1. 38CO2000 Economics of Intellectual Property Rights (IPRs) Spring 2006: Lecture 5 Practical issues: I said in the beginning of the course that you need to understand some game theory and IO concepts such as Nash and subgameperfect equilibrium Cournot and Bertrand competition But now I think this is not necessary. Nash will be needed but I will teach it. Of course it would be helpful to understand the other concepts, too, as they will be referred to.

2. A Digression on Methodology (Cf. “Arm-chair economist” by Landsburg or “Hidden order” by Friedman or Economics crime novels by Marsall&Jevons): TMP/IO-economics uses models not because they are realistic but because they offer a consistent way to understand firms’ & policy- makers’ behavior Models are like maps –One-to-one map pretty useless in orienteering –The real-world aspects left out depends on the purpose of the map Models build on assumptions: –Some of them are plainly false but the literate truth of assumptions is never a prerequisite for scientific inquiry –Assumptions are tested not by their literal truth but by the quality of their implications Good models and their predictions often generalize

3. Recap: 1) Optimal patent life minimizes ex post DWL but provides enough protection to justify the investment  T*=C/  p 2) Requires that society commits to T*. Ex post government has an incentive to cheat and put T=0: this is optimal ex post. However, if this were possible, the inventor would realize it, and would not invest. - The same logic applies to many policy debates related to IPs even if they do not invoke the duration of IPRs explicitly - e.g. Apple ITunes vs France, EU vs Microsoft

4. 3) Patent breadth is an elusive concept - legally breadth is governed by the “doctrine of equivalents” and claims - economically the breadth measures how difficult it is to bring a non- infringing substitute in the market - Determines the pricing power of the patent holder over the patent life - Product space (shift in demand) and technology space (the costs of inventing around)

5. Generic Economic Effects of Patent Breadth let b denote patent breadth, b  [0,1] let  (b) the profit flow after the successful innovation as a function of patent breadth (when the patent is in force) assume  ’>0,  (1) =  p and  (0) =  c =0 similarly, welfare flow as a function of the patent breadth W(b), over the duration of the patent: W’<0, W(1)=W p, W(0)=W c

6. The firm’s profits on an existing innovation as a function of the patent life & breadth, P(T, b): Ex post social welfare as a function of patent life and breadth, S(T, b): i.e, full social value minus DWL as before. Now DWL is a function of patent breadth DWL(b) Note For b S(T,1), where P(T,1) and S(T,1) are as in the analysis of optimal life i.e.,  P(T,b)/  b=T  ’>0 and  S(T,b)/  b=-TDWL’=TW’<0 i.e., the tradeoff between ex ante and ex post inefficiencies.

7. Seek optimal patent breadth-length mix Stage 1) The policy-maker chooses patent policy (T, b) Stage 2) The firm invest in R&D (  ) Proceed backwards, look for a SPE Stage 2) The firm chooses C so as to maximize  P(T,b)-C  the firm invests only if P(T,b)  C/  The effect of the patent breadth:  P(T,b)/  b=T  ’>0 i.e., increasing the breadth reduces ex post competition in the market, increasing the incentive to innovate.

8. Stage 1) The optimal patent policy (T*, b*) maximizes the ex ante social welfare:  S(T,b)- C subject to the firm’s incentive constraint By the same logic as before, the optimal policy satisfies P(T*,b*)=C/    (b*)T* =C/  Note Differentiating the optimal rule with respect to T* and b* yields: The patent life and breadth are policy substitutes If you decrease breadth (life) the life (breadth) should be increased to compensate to maintain the incentive to innovate

9. The optimal policy satisfies P(T*,b*)=C/    (b*)T* =C/  Problem: two substitute variables – what is their optimal mix? Consider two policies Policy 1: narrow but long patents (b1, T1) Policy 2: broad but short patents (b2, T2) such that T1>T2, b1<b2 and  (b1)T1=  (b2)T2=C/  Which policy is better?

10. Answer: choose the one with higher ex post social welfare S(T, b). So if S(T1,b1)> S(T2,b2), then policy 1 (narrow but long) is optimal and vice versa if S(T1,b1)<S(T2,b2) When S(T1,b1)>S(T2,b2)?  Wc/r-T1DWL(b1)>Wc/r-T2DWL(b2)  -T1DWL(b1)>-T2DWL(b2)  T1/T2<DWL(b2)/DWL(b1)   (b2)/  (b1)<DWL(b2)/DWL(b1)  (b2)/ DWL(b2)<  (b1)/DWL(b1) i.e., the choice of optimal policy mix boils down to a ratio test Choose the breadth that maximizes  (b)/DWL(b) i.e, if  (b1)/DWL(b1)>  (b2)/DWL(b2) then policy 1 (narrow but long) is optimal and vice versa if  (b1)/DWL(b1)<  (b2)/DWL(b2)

11. Why does the ratio test work? In fact the problem is to choose T and b so as to minimize TDWL(b) s.t.  (b)T=C/  The benefit of IP is that it provides the incentive to innovate The cost of IP is dead-weight loss Lower ratio means that the cost of creating the required incentive to innovate is higher The effect of T is linear both on the incentive to innovate and DWL. Therefore we can just focus on the effects of b and use T to compensate (recall, variables are policy substitutes)

12. Example: Let us consider technology space more closely The imitator can copy the innovator’s product with cost K(b) where b is patent breadth and K’>0. The more there are rivals in the market, the lower is the profit of a single firm, i.e.,  (1)>  (2)>… >  (n-1)>  (n)>  (n+1) - e.g., Cournot/quantity competition Free entry with cost K(b)  market will consist of n firms defined by  (n)>K(b)>  (n+1).  the profit of the patent holder is at most K(b)

13. P Q P(Q)=a-Q QnQn PnPn a PS n =nπ n Effect of Breadth in Technology Space: Market when the patent is narrow and there is n-1 entrants π n =K(b)

14. However, the costs of entry are pure waste. The firms can profit by avoiding these costs, e.g, via licensing Suppose the patent holder offers her technology to rivals with a price that is slightly less than K(b) (K(b)-  where  0). n-1 of rivals are willing accept. There will be n firms as before but the patent holder collects nK(b) i.e., the full producer surplus! Even from comparing the figures it is easy to see that narrow patent yields larger ratio of the patent holder profit to DWL  narrow but long patents are optimal

15. P Q P(Q)=a-Q QnQn PnPn a n  n =K(b) Effect of Breadth in Technology Space: Market when the patent is narrow DWL(n)

16. P Q P(Q)=a-Q QpQp PpPp Effect of Breadth in Technology Space: Market when the patent is broad a pp DWL(p)

17. Notes: 1)Both DWL and incentive to innovate crucially hinge on b and the shape of the demand. With very convex demand the finding can be reversed (broad but short are optimal) 2) If, for some reason, licensing to imitators in the technology space example is not feasible, the finding is reversed (broad but short are optimal) as the costs of entry are avoided - the imitators prefer to wait that the patent expires rather than engage in costly inventing around 3) If one interprets copyright as “narrow but long”, and patents as “broad but short” we have a prediction in which industries one should apply patents and in which industries one should apply copyrights - We have a prediction concerning the welfare impacts of the expansion of patentable subject matter 4) Management of IPRs: better to license than let rivals invent around

18. Notes: 5) Political Economy of IPRs. Recall the Mickey Mouse example and France vs. Apple Itunes. - The lobby typically based on the current/static context -In the static context strong IPRs are good for producers/innovators and bad for consumers/imitators - Countries that produce/innovate lobby for strong IPRs and countries that consume/imitate lobby for weaker IPRs - IPRs tend to be strong because a) producer side is typically more concentrated than consumer side  easier to organize lobby b) countries that produce/are technologically advanced have more political power - In more dynamic context this is not necessarily bad as strong IPRs may be also good for consumers

19. I.Intro II.Patent (and other IP) Policy II.1 Horizontal competition II.2 Cumulative competition III.Management of Patents (and other IPs) III. 1. Patenting vs secrecy III. 2. Licensing IV. Competition Policy and IPRs V. Innovation and IPRs in Financial Services Sector Other… Overview

20. Lectures so far: innovations are isolated events with no bearing on future innovations In practice, innovation is cumulative, building on the existing knowledge and previous innovations Cumulative innovation can consist of improvements on the existing products creation of basic technologies (e.g., research tools) and their (commercial) applications cost reductions for producing earlier products e.g., Windows-95 included e-mail, Windows-98 e-mail + internet explorer, Windows-XP e-mail + internet explorer + media player, Windows Vista e-mail + internet explorer + media player+ search engine.. II.2 Patents (IPRs) when Innovation is Cumulative: Policy and Some Management

21. The basic (Nordhaus) trade-off between the creation of incentives (ex ante) and the use of innovation (ex post) typically hangs around, but –it will be modified –it can even vanish (Bessen&Maskin, 2002) The basic trade-off of the static model: how to render social value of innovation private If innovation is cumulative, –social value of an innovation includes the incremental value of future discoveries the innovation makes possible –private value of an innovation depends on the ability of its inventor to appropriate the value of future discoveries  the “basic” trade-off of the cumulative innovation: how to render social value private to secure the incentives to innovate the first innovation without stifling the incentives to create future discoveries?

22. Bath breaking innovation Basic innovation Improvement 1 Improvement 2 Improvement 3

23. Basic innovation 2 Improvement Basic innovation 1 Basic innovation 3 Problem of fragmentation

24. ImprovementState of the art Quality q0 Quality Ladder Quality q1  =q1-q0 Quality

25. Other new problems –the later innovations may render the previous ones obsolete (quality ladder) –the ‘first’ and ‘second’ innovations can be made by different firms  how to split the cake?  what about if the second one infringes the patent of the first innovator? Effects depend much on what we will assume about breadth and inventive step A new concept: forward protection measuring how well an innovation is protected against future ones. It pools Inventive step determining patentability (Leading) breadth determining whether there is an infringement or not –Recall: An invention can be unpatentable, infringing patentable, non-infringing patentable, infringing (blocking patents) unpatentable, non-infringing

26. related concepts: patent quality affects, among other things, how clearly forward protection is defined, patent validity patent strength = probability that both the patent validity and the infringement holds in the court Patents (IPRs) are probabilistic property rights (Lemley- Shapiro)