Government Bonds in Emerging Asia: Term Structure Models and Style Factors

Effects of the US and China on Asian Government Bond Markets: We incorporate factors from the US, China and other Asian markets in the term structure models of local currency Asian government bond markets of China, India, Indonesia and Singapore, by using a time-varying parameter vector auto-regression to extend the Nelson-Siegel yield curve model. Our approach shows that incorporating these external factors improves the in-sample model fit for India and Indonesia, showing that the conventional understanding and term structure modelling approach that local yield curve factors contain all relevant information for modelling the term structure in developed markets may not be the case for developing markets.

Our models also accommodate structural change by computing the probabilities of different model structures at each point in time. We find that, on one end ofthe spectrum, for the US, the local-only model without external factors is most probable most of the time, while, on the other end of the spectrum, for Indonesia, it is less clear which model structures are dominant, with the other markets falling in between. Spikes in probability of the model incorporating factors from all markets also reflect shocks to the countries.

Our empirical findings show that apart from the US level and slope factors, China's level and curvature factors also have significant corresponding yield curve movements in various markets, which, if the China factors have analogous meanings to their US counterparts, may mean transmission of China expected inflation, real activity and asset risk in the Chinese economy. At the same time, we see the US effects decreasing over time, which may mean either these markets increasingly being viewed as standalone markets evaluated on their fundamentals, or declining influence of US expected inflation and monetary policy. A third finding is the differences in dynamics and persistence of changes in response, with Indonesia, at one end of the spectrum, with equal level shifts, i.e., same magnitude of change across every part of the yield curve, and information being incorporated relatively quickly, and Singapore, on the other end of the spectrum, where the yields do not change in lockstep, and changes being persistent across time and showing a stronger "rippling" effect than in the other markets.

Style Factors in Asian Government Bonds: We study the extent to which style factors - Value, Carry and Momentum - explain yield curve dynamics, in the Asian local currency government bond markets of China, India, Indonesia and Singapore. We form level, slope and butterfly portfolios that mirror the factors that typically describe yield curves { level, slope and curvature - and find that, locally, the style factors together explain a sizeable portion of excess returns from these portfolios (R-squared of 6.14% - 20.70%, compared to just 0.06% - 2.94% in Singapore and the US) with Value significant across most portfolios and explaining most of the excess returns (R-squared of up to 20.0%), followed by Carry (up to 15.9%).

However, we find that Value is not a significant common factor across the markets. Rather, Carry and Momentum, are the signi cant common factors, for slope and butterfly (t-stats of 3.251 and 6.168), and level and butterfly (t-stats of 3.876 and 2.832) portfolios respectively. This finding that the strongest factor locally is the same factor in each market, yet not a common factor globally, i.e., across markets, sheds light on how investor behaviour are similar yet not synchronous.

Our other key finding is that macroeconomic factors in these Asian markets are conditional on the style factors, i.e., the macroeconomic effects become apparent only in conjunction with the style factors. The shows that the effects of these macroeconomic factors in these Asian markets are obscured by investor behaviour, and controlling for investor behaviour through style factors allows us to better identify their effects.