This project develops a novel framework for estimating joint production functions of multi-product firms. Theoretically, it provides a new characterization linking separable functions to homothetic production correspondences and shows how common forms - such as Cobb-Douglas and CES - fit within this framework in multi-product environments. Empirically, the project proposes a structural model that leverages demand-side information to estimate production parameters. This approach improves on previous methods by (1) demonstrating the inconsistency of control function estimators when output market power is ignored, and establishing monotonicity under specific conditions; and (2) relaxing identification assumptions, such as rigid timing or adjustment frictions. Monte Carlo simulations confirm the robustness of the estimation strategy in the presence of adjustment shocks.
