A0245
Title: Continuously updating GMM estimator: Bridging theory and practice
Authors: Whitney Newey - MIT (United States)
Marcelo J Moreira - FGV (Brazil)
Mahrad Sharifvaghefi - University of Pittsburgh (United States) [presenting]
Abstract: The continuously updating generalized method of moments (CU-GMM) estimator has appealing theoretical properties. However, it lacks a closed-form expression for HAC errors and requires solving a non-convex optimization problem. It is demonstrated that the objective function of the CU-GMM estimator is a ratio of polynomials. Consequently, the optimization problem is translated into finding roots for polynomial equations. The analysis reveals the sensitivity of the CU-GMM estimator to different optimization methods, showcasing significantly reduced dispersion compared to prior literature.
