Title: Nearly exact Bayesian estimation of non-linear no-arbitrage term structure models
Authors: Marco Taboga - Banca d'Italia (Italy) [presenting]
Abstract: A general method is proposed for the Bayesian estimation of nonlinear no-arbitrage term structure models. The main innovations we introduce are: 1) a computationally efficient method, based on deep learning techniques, for approximating no-arbitrage model-implied bond yields to any desired degree of accuracy; 2) computational graph optimizations for the acceleration of the MCMC sampling of the model parameters and of the unobservable state variables that drive the short-term rate. We apply the proposed techniques to the estimation of a shadow rate model with time-varying lower bound, where the shadow rate can be driven both by spanned unobservable factors and by unspanned macroeconomic factors.