In this work, we show how scheduling and control problems can be treated within a unified optimization formulation. By transcribing discrete-time scheduling models into state-space form, a single dynamic model encompassing both layers can be optimized online as an instance of economic model predictive control (MPC), which inherits theoretical closed-loop properties from MPC theory. With the addition of suitable terminal constraints, a worst-case bound can be derived for the economic performance of the nominal closed-loop system, which helps to eliminate pathological behavior that can result from rescheduling too infrequently or without regard to recursive feasibility. Via an example system, we illustrate how constant feedback (i.e., reoptimization) and the inclusion of terminal constraints leads to better performance, even in the nominal case when no disturbances are present. With a unified treatment, the system is able to respond optimally and in real time to changing market conditions or process constraints, leading to improved economic performance and stability properties.