Journal on Policy and Complex Systems
discuss the rise of long memory properties taken together behavior of the agents , by simulating their individual actions and interactions . These three experiments were controlled for in order to remove any possible perspective of conception biases , which may lead to the construction of systems with long memory properties . In other words , the present authors choose experiments built by other authors without such analysis perspective , in order to show that such properties may arise according to system features , instead of building models aiming to obtain long-range dependency .
Thus , in the first experiment , adapted from Jumadinova et al . ( 2011 ), it was shown that long memory properties usually arise from heterogeneity between agents in a world where space and topology / local interactions do not have any role .
In the second experiment , it was studied a class of individual computational simulations called Cellular Automata , which aims to complement the analysis done in the first experiment . Instead of having heterogeneous agents and a network that does not allow local interactions , it was chosen an experiment where all agents are homogenous ( i . e . act the same way , using the same set of rules ) and they interact locally , producing a global behavior . Consequently , it was shown that local interactions of individual homogeneous agents may produce long memory patterns observed in the global behavior , as a consequence of the implicit dynamics in the transition rules that drives each agent behavior , related to the possibility of generating behaviors compared to smooth regime switches , which is known to generate a spectral density similar to a long-range dependent phenomenon .
Furthermore , in the third experiment , in order to complement both previous experiments , it was chosen another model — adapted from Epstein and Axtell ( 1996 ), where heterogeneity from the agents was removed and direct interactions between them were removed . Therefore , they only interact with a space that depends on the spatial configuration that the authors set up in order to control the presence of spatial complexity . Thus , such experiment suggests that spatial complexity is another interesting feature that plays an important role in the rise of long memory properties .
Therefore , keeping in mind these three different results , it is possible to show that long-range dependency is a feature that emerges from the system complexity and not necessarily arises from individual long memory capabilities nor economic / social frictions and extends the idea of heterogeneity as a source of complexity , as usually discussed in the traditional economic literature ( see Teyssière & Kirman , 2006 ).
Nonetheless , it is worth making a final discussion here , as part of perspectives for future work and analysis . The traditional literature is focused on building specific models that matches the spectral density of long memory processes , showing possible “ confusions ” and “ misunderstandings ” between what is a real “ long-range depen-
46