| The 2008 financial crisis was a very unpredictable event. To analyze this event, we use Agent-based network models. Our models suggest crises are very-rare-but-extremely-dangerous events, which are called "Dragon kings". Due to insufficient information, policies for a resilient system have limitations. Paradoxically, the safer a system is, the more likely it can face the Dragon king. Furthermore, our multiple networks model proposes that once systemic risks spread in interdependent networks, the risk patterns are more unpredictable, because the risks hiddenly spread. As a result, the high level of complexity of the multiple networks produces the pop-corn effect of the 2008 crisis. We also investigate the slow recovery after the financial crisis. Our evolutionary prisoner's dilemma game model shows cooperation strategy outperforms defection in a small-world network. But, when the small-world network becomes a random network, defection is new dominant strategy. Thus, under high level of randomness, cooperators are exploited by defectors. In the real world, governmental polices and investments of other agents are also exploited under high uncertainty. It leads to the slow recovery. |
