2001年-世界发展银行全球_On_the_Duration_of_Civil_War_34页_1mb

Summary of "On the Duration of Civil War" by Paul Collier, Anke Hoeffler, and Måns Söderbom

Core Content

This working paper investigates the factors that influence the duration of large-scale, violent civil wars, distinguishing it from the factors that influence the initiation of such conflicts. The authors analyze data from 1960 to 1999 to model how long conflicts last, using hazard functions and regression analysis. They find that the variables affecting conflict duration are largely different from those affecting conflict initiation.

Main Findings

• Duration of conflict is influenced by:

  • A few large ethnic groups
  • Extensive forest cover
  • Conflicts that began after 1980

• Initiation of conflict is not affected by:

  • Initial inequality
  • Political repression

• The initiation model (Collier and Hoeffler, 2001) includes the following significant variables:

  • Economic factors: per capita income, GDP growth, and primary commodity exports
  • Social factors: ethnic and religious diversity
  • Geographic factors: population dispersion, mountainous terrain, and forest cover
  • Historical factors: previous conflict experience

• Ethnic diversity is associated with lower conflict risk, while ethnic dominance (where the largest ethnic group constitutes between 45% and 90% of the population) is linked to higher risk.

• Forest cover increases the duration of conflict, while population dispersion also contributes to longer wars.

• Conflicts initiated after 1980 tend to last longer than those before.

• The hazard function model for duration is based on a piecewise exponential specification, allowing for stepwise estimation and controlling for unobserved heterogeneity.

• The baseline model of duration includes:

  • Economic variables: per capita income, male secondary school enrollment
  • Social variables: ethnic fractionalization, population size
  • Geographic variables: forest cover
  • Historical variables: decade of conflict onset (especially post-1980), neighboring wars

• The effect of male secondary school enrollment on conflict duration is non-linear: higher enrollment reduces the risk of conflict, but this effect diminishes over time as educated individuals may emigrate during prolonged conflicts.

• The hazard of peace is lower in conflicts initiated after 1980, which explains their longer duration.

Key Variables and Their Effects

Policy Implications

• Understanding the determinants of conflict duration is crucial for policy interventions aimed at ending wars rather than just containing them.
• The persistence of peace and conflict are both dynamic processes, with longer peace periods increasing the likelihood of continued peace.
• The role of education and economic development in reducing conflict risk is important, but these effects may be temporary due to population dynamics during prolonged conflicts.
• Geographic and social factors play a significant role in shaping the length of conflicts, highlighting the need for region-specific and culturally sensitive policies.

Conclusion

The paper concludes that the initiation and duration of civil wars are governed by different sets of variables, with the latter being influenced by ethnic composition, forest cover, and the time period in which the conflict began. These findings suggest that policy approaches to conflict resolution should focus on factors that prolong wars, such as ethnic fragmentation and natural resource availability, rather than on those that initiate them. The model of duration provides a robust framework for understanding the evolution of conflicts over time, offering insights for post-conflict stabilization and peacebuilding efforts.