• Keine Ergebnisse gefunden

5. Simulations

5.3 Results of the Simulations

In the following I will present the results of 1000 simulations with 60 time periods (for roughly representing 60 negotiating rounds40) for the states when there are 10 member states, 15 member states and 27 member states. For all simulation runs I have decided to present the results by first looking at the percentage of actors who choose pro-liberalization at time point t (“percentage of member states pro liberalization”). This roughly represents the probability of policy change (the more actors are in favour of liberalization the more probable is policy change). The second interesting result is the overall network structure – for this I have decided to look at the network density (the edges realized divided by all possible edges). And finally, since I am particularly interested in the role of the European Commission, I have looked at the centrality of the European Commission defined as the edges realized by the European Commission divided by all possible edges. The thick lines represent the average result across all simulations and the dashed lines the 95 per cent confidence intervals (similar representation by Leifeld (2013a)). For all models I include the mechanisms of preferential attachment and keep the coefficient constant as defined in the Equation 4).

I will start with a simple model, where the probability of edge creation is 5 per cent and all other mechanisms are inactive (as suggested by Leifeld (2013a)). The Figure 13 presents the results of this simulation run. Even though there are certain runs in which almost all member states can become in favour of liberalization, on average, all member states remain against liberalization.

So the preference of the European Commission does not spread at all.

Figure 13 Simulation Results edges with a probability 5 per cent

40 If we assume that the European community had roughly 2 negotiation rounds per year during the liberalization of

the European gas market, this could be interpreted as 30 years of negotiations.

The next possible mechanism is just the role of geography. The Figure 14 shows the results of this model. The in this case is set to 0.001 and other coefficients are set to zero (except for , which is always constant).

Figure 14 Simulation Results edges based on geography (0.001)

From these simulations we can see that there are not many dynamics within the system. In particularly when there are only 10 member states, their preferences remain stable and do not change at all. When there are 27 member states and accordingly more interest groups, after around 40 rounds the choice of “pro liberalization” even completely dies out. This is probably because the whole network becomes very dense but fewer actors choose liberalization.

In the next setting I have looked at the dynamics of the systems if only geography ( )

and resources ( ) are important for the creation of the ties. Figure 15 displays the results of this simulation runs.

Figure 15 Simulation Results edges dependent on resources (0.01) and geography (0.001)

The process now displays much more dynamics. The system with 27 member states even leads to 60 percent of member states being in favor of liberalization after 60 runs. Especially interesting are the observations of network densities and the centrality of the Commission for the different dynamics in the choices. It seems that albeit the network with 27 is less dense than with 10 or 15

member states, the power of the European Commission and the fact that it realizes more edges when there are more member states leads to a higher number of the pro-liberalization choice.

A slightly different principle of edge creation is based on the similarity of preferences. The role of geography ( ) and preferences ( ) is shown in the Figure 16. In this setting one would expect only marginal changes of policy, given that the number of actors who are in favor of liberalization is only 10 percent. The pro-liberalization choice therefore cannot spread and the EC remains totally isolated. This is a highly unrealistic setting, since it does not take into account the institutional setting.

Figure 16 Simulation Results edges dependent on preferences (0.01) and geography (0.001)

Finally, I combine the role of preferences and resources but first make them equally influential as is depicted in Figure 17. This scenario is very similar to the results when edge creation depends only on resources but this system is slightly more stable. This simulation setting comes very close to the real world dynamics as depicted in Chapter 1.3.

Figure 17 Preferences (0.01), resources (0.01) and geography (0.001)

As suggested in the principle 4, in the final model I assume that resources are more important than preferences. The simulations of this scenario are shown in the Figure 18.

Figure 18 Preferences (0.01), resources (0.02) and geography (0.001)

In particularly in the setting with 27 member states the choices in favor of liberalization converge even though the overall network is less dense than with 10 member states. These results suggest that the dynamics of the process mainly depend on the resources of the Commission. I therefore simulate two additional settings; one in which the Commission is less powerful (Figure 19) and one in which the Commission is more powerful (Figure 20).

Figure 19 Commission’s resources smaller than in the original setting

Figure 20 Commission’s resources higher than in the original setting

The last simulations make it clear (based on the rules developed for the simulations) that the power of the Commission is decisive for the dynamics of the liberalization process. Given the low number of the initial pro-liberalization agents, the policy change would not be possible even

in a highly dense network without a powerful institutional actor who has the capacity to build links to the majority of other actors.