This study develops a reduced-order power plant dispatch model and uses it to simulate marginal emissions factors (MEFs) for the 2014-2017 United States (U.S.) electric grid at the North American Electric Reliability Corporation (NERC) regional level. MEFs help quantify the health, environmental, and climate change impacts caused by changes in marginal net electricity consumption, which could result, for example, from new technologies or policies. This study develops the model, validates it against historical data, and compares its simulated MEFs against historically derived regression-based MEFs. Our method accurately reproduces CO2, SO2, and NOx emissions for multiple U.S. NERC regions and years and enables us to analyze future scenarios that are absent from the historical data. Though historically derived regression-based MEFs are generally more accurate, our simulated MEFs provide a more nuanced picture of how clusters of low- or high-emitting power plants of similar production cost create large swings in MEFs throughout the day. Policymakers could use these dynamic MEFs to target demand-reduction strategies at high-emissions portions of the power plant merit order.