Daniel Luo, Northwestern University
This paper studies “two-stage” perfect information dynamic bargaining. In the first stage, players cannot split surplus, but only agree whether or not to proceed to the second stage, where canonical Rubinstein bargaining occurs. Bargaining power is realized both through an exogenously evolving state variable and an endogenous choice of one player (the activist) to destroy some share of the other player’s (the government) surplus in a costly fashion. This second-order framework formalizes the intuition offered by activists during interviews that rioting is a justified response to repeated state ignorance of their movements and demands as a way to force engagement and secure a “seat at the table.” I prove existence of and characterize Markov perfect equilibrium in this generalized setting under the assumption the first stage must end in T periods. Under some technical assumptions on the opportunity cost of rioting, the equilibrium is unique, and moreover, under certain boundary conditions, the appearance of rioting is volatile and discontinuous even as its intensity decays monotonically in support, reflecting the volatile nature of rioting. Finally, I characterize the welfare effects of endogenous surplus destruction and show the option to riot strictly improves activist welfare, even if it is almost never exercised.
Read the full paper here.