Macro-Finance Shadow-Rate Modelling: Estimating the term structure with macroeconomic factors in a zero lower bound environment
Although shadow-rate term structure models can replicate the yield curve’s characteristics in a zero-lower bound environment, there could be additional information which the yield curve that is constrained by the lower bound cannot incorporate. I thus develop a macrofinance shadow-rate model with three macro factors: economic activity, inflation, and the policy rate. I estimate the model on monthly U.S. interest rates under various restrictions using a Maximum Likelihood-based extended Kalman filter. The key findings are as follows. First, the results are sensitive to initialisation values and parameter restrictions, particularly with respect to the factors’ persistence and yield dynamics. Second, incorporating macroeconomic information can improve in-sample fit and help to mitigate underestimation of persistence, while replicating relevant term structure dynamics and macro-finance linkages. Third, relative to the yields-only shadow-rate model and the macro-finance affine model, this model is better at replicating two key stylised facts near the lower bound: the nonnegativity of yield rates and the compression of yield volatilities for short and intermediate maturities. The evidence suggests that the macro-finance shadow-rate model is preferred over affine and yields-only models in a zero-lower bound environment.
|Keywords||shadow rate, zero lower bound, macro-finance, state space, maximum likelihood, extended Kalman filter, arbitrage-free Nelson-Siegel model|
|Thesis Advisor||Wel, M. van der|
Nieuwstad, N. (2019, July 30). Macro-Finance Shadow-Rate Modelling: Estimating the term structure with macroeconomic factors in a zero lower bound environment. Econometrie. Retrieved from http://hdl.handle.net/2105/47733