Strategic Success
Factors in CounterInsurgency Campaigns:
With Discussion of the Modelling Approach
Dstl/CP23836/1.2 (ORS10)
Presentation to ORS Defence Special Interest Group,
24 Feb 2010
Dr Andrew Hossack
Dstl Policy & Capability Studies Dept
© Crown Copyright 2010
Before I Start…
Recycled Presentation!
• This is a recycled presentation:
– Most of it has been previously presented at:
• Cornwallis XII, Nova Scotia, Apr 07 (CP23836) (Best Paper)
• UCSD IGCC CT Conference, San Diego, May 07 (CP24670)
• MORS IW Workshop, Monterey, Dec 07 07 (CP26591)
• Etc…
– The central research is (mostly) published in the Cornwallis XII Proceedings
– The background terms, definitions, scope etc are published in the Cornwallis
X Proceedings similarly
• My apologies to anyone who has
heard this all before…
…Please feel free to doze quietly for the next 4060 minutes!
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
2
Background to this Presentation
The Overall COIN Research Study
• FYs05-07: An ongoing multi-year research programme into
Counter-Terrorist/Counter-insurgency (CT/COIN) campaigns:
– Building upon previous HA studies of CT, stabilisation etc
– Undertaken to inform development of PSO/OOTW models by Dstl etc
– Phase I
(Aug 04 – Jul 05):
– Phase II
(Aug 05 – Sep 06): Identify generic RED success factors
– Phase IIIA (Oct 06 – Mar 07):
Identify generic BLUE success factors
Initial Study of Campaign Evolution
• A possible Phase IIIB (analysis of the strategic dynamics
within campaigns) has not yet been agreed
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
4
Previous Historical Analysis of OOTW
• This work builds upon a number of prior HA OOTW studies:
– Operations Other Than War (OOTW) Study (1999 - 2001)
• tactical analyses of patrols, ambushes etc in 8 COIN campaigns
– Counter-Terrorism (CT) Strategies “Fastball” (2001)
– Counter-Terrorism Overseas (CTO) Study (2002 - 2004) *
• Focused on structure, attributes of “International Terrorist” groups
– Iraq Campaign Assessment I (Nov 2003) *
– Stabilisation Operations Study (2004)
• Identified success factors for external stabilisation of states
*Key results included in Cornwallis IX presentation
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
5
Review of Analysis
Methodology
Study Approach
• Undertaken as a “historical analysis” (HA) study, i.e.
• Historical Analysis:
– Is operational analysis of quantified data describing the actual
behaviour of systems across a wide range of historical cases
– Is empirical, statistical and holistic
– Involves the testing of hypotheses using established
statistical techniques
– Focuses on understanding the enduring, underlying
mechanisms of conflict
– Is about the “analysis of real operations” …..ongoing, recent
and historical!
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
7
The Spectrum of Conflict
Low
Intensity
Conflict
Peacetime &
Crisis
US:
Mid
Intensity
Conflict
High
Intensity
Conflict
Disaster
Peace
Show of (Counter)- Strikes (Counter)- Civil Limited Theatre Global
Relief
Enforcement Force Terrorism & Raids Insurgency War
War
War
War
“Conflict Intensity” (Level of Violence)
UK:
Humanitarian Peacekeeping/
Relief Ops Enforcement
Non-Conflict
(Counter)Terrorism
(Counter)Insurgency
Asymmetric
(unconventional) conflict
Civil
War
Regional
Conflict
General
War
Symmetric
(conventional) conflict
Scope of CT/COIN Study
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
8
Study Scope
• No a priori distinction
made between major
terrorist & minor
insurgent campaigns
Cumulative Frequency
• Counter-Terrorism (CT) & Counter-Insurgency (COIN)
assumed to represent adjacent & overlapping regions of
some spectrum of “asymmetric conflict”
100%
Iraq 2003 (Estimate) c. 1.2 Terrorists/Sympathisers per 1,000 Popn
80%
60%
"Insurgent" Campaigns
"Terrorist" Campaigns
40%
20%
0%
0.0001
0.01
1
100
(Terrorists/Insurgents + Sympathisers) per 1,000 Popn
• Boundary Issues:
– Micro/Urban Terrorist Campaigns excluded (E.g.: November 17)
– Genuine “Civil Wars” excluded (E.g.: Chinese Communists)
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
9
Terrorism & Insurgency (1)
• “Terrorist”, “Terrorism” are emotive words, open to
multiple (mis-) interpretations and abuse
• “Terrorist” typically used to refer to groups that
predominantly use terrorism tactics……e.g.:
– bombings
– shootings
– assassinations
• …but identical tactics also often present in “insurgency”
campaigns as well to lesser degree
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
10
Terrorism & Insurgency (2)
• UK / NATO doctrine defines “terrorism” to be:
“…..the unlawful use or threatened use of violence against individuals
in an attempt to coerce or intimidate governments or societies to
achieve political, religious or ideological objectives”
• UK / NATO doctrine defines “insurgency” to be:
“….an organised movement aimed at the overthrow of a constituted
government through the use of subversion and armed conflict”
• Only terrorism by Non-State actors against the State is
considered in this study
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
11
Study Definition
CT/COIN Campaign
Any extended, essentially two-sided, asymmetric conflict in which
some non-state player largely within a (possibly de facto) state
attempts to force some change in either the nature and/or leadership
of said state predominantly through some mixture of terrorist and/or
insurgent tactics.
As defined from UK / NATO terminology!
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
12
Conceptual Model of Campaign
Neighbouring State
Neighbouring State
State in Conflict (SiC)
Terrorist /
Insurgent
Group (TIG)
State
Security
Forces
(SyF)
Neighbouring
State
23 March 2016
© Dstl 2010
External Intervening
State (EIS)
Area of Conflict Location (ACL)
Dstl is part of the
Ministry of Defence
13
Examples of Campaign Structure
Years
External
State in Conflict
Intervening State
Area of Conflict
Location
Terrorist /
Insurgent Group
1941 - 1945
N/R
German 3rd
Reich
Former Kingdom
of Yugoslavia
Yugoslav
Partisans
1955 - 1960
UK
Crown Colony
of Cyprus
Crown Colony of
Cyprus
EOKA
1954 - 1962
France
USA
Department of
Algeria
Republic of
Vietnam
ALN
1965 - 1972
Department of
Algeria
Republic of
Vietnam
1969 - 1994
N/R
UK
Northern Ireland
PIRA
1975 - 1999
N/R
Indonesia
E Timor
FALANTIL
23 March 2016
© Dstl 2010
Viet Cong
Dstl is part of the
Ministry of Defence
14
Outline of Analysis Method
• C. 100 possible success factors identified and tested for
statistically significant association with campaign outcome:
– FY05: c. 40 possible SyF success factors tested
– FY06: c. 60 possible Insurgent success factors tested
• Assessment criteria for ordered categories defined for each
factor:
– Category “-1 ”: “Poor, incompetent or no usage/presence”
– Category “ 0 ”: “Moderate, mixed or occasional usage/presence”
– Category “+1”: “Good, competent or extensive usage/presence”
• Assessment criteria, categories for outcome measures
defined similarly
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
15
Coding of Success Factors
• Factors judged on 3 point ordinal scale:
Field Title
Flexibility of Security Forces C2
Field Type
Security Forces Capability
Degradation of Key State Economic
Sectors
Insurgent Tactic
Field Value
-1
0
1
The Security Forces were unresponsive,
The Insurgents did not significantly
inflexible and incapable of either
seek to destroy or seriously degrade
achieving surprise in planned operations
any part of the State’s economy
or of exploiting targets of opportunity.
The Security Forces had limited
responsiveness and flexibility and were
sometimes able to act sufficiently rapidly
to achieve surprise in planned
operations or to exploit targets of
opportunity
The Security Forces were responsive,
flexible, capable of planning
innovatatively and of acting rapidly to
achieve tactical/operational surprise or
to exploit targets of opportunity
23 March 2016
© Dstl 2010
The Insurgents occasionally or
inconsistently sought to destroy or
seriously degrade key sectors of the
State’s economy
The Insurgents constantly and
consistently sought to destroy or
seriously degrade key sectors of the
State’s economy
Dstl is part of the
Ministry of Defence
16
Outcome Measures Used
• Two types of campaign success considered:
– Military Success (judged on achievement of monopoly of violence)
– Political Success (judged on achievement of initial objectives)
• Military Success assumed zero-sum a priori:
– ‘(State) Success’
State monopoly of violence at end campaign
– ‘(State) Partial Success’ Both State & Insurgents retain capacity for violence
– ‘(State) Failure’
Insurgent monopoly of violence at end campaign
• Political Success coded independently for all Actors (Internal
State, External State, Insurgents):
– ‘Success’
Most initial objectives achieved at end campaign
– ‘Partial Success’
Some initial objectives achieved at end campaign
– ‘Failure’
Few initial objectives achieved at end campaign
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
17
Data Collection & Coding
• Data for 44 COIN campaigns collected & categorised by
external researchers
– FY05: Data collected on overall campaign, context, SyF factors etc
– FY06: Additional data on Insurgent & social factors only
• Case selection was pseudo-random:
– Stratified sampling across geographic regions
– Spread over post WWI-period
• Each data point currently represents an entire campaign:
– Static analysis only; takes no account of development of campaign
over time
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
18
Statistical Analysis
• Candidate success factors tested for association with
success using Somers’ D statistic at 90% confidence
• Essentially assessing distributions of data in 3 x 3
Contingency Tables:
Criterion X Well Applied
Criterion X Averagely Applied
Full SyF Military
Success
6
1
Partial SyF
Military Success
2
4
SyF Military
Failure
0
1
Criterion X Badly Applied
1
0
7
“Is there a tendency for better (or worse) values of variable X to be
consistently paired with better (or worse) campaign outcomes?
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
19
Analysis Results (1):
(Attritional) Efficiency of SyF, Insurgents
Efficiency of SyF & TIG
10000
Security Forces
“Efficiency”
1000
Terrorists/Insurgents
100
10
1
0.1
0.1
1
10
100
"Force Ratio“ (Ratio of Median Annual Forces)
• Plot shows efficiency of each combatant vs Force Ratio:
– Efficiency: No. Opponents killed per 1,000 Man-Years force deployed
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
21
Efficiency of SyF & TIG
10000
Security Forces
“Efficiency”
1000
Terrorists/Insurgents
100
10
y = 56.5x
1
R
2
-0.9
= 0.4
0.1
0.1
1
10
100
"Force Ratio“ (Ratio of Median Annual Forces)
• SyF efficiency decreases significantly with increasing FR
– Each additional soldier/policeman adds less than the one before
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
22
Efficiency of SyF & TIG
10000
Security Forces
1000
Terrorists/Insurgents
“Efficiency”
y = 19.0x
100
R
2
0.3
= 0.1
10
1
0.1
0.1
1
10
100
"Force Ratio“ (Ratio of Median Annual Forces)
• No evidence that Insurgent efficiency changes with FR:
– Each Insurgent is as efficient as the one before
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
23
Efficiency of SyF & TIG
10000
Security Forces
1000
Terrorists/Insurgents
“Efficiency”
y = 19.0x
100
R
2
0.3
= 0.1
10
y = 56.5x
1
R
2
-0.9
= 0.4
0.1
0.1
1
10
100
"Force Ratio“ (Ratio of Median Annual Forces)
• Possibly this is because Insurgents are typically too small to
experience effects of diminishing returns on scale?
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
24
Analysis Results (2):
The relationship between Force Ratio &
Campaign Outcome
Force Ratio & Campaign Outcome (1)
• There is a weak relationship between odds of military
campaign success and whole-campaign Force Ratio:
– Statistically significant at 93% confidence
– A conservative and reasonable result given expected “noise” in data
– Military success used because a zero-sum outcome
– Strictly, tested against the Ratio of Median Annual Forces (used as
a surrogate approximation to Force Ratio)
• Can use ordinal logistic regression to estimate the rate of
change of odds with change in Force Ratio
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
26
Force Ratio & Campaign Outcome (2)
Military Success
100%
90%
Outcome probability
80%
70%
60%
SyF Failure
SyF Partial Success
SyF Success
50%
40%
30%
x3 increase in
Equivalent Odds
20%
x10 increase in
FR
10%
+0
6
1E
0
10
00
0
00
10
00
10
0
10
10
1
0.
1
0.
01
0.
00
1
0%
"Force Ratio" (Log Axis)
• Ordinal Logistic regression estimates that:
– Odds(Win), Odds(Lose) change by x3 with each x10 change in FR
– 90% Confidence that true rate of change is between x1+ - x8
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
27
Force Ratio & Campaign Outcome (4)
• The Force Ratio-Outcome Model may provide:
– A basis for future development of a model incorporating key success
factor values as well as Force Ratio
– A method of quantifying the value of changing outcome odds in terms
of the change in Force Ratio required to give an equivalent effect
• That is, each x2 change in campaign odds is:
– Estimated to be equivalent to a x4 change in Force Ratio
– At least equivalent to a x2+ change in Force Ratio (90% confidence)
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
28
Force Ratio & Campaign Outcome (2)
Military Success
100%
90%
Outcome probability
80%
70%
60%
50%
x4 Change in
Odds
SyF Failure
SyF Partial Success
SyF Success
40%
30%
20%
x16 Change in
FR
10%
+0
6
1E
0
10
00
0
00
10
00
10
0
10
10
1
0.
1
0.
01
0.
00
1
0%
"Force Ratio" (Log Axis)
• Each x2 (x4) change in campaign odds is:
– Estimated to be equivalent to a x4 (x16) change in Force Ratio
– At least equivalent to a x2+ (x4+) change in Force Ratio (90% confidence)
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
29
Analysis Results (3):
Generic Security Forces’ Success Factors
SyF Success Factors (1)These are the
same
factors
thatSuccess for
Magnitude of Estimated
Military Campaign Success for State &
Political
Campaign
Equivalent Force Multiplier
SyF
State &toSyF
were reported
x50 < FM
x20 < FM ≤ x50
x10 < FM ≤ x20
x5 < FM ≤ x10
They only tell us
the What? not the
x1 < FM ≤ x5
How?
(Context specific?)
23 March 2016
© Dstl 2010
Cornwallis
X!for Conflict
Pop Support
-
Pop Support for SyF
SyF Doctrine
-
Pop Support for SyF
-
Pop Support for Conflict
SyF Doctrine
SyF Training
Flexibility of SyF C2
Targeting TIG Leadership (-)
Resource Availability
SiC Legitimacy
SyF Counter-Intelligence
Winning “Hearts & Minds”
Overall SyF Intelligence
SyF Strategic Posture
Integration of SyF C2
SyF Counter-Intelligence
Winning “Hearts & Minds”
Flexibility of SyF C2
Overall SyF Intelligence
State Legitimacy
Integration of SyF C2
Legal Status of Campaign
Dstl is part of the
Ministry of Defence
31
SyF Success Factors (2)
• Generically, the factors most important to State military and
political success in COIN at the campaign level are those
concerned with:
– Wresting the campaign initiative away from the Insurgents:
• good Counter-Int; proactive Strategic Posture; Hearts & Minds
– The directed & precise application of Security Forces’ capability:
• flexible & integrated C2; good Int; good Training and Doctrine
– Creating/maintaining the necessary political context for success:
• Popular Support for Security Forces; Hearts & Minds; good
Training and Doctrine
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
32
SyF Success Factors (3)
• These generic, campaign level success factors include no
factors relating to:
– Security Forces’ capability itself:
• Firepower, mobility, use of ‘Special Forces’ etc
– the specific methods by which Security Forces’ capability is applied
against Insurgents
• degradation of infrastructure, direct attrition, control of population
• These factors may still be significant in specific contexts
• There is some requirement for boots-on-the-ground:
– force ratio advantage & resource availability
– Needed to avoid losing, but not in itself sufficient for winning?
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
33
Analysis Results (4):
Generic Insurgents’ Success Factors
Insurgent Success Factors (1)
Magnitude of Estimated
Equivalent Force Multiplier
x50 < FM
Military Campaign Success for
Insurgents
Pop Support for Insurgents
x20 < FM ≤ x50
x10 < FM ≤ x20
x5 < FM ≤ x10
x1 < FM ≤ x5
Political Campaign Success for Insurgents
(Insurgent Doctrine)
-
Pop Support for Insurgents
Destruction of SyF
Insurgent Firepower
Insurgent Counter-Intelligence
Overextension of SyF
(Insurgent Doctrine)
Attempted Destruction of State Infrastructure
-
Overthrow State Leaders
Insurgent Training
Insurgent Experience (-)
Legitimise Claim to Power
Insurgent Internal Structure (-)
Attract External Intervention
Attempted Destruction of State Institutions (-)
Insurgent Concealment in Terrain
Subversion of Population
Legitimise Claim to Power
Overall Insurgent Intelligence
Subversion of Population
• CAVEAT: The Insurgent Doctrine Factor is unsafe – use with caution!
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
35
Insurgent Success Factors (2)
• Proportionately fewer factors are identified as generically
associated with success for Insurgents compared to SyF
• Is Insurgent success more context-specific than for SyF?
– Fewer generic “rules of thumb” for guidance?
• Emphasises need for SyF Int to “get inside Insurgents’
heads”?
• Or a reflection of reduced robustness of Insurgent data
compared to SyF data?
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
36
Insurgent Success Factors (3)
• Possibly 3 broad themes discernible among Insurgent
Success Factors:
– Possession of favourable political context for success:
• Having popular support;
• establishing legitimacy of cause;
• subverting population into detaching itself from the State
– Maintaining an Intelligence Superiority over SyF
– Possession of ‘kinetic’ military strength:
• Firepower;
• Trained insurgents;
• Relevant Doctrine?
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
37
Insurgent Success Factors (4)
• Firepower / kinetic military strength may be more important to
Insurgents than to SyF
– Possession of heavy weapons firepower is a military success factor
for Insurgents but not for SyF!
– Insurgents do not suffer diminishing returns on increasing forces
• Presumably because Insurgents are much smaller than SyF
so gain greater benefit from improvement in military capability
• As much a hypothesis as an inference!
• Suggests the necessity of modelling both the military as well
as the political component of CT/COIN!
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
38
Insurgent Success Factors (5)
• Importance of military capability may explain some of the less
immediately understandable Insurgent success factors:
– Insurgent Internal Structures (with hierarchical better than cellular)
– Destruction of SyF
– Overextension of SyF
– Overthrow of State Leadership
– Concealment within Terrain (but not within populations)
• These all require (relatively) large Insurgency size to be worth
pursuing
• Or, they may be spurious results (“False Positives”)
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
39
Caveats on Results
Caveats on these Results
• Static, whole-campaign analysis:
– Effectively considers the average state of factors across geography and time
• Robustness of Insurgent factor design:
– Wide range of possible Insurgent strategies for different contexts / campaigns
– Hard to write generic definitions that cover 44 campaigns over 85 years!
– Problem of Intent vs Action (Threatening vs Actual Destruction etc…)
• Spurious or “False Positive” Results:
– An inherent limitation of inferential statistics!
– At 90% confidence, it is statistically likely that up to:
• 4 of the SyF “success factors” found for each outcome type
• 6 of the Insurgent “success factors” found for each outcome type
are false
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
41
Summary & Conclusions
Conclusions on Success Factors
• The Political Component of CT/COIN: Popular support factors for
Insurgents and SyF potentially offer some of the largest modifiers to
campaign odds of success:
– Affecting both military and political campaign outcomes
– Up to twice as important to political as to military success
• The Military Component of CT/COIN: Although Force Ratio has a
weaker effect upon campaign outcome than key SyF and Insurgent
Success Factors:
– There is still some requirement for boots-on-the-ground
– SyF suffer diminishing returns on manpower at typical FRs
– Insurgents gain at least some benefit from numbers, training, firepower
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
43
Conclusions on Research
• Method provides technique for quantifying the payoff from
adopting given strategies / capabilities….if done appropriately
• Results demonstrate the necessity for modelling the military
and the political components of CT/COIN together
– Endorses a Rational Expectations perspective of PSO/COIN
• There is potential for further exploitation of existing, static
campaign data
• Further major research will require more detailed, campaignphase resolution
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
44
Any Questions?
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
45
Published Research
• A HOSSACK. 2005. Historical Analysis of Terrorist Campaigns, with observations
on Current Operations in Iraq (Dstl/CP10135). Published In: A WOODCOCK, G
ROSE, eds., The Cornwallis Group IX: Analysis for Stabilization and CounterTerrorist Operations (2004) pp 393 – 417.
• C IRWIN, A S MORLEY, 2005. Drawing lessons from the past. A historical analysis
of stabilization operations. Royal United Services Institute (RUSI) Journal, Vol. 150
No. 1, February 2005, pp49 – 53
• A HOSSACK, K SIVASANKARAN. Success Factors in CT/COIN Campaigns:
Preliminary Results arising from Current Research (Dstl/CP14230). Published In:
A WOODCOCK and G ROSE, eds., The Cornwallis Group X: Analysis for New
and Emerging Societal Conflicts (2005) pp 469 - 491.
• A HOSSACK. Security Force & Insurgent Success Factors in Counter-Insurgency
Campaigns (Dstl/CP23836). Published In: A WOODCOCK and G ROSE, eds.,
The Cornwallis Group XII: Analysis for Multi-Agency Support (2008).
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
46
Contact Details
Dr Andrew Hossack FORS CMath FIMA
Principal Analyst & Associate Fellow
Historical and Operational Data Analysis (HODA) Team
Analysis, Experimentation & Simulation (AES) Group
Policy & Capability Studies (PCS) Department
Defence Science & Technology Laboratory (Dstl)
iSAT “K”, Rm C036, Grenville Bldg West Court,
Portsdown Hill Road, Fareham, HANTS PO17 6AD
Tel: +44(0) 2392 532889
Fax: +44(0) 2392 533458
Email: adhossack<at>dstl.gov.uk
[dstl] is part of the UK Ministry of Defence
23 March 2016
© Dstl 2010
Dstl is part of the
Ministry of Defence
47