Home Computer Science
The innovation system comprises a wide range of players with different strategies and behaviours. For instance, some firms are pure manufacturing facilities and hence do not engage in any research activities. On the contrary, some firms perform research intensively and play the role of highly specialised technology leaders in their traditional field. Other firms aim at diversifying to a broad range of research fields, either through incremental changes or, more riskily, through shifting to entirely different technology fields. This may enable firms to change to less competitive areas of research or spread their risk to reduce their vulnerability regarding external shocks.
To account for this heterogeneity, the knowledge creation model foresees two groups of agent strategies: The first group refers to choosing a research target, whereby an agent uses one out of four available search strategies. The second group, the agent’s research strategies, relates to the possible ways how the chosen research target is striven for. Hence, each time step (i.e. a quarter of a year), an agent ai engages in research activities by defining a research target kji. For that purpose, the agent randomly chooses one of the three existing kenes kji and modifies it according to its given search strategy:
• Gridlock: With this strategy the agent does not perform any research and hence, no new research target is formed; the ‘new’ research target equals the old target:
• Conservative: If the agent follows a conservative search strategy, it aims at increasing its expertise level Ej in a certain research field and subfield. Thus, its new research target is set to the following:
While the technology class and the subfield remain the same, the expertise level is increased by one. If the expertise level has reached its maximum level Ej = 10, the agent remains at this level of expertise.
• Incremental: An agent with an incremental search strategy tries to modify its research orientation represented by a change of its subfield value Sj, while staying in the same technology class. Since a new area of research has opened up for the agent, the corresponding expertise level is set to a beginner’s level
Ej = 1:
Every time the incremental strategy is applied, the subfield value is increased by one. If an agent with the maximum value Sj = 10 pursues the incremental strategy, its new subfield is set to Sj = 1.
• Radical: An agent with a radical search strategy goes for repositioning itself in a new technology class, associated with a diversification strategy. Thus, to define the new research target, a new technology class is chosen, whereby technological similarity is taken into account: Similar technology classes (to one of its current technology classes) are chosen more likely than distant ones.
The subfield of the old kene is inherited and hence remains unchanged. Again, due to the newly entered research field, the expertise level of the agent is set to one.
Once the research target is chosen, the research strategy defines the way how the agent tries to obtain it. The model provides three kinds of research strategies:
• Spillover: The agent receives the new kene if its research target kji is similar to another agent’s kene k' in the population, i.e. if the technology class and subfield are identical and the expertise level fulfils the condition:
• Internal research: An agent tries to achieve its research target without depending on other agents.
• Cooperative research: An agent looks for a partner with cooperative strategy that holds an equal or similar kene k' as its research target kj, set afore. The similarity of the desired kene depends on the agent’s search strategy. Agents with a conservative search strategy only have a tolerance level 8E regarding the expertise level,
whereas agents with incremental or radical search strategy additionally have a tolerance level 8s regarding the subfield:
In addition to the knowledge endowment and the strategies, each agent is individually characterised by organisational figures, namely research expenditures R,, number of employees L,, assets I, and age A,. These figures, representing the fitness of an agent, are essential for the evaluation of the agent’s research success in the output part.