A new time-varying Generalized Method of Moments: Theory and application to the Taylor rule

Estimating the policy coefficients of the Taylor rule raises well-known issues of endogeneity and parameter stability, the technical resolution of which has been the subject of extensive debate in the literature. The authors of this manuscript propose a Generalized Method of Moments estimator with time-varying coefficients (TV-GMM), specifically designed to ensure the validity of the exogeneity condition and the robustness of the estimated parameters to structural breaks. By applying the proposed methodology to U.S. time-series data over the sample period 1984Q4-2025Q2, it is shown that the TV-GMM estimator outperforms both two-stage least squares and static GMM models in terms of predictive accuracy and robustness to structural instability. The authors further discuss the practical implications of these findings and derive policy-relevant recommendations.