When Do Citizens Support Peace-Building? Economic Hardship and Civilian Support for Rebel Reintegration
When Do Citizens Support Peace-Building? Economic Hardship and Civilian Support for Rebel Reintegration (DOI: https://doi.org/10.1017/S002081832510113) was written by Amanda Kennard, Konstantin Sonin, and Austin L. Wright in 2025. It was published in International Organization (vol. 79, no. 4).
Building off of the literature exploring how economic conditions affect rebels' willingness to re-integrate, the authors investigate how they affect civilians' willingness.
They hypothesize that civilians will anticipate greater competition in labor markets from the influx of 'new' workers. This anticipated wage depression creates a disincentive for re-integration. More formally:
Every household i splits its time between productive agricultural work (li) or household work. The latter is a form of leisure or self-sufficience such that there is a constant per-unit cost c to li.
Households also have a stock of productive capital (ki) distributed according to f.
For the supply side, assume Leontieff production as qi = min{ki ,li}.
Furthermore assume that capital is the limiting factor, i.e. l*i = ki and qi = h(ki).
It follows that aggregate production is ∫ k h(k) dk
For the demand side, assume q(p) = a - bp.
Furthermore assume that a is small.
Equilibrium price is determined by setting the two sides equal: p* = (1/b) (a - ∫ k h(k) dk).
Assume utility from the market is described by a function u where u' > 0, u'' < 0, and u''' < 0.
Plainly speaking, u is a monotone concave down function.
So initially, household utility is given by u(p(q) * h(ki)) - c * li
Now introduce a number of re-integrees m and a preference for peace as φ(m).
Final household utility is Ui = u(p(q) * h(ki)) - c * li + φ(m).
Because re-integrees have equivalent distribution of capital, aggregate production is simply scaled as (1 + m)∫ k h(k) dk. The equilibrium price can be expressed now as a function of m: p*(m) = (1/b) (a - (1 + m)∫ k h(k) dk).
Households' selection of l*i = ki is not in terms of m or p, but p* decreases as m increases. Clearly households take less revenue after re-integration.
Each household has an optimal level of re-integrees (mi) that is determined by its level of productive capital (and therefore can be notated mi(ki)).
- Furthermore, because the utility function is assumed monotone concave down, it is known that households with lower productive capital stocks have lower optimal level of re-integrees.
Finally, assume an exogenous shock to capital stocks such that f first-order stochastically dominates f '.
- In the simplest case, only one household is changed and their capital stock is reduced.
Clearly then Ef '[mi(ki)] < Ef[mi(ki)]
The authors study civilian attitudes toward re-integration of the Taliban. They leverage the 2015 Hindu Kush earthquake as an adverse exogenous shock. They use data from the Afghanistan Nationwide Quarterly Research (ANQAR) survey.
- waves 29-30, fielded August and November 2015.
PSUs are administrative districts selected by PPS sampling.
- "Like Condra and Wright, we rectify ACSOR’s sampling frame using the administrative map produced by the Empirical Studies of Conflict group."
- SSUs are villages/settlements randomly selected.
- In-person interviews; interviewers were thoroughly trained, used random walks and Kish grids to select respondents in the field.
- Distance from the earthquake is the treatment. More specifically, being within 300km of the epicenter.
- Measurements of support for peace: “If an insurgent were to stop fighting against the government, would you accept him back into the community if he came back as a farmer?"
- Authors also use similar measures as placebos (overall support, support in unaffected sectors (like police officers), perception of security conditions).
The authors leverage a DID design to estimate causal effect. For the primary measure, they find a significant negative effect. For the placebos, there is no significant and robust effect.
- Control for ethnicity, gender, socioeconomic status, age, and educational attainment.
Clustered standard errors on district (PSU).
- The authors then test on the interaction of treatment and being in an agriculture-dependent district; they again find a significant negative effect.
Reading notes
A master class on formalizing the statement 'poor people have a higher marginal product of capital'. Something which practitioners of other fields would simply state as an assumption, these authors have turned into three pages of content. Beautiful.