Religion, Terrorism and Public Goodselib/250C/religion_terrorism.pdf · 2008-04-26 · Religion, Terrorism and Public Goods . 2 Prime Minister’s Office, Algeria – 4/11/07

1

Religion, Terrorism and Public Goods

2

Prime Minister’s Office, Algeria – 4/11/071 Al Qaeda suicide bomber, 12 dead, 112 injured

3

The Global War on Terrorism

167* 10,000* Post 9/11

October 01 –September 06

1094,800Pre 9/11January 98 –August 01

Fatalities / monthFatalities

Fatalities due to Terrorist Attacks Worldwide1998-2006

Source: Memorial Institute for the Prevention of Terrorism, Oklahoma City, www.tkb.org•Excluding casualties in Iraq and Afghanistan. Including those fatalities the 10/01 – 9/06 figure is447 rather than 167.

• Note change in organizations threatening civilians and governments.

4

Outline

0. Introduction: The Violent Puzzles1. Background: Hamas, Hezbollah, Taliban, Insurgency

and Suicide Attacks2. Framework: Terrorist Clubs vs. Hard Targets 3. Testing: Clubs, Hard Targets and Suicide Attacks4. Policy Implications: Counterterrorism, Economic

Development and Nation Building5. Clubs vs. Rational Peasants, (or Gangs and Community

Policing (A&Y))6. Street Lights: Some results from Iraq

5

Hamas

6

Hamas 2006

7

Taliban

8

Hezbollah

9

Conclusions• Why so few terrorist organizations?

Defection constraint.• Why are religious radicals effective terrorists?

Solve defection constraint in benign activities.Not necessary theology.

• Why suicide attacks?Hard targets.

• What to do about it? Compete in providing benign services and competent governance, with muscular protection.

• Future WorkNeed research and evaluation

Distinguish club model from standard “rational peasant”approach to “winning hearts and minds”

10

Jewish Underground

11

Muqtada al-Sadr’s Mahdi Army

12

A Violent Puzzle Among Religious Sects

The Violent Puzzles:• Why are radical religious militias and terrorists so efficient at

violence?HamasHizbullahTalibanAl SadrAl Qaeda

They make the secular terrorist organizations of the 60s-90s look lame

• Why Suicide Attacks? • Our approach: Draw on

a) insurgency literature in IR, b) economics/sociology of religion, c) agency and collective action in organizations

13

What Motivates Terrorists? The Afterlife and Other Myths

• Is the advantage of radical religious terrorists due to the superior motivation that stems from theology and beliefs?- An ideology of hate? - Promises in the afterlife?

• Israeli psychiatrist Ariel Merari has spent years interviewing suicide attackers, their families and friends.

• He finds that: - Hamas and Jihad suicide attackers never mention religion or virgins in heaven as their primary motivation

• Consistent with experience in other countries- many suicide attackers worldwide are not religiousradicals, including the Tamil “Tigers” and the majority of attackers in Lebanon in the 1990s.

14

So what does motivate suicide attackers?

• Merari finds that there is no specific primary motivation- not economic depravity, - not depressed or suicidal or mentally ill

- consistent w/ research on Bader-Meinhoff, Red Brigade, ETA

- not ignorant - generally not seeking revenge

• Might be best thought of as well adjusted altruists, who truly believe that their courageous act will help their communities- combination of altruism and delusions of self-importance- close to profile of rational recruits to sects; triage needed

• Now that’s a frightening thought, because the world is full of self-motivated altruists who are willing to give their lives for some cause- and indeed there seems an ample supply of suicidal terrorists

15

Suicide Attacks as a Rebel Tactic

• Civil wars 1945-1999 (Fearon-Laitin)127 in 69 countriesdirectly account for 16m fatalities

• Rebel tactic is usually rural insurgency• Suicide attacks are very rare, but becoming more

common

16

Table 1: Suicide Attacks by Country of Perpetrator

17

Table 1: Suicide Attacks by Country of Perpetrator

18

Income, Insurgency and Suicide Terrorism

19

Religious Differences between Insurgents (Suicide Attackers) and Targeted Victims

(Table 3)

Insurgency

16.50%

83.50%

Suicide Attack

13%

87.4%

DifferentReligion

SameReligion

Unlike civil wars, this is the only consistent predictor.Why? Because coreligionists are soft targets.

20

2. Terrorist Clubs vs. Hard Targets

Militia activity - Coordinated rent captured involving violence.

• e.g. attacking occupying army, providing law and order, organizing and carrying out a clandestine activity (like terrorism).

• often involves personal risk. • Key aspect is sensitivity to defection.

21

Map of Afghanistan

22

A. Securing a Trade Route

31 2 4... ..N-1 N1

Convoy Destination

Checkposts

$B

23

Securing a Trade Route

C Ri C Rii

N

({ }) ( )==

∏1

Convoy will choose to set out only if all Ri = 1.

Payoffs: Club extracts surplus B and shares it equally among members, who buy goods at price P.

Benign local public goods provided by govt., G and club, A.Defector’s outside option is wi, but no access to C or A.

Incentive compatibility: Member loyal iff

(ICC) U( B

N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)

24

Securing a Trade RouteC Ri C Ri

i

N

({ }) ( )==

∏1

If ICC fails this is an N player prisoner’s dilemma, resulting in an unsafe route, no convoy and no rents.

Adverse selection: Imagine two unobserved types (as above), such that

ICC holds if wi = wH (> wL) for all i. A club with a costly sacrifice as an initiation rite which successfully excludes all low wage types can secure the route and extract the rent.E.g., The Taliban

(ICC) U( B

N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)

25

Figure 2: Selecting Low Wage Membership Allows Larger Projects

Project Value (B)

High wagedefector

Low wagedefector

Loyalmember

B*B* B**

Incentivecompatibilitylimit with lowwage members

26

B. Capturing a Hill

2

3

N

1

$B

27

3. Clandestine Violence

2

N

1

Conspirators

Target

Defection – The reason so few militias and terrorist organizationssurvive.. defection is common.

How do successful militias and terrorist organizations prevent defection?

They have an organizational advantage.

28

An analogous problem: Reducing free riding in

Religious Sects• Sect – a religious group that:

– imposes extreme prohibitions and requires distinctive sacrifices

– views secular society as corrupt, dangerous, and threatening – economic life: high levels of mutual aid, and local public goods

provided through volunteer work, e.g. education, health care, law and order, welfare services, orphanages, day care, soccer clubs

How can you trust members to apply full effort?

• Internal economies of sects rely on trust-based transactions- sacrifices are elicited early in life to signal commitment

e.g., education, missionary work, jail time- prohibitions distance members from market culture

29

30

Rational choice approach to religious sectsIannaccone (1992)

Formally..

(1) Ui = U (Si, Ri, C({Rj}) ), where S – consumption, R – religious activity,C – local public good .

(2) C({Rj}) = for j=1 to J .

C could be mutual insurance, health care, education.

(3) R = T – H . Budget constraint for time.(4) wH = S . Budget constraint for money.

Figure 1 illustrates optimal religious prohibitions.

∑=

=J

j

j

J

RR

1

31

Analogy: Seminar as a Club

• A seminar (like this one) is a club, where participants benefit from their own effort “R” and the average R of colleagues .

• A good citizen comes prepared, asks questions, provides good answers, all because she studies.

• Lacking a way to subsidize R, the club would like to tax outside activity of members.

• In principle, a research club should tax, or tithe, if it can.But it typically lacks tax authority.

32

Optimal Prohibitions for Seminar Participants

• Efficient proxy taxes on outside options might be:

• Prohibit alcohol with nonmembers

• Prohibit beach on Sabbath• Dress strangely• Limit eating with nonmembers

through dietary restrictions• Limit communication with

outsiders by speaking arcane language

• With enough prohibitions seminar participants would have nothing better to do with their time than study

• Enforcement could be through threat of expulsion or through peer pressure

If this example doesn’t work for you, think of a fraternity (or a team),where R is partying (training) and helping out other members.

33

Fig 1: Optimal Taxation Through Prohibition

Work hours

Wages

Religious activity

34

Maimonides Rationale for Circumcision• Twelfth century philosopher Rav Moses Maimonides

explaining circumcision..• “It gives to all members of the same faith, i.e., to all believers

in the Unity of God, a common bodily sign, so that it is impossible for any one that is a stranger, to say that he belongs to them. For sometimes people say so for the purpose of obtaining some advantage ... ....It is also a fact that there is much mutual love and assistance among people that are united by the same sign when they consider it as [the symbol of] a covenant.

• [The Guide for the Perplexed, late 12th century, translated 1904. Chapter XLIX. Brackets are those of the translator. Italics are my own.]

• Signaling theory won a Nobel Prize in Economics, 9 centuries later

35

Rationalizing Sacrifices

Imagine heterogeneity in participants’ outside options, wj ,(call them “wages).

Members would prefer that other members have low wages wj , since that implies higher R and larger externalities .

Low R members are free-riders who it would be efficient to exclude, but wj is unobserved.

36

Rational Sacrifice (cont.)

• Voluntary sacrifices of time might exclude high wage individuals but include low wage for an efficient separating equilibrium.

e.g.s Insist on an arcane language that takes years to learn

Religious education with no market value

370

0

�

�

B1

A1

��A2B2

High wage, low C

Low wage, low C

High wage, high C,sacrifice

Low wage, high Csacrifice

Work Hours (H) º » R+Kκ∗H1

Figure 1: Rationalizing Sacrifices

38

Iannaconne (92): Sects and Churches in National Data

39

Subsidizing Sacrifice

Source: Berman (2000)

40

Evidence: Fertility and Schooling Results• Data : extensive search yielded household surveys in

Indonesia, India, Bangladesh, Cote D’Ivoire, Pakistan (B&S ‘04) and Israel (B ‘00)

• Women in families with Islamic and Ultra-Orthodox religious education have higher fertility in all 6 countries, by 2/3 to one more expected lifetime child.

• Islamic and Ultra-Orthodox education have significantly lower rates of return than secular education in 3 of 6 countries; insignificant results in other 3 countries

• Prevalence of radical religious schooling: 2-5% of Muslims in Rural Bangladesh, Pakistan, Cote

D’Ivoire, 5% of Israeli Jews14-25% in Indonesia and two Indian States

(Uttar Pradesh and Bihar)

41

Fig 1: Optimal Taxation Through Prohibition, and Fertility

Work hours

Wages

Religious activity& Fertility

42

Differential Fertility by Sect MembershipSix countries – Berman and Stepanyan (04)

AnyAnyOwnAnyOwnAnySect indicator

0.66(.39)

1.34(.46)

0.58(.27)

0.77(.43)

0.67(.26)

5.34(.30)

Diff. fertility

PakistanCote D’Ivoire

Bangla-desh

UP & Bihar

Indo-nesia

Israel

43

Differential Returns to Education by Sect Membership - Six countries (Berman and Stepanyan ’04)

-.048(.026)

.132(.006)

Pakistan

.175(.010)

.097(.007)

.122(.008)

.116(.005)

.094(.002)

Secular schooling

-.029(.070)

-.073(.034)

-.051(.229)

-.022(.013)

-.076(.006)

Religious schooling

Cote D’Ivoire

Bangla-desh

UP & Bihar

Indo-nesia

Israel

44

Fertility Differential for Ultra-Orthodox Jews (Berman ’03)

TABLE V TOTAL FERTILITY RATES OF ISRAELI SUBPOPULATIONSA. Source: Labour Force Survey

Period Full Population Jews Ultra-0rthodox Jewsc

All other Jews

1980-1982 2.99a 2.76 6.49 2.61

(0.04)b (0.04) (0.31) (0.04)

obs. 31347 27635 1040 26569

1995/96 2.66 2.53 7.61 2.27

(0.04) (0.05) (0.30) (0.05)

obs. 27866 22776 1021 21755

Change -0.33 -0.23 1.13 -0.34

(0.06) (0.06) (0.44) (0.06)

B. Source: Population Registry

Period Full Population Jews Christians Muslims

1980 3.14 2.76 2.66 5.98

1995/96 2.90 2.57 2.19 4.65

Change -0.24 -0.19 -0.47 -1.33

45

Analogy: Military unit as a Club• Like a sect, a unit is involved in cooperative

production, i.e., participants benefit from their own effort and the average effort of other members.

• A good soldier/member comes prepared, trains, works hard, covers his/her buddies, would never defect.. because they are devoted.

• Lacking a way to subsidize devotion, the unit would like to a) tax outside activity of members, b) select devoted members.

46

Why are sects effective at violence?

• Recall that militias and terrorist groups are organizations extremely sensitive to defection

• Sects have a strong advantage at coordinated violence because their benign service provision activities help thema) select operatives unlikely to defectb) influence operatives through their support of friends and family

• Testable implication: a sect will be more effective the stronger its’ social service provision

47

Figure 2: Selecting on Low Wage Membership Allows Larger Projects

Project Value (B)

High wagedefector

Low wagedefector

Loyalmember

B*B* B**


48

Figure 3: Benign Activity Increases a Militia’s Potential

(ICC) U( B

N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)

Project Value (B)

Utility ofdefector

Utility of loyalmember

B**

Incentive CompatibilityLimit - augmented clubgood

B*

Utility of loyalmember - augmentedclub good

IncentiveCompatibility Limit

49

Resolving the Puzzle

(ICC) U( B

N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)

• Taliban, Hamas, Hizbullah, Sadr’s Militia are all examples of remarkably effective violent radical Islamic organizations which started out as classic sects providing social services.

• Cooperative production of social services has the same “free-rider” problem, though less extreme e.g., mutual insurance is sensitive to defection

• An organization designed to limit defection in a benign context will have a huge advantage in the cooperative production of violence

50

TABLE 4: SOCIAL SERVICE PROVISION AND LETHALITY OF TERRORIST ATTACKS

ISRAEL AND LEBANON: 1968-2006

Group nameAttacks Injuries Fatalities

Injuries per attack

Fatalities per attack

(std. error)

Hamas 70 2202 413 30.2 5.9 0.87Hezbollah 90 387 449 4.3 5.0 2.82Palestinian Islamic Jihad 38 722 111 12.7 2.9 0.81Popular Front for the Liberation ofPalestine

38 376 107 9.9 2.8 1.03

Fatah/PLO 131 1465 279 11.20 2.1 0.48Democratic Front for the Liberationof Palestine 21 240 22 10.4 1.0 0.37

Unknown 427 1055 351 2.2 0.8 0.28

Social Service Providers: Hamas andHezbollah 160 2589 862 15.8 5.4 1.62

Others: DFLP, Fatah/PLO, PIJ, PFLP 228 2632 519 11.6 2.3 0.35Difference 3.1 1.67*

51

Suicide

SuicideSuicide Suicide

Suicide Suicide Suicide0

20

40

60

80

100

120

Att

acks

Hamas

Tanz

im

Al Aq

sa M

artyrs

PIJFa

tahPF

LP

Force

17

Terrorist Organization

Who Selects Suicide Attacks?

Suicide attacks are so damaging that only defection proof organizationscan succeed at them. Most do not try.

52

Efficiency of Suicide Attacks in Israel and Lebanon

63

4437

7

31

2 17.2

17.3

4 2.9 2.8 0.5 00

10

20

30

40

50

60

70

Hamas

Hizbulla

h PIJPF

LP

Al Aq

sa M

artyrs Fa

tahSS

NP


Suicide Attacks Ave. Fatal./Attack

53

Lethality of Suicide Attacks in Israel and Lebanon

63

4437

7

31

2 17.2

17.3

4 2.9 2.8 0.5 00

10

20

30

40

50

60

70

Hamas

Hizbulla

h PIJPF

LP

Al Aq

sa M

artyrs Fa

tahSS

NP


Suicide Attacks Ave. Fatal./Attack

54

Recap: Sects and Violence

• Why so few terrorist organizations?Defection constraint.

• Why are religious radicals effective terrorists?Solve defection constraint in benign activities.Not theology.



• How can we be sure it will work?Need research and evaluation, just like any weapon.

55

Sect CharacteristicsHamas Taliban

Localpublicgoods

schools,hospitalswelfare, militias

law and order,militias

Militia activity

assassination ofinformants,attacks onIsraeli civiliansand Israeli military

guardedsmuggling routes, law & order,conquered Afghanistan

Increasedstringency

dress codes, personal piety, worship,

personal piety

Sacrifice risking arrest,injury or death

madrassaattendance

These benign activities are the norm among religious sects for Christians (Iannaccone 92), Muslims and Jews (Berman 00, 03)

56

Constructive Efficiency : [Figure 7]

If ICC does not hold it may be efficient for a club to make it hold by

• 1) raising C(1), through some other investment that augments local public goods (e.g. welfare, hospitals, etc.),

• 2) making cash payments to members (e.g. families of martyrs),

57

Figure 8: Govt. Provision of Public Goods Reduces Militia’s Potential

(ICC) U( BN

,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)

Project Value (B)

Defector - high govt services

Loyal member -high govt. services

B** B*

Defector - low govt. servicesLoyal member -low

govt. services

Incentivecompatibility limit -high govt services

E.g., Malaya, Phillipines, Egypt

58

Destructive Efficiency: [Figure 8]

• 3) reducing G , the public good available to members and nonmembers, (assassination of public officials),

• 4) limiting B (ban on heroin cultivation), • 5) raising P (general strikes, access to goods markets),• 6) lowering wi , the outside options of members,

(Madrassah, jail time, secluding women, harassing nonmembers, destroying or banning access to Israeli labor markets).

59

Implications

Violent radical religious groups thrive where.. • a) govt. provision of local public goods is weak

- Somalia, Nigeria, Pakistan, Algeria, Lebanon, Afghanistan, Palestine, Iraq

• b) local militias are popular..- Chechnya, Afghanistan during war, Kashmir, Palestine, Jordan (“Black September”), Palestine, Iraq

• c) wages are low.. - all of the above,

• d) where outside subsidies are available,- Kashmir, Lebanon, Palestine, Afghanistan, Iraq

60

Suicide Attacks, Terrorism and Insurgency

• Why suicide attacks?• Deadliest method of delivering explosives to a

target- precise- leaves no operative to interrogate

• Method of choice vs. “hard” targetsi.e., targets whose destruction implies a high probability of death or capture

61

Hard Targets• p(h) – probability of apprehension increases in

govt. investment in “hardening” target• Expected utility, loyal operative, suicide attack..

where D is damage, benefit B is proportional to D• Utility from defection:

• Utility from conventional attack:

• Choose suicide attack if (7a) > (7b), (7a) > (7c)

62

Figure 4: Strong Governments Harden Targets, Insurgents respond with Suicide Attacks

Utility of Defector -

Damage to Target (D)

DC

Utility of LoyalMember -Suicide Attack,

E

O

D

DD DE

Utility of LoyalMember -ConventionalAttack, low p

Utility of LoyalMember -ConventionalAttack, high p

C

Notes: a) Global decline in insurgency;b) Coreligionistsare usuallysoft targets; c) Foreign allies of govt. are hard targets – switch to suicide attacks puts them at risk.

63

TABLE 5: ATTACKS ON ISRAELI RESIDENTS BY LOCATION AND TACTIC

Sept 2000 through July 2003

64

Attacks on Israelis by Location and Tactic

Attacks on Israeli Residents

17405

730

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

Palestine Israel

Attacks

Attacks on Israel Residents

341

511

8

401

0

100

200

300

400

500

600

Palestine Israel

Fatalities Suicide Attack Fatalities

Why? Because targets in Palestine are soft, whereas targets in Israel are hard.

65

Religious Differences between Insurgents (Suicide Attackers) and Targeted Victims

(Table 3)

Insurgency

16.50%

83.50%

Suicide Attack

13%

87.4%

DifferentReligion

SameReligion

What about this? Coreligionists are soft targets.

66

Coreligionists are soft targets

• Insurgents and terrorists often target coreligionists: political rivals, members of rival militias, collaborators, targets of extortion.

• They seldom use the suicide tactic to do so..• .. probably because it’s not necessary. A coreligionist

assailant can defeat profiling.• Exception are target well defended by means beyond

profiling: e.g., Sadat, Massoud, Rajiv Ghandi.• When members of other religions have similar appearance

suicide attacks are not used: N. Ireland.

67

Figure 5: Strong clubs choose more suicide attacks and do more damage

Utility of Defector -weak club

Damage to Target (D)DF

Utility of LoyalMember -Suicide Attack,

E

O

D

DD DE

Utility of LoyalMember -ConventionalAttack, high p

F

C

Utility of Defector -strong club

Note: Most benign policies vs. insurgents will affect the high damage margin, so they will reduce suicide attacks.

68

Evidence on Benign Activity

and a Militia’s Potential

69

70

Suicide

SuicideSuicide Suicide

Suicide Suicide Suicide0

20

40

60

80

100

120

Att

acks

Hamas

Tanz

im

Al Aq

sa M

artyrs

PIJFa

tahPF

LP

Force

17


Who Selects Suicide Attacks?

• Suicide attacks are so damaging that only defection proof organizations can succeed at them. Most do not try.• Nonsystematic evidence from Iraq, Afghanistan, Sri Lanka and Chechnya is consistent with suicide attacks being reserved for hard targets.• Religious radicals specialize in s. attacks in Iraq, Afghanistan, Chechnya, but evidence on soc. service provision is weak.

71

Application: Jewish Underground

• Violent militia which drew members from Gush Emumin,a messianic settler movement but a weak sect, (weak prohibitions and sacrifices).

• Began settling illegally in West Bank in mid 1970s. • - vigilante activity had local public good aspect

• After Camp David (I) agreements frustrated settlers organized conspiracy to destroy Muslim holy sites on Temple Mount / Haram A-Sharif.

• - project aborted for lack of rabbinical authorization.

72

Testable Implications: Other countries

• Richer countries are less likely to have insurgencies and civil wars (Fearon-Laitin)

73

Why the Increase in Suicide Attacks?

• .. Because of the decrease in viable options for insurgents

• Insurgents attempt conventional tactics first, including against coreligionists. When these fail they turn to suicide attacks, generally reserving them for “hard” targets.

• As governments improve at counterinsurgency we will see more terrorism and suicide attacks- directed against both local targets and allies of govt.

74

4. Mosque and State –Implications for Counterinsurgency

Activity can be rationalized, so incentives matter, which implies that subtle instruments could work.

Subtle Policies: Governments, Economic Rents and MarketsA. Improve Provision of Local Public Goods

by Secular Governments- the Kilcullen / Petraeus / SOC approach

B. Fiscally Separate Church and State if that Government is Radical Religious - so that it cannot lock itself into power

C. Reduce Rents Available to Militias and Smugglers- e.g., demand for Heroin, Cocaine and Oil

D. Improve Market provision of substitutes.

75

Once suicide attacks are being used, constructive intervention operates at that margin

76

The Kilcullen Approach

“23. Practise armed civil affairs. Counterinsurgency is armed social work; an attempt to redress basic social and political problems while being shot at. This makes civil affairs a central counterinsurgency activity, not an afterthought.”

“You need intimate cooperation with inter-agency partners here, national, international and local. You will not be able to control these partners . Many NGOs, for example, do not want to be too closely associated with you because they need to preserve their perceived neutrality.”

“Thus, there is no such thing as impartial humanitarian assistance or civil affairs in counterinsurgency. Every time you help someone, you hurt someone else . not least the insurgents. So civil and humanitarian assistance personnel will be targeted.”

Source: “Twenty-Eight Articles: Fundamentals of Company-level Counterinsurgency” (2006)Lt. Col. David Kilcullen, Ph.d. in Political Anthropology, Australian advising the Pentagon

77

DoD Policy Shifts Towards Social Science

• “Irregular warfare is about people, not platforms. IW depends not just on our military prowess, but also our understanding of such social dynamics as tribal politics, social networks, religious influences, and cultural mores. People, not platforms, and advanced technologies will be the key to IW success.”

Department of Defence, Irregular Warfare Joint Operating Concept, September 2007.

78

Nonstandard Development Economics

• Not designed to maximize growth or social welfare, but to undermine rebels

• Targeted at likely defectors and likely sources of intelligence

• Focus on programs that compete with services offered by clubs

• Benign programs will be targeted by rebels• Standard development programs can be captured by rebels

e.g., -“heros village”, LTTE controlled Sri Lanka- Sadr city garbage cleanup

Increased welfare, but reinforced rebels• Includes political development

79

5. Conclusion

• A “rational choice” economic model can explain the behavior of violent religious radicals- it succeeds on testable implications where the conventional wisdom about theological motivation fails

• That’s a relief: it provides benign options for dealing with violent religious radicals

• Those options are practical but are poorly understood and current implementation is awful

80

Conclusions• Why so few terrorist organizations?

Defection constraint.• Why are religious radicals effective terrorists?

Solve defection constraint in benign activities.Not necessary theology.



• Future WorkNeed research and evaluation

Distinguish club model from standard “rational peasant”approach to “winning hearts and minds”

81

Is it Ideology?

• Well.. Ideological Shifts: • Taliban: from personal piety, local Islamic govt.

to international Jihad• Hamas: from personal piety, local Islamic govt. to

nationalist territorial struggle• Jewish Underground: from glorifying the state to

undermining it’s authority

82

Rational Choice Matters

• Can this fit in a rational choice model? - is that model helpful in predicting behavior?

• Policy implication: - can the shift to militia activity be reversed?- does behavior respond to incentives?- what could we recommend if it did not?

83

ReferencesBerman, Eli “Sect, Subsidy and Sacrifice: An Economist’s

View of Ultra-Orthodox Jews,” Quarterly Journal of Economics, 115(3) (August, 2000).

__________ “Hamas, Taliban and the Jewish Underground: An Economist’s View of Radical Religious Militias,”NBER WP 10004, (October 2003).

__________ and David Laitin, “Rational Martyrs: International Evidence on Suicide Attacks,” UC San Diego mimeo, (October 2003).

__________ and Ara Stepanyan, “How Many Radical Islamists? Evidence from Asia and Africa.” UCSD mimeo, 2003.

Iannaccone, Laurence R. “Sacrifice and Stigma: Reducing Free-riding in Cults, Communes, and Other Collectives,” Journal of Political Economy, C(1992), 271-291.

84

VI. Where Research can HelpLessons from Pacific Special Operations Command

1. Development and counterinsurgency in poorly governed spaces - now a Nat. Security concern- directly + indirectly through allies

2. USAID alone spending at ~$4B annually on this development effort, DOD spending more

3. USAID and DOD lack capability to do economic and political development in dangerous spaces. World Bank and NGOs not much better.Specifically, they lack

a) basic research on counterinsurgency & developmentb) a way to evaluate their development efforts

4 . Need: research and evaluation to guide development, governance and political violence- what DOD calls soc. sci research is mostly purchased validation of what they think is true already.

- like Great Society project, which Moynihan fixed- solution is intellectually independent research:

own core funding, university based

85

Nuclear Terrorism

• Defection constraints indicate that nuclear terrorism is very unlikely to come from most of the terrorists we see today

• Look for combination of state-backed expertise and terrorism: - Iran & Hezbollah, Pakistan (ISI) & Islamists

5. Streetlights and ViolenceEli Berman UCSD

(joint research with Jacob Shapiro PrincetonJoe Felter CTC West Point)

Results are Preliminary – Not for circulation

87

Outline

A. Two models that link public goods to violence• Clubs • Rational Peasants, Counterinsurgency &

Community Policing

B. Street Lights: Public goods and violence in Iraq

88

5.1 Clubs and Terrorism

2

N

1

Conspirators

Target

Defection – The reason so few rebel and terrorist organizationssurvive.. defection is common.

How do successful rebel and terrorist organizations prevent defection?

They have an organizational advantage.

89

Clubs and Terrorism

C Ri C Rii

N

({ }) ( )==

∏1

Operation initiated only if all Ri = 1.

Payoffs: Club extracts surplus B and shares it equally among members, who buy goods at price P.

Benign local public goods provided by govt., G and club, A.Defector’s outside option is wi, but no access to C or A.

Incentive compatibility: Member loyal iff

(ICC) U( B

N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)

90

Clubs: Selecting Low Wage Membership Allows Larger Projects

Project Value (B)

High wagedefector

Low wagedefector

Loyalmember

B*B* B**


91

Clubs and TerrorismC Ri C Ri

i

N

({ }) ( )==

∏1

If ICC fails this is an N player prisoner’s dilemma, resulting in an unsafe route, no convoy and no rents.

Adverse selection: Imagine two unobserved types (as above), such that

ICC holds if wi = wH (> wL) for all i. A club with a costly sacrifice as an initiation rite which successfully excludes all low wage types can keep members loyal in larger projects. So religious radicals who operate mutual aid clubs have an advantage.

(ICC) U( B

N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)

92

Govt. Provision of Public Goods Reduces Militia’s Potential

(ICC) U( BN

,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)

Project Value (B)

Defector - high govt services

Loyal member -high govt. services

B** B*

Defector - low govt. servicesLoyal member -low

govt. services

Incentivecompatibility limit -high govt services

E.g., Malaya, Phillipines, Egypt

93

The Kilcullen Approach

“23. Practise armed civil affairs. Counterinsurgency is armed social work; an attempt to redress basic social and political problems while being shot at. This makes civil affairs a central counterinsurgency activity, not an afterthought.”

“You need intimate cooperation with inter-agency partners here, national, international and local. You will not be able to control these partners . Many NGOs, for example, do not want to be too closely associated with you because they need to preserve their perceived neutrality.”

“Thus, there is no such thing as impartial humanitarian assistance or civil affairs in counterinsurgency. Every time you help someone, you hurt someone else . not least the insurgents. So civil and humanitarian assistance personnel will be targeted.”

Source: “Twenty-Eight Articles: Fundamentals of Company-level Counterinsurgency” (2006)Lt. Col. David Kilcullen, Ph.d. in Political Anthropology, Australian advising the Pentagon

94

5.2 “Rational Peasant” ModelMotivation – “Hearts and Minds”

• Operating procedure of US and Allied Special Forces includes providing local public goods:

1. Control some territory2. Ask population what services they want3. Provide them4. Ask population for information about insurgents5. Use information to ambush or capture insurgents, allowing control of more

territory6. Repeat (1)-(5) until entire country is controlled

Note:a) This is less prosaic than “hearts and minds”b) Even the most disenfranchised population gets servicesc) “Rational Peasant” is due to Popkin

95

Akerloff-Yellen (94) Gang ModelMotivation

• Gangs are limited by community “norms” of behaviore.g., if the parents complain that a member is selling drugs in a primary school, the member is disciplined by the gang“…now I understand that if you ain’t got the community with you it’s just a matter of time before you got to close up shop.”- Duck (the gang member who learned his lesson)

96

A-Y model with public goods(and otherwise stripped down)

• Community: representative agent chooses whether to snitch to police: 0 = c = 1

• Gang: chooses level of crime: s = 0• Police: monitor criminals, m, and provide public

goods, g• Police move first, then gangs, then community

97

CommunityUtility from cooperating with police

Uc = (Bcsg - ac R) cwhere g = 0 is the level of govt. provided public goods, which complement public safety, and R is the expected value of gang retaliation, a constant.

It’s linear, so at the corner solution,c=0 iff ac R = Bcsg

“noncooperation constraint” ;

otherwise c=1

98

Gang

• Maximizes Ug = (Bg – Agmc) swhere Bg and Ag are positive constants,m is enforcement effort set by police, s is crime.

If Bg – Agmc = 0 when c=1, then gang will choose s so that the noncooperation constraint just binds, at s* = ac R / Bcg , so s > 0 and s decreases in g .

If Bg – Agmc > 0 when c=1, then gang chooses infinite s ..police will set m high enough to avoid that.

99

Police

• Minimize Ap s + Bp m + Cp g

.. which yields a corner solution for m = Bg / (Ag (avoiding infinite crime at c=1),

and an interior solution for g2 = (Ap ac R)/(BcCm).

Note: reducing cost of public goods Cm raises g & reduces s That could include reducing corruption in public good provision, or increasing the efficiency of taxation.

100

Back to Rational Peasant

Analogy to Rational Peasant model of insurgency, where even the disenfranchised noncombatants are favored with public goods by Special Forces.

Extensions• s and g need not be complements• Enrich g so it differs from a transfer?• If gangs can provide public goods can there be a race to the

top in public good provision?• If government can retaliate can there be a race to the

bottom in extortion?

101

Clubs or Rational Peasants?

• The two models share defection or snitching (cooperation with authorities)

• In the club model a combatant defects, while in the rational peasant model a noncombatant snitches

• Who cares which model is relevant? - predicting tactic choice by rebels:

- conventional tactics are cheaper, but share information with noncombatants (ambush, IED) while modern (club) tactics do not (suicide attack)- most organizations are not capable of club tactics because of defection constraints- targeting of police efforts- only club model relevant for domestic terrorism

102

Contents

1. Street Lights: Public services and violence2. What can we learn from CERP?3. Operational Initiative in the Philippines4. Other potential projects5. Research on Conflict: Building institutions6. Training: building in house analytical capability7. Theory and Doctrine development – upcoming

conferences8. Joint implementation plan

103

Why Statistics Can Help

6. SIGACTS (Significant Actions) in Iraq

104

6.1 Street Lights – a public good

Street lights are pretty durable, but some places improved and some got worse- These are measured on a scale of 1-bad to 3-good in a 2004 survey.

Al-Ka'im

Ana

Falluja

Haditha

Heet

Ramadi

Al-Mahawil

Al-Musayab

HashimiyaHillaAbu Ghraib

Adhamiya

Al Resafa

Al Sadr

Karkh

Khadamiya

Mada'in

Mahmoudiya

Tarmia

Abu Al-KhaseebAl-MidainaAl-Qurna

Al-Zubair

BasrahFao

Shatt Al-Arab

Amedi

Dahuk

Sumel

Zakho

Al-KhalisAl-Muqdadiya

Ba'qubaBaladroozKhanaqinKifriChoman

Erbil

KoisnjaqMakhmurMergasur

ShaqlawaSoran

Ain Al-Tamur

Al-Hindiya

Kerbala

Al-KahlaAl-Maimouna

Al-Mejar Al-Kabi

Ali Al-Gharbi

AmaraQal'at Saleh

Al-KhidhirAl-RumaithaAl-Salman

Al-Samawa

Al-ManatheraKufa

Najaf

Akre

Al-Hamdaniya

Al-Shikhan

Mosul

SinjarTelafar

Tilkaif

AfaqAl-Shamiya

Diwaniya

Hamza Al-Daur

Al-Shirqat

Baiji

BaladSamarra

Tikrit

Tooz

Chamchamal

Darbandihkan

DokanHalabja

KalarPenjwinPshdarRaniaSharbazher

SulaymaniyaAl-HawigaDaquq

Kirkuk

Al-Chibayish

Al-Rifa'i

Al-Shatra

NassriyaSuq Al-Shoyokh

Al-Hai

Al-Na'maniya

Al-SuwairaBadra

Kut

Al-Ka'im

Ana

Falluja

Haditha

Heet

Ramadi

Al-Mahawil

Al-Musayab


Adhamiya

Al Resafa

Al Sadr

Karkh

Khadamiya

Mada'in

Mahmoudiya

Tarmia


Al-Zubair

BasrahFao

Shatt Al-Arab

Amedi

Dahuk

Sumel

Zakho



Erbil


ShaqlawaSoran

Ain Al-Tamur

Al-Hindiya

Kerbala

Al-KahlaAl-Maimouna

Al-Mejar Al-Kabi

Ali Al-Gharbi

AmaraQal'at Saleh


Al-Samawa

Al-ManatheraKufa

Najaf

Akre

Al-Hamdaniya

Al-Shikhan

Mosul

SinjarTelafar

Tilkaif

AfaqAl-Shamiya

Diwaniya

Hamza Al-Daur

Al-Shirqat

Baiji

BaladSamarra

Tikrit

Tooz

Chamchamal

Darbandihkan

DokanHalabja



Kirkuk

Al-Chibayish

Al-Rifa'i

Al-Shatra


Al-Hai

Al-Na'maniya

Al-SuwairaBadra

Kut

Al-Ka'im

Ana

Falluja

Haditha

Heet

Ramadi

Al-Mahawil

Al-Musayab


Adhamiya

Al Resafa

Al Sadr

Karkh

Khadamiya

Mada'in

Mahmoudiya

Tarmia


Al-Zubair

BasrahFao

Shatt Al-Arab

Amedi

Dahuk

Sumel

Zakho



Erbil


ShaqlawaSoran

Ain Al-Tamur

Al-Hindiya

Kerbala

Al-KahlaAl-Maimouna

Al-Mejar Al-Kabi

Ali Al-Gharbi

AmaraQal'at Saleh


Al-Samawa

Al-ManatheraKufa

Najaf

Akre

Al-Hamdaniya

Al-Shikhan

Mosul

SinjarTelafar

Tilkaif

AfaqAl-Shamiya

Diwaniya

Hamza Al-Daur

Al-Shirqat

Baiji

BaladSamarra

Tikrit

Tooz

Chamchamal

Darbandihkan

DokanHalabja



Kirkuk

Al-Chibayish

Al-Rifa'i

Al-Shatra


Al-Hai

Al-Na'maniya

Al-SuwairaBadra

Kut

Al-Ka'im

Ana

Falluja

Haditha

Heet

Ramadi

Al-Mahawil

Al-Musayab


Adhamiya

Al Resafa

Al Sadr

Karkh

Khadamiya

Mada'in

Mahmoudiya

Tarmia


Al-Zubair

BasrahFao

Shatt Al-Arab

Amedi

Dahuk

Sumel

Zakho



Erbil


ShaqlawaSoran

Ain Al-Tamur

Al-Hindiya

Kerbala

Al-KahlaAl-Maimouna

Al-Mejar Al-Kabi

Ali Al-Gharbi

AmaraQal'at Saleh


Al-Samawa

Al-ManatheraKufa

Najaf

Akre

Al-Hamdaniya

Al-Shikhan

Mosul

SinjarTelafar

Tilkaif

AfaqAl-Shamiya

Diwaniya

Hamza Al-Daur

Al-Shirqat

Baiji

BaladSamarra

Tikrit

Tooz

Chamchamal

Darbandihkan

DokanHalabja



Kirkuk

Al-Chibayish

Al-Rifa'i

Al-Shatra


Al-Hai

Al-Na'maniya

Al-SuwairaBadra

Kut

11.

52

2.5

Stre

et li

ghts

04

1 1.5 2 2.5 3Street lights 02

Street lights 04 Street lights 04

105

Who improved? st

reet

ligh

t 04

street light in Dec 20021 1.5 2

1

1.5

2

Abu Al-K

Abu Ghra

Adhamiya

Afaq Ain Al-T

Akre

Al Resaf

Al Sadr

Al-Daur

Al-Hai

Al-Hamda

Al-Hawig

Al-Hindi

Al-Ka'im

Al-Kahla

Al-KhaliAl-Khidh

Al-Mahaw

Al-Maimo

Al-Manat

Al-Mejar

Al-Midai

Al-Muqda

Al-Musay

Al-Na'ma

Al-Qurna

Al-Rifa'

Al-RumaiAl-Salma

Al-Samaw

Al-Shami

Al-Shatr

Al-ShikhAl-Shirq

Al-SuwaiAl-ZubaiAli Al-G

Amara

Amedi

Ba'quba

BadraBaiji

BaladBaladroo

Basrah

ChamchamChoman

Dahuk

Daquq

DarbandiDiwaniya

Dokan

Erbil

Falluja

Fao

Haditha

Halabja

Hamza

HashimiyHeet

Hilla Kalar

Karkh

KerbalaKhadamiy

KhanaqinKifri

Kirkuk

Koisnjaq

Kufa

Kut

Mada'in

Mahmoudi

MakhmurMergasur

Mosul

Najaf

Nassriya

Penjwin

Pshdar

Qal'at S

Ramadi

RaniaSamarra

Shaqlawa

Sharbazh

Shatt AlShekhan

Sinjar

SoranSulayman

Sumel

Suq Al-S

Tarmia

Telafar

Tikrit

TilkaifTooz

Zakho

Abu Al-K

Abu Ghra

Adhamiya

Afaq Ain Al-T

Akre

Al Resaf

Al Sadr

Al-Daur

Al-Hai

Al-Hamda

Al-Hawig

Al-Hindi

Al-Ka'im

Al-Kahla

Al-KhaliAl-Khidh

Al-Mahaw

Al-Maimo

Al-Manat

Al-Mejar

Al-Midai

Al-Muqda

Al-Musay

Al-Na'ma

Al-Qurna

Al-Rifa'

Al-RumaiAl-Salma

Al-Samaw

Al-Shami

Al-Shatr

Al-ShikhAl-Shirq


Amara

Amedi

Ba'quba

BadraBaiji

BaladBaladroo

Basrah

ChamchamChoman

Dahuk

Daquq

DarbandiDiwaniya

Dokan

Erbil

Falluja

Fao

Haditha

Halabja

Hamza

HashimiyHeet

Hilla Kalar

Karkh

KerbalaKhadamiy

KhanaqinKifri

Kirkuk

Koisnjaq

Kufa

Kut

Mada'in

Mahmoudi

MakhmurMergasur

Mosul

Najaf

Nassriya

Penjwin

Pshdar

Qal'at S

Ramadi

RaniaSamarra

Shaqlawa

Sharbazh

Shatt AlShekhan

Sinjar

SoranSulayman

Sumel

Suq Al-S

Tarmia

Telafar

Tikrit

TilkaifTooz

Zakho

Abu Al-K

Abu Ghra

Adhamiya

Afaq Ain Al-T

Akre

Al Resaf

Al Sadr

Al-Daur

Al-Hai

Al-Hamda

Al-Hawig

Al-Hindi

Al-Ka'im

Al-Kahla

Al-KhaliAl-Khidh

Al-Mahaw

Al-Maimo

Al-Manat

Al-Mejar

Al-Midai

Al-Muqda

Al-Musay

Al-Na'ma

Al-Qurna

Al-Rifa'

Al-RumaiAl-Salma

Al-Samaw

Al-Shami

Al-Shatr

Al-ShikhAl-Shirq


Amara

Amedi

Ba'quba

BadraBaiji

BaladBaladroo

Basrah

ChamchamChoman

Dahuk

Daquq

DarbandiDiwaniya

Dokan

Erbil

Falluja

Fao

Haditha

Halabja

Hamza

HashimiyHeet

Hilla Kalar

Karkh

KerbalaKhadamiy

KhanaqinKifri

Kirkuk

Koisnjaq

Kufa

Kut

Mada'in

Mahmoudi

MakhmurMergasur

Mosul

Najaf

Nassriya

Penjwin

Pshdar

Qal'at S

Ramadi

RaniaSamarra

Shaqlawa

Sharbazh

Shatt AlShekhan

Sinjar

SoranSulayman

Sumel

Suq Al-S

Tarmia

Telafar

Tikrit

TilkaifTooz

Zakho

Abu Al-K

Abu Ghra

Adhamiya

Afaq Ain Al-T

Akre

Al Resaf

Al Sadr

Al-Daur

Al-Hai

Al-Hamda

Al-Hawig

Al-Hindi

Al-Ka'im

Al-Kahla

Al-KhaliAl-Khidh

Al-Mahaw

Al-Maimo

Al-Manat

Al-Mejar

Al-Midai

Al-Muqda

Al-Musay

Al-Na'ma

Al-Qurna

Al-Rifa'

Al-RumaiAl-Salma

Al-Samaw

Al-Shami

Al-Shatr

Al-ShikhAl-Shirq


Amara

Amedi

Ba'quba

BadraBaiji

BaladBaladroo

Basrah

ChamchamChoman

Dahuk

Daquq

DarbandiDiwaniya

Dokan

Erbil

Falluja

Fao

Haditha

Halabja

Hamza

HashimiyHeet

Hilla Kalar

Karkh

KerbalaKhadamiy

KhanaqinKifri

Kirkuk

Koisnjaq

Kufa

Kut

Mada'in

Mahmoudi

MakhmurMergasur

Mosul

Najaf

Nassriya

Penjwin

Pshdar

Qal'at S

Ramadi

RaniaSamarra

Shaqlawa

Sharbazh

Shatt AlShekhan

Sinjar

SoranSulayman

Sumel

Suq Al-S

Tarmia

Telafar

Tikrit

TilkaifTooz

Zakho

106

Street lights by themselves don’t predict much

Mean of Sig. Acts in ‘04 is 0.60 per 1000 per yearRegression slope is - 0.19 (not stat. significant)

Al-Ka'im

Ana

FallujaHadithaHeet

RamadiAl-Mahawil

Al-MusayabHashimiyaHilla

Abu Ghraib

Adhamiya Al ResafaAl Sadr

KarkhKhadamiyaMada'in

MahmoudiyaTarmia

Abu Al-KhaseebAl-MidainaAl-Qurna Al-Zubair BasrahFaoShatt Al-Arab Amedi DahukSumelZakhoAl-Khalis

Al-Muqdadiya

Ba'quba

BaladroozKhanaqinKifri

Choman ErbilKoisnjaqMakhmurMergasurShaqlawaSoranAin Al-TamurAl-Hindiya KerbalaAl-Kahla

Al-Maimouna

Al-Mejar Al-KabiAli Al-GharbiAmara

Qal'at SalehAl-KhidhirAl-RumaithaAl-Salman

Al-SamawaAl-ManatheraKufa NajafAkre

Al-Hamdaniya

Al-Shikhan

Mosul

SinjarTelafar

TilkaifAfaqAl-ShamiyaDiwaniyaHamza

Al-Daur

Al-Shirqat

Baiji

Balad

Samarra

Tikrit

ToozChamchamalDarbandihkanDokanHalabjaKalarPenjwinPshdarRaniaSharbazherSulaymaniya

Al-HawigaDaquq Kirkuk Al-ChibayishAl-Rifa'iAl-Shatra NassriyaSuq Al-ShoyokhAl-HaiAl-Na'maniya

Al-SuwairaBadra Kut0

510

15

1 1.5 2 2.5Street lights 04

Incident per 1000 persons Fitted values

107

Multivariate regression: 2004

• All statistically significant, F=• - conditional on service provision after invasion, good services under Saddam predict violence vs.

coalition forces- conditional on service provision under Saddam, post-invasion service provision predicts less violence vs. coalition forces

05

10

15e(

p_i

nc |

X )

-.5 0 .5e( street_light | X )

coef = -2.2283984, (robust) se = .715622, t = -3.11

(conditioning on street lights 02)Incidents 04 vs Street lights 04

05

10

15e(

p_i

nc |

X )

-.5 0 .5e( street_light_02 | X )

coef = 1.95034, (robust) se = .6454984, t = 3.02


108

Street Lights and Violence: 2004, 05, 06, 07

Could it be reverse causality?

Violence could be bad for street lights.… so look at the timing:

Interpretation: Granger causality runs from lights in 02 and 04 to subsequent violence?

Tactical implication:Location of escalation was predictable based on 02 and 04 data.

010

2030

40e(

p_i

nc |

X )

-.6 -.4 -.2 0 .2 .4e( street_light | X )

coef = -2.2283984, (robust) se = .715622, t = -3.11

Incidents 2004

010

2030

40e(

p_i

nc |

X )

-.6 -.4 -.2 0 .2 .4e( street_light | X )

coef = -4.8966916, (robust) se = 1.1795619, t = -4.15

Incidents 2005

01

020

3040

e( p

_inc

| X

)

-.6 -.4 -.2 0 .2 .4e( street_light | X )

coef = -9.4834806, (robust) se = 2.3861742, t = -3.97

Incidents 2006

01

020

3040

e( p

_inc

| X

)

-.6 -.4 -.2 0 .2 .4e( street_light | X )

coef = -10.599431, (robust) se = 2.8844079, t = -3.67

Incidents 2007

(conditioning on street lights 02)Incidents vs Street Lights 04

109

Street Lights and Violence: 2004, 05, 06, 07

OLS results for Incidents per 1000 persons, 2004-2007Dependent variable:Incident per 1000 persons 2004 2005 2006 2007Street lights -2.23 -4.90 -9.48 -10.6

(0.72) (1.18) (2.39) (2.88)Street lights 02 1.95 3.56 7.56 8.70

(0.64) (1.08) (2.24) (2.81)Constant 0.98 2.89 4.89 5.07

(0.55) (1.12) (2.28) (2.14)

R squared 0.05 0.05 0.06 0.06Sample size 100 100 100 100Average incident per 1000 persons 0.63 1.13 2.38 2.60

110

Is it literally Street lights?

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

Tot

al N

umbe

r of

Inci

dent

s

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Hour of the day

Mar 2006 - Dec 2007

Total Number of Incidents

111

Is it literally Street lights?OLS results for Day vs Night Incidents per 1000 persons, 2006-2007Dependent variable:

2006 2007 2006 2007Street lights -6.00 -7.71 -6.43 -6.50

(1.44) (2.07) (2.33) (2.68)Street lights X night 3.43 4.96 -0.09 0.23

(1.64) (2.24) (3.71) (4.13)Street lights 02 5.00 6.54 6.55 7.08

(1.42) (2.06) (1.81) (2.08)Street lights 02 X night -3.05 -4.49 -0.58 -1.14

(1.57) (2.20) (2.89) (3.17)Night (= 1) -1.24 -1.64 0.29 0.59

(1.52) (1.62) (1.88) (2.25)Constant 2.75 3.32 -1.43 -1.97

(1.36) (1.48) (1.30) (1.53)

R squared 0.09 0.10 0.19 0.19Sample size 200 200 170 176Average day-time incident per 1000 persons 1.46 1.81Average night-time incident per 1000 persons 0.68 0.76

Incident per 1000 persons Log of incident per 1000 persons

112

Is it literally Street lights?

Similar results if public garbage collection is used instead of 04 street lights

Similar results if road quality is used instead of 02 street lights

Implication: It looks likePast (current) public service provision predicts (reduced) violence

05

1015

e( p

_inc

| X

)

-.6 -.4 -.2 0 .2 .4e( street_light | X )

coef = -2.2283984, (robust) se = .715622, t = -3.11


05

1015

e( p

_inc

| X

)

-.5 0 .5 1e( pub_garbage | X )

coef = -.47651547, (robust) se = .30347421, t = -1.57

(conditioning on street lights 02)Incidents 04 vs Garbage collection 04

05

1015

e( p

_inc

| X

)

-.4 -.2 0 .2 .4 .6e( street_light_02 | X )

coef = 1.95034, (robust) se = .6454984, t = 3.02


05

1015

e( p

_inc

| X

)

-1 -.5 0 .5 1 1.5e( road_qual | X )

coef = .45150932, (robust) se = .12967818, t = 3.48

(conditioning on street lights 04)Incidents 04 vs Road quality 04

113

Recap: Data supports both club and rational peasant models

The next steps: • Models have an additional prediction about tactics and

geography: - In areas where noncombatants do not cooperate with government we expect conventional attacks, - while in areas where noncombatants cooperate, suicide attacks and other tactics which do not share information with noncombatants are required.

114

Where did these data come from?

• Capable of pulling classified data from military sources, declassifying it, and making it available for civilian researchers.

• In this case, the SIGACTs and survey data come from the US Central Command- don’t ask for these two datasets until we’ve examined, cleaned and written a paper

www.ctc.usma.edu

115

What can we learn from CERP?(Commanders Emergency Response Program)

• CERP is spent on public service provision, among other things

• CERP per capita and sig. acts:• But CERP is also directed disproportionately at the most troublesome areas

Al-Ka'im Al-RutbaAnaFallujaHadithaHeetRamadiAl-MahawilAl-MusayabHashimiyaHilla

Abu Ghraib

AdhamiyaAl ResafaAl Sadr KarkhKhadamiyaMada'inMahmoudiyaTarmia

Abu Al-KhaseebAl-MidainaAl-QurnaAl-ZubairBasrahFaoShatt Al-ArabAmediDahukSumelZakhoAl-Khalis

Al-Muqdadiya

Ba'qubaBaladroozKhanaqinKifriChomanErbilKoisnjaqMakhmurMergasurShaqlawaSoranAin Al-TamurAl-HindiyaKerbalaAl-Kahla

Al-MaimounaAl-Mejar Al-KabiAli Al-GharbiAmaraQal'at SalehAl-KhidhirAl-RumaithaAl-SalmanAl-SamawaAl-ManatheraKufaNajafAkreAl-Ba'aj

Al-Hamdaniya

Al-ShikhanHatra

Mosul

ShekhanSinjarTelafarTilkaifAfaqAl-ShamiyaDiwaniyaHamza

Al-Daur

Al-ShirqatBaiji

Balad

SamarraTikrit

ToozChamchamalDarbandihkanDokanHalabjaKalarPenjwinPshdarRaniaSharbazherSulaymaniyaAl-HawigaDaquqKirkukAl-ChibayishAl-Rifa'iAl-ShatraNassriyaSuq Al-ShoyokhAl-HaiAl-Na'maniyaAl-SuwairaBadraKut

Al-Ka'im

Al-Rutba

Ana FallujaHaditha

HeetRamadi

Al-MahawilAl-MusayabHashimiyaHilla

Abu Ghraib

AdhamiyaAl ResafaAl SadrKarkhKhadamiyaMada'in

Mahmoudiya

Tarmia

Abu Al-KhaseebAl-MidainaAl-QurnaAl-ZubairBasrahFaoShatt Al-ArabAmediDahukSumelZakhoAl-Khalis

Al-Muqdadiya

Ba'qubaBaladroozKhanaqin Kifri

ChomanErbilKoisnjaqMakhmurMergasurShaqlawaSoran Ain Al-TamurAl-HindiyaKerbalaAl-KahlaAl-MaimounaAl-Mejar Al-KabiAli Al-GharbiAmaraQal'at SalehAl-KhidhirAl-Rumaitha Al-SalmanAl-SamawaAl-ManatheraKufaNajafAkreAl-Ba'aj

Al-Hamdaniya

Al-ShikhanHatra

Mosul

ShekhanSinjarTelafar


Al-Daur

Al-Shirqat

Baiji

Balad

Samarra

Tikrit

ToozChamchamalDarbandihkan DokanHalabjaKalarPenjwinPshdarRaniaSharbazherSulaymaniya

Al-Hawiga

DaquqKirkukAl-ChibayishAl-Rifa'iAl-ShatraNassriyaSuq Al-ShoyokhAl-HaiAl-Na'maniyaAl-SuwairaBadraKut

Al-Ka'im

Al-RutbaAnaFalluja

HadithaHeet

Ramadi

Al-MahawilAl-MusayabHashimiyaHilla

Abu Ghraib

AdhamiyaAl ResafaAl Sadr

KarkhKhadamiya

Mada'in

Mahmoudiya

Tarmia

Abu Al-KhaseebAl-MidainaAl-QurnaAl-ZubairBasrahFaoShatt Al-ArabAmediDahukSumelZakho

Al-Khalis

Al-Muqdadiya

Ba'quba

BaladroozKhanaqin

KifriChomanErbil KoisnjaqMakhmurMergasurShaqlawaSoranAin Al-Tamur

Al-HindiyaKerbalaAl-KahlaAl-MaimounaAl-Mejar Al-KabiAli Al-GharbiAmaraQal'at SalehAl-KhidhirAl-Rumaitha Al-SalmanAl-SamawaAl-ManatheraKufaNajaf AkreAl-Ba'aj

Al-Hamdaniya

Al-ShikhanHatra

Mosul

ShekhanSinjarTelafar


Al-Daur

Al-Shirqat

Baiji

Balad

Samarra

Tikrit


Al-Hawiga

DaquqKirkukAl-ChibayishAl-Rifa'iAl-ShatraNassriyaSuq Al-ShoyokhAl-HaiAl-Na'maniyaAl-Suwaira

BadraKutAl-Ka'im

Al-RutbaAna FallujaHaditha

HeetRamadi

Al-Mahawil

Al-MusayabHashimiyaHilla

Abu Ghraib

AdhamiyaAl ResafaAl Sadr

Karkh

Khadamiya

Mada'in

Mahmoudiya

Tarmia

Abu Al-Khaseeb

Al-MidainaAl-Qurna Al-ZubairBasrahFaoShatt Al-ArabAmediDahukSumelZakhoAl-Khalis

Al-Muqdadiya

Ba'quba

BaladroozKhanaqinKifri

ChomanErbilKoisnjaq MakhmurMergasurShaqlawaSoranAin Al-Tamur Al-HindiyaKerbalaAl-KahlaAl-MaimounaAl-Mejar Al-KabiAli Al-GharbiAmaraQal'at SalehAl-KhidhirAl-RumaithaAl-Salman

Al-SamawaAl-ManatheraKufaNajafAkreAl-Ba'aj

Al-Hamdaniya

Al-ShikhanHatra

Mosul

ShekhanSinjarTelafarTilkaifAfaqAl-Shamiya

DiwaniyaHamza

Al-Daur

Al-Shirqat

Baiji

Balad

Samarra

Tikrit


Al-Hawiga

DaquqKirkukAl-ChibayishAl-Rifa'iAl-ShatraNassriyaSuq Al-ShoyokhAl-HaiAl-Na'maniyaAl-Suwaira

BadraKut010

2030

40

0 100 200 300 400 500CERP spending per capita

Incident per 1000 persons Fitted values

116

CERP per capita and sig. acts. 06(zoom on CERP pc <100)

CERP per capita

sig acts / 1000 Linear prediction

0 50 100

0

5

10

15

20

Abu Al-K

Adhamiya

Afaq

A i n A l - T

Akre

Al ResafAl SadrAl-Ba'aj

Al-Chiba

Al-Hai

Al-Hamda

Al-Hawig

Al-Hindi

Al-Ka'im

Al-Kahla

Al-Khali

Al-Khidh

Al-Mahaw

Al-Maimo

Al-ManatAl-Mejar

Al-Midai

Al-Muqda

Al-Musay

Al-Na'maAl-Qurna Al-Rifa'Al-Rumai

Al-Rutba

Al-SamawAl-ShamiAl-Shatr

Al-Shikh

Al-ShirqAl-Suwai

Al-ZubaiA l i A l - GAmara Amedi

A n a

Ba'quba Badra

Baiji

Baladroo

Basrah

ChamchamChoman

Dahuk

Daquq

Darbandi

DiwaniyaDokan

Erbil

Falluja

Fao

Haditha

Halabja HamzaHashimiy

Hatra

HillaKalar

Karkh

Kerbala

Khadamiy

Khanaqin

Kifri

Kirkuk

Kufa Kut

Mada'in

Mahmoudi

MakhmurM e r g a s u r

Mosul

NajafNassriyaP e n j w i n

PshdarQal'at S

Ramadi

Rania

Samarra

ShaqlawaS h a r b a z h

Shatt Al

Shekhan

SinjarSoranSulayman SumelSuq Al-S

Telafar

Tikrit

Tilkaif

Tooz

Zakho

117

Can Hearts & MindsBe Bought?

• Huge investment ~ $41B • Uncertain impact• Reconstruction spending

– IRRF– CERP/CHRRP

• Community characteristics – WFP– ILCS

118

How to use CERP efficiently?

• To answer that research question we need access to more data

• In particular, we need a way to filter out the reverse causality due to CERP spending being directed at locations expected to be dangerous.

• One solution: Troop rotation data (declassified in the same way these data have been)

Documents

Religion, Terrorism and Public Goodselib/250C/religion_terrorism.pdf · 2008-04-26 · Religion, Terrorism and Public Goods . 2 Prime Minister’s Office, Algeria – 4/11/07