I am reading Predictocracy, and I particularly liked this sentence (p. 66)

The value of experimentation accrues largely to the public , not the individual business, and businesses will hesitate to subsidize these experiments […]

Abramowicz is talking about experimentation with prediction market, but it seems to me that this applies to experimentation with institutions in general. Businesses that experiment with new institutions can create large social benefits when such experimentation is successful and others copy them. However, such businesses bear all of the costs and risks associated with such experimentation but reap only a fraction of the reward, so we should expect such experimentation to be under-provided.

To some extent, relatively forgiving bankruptcy law should encourage private individuals to produce such experimentation. The government can also directly produce some experimentation by funding basic social research into topics like prediction markets. Sadly, there doesn’t seem to be a good way to subsidize useful experimentation with institutions.