Religious Extremism, Institutions, and Income: Theory and Evidence ESHIA June,2007 Michael Makowsky H.B. Earhart Fellow Center for the Economic Study of Religion Department of Economics George Mason University
Extremism, Religious or Otherwise
For better or worse, religious extremism is associated with violence
Are there social and economic conditions under which extremist groups thrive or fail?
Can we model this emergence of viable extremist groups within a population?
Can we gain insight into peculiar characteristics of violent extremist groups
“ Approximately 90 percent of [Hamas’s] work is in social, welfare, cultural and educational activities. These are important elements of Hamas's popularity that keep it closely tied to the public.” - Reuven Paz(2001)
Extremist Groups and Violence
Extreme = High Sacrifice
Extreme Sacrifice Groups are not violent by nature (the heavy majority are entirely peaceful)
But…extremist groups possess the characteristic group loyalty and commitment necessary for the recruitment and training of martyrs (Iannaccone IJRR 2005)
Why Can Groups Demand Sacrifice?
Prisoner’s Dilemmas, Free-Riders, Principal-Agent problems – lots of reason’s why large groups should fail (Olson 1965)
Religious/Ideological groups manage to survive without contracts, high-cost monitoring, or wage incentives
Solution: Sacrifice and Stigma (Iannaccone JPE 1992)
Sacrifice And Stigma Model
Individuals use time and money to produce Private/Secular (S) and Group/Religious (K) goods
Religious groups prevent free-riding by Sacrifice & Stigma (S&S)
Effectiveness of S&S depends on relative prices (via income) and the substitutability of S for K
Quick Version High Beta= Happy with one or the other Low Beta= Need a combination of both
The Functions (key on social interactions)
An Agent-Based Model of a Religious Economy (Built with the lovely MASON java library)
Sacrifice and Stigma – single group
Agent-based extension - macroscopic level, winners and losers
Dynamic, interactive, rules-based agents operating in a spatial and temporal world
Max( U )
MARS – S&S
Agents color- and shape-coded by group
Groups identifiable along a sacrifice spectrum from 0 to
100 % of secular productivity loss
Lognormal Income Distribution
Extremism as a Percent of the Population Income Substitutability
Empirical Connections – Krueger and Maleckova 2003
Terrorists from a country are decreasing in number as that nation’s civil liberties, as measured by Freedom House Civil Liberties Index, are increasing.
When Civil Liberties are controlled for, the relative wealth of the country and its illiteracy rate have no statistically significant effect.
Regression Analysis of the Model Quantile Regression Estimates: Fraction Log Full Income Dedicated to Club Production Standard errors in parentheses, all coefficients significant at 0.1% level 52272 52272 52272 52272 52272 N (0.014) (0.032) (0.014) (0.022) (0.001) 1.519 1.173 -2.543 -4.959 -9.12 Constant (0.013) (0.032) (0.017) (0.040) (0.002) 1.928 0.184 -3.539 -5.940 -8.478 Log Substitutability (β) (0.003) (0.007) (0.004) (0.007) (0.000) -0.519 -0.961 -0.817 -0.828 -0.250 Log Full Income 90% 75% 50% 25% 10% Quantile
A public good competitive with the club good reduces the appeal of extremism
The popular shift is towards the strict groups, and not the most liberal (secularists)
BUT there is no impact on the sacrifice profile of a population unless the public good is subject to group sacrifice requirements
This is potentially a good thing – sacrifice of the public good is most likely to be realized by groups operating outside the law.
Substitutability is necessary and quasi-sufficient for Extremist groups to be sustainable. Population Income matters, but is dominated by Subsitutability.
Extremist groups have incentive to package themselves as strong substitutes for privately productive activities
This correlates to Hamas and Hezbollah dedicating the bulk of their resources to social services
Public Goods competitive with the club good are a potentially viable mechanism for reducing the appeal of extremism
The Rest are just Appendices
Test Variables – Beta and Mean Income
1 (prices of good) p S, p R 60 (number of Groups) G 0.3 b 0.7 a 0.3 α 1.25 s 1 b s , b k Value Related Function Parameter
Nash starting condition
iterated pure strategy interaction between an agent i and a generic neighbor j .
is assumed to be common knowledge, as is the rationality of both players. Player turns are executed sequentially (as opposed to simultaneously) with t incrementing by one when an agent makes a calculation. The starting value is a parameter value assigned to avoid indefinite solutions.