I In this paper we have simulated the behaviour of an RJV network starting from a very simple model where cost and benefit of ties depends from the number of actors that participate to a R&D project. Given the revenue and cost function there is an optimum number of actors for each project. The size of the network will depend from the number of projects that are realized and the number of actors.. In our model the equilibrium is reached when the profit for a new cooperative project is equal to zero for all actors. The optimal structure of the networks is the result of the optimal number of projects and the optimal number of actors for project. In our model the probability of collaboration is positively affected by heterogeneity in markets .The reason is that the slope of the revenue curve line of a firm decrease as market heterogeneity of the firms that collaborate in the project increase .On the other hand the probability of collaboration increase when firms belong to the same sector .The reason is that when firms belong to the same sector is more easy that they have developed complementary assets and therefore the cost curve decrease at an higher rate. We have shown that there are situations in which the incentives to the firms to cooperate are sufficiently high and therefore there is no need for a policy to finance a RJV. This situation happen when firms sell in non‐homogeneous markets but belong to the same sector. When firms sell in a relatively homogeneous market but belong to different sectors probably the incentive to cooperate are low and welfare will be not positively affected by cooperation .In an intermediate situation incentive to cooperate could be too low but a policy to increase cooperation could be welfare enhancing The project is additional and is worth to subsidize the project. Starting from our theoretical model is possible to see that it is compatible with different kind of network structure . Therefore we have built a simulation model that starting from the developed theoretical model shows what kind of equilibrium we could expect. After that starting from a real technological district we have studied the evolution of such network and how it differs from the simulated model. The computational model which simulate a R&D networks allows to compare quantitatively the structure of simulate dynamics against the real data. The exploratory data analysis in this sense is very useful and can allow to replicate important aspects of the reality(for example the structure of Freeman degree in a network )A model which have a string because allows to detect some relevant features which could be important to insert in the simulation and can allow to replicate important aspects of reality. A model which have a strong validation and could be based also on a theoretical framework can allow to obtain better forecasts using data and can allow to detect structural change over time.
- Social Network Analysis,
- Innovative Networks
Available at: http://works.bepress.com/carlo_drago/121/