ABSTRACT
As a central issue in macro-finance studies, the spanning hypothesis has always been the focus of research. Previous studies have focused on whether this hypothesis holds true in developed markets, while paying little attention to that in emerging markets. Because of their unique monetary systems, governments in most emerging markets play a key role in bond returns. This study identifies macroeconomic factors for forecasting excess returns in emerging government bond markets under spanning hypothesis. We find that in previous research, government intervention factors employed in excess returns forecasting have no additional predictive ability, as they are already incorporated in current yields. Using dynamic factor analysis, we find that macroeconomic information, including pure macroeconomic activities and financial factors, has robust incremental predictive power for in-sample and out-of-sample bond excess returns.
Notes
1. Interest rate liberalization means that interest rates are determined by market factors rather than being set by the regulators. Through market competition mechanisms, financial institutions are allowed to set interest rates independently.
2. The data period is from March 2006 to December 2018.
3. The three principal components in the US could explain the 1-year-ahead variation in excess returns of 17% for the sample from January 1985 to December 2017, and of 25% from January 2001 to December 2017.
4. To obtain a robust result, in Appendix I we investigate the same issue in the case of India for a cross-comparison of our results for China. The monetary system of India, another major emerging market, also has a certain impact on yield curves. By adjusting the repo and reserve repo rates, the Indian government influences the deposits and loans of commercial banks, and thereby restricts investors’ behavior. The results for India’s bond returns support our conclusions obtained from China. Government activities have little incremental predictive power, while market activities, especially the inflation level, provide robust information for excess return forecasting beyond the yield curve.