中国债券市场时变风险溢价——远期利率潜在信息

 近年来，中国债券市场发展迅速，在全球债券市场排行中紧跟美国和日本债券市场，已跃居世界第三。与此同时，中国债券市场亟需得到更多的关注和研究。       在2005年之前，中国债券市场被认为符合预期假说，即长短期债券间不存在风险溢价。 由于预期假说假设投资者偏好为风险中性，而实际市场中的投资者偏好往往存在较大差异，因而债券市场风险溢价应长期存在。为研究该问题，直接关注零息债券持有期超额收益，力求捕捉风险溢价的时变特性。在Fama-Bliss和Cochrane-Piazzesi溢价预测模型的研究框架下，利用中国远期利率特性，构建远期利率差和远期利率组合两种预测因子。采用两种因子分别对中国债券市场风险溢价进行预测，探讨中国债券风险溢价的时变性。选取2006年至2015年中国零息国债即期利率数据，该区间能够完整覆盖中国债券市场的发展期，并涵盖金融危机时期或货币政策松、紧期。在此基础上，将宏观经济和货币政策代理变量引入预测模型，与远期利率组合进行多元预测对比，揭示远期利率所暗含的经济信息。此外，为充分验证预测结果的鲁棒性，进行多重共线性分析和样本外检验。       研究结果表明，预期假说在中国债券市场不成立，即中国债券市场存在明显的时变风险溢价，并且风险溢价随着期限的增加而升高。研究还发现远期利率组合的预测能力来源于两方面，一方面，自身蕴含了大量的宏观经济和货币政策信息，能够反映出经济状况对风险溢价的影响；另一方面，该组合属于水平型因子，能够很好地解释风险溢价中占比最高的成分，因此占优于远期利率差这种斜率型因子，更好地刻画风险溢价中的系统性部分。       准确刻画时变风险溢价不仅可以辅助投资者进行交易决策，而且有利于更精确地构建中国债券理论期限结构。一个合适的利率期限结构能够指导中国政府制定正确的货币政策，有助于促进中国债券市场的发展以及完善中国债券市场结构。

Time-varying Risk Premia in Chinese Bond Market: Latent Information in the Forward Rates

 In recent years, Chinese bond market has grown rapidly. Aside from US and Japanese bond markets, Chinese bond market becomes the third largest bond market in the worldwide, and calls for more attention for the further study.       Before 2005, it was considered that the expectation hypothesis succeeded in Chinese bond market, and there was no risk premia between the long-term and short-term bonds. Since the the real investor preference obeys risk neutral, which expectation hypothesis assumes, the risk premia should exist in the bond market. To investigate this, this work pays attention to the holding period return of default-free zero-coupon bond and tries to capture the time-varying feature of the risk premia. Based on the Fama-Bliss and Cochrane-Piazzesi regressions, this work constructs two types of forecasting factors including the forward spreads and forward rate portfolio, using the feature of forward rates in China. To investigate the bond time-varying risk premia, this work uses the two factors to predict the risk premia respectively. The date from 2006 to 2015 covers the developing periods of Chinese bond market including the financial crisis episodes. Furthermore, this work exams the economic information hidden behind the forward rate by introducing some proxy variables of macroeconomy and monetary policy into the forecasting model. To check the robustness of these outcomes, this work tests the multicollinearity issue and out-of-sample performance.       The results show that the expectation hypothesis fails in China, namely, the risk premia exist and obviously increase in maturities. In addition, we find two main reasons to explain the forecast power of the forward rate portfolio: This portfolio holds substantial latent information of macroeconomy and monetary policy which affect the risk premia; a level factor that can explain the largest component of risk premia, and this portfolio is superior to the forward spread for capturing the systematical part of risk premia.       Capturing bond risk premia more precisely can not only help to make right trading decision for investors, but also help to build a more accurate term structure of Chinese bond market. Consequently, an appropriate term structure will guide the Chinese government to make a correct monetary policy, so as to promote the bond market development as well as improve the bond market structure in China.