We consider a clustered wireless network composed of energy-harvesting access points (APs) to improve both spectrum efficiency and energy efficiency. Each AP only uses the harvested energy for transmission and becomes inactive when the harvested energy is less than the required energy for transmission. The number of inactive APs increases as the number of users being served by an AP increases. Considering the relationship between the number of served users and the active AP density, we propose a spectrum-efficient operating policy designed to ensure that an AP serves the closest n users rather than all users in the cluster. We derive the outage probability and the area spectral efficiency (ASE) for an energy-harvesting clustered wireless network in closed form using stochastic geometry. Numerical results show that there is an optimal number of served users per AP that maximizes the ASE, which verifies that the ASE can be increased by the proposed spectrum-efficient operating policy.