R&D and competition

� Motivation: to investigate the relation between competition and R&D expenditures

� Schumpeterian hypotheses� Incentives� A case: steam engines� Patent races� Common pool problem� Recent developments

Schumpeterian hypotheses:

� Monopoly power and firm size are conducive for R&D– A monopolist is well-placed (distribution, reaction

to rivals)– Internal finance of innovation (moral hazard)– Advantage on labour market for engineers– More internal spillovers in a large R&D team– Technology push favours firms with large

resources for R&D– Demand pull favours firms with a deep knowledge

of market and large scope

Incentives

� Same context as in the patent length model: a process innovation

� Compare profits with and without innovation: the difference is what a firm would maximally pay to get the innovation (=maximum amount of R&D it is willing to spend)

Incentives – mathematics

q�A�Bp,

p ��A

2B�

12

c, qp ��12

(A�Bc), ���14

(A�Bc)2

B�

q 2p �

B,

��(c)���(c)��c

c

q(p �(c))dc� 12

A(c�c)� 14

B(c2�c2)

�minor��c

c

q(c) dc� (A�Bc) (c�c).

�c

c

q(c)dc.

Monopolist:

Competitive firm:

Social planner:

Incentives: conclusion

� The incentive for a monopolist is strictly smaller than that for a competitive firm with a minor innovation, which is again strictly smaller than that for a social planner

Steam engines – An illustration

P C HP H240 2240× × × ,Fuel consumption:

P=266 pence, HP=3250, H=4000 (Cornwall, 1800)

Newcomen engine: C=24.3

Watt engine: C=6.1

Woolf engine: C=3.3

Trevithick engine: C=4.3

266240

24 3 612240 3250 4000 117068× − × × ≈( . . )

266240

24 3 4 32240 3250 4000 128646× − × × ≈( . . )

266240

24 3 332240 3250 4000 135078× − × × ≈( . . )

� Engine builders in Cornwall in 1800 could charge 1/3rd of the total fuel savings relative to a Newcomen engine

Patent race Incumbent vs entrant�m(c)��d(c, c)��d(c, c).

V1���

0

e ���e �(R �1 �R�2 )��m(c)�R1�R �

1�m(c)�

�R �

2�d(c, c)�

�

�m(c)�R1�R �

1 �m(c) /��R �

2 �d(c,c) /�

��R �

1 �R �

2

.

Incumbent:

Entrant:

V2���

0

e ���e �(R �1 �R �2 )� R �

2�d(c,c)�

�R2 �R �

2 �d(c,c) /��R2

��R �1 �R �

2

.

Efficiency effect

Patent race Incumbent vs entrant

�V1

�R1

��[��R �

1 �R �

2 ]��R ��11 [�m(c)��m(c)�R �

2 [�m(c)��d(c,c)] /�� R1]

(��R�1 �R �

2 )2�0�

�V2

�R2

���R �

1 �R �

2 ��R��12 [�d(c,c)�R �

2 �d(c,c) /��R2]

(��R �

1 �R �

2 )2�0.

Nash equilibrium:

π π πd c c m c d c c( , ) ( ), ( , ,)= = 0Drastic innovation:

��m(c)R ��11 ���m(c) [R ��1

1 �R ��12 ]��R �

1 ��R �

2 ��R ��1

1 R �

2

���R ��1

2 R �

1

�.

Patent race Incumbent vs entrant

Symmetric patent race

V���

0

e ���e �(R�1 � (n�1)R �)�(R �

1 ��R1)d��R �

1 ��R1

R �

1 � (n�1)R ���

.

[(n�1)R ���][�R ��11 ��1]�R �

1 (1��)�0.

[(n�1)R �

1 ��][�R ��11 ��1]�R �

1 (1��)�0.

Symmetric patent race

Symmetric patent race

V�n��

0

e ���e �(nR�m)�(R �m��Rm) d��nR �

m��Rm

nR �

m��

.

�R ��1m ��1�

nR �

m (1��)

�.

�R ��11 ��1�

R �1 (1��)

(n�1)R �1 ��.

Monopolist (centralized):

Decentralized:

Patent race as a common pool problem

Number of fishermen Catch per fisherman Total catch Marginal increase intotal catch

0 - 0 -

1 100 100 100

2 90 180 80

3 80 240 60

4 70 280 40

5 60 300 20

6 50 300 0

7 40 280 -20

Recent developments

� Patent races with memory (leads to a single dominant firm)

� Mixed strategies� repeated games: if leapfrogging is possible

this favours a large number of firms undertaking R&D

recent developments

� Invariance theorem– if all players in a race are allowed to engage in

multiple projects without externalities between them, it does not matter how many players there are (and otherwise it does!)

� Second-mover advantages (it may pay to wait, see inventing around example)

� Size of prize depends on R&D

R&D and competition

� An alternative view: technology regimes in a boundedly rational, evolutionary world

� When uncertainty is strong, the rational behaviour of patent race and patent protection models is not very realistic

� Nelson & Winter model

Technology regimes

� Basic idea is that the nature of the knowledge base has an impact on the way innovation and technological change takes place, and this influences market structure

� Schumpeter Mark I vs Mark II� Different regimes within the same industry?

(Saxenian)� The Nelson & Winter model can generate

technology regimes

A (more) realistic Model: Nelson & Winter

� Firms differ, they are characterized by– A labour coefficient (labour output ratio)– A capital coefficient (capital output ratio)– A capital stock

� Firms sell a homogenous output, and hire homogenous labour

� Rationality is bounded: based on rules of thumb and routines

� Firms engage in search (R&D) to improve their technology (input coefficients)

R&D in the Nelson & Winter model

� R&D can be imitative (trying to imitate technology of other firms) or innovative (trying to find new techniques

� R&D is local search, it starts from the technology that the firm has now; this is in line with the notion of a technological paradigm

� In one version of the model, R&D depends on the nature of the knowledge base

The market structure model

� Two technological regimes:– Routinized– Entrepreneurial

� This is modeled after Schumpeter Mark I vsMark II

The routinized regime

� Lower probability of innovation (for equal levels of R&D spending)

� Innovative step depends on external (scientific) developments and the firm’s own performance (innovation in the past breeds more success): cumulativeness

� As a result outsiders (potential entrants) have a low probability of success (they have no experience)

The entrepreneurial regime

� A high probability for innovation� Outsiders (potential entrants) are as good in

innovation as incumbents

Simulation analysis

� The usual way of analysis of a model is to find the equilibrium

� The evolutionary Nelson and Winter model does not have a clear-cut equilibrium

� Outcomes of the model depend on random factors (e.g., innovative success)

� The model can be analyzed by running it on a computer

Guidance for simulation analysis

� A single simulation run is not enough to show a result

� Monte Carlo experiment: run multiple times with different parameter settings, and then compare the outcomes by means of a statistical analysis

Technology regimes – Nelson & Winter

� http://www.tm.tue.nl/ecis/bart/nwreg.zip

Simulation results – Nelson & Winter

Simulation results – a systematic comparison of the regimes

Technology regimes – Nelson & Winter

� A conclusion: differences in knowledge base cause differences in R&D spending and market structure

� Which regime generates which market structure?

� How can the changes between the regimes be influenced?– Explorations of parameter space