Developed market government bonds are viewed as “safe havens”, but in reality they have been prone to sudden outsized price declines, similar to FX carry trades, even during the past 20 years of modest inflation. Drawdowns are worse in poor liquidity. This empirical finding is not new but more relevant as bond yields have been compressed and institutional duration exposure has surged relative to banks’ market making capacity.

Fujiwara, Ippei, Lena Mareen Korber, and Daisuke Nagakura, “Asymmetry in Government Bond Returns”, AJRC Working Papers 01/ 2013.

The below are excerpts from the paper. Emphasis and cursive text have been added.

What has been analyzed?

“We study asymmetries in government bond excess returns for 5 developed countries – Canada, Germany, Japan, UK and the U.S… for the period from 1997 to 2011…we measure unconditional asymmetry in government bond returns…we investigate if there is conditional asymmetry in government bond excess returns … we analyze how quarterly asymmetry co-moves across countries and how it is related to macroeconomic and financial variables.”

What has been found?

“The tails in the distribution of government bond excess returns are thicker than those of a normal distribution…We find that the coefficient of skewness is negative for almost all countries and maturities, or that there is a small probability of a large and negative excess return…Skewness in bond returns are mainly driven by extreme observations. Although government bonds are usually considered as risk free assets, they are subject to a tail risk in form of a sudden drop in bond prices, similar to other risky assets such as equity returns and exchange rates.”

“We provide evidence for conditional asymmetry [asymmetry of returns under consideration of the concurrent information set]… For bonds with a maturity of 2 years, conditional asymmetry is statistically significant in all countries… there is also evidence for significant conditional asymmetry in 10 year bonds for Canada, Germany and Japan…there is less evidence for asymmetry in long term bonds for the U.S. and the UK…One situation that gives rise to conditional symmetry is when a time series reacts more strongly to negative announcements when compared to positive announcements. There is empirical evidence [in other papers] that this is the case for government bond yields.”

“We find that the cross-country correlation of asymmetry is increasing in the maturity of the bond. This finding implies that for longer maturities, common factors play a larger role in explaining asymmetries, while idiosyncratic factors are more important at short maturities… there is tail correlation in government bond excess returns.”

“We find that liquidity in government bond markets predicts the coefficient of skewness with a positive sign, meaning that the probability of a large and negative excess return is more likely in a less liquid market.”

“A positive realized return is associated with a negative coefficient of skewness or a small probability of a large and negative return in the future.”

What are the conclusions?

“Affine term structure models… assume normally and thus symmetrically distributed innovations to bond returns to make the computation of the term premium easier. This is at odds with our findings.“

“Concerning forecasting, the presence of asymmetries calls for new models that allow for the possibility that positive and negative forecast errors are not equally likely. “

“Additionally, asymmetry in government bond returns implies that mean and variance are not sufficient to characterize the risk in government bond returns, and has thus important implications for optimal portfolio allocation.”