The Trend Real Interest Rate and Stagnation Risk: Bayesian Exponential Tilting with Survey Data

Download Article

RWP 17-08, July 2017; updated May 2019

The decline in the real interest rate during the recent few decades coupled with the Great Recession of 2007-2009 raised a concern that the U.S. econ-omy might face stagnation like the Japanese economy since the late 1990s. The increased likelihood of the zero lower bound (ZLB) on the nominal in-terest rate that constrains the effectiveness of monetary policy at the low equilibrium real interest rate is often cited as a cause of stagnation. However, the central bank’s unconventional policies such as large scale asset purchases and forward guidance can mitigate stagnation risk arising from the ZLB con-straint. To empirically assess the impact of these opposing forces on the risk of stagnation, this paper uses long-horizon predictive distributions of macro variables from a time-varying parameter vectorautoregression (TVP-VAR) model. While the concern for long-term (five-year ahead) stagnation risk due to the ZLB constraint on monetary policy appears to be justified from the purely model-based predictive distributions for macro variables, the risk substantially declines when these predictive distributions are tilted to match both cross-sectional means and variances from survey forecasts of inflation and the nominal interest rate. And the probability score ranking based on the prediction accuracy of downside tail events favors the predictive dis-tribution with tilting. This finding affirms the view that unconventional monetary policies as conducted by the Federal Reserve were effective in re-ducing stagnation risk by influencing the private sector’s expectations.

JEL Classification: C11; E43