A variety of interventions can affect outcomes. The important one are, intervention to solve coordination problem, information as an interventions, interventions to change the dynamics of political process and interventions change distribution of wealth. As the equilibrium set of behaviours in a decentralized economy may not be Pareto efficient, it can not be concluded that political process can be a suitable alternative to attain required improvements. We have seen that policy process is a dynamic phenomena so, it make an analysis of policy intervention would also require a dynamic framework.

The critical question that has been investigated in this respect may be whether and how an initial coordination failure will in fact transmit itself over there? Why would not a forward looking agents, with positive expectations, adopt a policy path, that may be self fulfilling to move away from a bad equilibrium? And most critically, we may ask, is there really any scope for policy? Adsera and Ray (1998) addressed these questions and obtained some striking results. They found that if positive externalities from moving to more favourable set of policies appear with a time lag. ‘then the final outcome depends entirely on the initial conditions unless there is some gain to being the first to switch'. In other words, unless there is some gain to being among first to experiment a policy switch, each agent will rationally wait for other to make a policy change and so no one will switch at all.