the MQM-H from
its areas of territorial control and targeted potential trouble-makers within the group (Gayer 2012,
78-9). Its leader, Afaq Ahmed, was jailed from 2004 to 2011. The Haqiqis had little
organizational capacity with which to fight back when the state decided to dispense with them.
The proxy period from 1992-2003 is explained by my theory, but this 2003 purge is not. The
theory predicts that should the MQM-H should have been incorporated once it was no longer
useful. This purge appears to have been used as a credible commitment to the MQM by the army,
which had the power to pursue its own interests in Karachi.
Street Fighters and Party Incorporation: PPP Competitive Arming
The MQM is not the only party that deploys violence in Karachi. In addition to the
growing insurgent threat of the Pakistani Taliban, the PPP has become increasingly armed in
recent years. Journalistic and analytical accounts point to street fighters, gunmen, and violent
student activists as key players in the PPP’s battles to defend itself and attack its enemies. As a
32
pair of journalists reported in 2011, “There is consensus among them [interviewees] that the
combatants of this gang war were the proxies of leadership figures of the ruling Pakistan People's
Party (PPP) and its Awami National Party (ANP) coalition partner, and their on-again, off-again
ally, the Muttahida Qaumi Movement (MQM)” (Hadayat and Hussain 2011). The Human Rights
Commission of Pakistan noted in 2011 that “[A]lmost everyone that the HRCP mission talked to
spoke of the involvement of the main political parties in Karachi and their armed/militant wings
in violence and other crimes” (HRCP 2011, 11). In 2013, over 2,700 people were killed in
Karachi, many of them in battles linked to elections and political parties (Khan and Ashraf 2014).
We have some evidence on PPP links to to electoral armed groups. The party is usually
part of the ruling coalition at the Sindh provincial level, and often at the national level, meaning
that its relationship with armed actors tends to approximate that between a government and
armed group. The general pattern is one of incorporating student networks and local gangs
among the Sindhi and, to a lesser extent, Baluch and Pashtun, ethnic populations.
These small and fragmented local groups were incorporated into the PPP in two contexts:
inter-ethnic boundary zones and university campus conflicts. They then become linked to the
party and patronage networks of the PPP, which can provide political cover, resources, and an
institutional home for protection and influence. Students’ associations on Karachi campuses –
above all the PPP’s Peoples Students Federation (PSF) – have been centrally important as
sources of violent activists to be channeled into the PPP rather than going rogue or establishing
parallel institutional structures (Gayer 2014, 65, 71-72).8
8
For a detailed history by an eminent left-wing Pakistani journalist, see Paracha N.d. See also
Gayer (2014).
33
These are not autonomous organizations; instead, they are conveyor belts for violent
activists into the PPP and its often-ruling party apparatus. As Zahid (1989, 59) noted in 1989,
this represents a broader pattern: “many of these student leaders have been supported and
nurtured by mainstream political parties and groups and no student group has been without such
backing or influence” (59). The PSF was particularly involved in clashes with the MQM in 1989
and 1990 (Abbas 1990a, 66). Criminal networks in ethnically tense areas have similarly provided
specialists in violence. It seems that most are groups of neighborhood thugs who are easily
controlled as proxies or sucked into the party structure (cells A and B).
However, not all such local groups are dependent. There have been conflicts in some
areas of Karachi, notably Lyari, as the PPP and the government forces it controls have ended up
clashing with its former ally, the People’s Aman Committee (PAC) (Yusuf 2012, 12-13, 30). The
PAC has its own Baloch organized crime networks on the ground that have allowed it to push
back against PPP control and later repression. The PPP’s Sindh Home Minister, Zulfiqar Mirza,
is alleged to have allied with the PAC in 2010-11 as protection against the MQM as part of a
broader strategy of “using the PAC as its [the PPP’s] own military wing” (Gayer 2014, 221).
However, his resignation, growing violence, and pressure from the MQM then led to armed
clashes between government forces and the PAC (Khan 2012). After initially being normalized,
the PAC became a resilient opponent in the face of crackdowns in Lyari in 2012-13 (Gayer 2014,
156-161). This case shows variation in the ability of the PPP and the security apparatus it
controls to manage local allies: localized street thugs and dependent student groups could be
absorbed or controlled, but the preexisting networks of the PAC gave it the power to resist
becoming a simple PPP proxy.
Limits of the Argument and Future Research
34
Electoral interests and ideological compatibility have frequently determined the political
roles of armed groups in Karachi and the strategies that elected officials have pursued toward
them, while these groups’ organizational power helps explain how they reacted. The evidence
makes it clear that we can operationalize the theory and assess its match with empirical reality.
Table 2 shows that most of the theoretical predictions are borne out in the historical record.
However, the argument assumes that state strategy in electoral environments is
influenced by electoral considerations. This approach fails to explain the timing of the first
crackdown on the MQM in 1992 and the ultimate fate of the MQM-H. Both actions were taken
by the Pakistan Army, which is far less electorally accountable than civilian governments, even
when it is holding (dubious) elections. The military has pursued a number of strategies toward
armed groups on its own accord (Fair 2014, Kapur and Ganguly 2012). Patterns of coercive
restraint and repression by the Sharif, Bhutto, and Zardari elected governments, by contrast, are
much closer to the theory’s predictions. The argument also explains the resistance of the Sharif
government to the 1992 crackdown in Karachi, even though the Pakistan Army’s power
prevented these preferences from being turned into actual policy.
Future research should grapple with when and how elected officials are thwarted by
security apparatuses in crafting policy. We particularly need to understand which political
domains militaries determine to be “off limits” to politicians, helping to explain when my
argument does and does not apply. In Pakistan, for instance, politicians are allowed to be
involved in Karachi’s electoral violence, but their room to maneuver is far more limited when it
comes to security policy on the northwest frontier or toward anti-India jihadist groups.
35
Armed Politics and Democracy
Militarized elections are part of democratic practice in much of the world. This article has
explored the relationships between governments and armed groups that emerge in these contexts.
The ideology and electoral needs of governments determine how they try to manage electoral
armed groups, and groups’ organizational power explains how these strategies play out over time.
This theory politicizes state coercion and restraint, revealing the fascinating variety of political
roles that armed groups can occupy, while integrating these groups’ ability to thwart or
manipulate government strategies.
Militarized elections are important beyond Pakistan. Violence and elections have been
intertwined in Libya, Iraq, Nigeria, and Afghanistan in recent years, undermining attempts to
consolidate meaningful democratization, and often complicating international intervention
policies. Policymakers and analysts are likely to be unpleasantly surprised if they do not take
electoral armed groups seriously: they are not simple pawns of regimes, but can instead dominate
patronage networks, challenge state power, or carve out enduring local fiefdoms. Understanding
the ideological and electoral roots of government policy can in turn make better sense of
strategies that seem counterproductive or irrational to outside observers.
These claims have broader theoretical implications. Conventional studies of political
violence distinguish between civil wars, electoral violence, and state building. This article
challenges these categories: militarized elections can lead to everything from state centralization
to civil war to violent bargaining. Researchers should explore interactions between states and
armed groups across a wide variety of related substantive areas, from counterinsurgency to
regime politics. We need new ways of studying the armed politics that define a contemporary era
in which elections have spread but political violence endures.
36
References
Abbas, Zaffar. 1990a. “Back From the Brink?” Herald, May: 65-70.
---. 1990b. “The Sindh Card,” Herald, September: 48-54.
Acemoglu, Daron, James A. Robinson, and Rafael Santos. 2013. The Monopoly of Violence:
Evidence from Colombia. Journal of the European Economic Association 11 (s1): 5–44.
Ahmed, Ishtiaq. 2013. The Pakistan Garrison State: Origins, Evolution, Consequences. Karachi:
Oxford University Press.
Ali, S. Mahmud. 2010. Understanding Bangladesh. Columbia University Press.
Bakhtiar, Idrees. 1992. “Is the Army here to stay?” Herald, August: 22-26.
---. 1994. “What does the MQM really want?” Herald, May: 27-33
Bakhtiar, Idrees, and Hasan Iqbal Jafri. 1994. “Eye for an Eye,” Herald, July: 27-35.
Bates, Robert H. 2008. When Things Fell Apart: State Failure in Late-Century Africa. New
York: Cambridge University Press.
Bekoe, Dorina Akosua Oduraa. 2012. Voting in fear: electoral violence in Sub-Saharan Africa.
Washington, D.C.: United States Institute of Peace.
Boudreau, Vincent. Resisting Dictatorship: Repression and Protest in Southeast Asia.
Cambridge ; New York: Cambridge University Press, 2004.
Brady, Henry E, and David Collier, eds. 2010. Rethinking Social Inquiry: Diverse Tools, Shared
Standards, 2nd ed. Lanham, Md: Rowman & Littlefield.
Brancati, Dawn, and Jack L. Snyder. 2011. Rushing to the Polls: The Causes of Premature
Postconflict Elections. Journal of Conflict Resolution 55 (3): 469–492.
Brass, Paul R. 1974. Language, Religion and Politics in North India. London: Cambridge
University Press.
37
Brass, Paul R. 1997. Theft of an Idol: Text and Context in the Representation of Collective
Violence. Princeton, NJ: Princeton University Press.
Brown, Michael and Sumit Ganguly, eds. 2003. Fighting Words: Language Policy and Ethnic
Relations in Asia. Cambridge, Mass.: MIT Press.
Byman, Daniel. 2005. Deadly Connections: States That Sponsor Terrorism. Cambridge, UK:
Cambridge University Press.
Cohen, Stephen P. 2004. The Idea of Pakistan. Washington, D.C: Brookings Institution Press.
Eckstein, Harry. 1975. “Case studies and theory in political science.” In Greenstein, F., and N.
Polsby, eds. Handbook of political science, vol. 7, Reading, MA: Addison-Wesley, 79–138.
Elman, Colin. 2005. “Explanatory Typologies in Qualitative Studies of International Politics.”
International Organization 59 (02): 293–326.
Fair, C. Christine. 2014. Fighting to the End: The Pakistan Army’s Way of War. New York:
Oxford University Press.
Fearon, James D. 1995. Rationalist Explanations for War. International Organization 49 (3):
379–414.
Foucault, Michel. 1995. Discipline & Punish: The Birth of the Prison. New York: Vintage
Books.
Frotscher, Ann. 2008. Claiming Pakistan: the MQM and the fight for belonging. 1. ed. BadenBaden: Nomos.
Gayer, Laurent. 2014. Karachi: Ordered Disorder and the Struggle for the City. Oxford: Oxford
University Press.
---. 2012. “Political Turmoil in Karachi.” Economic and Political Weekly 47(31): 76–84.
Gazdar, Haris. 2011. “Karachi’s Violence: Duality and Negotiation.” SPO Discussion Paper
38
Series, No 10.
George, Alexander L., and Andrew Bennett. Case Studies and Theory Development in the Social
Sciences. Cambridge, MA: MIT Press, 2005.
Hadayat, Amjad and Tom Hussain. 2011. Political violence shifts to Karachi. The National,
November 11 < http://www.thenational.ae/news/world/south-asia/political-violence-shiftsto-karachi#full>.
Hafner-Burton, Emilie M., Susan D. Hyde, and Ryan S. Jablonski. 2014. When Do Governments
Resort to Election Violence? British Journal of Political Science 44 (01): 149–179.
Haq, Farhat. 1995. “Rise of the MQM in Pakistan: Politics of Ethnic Mobilization.” Asian Survey
35(11): 990–1004.
Hyde, Susan D., and Nikolay Marinov. 2012. Which Elections Can Be Lost? Political Analysis
20 (2): 191–210.
Human Rights Commission of Pakistan (HRCP). 2011. "Karachi: Unholy Alliances for Mayhem.”
Lahore: Human Rights Commission of Pakistan.
Human Rights Watch. 2012. “Philippines: Keep Promise to Disband Paramilitaries.”
http://www.hrw.org/news/2012/03/30/philippines-keep-promise-disband-paramilitaries.
Hyde, Susan D. 2011. The Pseudo-Democrat’s Dilemma: Why Election Monitoring Became an
International Norm. Cornell University Press.
Hyde, Susan D., and Nikolay Marinov. 2012. Which Elections Can Be Lost? Political Analysis
20 (2): 191–210.
Jafri, Hasan. 1990. “MQM: A Fight to the Finish?” Herald, July: 25-30.
Kaplan, Morgan. 2014. “How Civilian Perceptions Affect Patterns of Violence and Competition
in Multi-Party Insurgencies.” Working paper, University of Chicago.
39
Kapur, S. Paul, and Sumit Ganguly. 2012. The Jihad Paradox: Pakistan and Islamist Militancy in
South Asia. International Security 37 (1): 111–141.
Khan, Faraz. “Violence 101: A primer to the Lyari gang war.” Express Tribune, May 2.
http://tribune.com.pk/story/372834/violence-101-a-primer-to-the-lyari-gang-war-karachi/
Khan, Faraz and Gibran Ashraf. 2014. “Karachi 2013: the Deadliest Year of all.” Express
Tribune, January 6.
Khan, Nichola. 2010. Mohajir militancy in Pakistan: violence and transformation in the Karachi
conflict. London ; New York: Routledge.
Kohli, Atul. 1990. Democracy and discontent: India’s growing crisis of governability.
Cambridge University Press.
Krause, Peter. “The Structure of Success: How the Internal Distribution of Power Drives Armed
Group Behavior and National Movement Effectiveness.” International Security 38, no. 3
(2013/2014): 72–116.
Lacina, Bethany. 2013. How Governments Shape the Risk of Civil Violence: India’s Federal
Reorganization, 1950–56. American Journal of Political Science. doi: 10.1111/ajps.12074.
Laitin, David. 1986. Hegemony and culture  : politics and religious change among the Yoruba.
Chicago: University of Chicago Press.
Levitsky, Steven, and Lucan A. Way. 2010. Competitive Authoritarianism: Hybrid Regimes
After the Cold War. Cambridge University Press.
---. 2012. “Beyond Patronage: Violent Struggle, Ruling Party Cohesion, and Authoritarian
Durability.” Perspectives on Politics 10 (4): 869–89.
Lieven, Anatol. 2011. Pakistan: A Hard Country. 1st ed. New York: PublicAffairs.
40
McAdam, Doug, Sidney Tarrow, and Charles Tilly. 2001. Dynamics of Contention. Cambridge:
Cambridge University Press.
Nasr, S.V.R. 2002. Islam, the State and the Rise of Sectarian Militancy in Pakistan. In Pakistan:
Nationalism Without a Nation?, ed. Christophe Jaffrelot, 85–114. New Delhi: Manohar.
Nawaz, Shuja. 2008. Crossed swords: Pakistan, its army, and the wars within. Oxford: Oxford
University Press.
Flores, Thomas Edward, and Irfan Nooruddin. 2012. The Effect of Elections on Postconflict
Peace and Reconstruction. The Journal of Politics 74 (02): 558–570.
Paracha, Nadeem F. N.d. Student politics in Pakistan: A celebration, lament, and history.
<http://nadeemfparacha.wordpress.com/student-politics-in-pakistan-a-celebration-lamenthistory/>.
Paris, Roland. 2004. At War’s End: Building Peace After Civil Conflict. Cambridge, U.K:
Cambridge University Press.
Pearlman, Wendy. 2011. Violence, Nonviolence, and the Palestinian National Movement. 1st ed.
Cambridge University Press.
Radin, Andrew. 2012. “The Limits of State-Building: The Politics of War and the Ideology of
Peace.” Ph.D dissertation, MIT.
Reno, William. 2011. Warfare in independent Africa. New York: Cambridge University Press.
Roessler, Philip G. 2005. “Donor-Induced Democratization and the Privatization of State
Violence in Kenya and Rwanda.” Comparative Politics 37(2): 207–227.
Scott, James C. 1998. Seeing like a state: how certain schemes to improve the human condition
have failed. New Haven, CT: London: Yale University Press.
41
Shah, Aqil. 2014. The Army and Democracy: Military Politics in Pakistan. Cambridge: Harvard
University Press.
Sidel, John. 1999. Capital, coercion, and crime: bossism in the Philippines. Stanford: Stanford
University Press.
Snyder, Jack L. 2000. From voting to violence: democratization and nationalist conflict. New
York: Norton.
Staniland, Paul. 2010. Cities on Fire: Social Mobilization, State Policy, and Urban Insurgency.
Comparative Political Studies 43 (12): 1623 –1649.
Staniland, Paul. 2012. States, Insurgents, and Wartime Political Orders. Perspectives on Politics
10 (02): 243–264.
Staniland, Paul. 2014. Networks of Rebellion: Explaining Insurgent Cohesion and Collapse.
Ithaca: Cornell University Press.
Straus, Scott, and Charlie Taylor. 2012. Democratization and Electoral Violence in Sub-Saharan
Africa, 1990-2008. In Voting in fear: electoral violence in Sub-Saharan Africa, edited by
Dorina Akosua Oduraa Bekoe, 15–38. Washington, D.C.: United States Institute of Peace.
Varshney, Ashutosh. 2002. Ethnic Conflict and Civic Life: Hindus and Muslims in India. New
Haven, CT: Yale University Press.
Varshney, Ashutosh, and Joshua R. Gubler. 2012. “Does the State Promote Communal Violence
for Electoral Reasons?” India Review 11 (3): 191–199.
Von Borzyskowski, Inken. 2014. A Double Edged Sword: International Influences on Election
Violence. PhD dissertation, University of Wisconsin-Madison.
Wilkinson, Steven. 2004. Votes and Violence: Electoral Competition and Ethnic Riots in India.
Cambridge, U.K: Cambridge University Press.
42
---. 2013. “Electoral Competition, the State, and Communal Violence: A Reply.” India Review
12 (2): 92–107.
Wolf, Eric R. 1999 [1969]. Peasant Wars of the Twentieth Century. Norman: University of
Oklahoma Press.
Wood, Elisabeth Jean. 2000. Forging Democracy from Below: Insurgent Transitions in South
Africa and El Salvador. Cambridge: Cambridge University Press.
Yusuf, Huma. 2012. Conflict Dynamics in Karachi. Peaceworks No. 82. Washington, DC:
United States Institute of Peace. http://www.usip.org/sites/default/files/PW82Conflict%20Dynamics%20in%20Karachi.pdf
Zahab, Mariam Abou, and Olivier Roy. 2004. Islamist Networks: The Afghan-Pakistan
Connection. Columbia University Press.
Zahid, Shahid. 1989. “Licence to Kill?” Herald, August, 57-59.