Expectations or Surprises: What Really Moves the U.S. Treasury Market?
51 Pages Posted: 1 Jun 2017
Date Written: May 30, 2017
The standard approach in asset pricing is to use information shocks to determine how markets react to news. We examine this paradigm empirically by decomposing high-frequency bond responses into ex-ante (expected) and ex-post (surprise) news components. Our analysis shows that the magnitude, direction and duration of reactions depend on the choice of measurement component. While bond returns barely react to news, volatility is closely linked to fundamentals. Ex-ante forecasts of investors generate significant jump (tail) clustering in the data, but we find no evidence for such effects with (ex-post) surprise measures. This suggests that considering ex-post surprises solely as proxy for shocks undermines the realized announcement impact, particularly for characterizing jump-type tail behavior in crisis periods. The news-implied reaction dispersion between expectations and shocks is sizable, related to trading volume and time-varying over the business cycle. Our findings provide relevant implications for macro-finance modeling and bond market microstructure.
Keywords: U.S. Treasury market, Volatility, Jumps, High-frequency data, Macroeconomic news announcements, Public information, Price discovery
JEL Classification: C14, C53, E44, G01, G17
Suggested Citation: Suggested Citation