1

PATENT RACING -THE GAME THEORETIC APPROACH

Assume two rms A and B who want to decide whether they should attempt to produce a new product with marginal cost c. The demand for the good is ! = " #$ and as for the R&D eort, the costs of setting up a research lab are K and the probability that the lab will successfully develop the product is %. If both rms successfully develop the product they will be a Cournot duopoly 2 ' ' (&! = &" = (#$) = 4& , ! ' = #+2$ , '1 = '2 = #$ ), while if only one of 9% 9 3 3% the two develops the new product she will be rewarded with monopoly prots 2 (( = (#$) , ! & = #+$ ) $& = #$ ). 4% 2 2% The expected prot net of setup costs if only A establishes an R&D lab while B does not is *(&" +&,! = 0) +&," - 0) = *(& ! j +&,! - 0) +&," = 0) = (" .)2 = % + (1 %)0 / = %( / 4# If both establish R&D labs, then +&,! - 0) +&," - 0) = *(& ! j +&,! - 0) +&," - 0) = (" .)2 5 (" .)2 = %(1 %) + %2 / = %( (1 %) / 4# 9# 9 No R&D 0) 0 %( /) 0 R&D 0) %( / %((1 5 %) /) %((1 5 %) / 9 9 j j

*(&"

No R&D R&D

1.1

Possible Nash equilibrium outcomes

1. Neither rm establishes an R&D lab if %( 0 /. 2. Only one rm establishes an R&D division in equilibrium, say A if a. (R&D, no R&D) - (no R&D, no R&D) (if %( - /) and b. (R&D, no R&D) - (R&D, R&D) (if 0 - %((1 5 %) / () / - %( (1 5 %). 9 9 Combining a and b gives: %( - / - %( (1 5 %) 9 3. Both rms will R&D in Nash equilibrium if %( (1 5 %) - /. In this 9 case B prefers to R&D irrespective of whether A R&Ds or not and the same applies for A. Hence the (R&D, R&D) is a dominant strategies equilibrium.

1.2

Complications

Note that (R&D,R&D) may be a Nash equilibrium,but it will not necessarily be a Pareto optimum from the rms perspective if the sum of prots under this equilibrium is less than the sum of prots under a (R&D, no R&D) or (no R&D, R&D) combination. This will be the case if 10 5 2%( (1 %) 2/ 0 %( / () / - %((1 %) 9 9 and combining this with the fact that (R&D,R&D) is a Nash equilibrium, gives: 10 5 %((1 %) - / - %( (1 %) 9 9 Hence the above condition gives us the case where both rms doing R&D is a Nash equilibrium which is not Pareto optimal from the rms point of view, in the sense that rms would be jointly better o with only one of the two rms having an R&D lab; in other words there is too much1 innovation.

1.3

Consumer surplus
("
#+$ #$ 2 )( 2% )

In the case of monopoly consumer surplus is equal to 12& = = ( 2

2 while in the Cournot case it is equal to 12' = 2 ("

#+2$ #$ 3 )( 3% )

4( 8( =2 = 2 9 9 The expected social surplus ignoring the R&D costs if only one rm, say A, established an R&D lab is equal to: *(&! j +&,! - 0) +&," = 0) + *(&" j +&,! - 0) +&," = 0) + %12& 3 ( = %( 2 2 The expected social surplus ignoring the R&D costs if both rms set up an R&D lab are = %( + 0 + % *(& ! j +&,! - 0) +&," - 0) + *(& " j +&,! - 0) +&," - 0) +%2 12' + 2%(1 %)2& =
1 In other words, too much is dened as the case where both rms innovate as it is a Nash equilibrium to do so, but this generates less aggregate prot than if only rm only performed R&D.

= 2[%(1 %)( + %2

( 4( 8( 11% ] + %2 + 2%(1 %) = %( (3 ) 9 9 2 9

The second lab is socially desirable only if %((3 11% 3 11% ) / / - %( / () / 0 %( (135 ) 9 2 9
5( 9 )

So a Nash equilibrium of both rms doing R&D (where / 0 %( (1 will be Pareto optimum from the societys point of view if / 0 %( (135 11% ) 9

for % - 0375 (since then %((135 11( ) 0 %((1 5( )), and always for 9 9 % 0 03753 On the other hand, both rms doing R&D will not be socially optimum (would be too much) from the societys point of view if %( (1 for % - 0375. Corollary 1 It is always socially optimum for both rms to do R&D if the probability of success is low. Corollary 2 If both rms doing R&D is desirable (Pareto optimal) from the rms point of view because / 0 %( (1 10 %), then since %((135 11 %) 9 9 %((1 10 %)) it always follows that double R&D will also be optimum from the 9 societys point of view. Corollary 3 It is possible to have too much innovation from the rms point of view, while it is socially optimum from the societys point of view if 10 11 %) - / - %( (1 %) 9 9 5% 11% ) - / - %((135 ) 9 9

%((135