price auction receives a patent with probability e and in that event pays the second-highest bid.
With probability (1 — e), the patent is invalid, and the information associated with the patent is
placed in the public domain. In either event, the innovator receives a prize which is a multiple (a
markup) of the second-highest bid.
Suppose that the value of θ is observed by the innovator and all market participants. Without
market manipulation, Kremer’s mechanism leads to better outcomes than under the patent system
with probability (1 — e) and to no worse outcomes with probability e, if the prize multiple is suffi-
ciently high. The argument for this result is that in the second price auction, the dominant strategy
equilibrium is for all bidders to bid the true value of the object. Thus, all bidders bid μπm (θ~), where
μ is the length of the patent and πm (θ~) denotes monopoly profits. As long as the prize markup
is greater than 1, the patent holder finds it optimal to accept the auction mechanism. Indeed,
this mechanism can lead some innovators to choose to innovate under the auction mechanism when
they have not chosen under the patent system. It is, however, possible to set the markup so as to
induce the efficient level of innovation. Note that for this mechanism to yield good outcomes, it is
important that e > 0, because if e = 0, the market participants have no incentive to submit any
bids. Equilibria in which the participants submit no bids are possible and likely to occur, especially
if bidding is costly.
Next consider the possibility of market manipulation. Formally, we assume that the innovator
can participate directly in Kremer’s auction or can designate an agent whom we call an accomplice
to bid on his behalf. We will argue that market manipulation leads to extremely inefficient outcomes
under the (realistic) assumption that the innovator is better informed than the competitors regarding
the quality of the innovation. We show that the extent of this informational advantage can be made
arbitrarily small. To make this argument, suppose that each other market participant г receives
a signal Si from a distribution G (s∣0). These signals are independent across market participants
conditional on the innovator’s type θ. Without loss of generality, suppose that E (s⅛∣0) = θ.
We will show that in any equilibrium of this game, the innovator will not submit his patent
to the auction. The argument is by contradiction. Suppose that the equilibrium specifies that the
innovator submits his innovation to the auction. In that event, we argue that in any equilibrium of
this subgame, the innovator (or his accomplice) wins the auction and that the second-highest bid
is zero. This argument is also by contradiction. To see that this outcome is an equilibrium of the
subgame, suppose that one of the market participants wins the auction. In order for this outcome
to be an equilibrium, the market participant must submit a bid higher than the innovator’s bid.
23