The collusion among various generating units has been a problematic issue affecting the fairness and transparency of electricity markets. Therefore, it is of great significance to assess the potential of such collusion in the electricity market. However, the previous assessment studies primarily focused on the bidding behaviors of collusive generating units, without considering the influences of generation flexibility, such as ramp rates. In this paper, a novel assessment method is proposed to evaluate the collusion potential in the electricity market considering generation flexibility. First, a bi-level optimization model is developed to simulate the collusive strategies of different generating units, including the withholding of generation capacities and ramp rates, as well as the uplifting of minimum outputs and bidding prices. In the upper-level problem, collusive generating units optimize their offering strategies to optimize the generation profits without violating the regulatory laws. The lower-level problem is a day-ahead economic dispatch model which minimizes the dispatching costs. Based on the optimal collusive strategies determined by the bi-level model, a framework is then proposed to assess the collusion potential in electricity markets. Moreover, price-based and profit-based indices are proposed to quantitatively evaluate the collusion potential of different generating units. Finally, the proposed assessment method is validated on a modified IEEE 39-node system. The numerical results demonstrated that generation flexibility can be exploited collusively for making excessive profits, particularly during load peaks and valleys.