Markovian analysis of U.S. Treasury volatility: Asymmetric responses to macroeconomic announcements

Abstract

Extensive is the literature supporting the announcement effect of the release of pre-scheduled macroeconomic news on U.S. Treasuries conditional moments; nevertheless, none of them considered possible asymmetric responses to these news arrival depending on the latent regimes characterizing the U.S. economy. Owing to the peculiar U.S. interest rate environment over the last 23 years, not allowing for stochastic-switching states even in the GARCH-class models may miss important volatility pattern. Hence, the therein model proposes the first partitioning of Treasuries excess returns in different regimes (normal, crash and low-rate) through hidden Markov chains to, only in a subsequent moment, apply a conditional volatility GARCH-class model augmented by dummy variables accounting for macroeconomic release days. The same model is then re-run to assess the explanatory power of U.S. Treasury auctions within volatility modeling. Results point to a statistically significant and asymmetric effects imparted by both prescheduled macroeconomic releases and new issuances on U.S. Treasury excess returns depending on both the regime and the securities’ maturities, while evidencing the very different characteristics of GARCH and ARCH coefficients depending on the regime.