Abstract:
We use daily data for the period 5 January 2000 to 31 October 2018 to analyse the impact of structural oil supply, oil demand and financial market risk shocks on the level, slope and curvature factors derived from the term structure of interest rates of the U.S. Treasury securities covering maturities of 1–30 years. Linear causality tests detect no evidence of predictability of these shocks on the three latent factors. However, statistical tests performed on the linear model provide evidence of structural breaks and nonlinearity, and hence indicate that the Granger causality test results are based on a misspecified framework, and cannot be relied upon. Given this, we use a nonparametric causality in-quantiles test to reconsider the predictive ability of the three shocks on the three latent factors, with this model being robust to misspecification due to regime changes and nonlinearity, as it is a data-driven approach. Moreover, this framework allows us to model the entire conditional distribution of the level, slope and curvature factors, and hence can accommodate, via the lower quantiles, the zero lower bound situation seen in our sample period. Using this robust model, we find overwhelming evidence of causality from the two oil shocks and the risk shock for the entire conditional distribution of the three factors, suggesting the predictability of the entire U.S. term structure based on information contained in oil and financial market innovations. Our results have important implications for academics, investors and policymakers.
