This paper proposes a novel event-triggered economic model predictive control (EMPC) scheme with shrinking prediction horizon of constrained nonlinear systems. Concretely, the triggered mechanism is designed by relaxing the stability constraint, while the involved parameter can be used to achieve a sensible compromise between system performance and computational complexity. The rule of shrinking prediction horizon shortens the length of the horizon as much as possible, while the length of triggered interval is ensured to avoid Zero behavior. The utilization of control sequence is improved. As the prediction horizon becomes shorter, the state always keeps in the terminal set. The sufficient conditions guaranteeing the recursive feasibility and stability of the EMPC are derived. Finally, the effectiveness of the algorithm is illustrated by a continuously stirred tank reactor example.