Information Cascades as an Explanation of Revolutions
Christopher Ellis and John Fender wrote a paper published in The Economic Journal detailing a model that predicts the likelihood of a “regime transition” based on the economic situation of the country and how the information about the economy are spread throughout two classes. The model splits society into a rich class that is an active participant in the economy and a poor class that chooses whether or not to revolt. The economy is set up as follows: the rich receive an income and the poor receive a proportionally smaller income, both pay taxes, and both then receive a transfer from the government. The transfer payment is set up to benefit the rich more than the poor. At the beginning of the game, the rich are the only group that can vote, and they make an initial decision to enfranchise the poor or to continue with the divided government. Both the rich and poor have perfect knowledge of how the economy functions, but only the rich know the potential permanent damage that would be caused by a revolution. The poor, on the other hand, receive a signal that is based on the probability that a revolution will occur but does not necessarily accurately represent the revolutionary damage. All actions and beliefs are assumed to be rational to limit any extra equilibria caused by “strange” beliefs.
If the rich choose not to enfranchise the poor to create an undivided government, each poor agent, as the paper calls them, has a chance of rebelling based on the signal received. The poor agents take all the information they have an make an educated decision about revolting. Will the benefits of a revolution outweigh the damages caused? What did the earlier agents think about revolution? The poor take the signal they receive and look at the actions of their peers and predecessors and choose to protest or do nothing. Based on the number of poor agents that chose to rebel, the revolution can either be a success or a failure, with a success overthrowing the oligarchy and implementing a fair system. A success will only occur if all agents choose to rebel. As time goes on, the probability of a revolution increases exponentially because as one agent chooses to revolt, everyone in the next round of the game will see his choice. In information cascades, you can’t see someone else’s information, but you can infer from his action what signals he received, which is what poor agents look at when deciding their next action. As in class, Bayes Theorem can be applied to this situation to determine the probability that an agent received a certain signal based on his choice to revolt or not. Agents can calculate this probability and use that information to influence their decision.
Source: http://onlinelibrary.wiley.com/doi/10.1111/j.1468-0297.2010.02401.x/full#ss23