Provoking insurgency in a federal state: theory and application to India

Abstract

This paper presents a model of provocation in a federation, wherein the state government triggers an insurgency with a view toward acquiring control of some economic assets with the help of the central government. Some econometric support for this model is found using data on the Naxalite conflict in eight states of India. The tests performed control for endogeneity of the state government’s police force interventions. They suggest that the latter are meant to trigger the violent activity of the insurgents, so as to lure the central government into intervening and helping clear the ground for mining purposes in the lands of tribal people.

Appendices

Appendix 1

1.1 Proof of Proposition 1

A glance at (6) shows that the potential rebels minimize their loss by choosing $R^{*} = 0$ if $A = 0$, because the constitutional constraint deters the central government from intervening in this case, yielding $L_{T} ^{*} = 0$. Looking again at (6) shows that setting $R = 0$ when $A > 0$ would not minimize the potential rebels’ loss as this would make $L_{T} \to \infty$.

Now, in order to prove that $R^{*}$ is an increasing function of $A{\text{ and }}\mu$ as soon as A is strictly positive, we can compute the first-order condition for problem (6), taking due account of (5), which yields:

$$ \gamma \,R^{2 + \xi } - (1 + \xi \,)Bh\,\left( {\mu ,\,A} \right) - \xi R\,(A^{\xi } + \alpha B) = 0 $$

(12)

where $B \equiv \delta [(A + M^{*} )^{\xi } - A^{\zeta } ] > 0$ For the sake of saving notation, define:

$$ \begin{aligned} E \equiv \delta [(A + M^{*} )^{\xi - 1} - A^{\xi - 1} ]\, > 0, \hfill \\ D \equiv \,A^{\xi - 1} > 0. \hfill \\ \end{aligned} $$

Then, taking the total differential of (12) and rearranging the terms yields:

$$ \frac{d\,R^{*}}{d\,A} = \frac{{\xi (1 + \xi )h(\mu ,A)E + \delta \xi (A + M^{*} )^{\xi - 1} [(1 + \xi )h(\mu ,A) + \xi \alpha R]\partial M/\partial A + B\left( {1 + \xi } \right)\,\partial h/\partial A + \xi^{2} R(D + \alpha E)}}{{\left( {1 + \xi } \right)\,[\gamma \,R^{1 + \xi } + {{B\,h\,\left( {\mu ,\;A} \right)} \mathord{\left/ {\vphantom {{B\,h\,\left( {\mu ,\;A} \right)} {R]}}} \right. \kern-0pt} {R]}}}} $$

(13)

and

$$ \frac{d\,R^{*}}{d\,\mu } = \frac{{B\,{{\left( {1 + \xi } \right)\,\partial \,h\,} \mathord{\left/ {\vphantom {{\left( {1 + \xi } \right)\,\partial \,h\,} {\partial \,\mu }}} \right. \kern-0pt} {\partial \,\mu }} + \xi \,\delta (A + M^{*} )^{\xi - 1} [(1 + \xi )\,h(\mu ,A) + \xi \alpha R]{{\,\partial M} \mathord{\left/ {\vphantom {{\partial M} {\partial \mu }}} \right. \kern-0pt} {\partial \mu}}}}{{\left( {1 + \xi } \right)\,[\gamma \,R^{1 + \xi } + {{B\,h\,\left( {\mu ,\;A} \right)} \mathord{\left/ {\vphantom {{B\,h\,\left( {\mu ,\;A} \right)} {R]}}} \right. \kern-0pt} {R]}}}}. $$

(14)

It is easily checked that these two derivatives are positive.

1.2 Proof of Proposition 2

If $R = 0$, the state government’s objective function reduces to $\lambda \,A - x\,\left( A \right)$, which is strictly increasing in the neighbourhood of A = 0 if $\lambda > \theta$. It takes some simple calculations to show that conditions (8) and (9) ensure that $p\,\left( {\mu ,\;A,\;R} \right)\,\,M\,\left( {\mu ,\;A} \right)$ is strictly concave in A for large enough values of μ. This property is used in the graphical analysis of Fig. 3.

Appendix 2: construction of the conflict variables from SATP

The data on killings by rebels, state forces and central forces were constructed by the authors from the SATP website. Below are 5 examples of daily summaries from the SATP website.

Case 1

The CPI-Maoist cadres killed a civilian, identified as Sodhi Darra, in the Chintalnar Police station area of Dantewada District. A pamphlet recovered from the incident site allegedly branded the victim as a Police informer.

Case 2

CPI-Maoist cadres killed two special Police officers in the Narayanpur District.

Case 3

Two women cadres of the CPI-Maoist were killed and another was arrested by the Police during an encounter at Badgai village in the Bastar District. Police also recovered an unspecified number of arms and ammunition from the encounter site.

Case 4

Three CPI-Maoist cadres were killed in separate encounters with Security Forces in the Bastar region.

Case 5

Two women of the CPI-Maoist were killed in an encounter when a joint party of the CRPF and the Police neutralized a Maoist hideout during a search operation in the Narayanpur District. An unspecified number of arms and ammunition were also recovered from the hideout.

Coding killings by rebels is simple. Both Cases 1 and 2 are coded as violence by Maoists. Police intervention is coded as state violence and fatal attacks by central Special Forces are coded as central violence.³⁴ In Case 3, the 2 Maoist deaths are coded as state violence. In Case 4, the 3 killings are attributed to central government. Case 5 is an example of joint operation between the police and central security forces. Here, we divide the number of Maoist killed equally between the state and central forces (if nothing has been specified).