Background
In this paper, Shavell and van Ypersele begin by considering the historically controversial nature of patents. Economists were often critical of the patent system in the past. Indeed, during the 19^th century, strong opposition was voiced against the patent system and several advanced countries abolished patents altogether for a time. Alternative tools to encourage innovation were discussed and used. One alternative was publicly funded awards, such as the reward offered for a reliable chronometer, popularized recently in Dava Sobel’s Longitude.

This paper provides a formalized economic consideration of the potential merits of reward systems compared to the patent system. The authors construct a simple economic model with "asymmetric information;" they assume that inventors have better information than the government about the prospects for a valuable invention.

Findings
The patent system encourages innovation by providing inventors with a limited product monopoly. However, patent monopolies may lead to sub-optimal investment in research and sub-optimal production ("deadweight loss"). On the other hand, in a "reward regime," the government offers a reward equal to the expected value of an invention. But, because the government lacks the specific information about each particular invention, the reward offered may encourage too little or too much R&D compared to the level that maximizes social welfare.

Shavell and van Ypersele find that reward systems tend to outperform patents provided that the distribution of invention values is not so dispersed that the government's investment decisions are extremely bad. They suggest later that a scheme in which the government bases its rewards on the ongoing sale of inventions (or goods that incorporate them) would go some way to guaranteeing this. Also, a regime in which inventors can choose between a reward and a patent (one in which governments offer to buy patents) is strictly superior to a pure patent system.

Assumptions
The model does make some strong assumptions. Firstly, it considers only inventors with a single possible invention. A problem of the naive (not sales-based) reward regime is that inventors have an incentive to discover cheaper (in research terms), less useful, things. The authors’ defense is that basing rewards on market sales would address this problem. Nevertheless, the real-world validity of this argument depends on the extent to which the limitations of this information-collection system outstrip the economic and administrative problems of a patent system.

Secondly, this paper uses the conventional economic measure of social welfare, the sum of consumer and producer surplus. This is a measure based on market demand and, consequently, the needs of a large fraction of the world’s population count for very little. The standard Coasean argument is that an efficient economic system will maximize joint wealth and then issues of the distribution of that wealth can be addressed separately. Since such redistribution is somewhat improbable in contexts and scales relative to international IP laws, Shavell and van Ypersele's case for reward systems could be bolstered significantly by consideration of the distribution of benefits. Indeed, this is why reward systems are particularly attractive for encouraging the development of drugs to counter malaria and other Third World diseases.

The limitations I have identified in Shavell and van Ypersele's analysis should perhaps not detract from the helpfulness of their contribution to debates about patents and intellectual property more generally. Reward mechanisms are clearly an important alternative to exclusive-rights models, and deserve a far richer literature than they currently possess.