Smoothing effect is an important characteristic of large scale wind power. In this paper we analyze the smoothing effect from the prospect of output variability. Specifically, the aggregated output variability of a wind farm cluster may be significantly lower than that of an independent wind farm, and this phenomenon is referred to as the variability smoothing effect. In order to quantitatively analyze the variability smoothing effect, this paper introduces the concept of variability costs and evaluates the variability costs of each wind farm and overall wind farm cluster based on an optimal scheduling model. It is found that the variability cost of a wind farm cluster as a whole is lower than the sum of variability costs of all wind farms. Moreover, the difference between wind farm cluster variability cost and the sum of variability costs of each wind farm is termed the variability smoothing benefit. Meanwhile, the Shapley value method is deployed to equitably allocate the variability smoothing benefits of the wind farm cluster. The results indicate that the combined wind farms have the additional benefits of reducing variability costs as well as encouraging the integration of large scale wind farms.