Modeling Ideas
Tuesday, January 7, 2014
Daron Acemoglu and James Robinson

So how can we model ideas and their impact on political economy?

One obvious way is to approach it as a Bayesian learning problem. Perhaps we can think of new ideas as states of nature which had very low probabilities before being proposed or developed by some agents (“idea entrepreneurs”, and once those proposed, people may come to believe that they are more likely through Bayesian updating. If this happens, behavior will change.

This sort of modeling would tap into a long tradition in economics and political economyThough the Bayesian approach is both useful and standard in general and may also be a promising research direction in the context of ideas, it does not feature idea innovations in the sense that from the beginning, even if some states of nature have low probability, all agents are aware of all states of nature, so there are no surprises and no eureka moments. 

An alternative would be to partially depart from the Bayesian framework and assume that agents start with a limited state space (a limited set of possible states of nature) and are unaware of the full set of states of nature. This type of unawareness has been modeled in economics by, among others, HalpernModica and Rustichini, Li, and Heifetz, Meier and Schipper.

These approaches make individuals unaware of states of nature. An alternative might be to make individuals unaware of potential political states. For example, people in the Middle Ages may not have been able to conceive of democracy because they were unaware of the political system in which political power is not monopolized by the king or the aristocracy. 

The implications for political dynamics of this type of approach could be quite interesting. If at some point a political entrepreneur uses the argument that a monarch’s power is not divinely granted but based on his ability to dominate society, then this can make people aware of a new political state (or in fact a whole new set of political states without such monarchical monopoly of power). Interestingly, when such possibilities are taken into account by potential clinical entrepreneurs and organizations, there will be other game theoretic interactions. For example, if the political entrepreneur who wants to empower the middle class is afraid of power shifting to poorer segments of society after they become aware of organized social groups to be able to increase their political power in the world no longer divinely dominated by the monarch, then he may not want to start using arguments and strategies that will change their awareness, even if these arguments and strategies would, in the short run, enable him to shift power to the middle class. This of course has a nice parallel to what Daron, Georgy Egorov and Konstantin Soninhave have called the “slippery slope” in the context of dynamic political economy (though in a model without this sort of unawareness issues and the possibility to make sense of the importance of ideas).

A complementary approach might be to relax the Bayesian approach to allow for manipulation of ideas. This could be, for example, through indoctrination and inculcation (and a few papers in economics have investigated this, for example, this and thisor through the ability of inflation allegiance or thought leaders to convince others of certain ideas (for example as in this work).

Yet another possibility is to relate ideas to arguments. It might well be that people are more likely to understand (less likely to misunderstand) arguments that they have encountered before. This would imply that political and thought leaders might have to use and combine arguments that have already been used a lot in order to avoid misunderstanding and be more effective. But then this may then restrict what they can communicate. At some point, this restriction may become too burdensome, inducing them to experiment with new arguments even if this will lead to some temporary misunderstanding. 

The bottom line is that there seem to be lots of interesting ways to approach ideas within political economy, and many of them use already existing tools and sometimes even insights in political economy even if their details and of course implications will be quite different once they come to the worked out systematically.

Article originally appeared on Why Nations Fail by Daron Acemoglu and James Robinson (
See website for complete article licensing information.