It is a major aspect of social network analysis to apply data mining techniques to analyze relationship pattern and reveal social phenomena. And the analysis of attack-resistance of national economy is critical to a country and her people's livelihood. Previous studies of attack resistance of social network mainly focus on analyzing the attack effect on the network's connectivity, when a few nodes and edges are removed, for instance the internet attack research. However, the method is not applicable to the social network that is full-connected, such as a population flow network. Therefore, the paper proposes a novel attack evaluation method that considers the changes of nodes' importance before and after the attack. The method is applicable to the more general network. First, we get the number of floating population between all provinces of China based on the 1% national population sample survey in 2005, and the population flow data is used to construct the social network that reflect economical connection between all provinces; Second, the PageRank algorithm is adopted to compute the importance of each regional economy; Third, we analyze the attack resistance of nation economy when random and deliberate attacks at some regional economies and the links of them occur. The experimental results show that the deliberate attack at important province and link is more virtual to national economy.