US Treasury Bond Betas: 1961–2019

Abstract

This study plumbs the limits of US Treasuries (USTs) as a “safe asset” through lenses neglected in the literature on the correlation between bond and equity returns. An asymmetric M-GARCH model confirms a shift from positive to negative correlations in recent decades. However, the variance around bond-stock covariance has increased, consistent with greater not lower covariance premiums. Spectral analysis shows that, like the contribution to overall variance in returns, high-frequency cycles of no longer than a week account for most of the covariance between the yardstick risk-free and risky assets, increasingly so over the years. However, there is no consistent evidence that USTs are better hedges against shorter-lived shocks. Quantile regressions suggest that USTs are not particularly convex hedges, either. Even amid very low yields in recent years, the distribution of 10-year UST returns is wider as well as more negatively skewed conditioned on stock market swoons.

TOPICS: Fixed income and structured finance, performance measurement, wealth management, equity portfolio management

Key Findings

• • An asymmetric M-GARCH model confirms a shift from positive to negative correlations between returns on US Treasuries (USTs) and the S&P 500 in recent decades. However, the variance around bond-stock covariance has increased, consistent with greater not lower covariance premiums.

• • Spectral analysis shows that, like the contribution to overall variance in returns, high-frequency cycles of no longer than a week account for most of the covariance between the yardstick risk-free and risky assets, increasingly so over the years. However, there is no consistent evidence that USTs are better hedges against shorter-lived shocks.

• • Quantile regressions suggest that USTs are not particularly convex hedges, either. Even amid very low yields in recent years, the distribution of 10-year UST returns is wider, as well as more negatively skewed, conditioned on stock market swoons.