Commodity modeling is like many other endeavors in science and in life—we continue to strive for completeness and perfection but may never be satisfied with the current level of our accomplishments. This is a productive attitude, because it always generates incentives for continued effort and progress. An alternative attitude adopted by some of our professional colleagues is agnostic: we do not know the underlying structure of commodity markets and perhaps cannot know it so we might as well give up and fit a reduced-form or a time series model.
Meyers, William H.; Devadoss, S.; and Helmar, Michael D., "The U.S. Export Response to Prices and the Impacts of Trade Liberalization: A Regional Trade Model Analysis" (1986). CARD Working Papers. 48.