Innovation, Imitation and IP: Existing Literature

  • 2006-03-17

Summary Table

Theorem 1: Increase in IP strength (always) increases innovation.

Theorem 1: With no IP have zero innovation. (Usually via some version of perfect non-rivalry, i.e. instantaneous costless imitation, which results in innovator’s rents being zero without IP monopoly).

Article Imitation Sequential Innovation Thm 1 Holds Thm 2 Holds Comments
gilbert_ea_1990 Y (concept of breadth) N Y Y (assuming no IP corresponds to p=0) Optimal policy is long but narrow patents (patents are infinitely long with minimum breadth necessary to pay costs). Define patent scope as the price p that an innovator can charge for a product embodying the innovation. (So patent policy is defined by (T,p) where T is patent length).
klemperer_1990 Y. Breadth of patent used as exclusion zone in a consumer preference space. Innovator's product is where all consumers are located and imitators location is controlled by patent breadth. Consumers incur 'cost' of travelling to imitator products. N Y Y (imitation costs less and w/o patent no breadth so imitators taken away innovators rent) In equilibrium no imitation at all since innovator limit prices under the optimal regime.
gallini_1992 Yes. Focus of the paper. Imitation both with and without patents. Idea of breadth increasing cost of imitation. N Y (stronger: increasing breadth increases innovation) N (will still have some innovation as imitation may not occur even without patents) Under optimal patent policy no imitation occurs (Prop 1). Usual result regarding limit pricing.
green_ea_1995 N Y Y (increases return to first stage innovator. Optimal policy in one circumstance calls for infinite patent breadth and is always gives non-zero breadth) N/A
odonoghue_ea_1998 N Y. Have leading and lagging breadth concepts. Y. More detailed: under homogenous preferences for quality lagging breadth is not enough since don't appropriate much of ongoing social value (every future innovation builds upon the quality improvements you introduce). Consider patent policies of form (K, T) where K is leading breadth and T is patent length. With T infinite they have (Prop 2) innovation rate approaching optimum as K -> infinity. N/A Paper covers alot
menell_ea_2005 N/A N/A Y. p.3 "In a competitive economy, profits will be driven to zero, not accounting for sunk costs such as research and development (R&D) or costs of authorship. From an ex post point of view, this is a good outcome, as it keeps prices low for consumers and avoids deadweight loss. But from an ex ante point of view, it produces a sub-optimal level of investment in R&D. Most firms would not invest in developing new technologies, and potential creators might not spend their time on creative works, if rivals could enter the market and dissipate the profit." N/A Summary paper.

Some Quotes on This

scotchmer_2004 p. 150

An improver has two options when he receives an idea for an improvement. The first option is to invest and try to market the good, knowing that it will not be patentable. The second is to cache the improvement until the same innovator gets a second improvement and qualifies for a patent. Marketing the unpatentable improvement is clearly not sensible, since an entrant will imitate the good and erode the improver’s profit [emphasis added].

RAND Journal of Economics, Vol. 32, No. 1, Spring 2001, pp. 152-166. Innovation variety and patent breadth; Hugo A. Hopenhayn, and Matthew F. Mitchell

p. 152

Patent protection is costly because it generates market power for the innovator; it is necessary because inventions are costly to produce but may be nonrival (costless to reproduce) after their invention, leaving the inventor without a means of benefiting without some protection.

State as part of their model’s assumptions:

(p. 152) … Innovators can make profits if and only if they are given a property right.


Matutes ea. 1996 (p. 62):

Finally, our definition of scope differs significantly from that of the previous literature. Gilbert and Shapiro (1990) define greater breadth as anything that increases the flow rate of the innovator’s profits uniformly during the period of protection. Relying on the doctrine of equivalents, Klemperer (1990) measures breadth as a quality advantage conferred on the patent holder. Gallini (1992) defines breadth in terms of the cost of imitation, and Scotchmer and Green define breath in terms of novelty. In terms of the law, our definition relates to the the leniency of courts in granting claims of innovations that are not fully developed.

This is useful summary but really they are all looking at the same elephant – just different parts. A patent can be seen as a defining some region in product space from which all except the owner of the patent are are excluded. For expository purposes imagine a 2-d product space with horizontal (the same but different) and vertical (better and different) axes:

          |       ,---.
          |      /     \
  quality |     (       )
          |      \     /  -- products covered by a patent
          |       `---'
           locational differentiation

Since Green and Scotchmer are concerned about cumulative innovation (i.e. movements up the quality ladder in the vertical direction) they talk about novelty. Meanwhile, Klemperer, for example, is only concerned with locational differentiation (though he also associates this with quality – or at least consumer preferences – by specifying a distribution of consumers that are concentrated in the area covered by the patent).

This visualization is fine as long as we are only considering products of the same type but different characteristics. But what happens if we were interested in the developments of tools and their applications. These products may exist in very different spaces: for example a new alloy might be developed which could be used in anything from ships to bicycles. In the case we would want to have several layers like the above corresponding to the different types of products. Matutes et. al. are then interested in allowing a patent in one layer (the tool layer) to give exclusion rights in another layer (the application layer). In the bio-tech industry this occurrence is so frequent that it has its own name: reach-through (in which a license to use a given tool entitles the owner of the original patent on the tool rights to the products developed using the tool).