| As the cornerstone asset of the financial system,the predictability of its future return has always been the focus of attention and controversy in the financial academic and industrial circles.The Treasury bond yield curve involves the pricing of all financial products and constitutes the pricing basis of all financial products.Since the resumption of the issuance of treasury bonds in 1981,China’s treasury bond market has experienced 40 years of tortuous but rapid development.Its function has developed from initially making up the fiscal deficit and raising construction funds to supporting the national macroeconomic regulation and control,accelerating the development of the real economy,meeting the investment needs at home and abroad,undertaking the pricing benchmark of the financial market,resisting various crises It has played an irreplaceable important role in promoting the internationalization of RMB.The report of the 18 th National Congress takes "accelerating the development of multi-level capital market" as an important aspect of comprehensively deepening economic system reform.The report of the 19 th national congress once again stressed that "deepen the reform of the financial system,enhance the ability of financial services to the real economy,increase the proportion of direct financing,and promote the healthy development of multi-level capital market".As the core component of the direct financing system,whether it is to promote the process of interest rate marketization,the opening of financial market,the internationalization of RMB,the development of offshore RMB market,the construction of multi-level capital market system,and support the entity economic development with government investment and finance,it all needs the strong support of the national debt market.In this context,it is of great policy significance to understand the fluctuation of national debt price and its relationship with finance and economy.At the theoretical level,as the cornerstone asset of the financial system,the predictability of its future return has always been the focus of attention and controversy in the financial academic and industrial circles.The traditional pricing theory holds that the returns of financial assets such as stocks,bonds and foreign exchange are close to unpredictable.For the bond market,the expectation hypothesis of interest rate term structure holds that the yield to maturity of long-term bonds is equal to the average of the expected future spot interest rate,the expected holding return of short-term or long-term bonds is the same in any period,and the excess return of bonds is almost unpredictable.However,empirical studies in recent three decades have revealed that the return on long-term treasury bonds has significant predictability,and the current yield curve covers a wealth of prediction information.The historical data of international financial markets such as the United States and Germany show that for bonds maturing for more than one year,forward spot spread,yield spread or forward interest rate can effectively predict their future holding return(for example,Fama & bliss,1987;Campbell & Shiller,1991;Cochrane & piazzesi,2005);Macroeconomic variables such as output gap,inflation and real economic cycle also contain the prediction information of future bond returns(for example,Ludvigson & Ng,2009;Cooper & Priestley,2009;Joslin et al,2014).On this basis,the current research on the predictability of bond return focuses on the prediction information outside the yield curve.The focus of the current debate is whether there are economic variables that can provide prediction information outside the term structure of interest rate? What macro financial risks are related to the predictability of national debt risk premium? What is the economic mechanism of macro financial risk prediction ability? Using the interest rate data of U.S.Treasury bonds,this paper empirically investigates the prediction ability of macro financial risks such as disaster risk and macro uncertainty on the risk premium of treasury bonds and the underlying economic mechanism from the perspective of income predictability,so as to contribute to the research on the predictability of treasury bond returns.The Treasury bond market is in the core position in the modern financial system.Understanding its income fluctuation is very important to promote the healthy development of multi-level capital market;The impact of rare disaster risk on financial and economic development is becoming increasingly prominent,and it is of great value to study its impact on the financial market;Under the great changes in the new era,the impact of macroeconomic uncertainty has increased rapidly,and the impact on the financial and economic system can not be ignored.In conclusion,this study has good theoretical and practical significance.Therefore,this paper first examines the relationship between rare disaster risk and time-varying bond risk premium.Existing literature studies introduce time-varying disaster risk into the asset pricing framework and explain many long-standing asset pricing problems,such as stock premium problem,risk-free interest rate problem,forward premium problem and return predictability(for example,Rietz,1988;Barro,2006;Gourio,2012;Gabaix,2012;Wachter,2013;Farhi & Gabaix,2015).In theory,forward-looking investors will demand a high risk premium when they perceive the possibility of rare but extreme disasters.Therefore,public disaster risk concerns should be able to predict the return on assets.Specifically,in the bond market,if investors are worried about the potential disaster risk,they will transfer their funds to safer and more liquid assets,such as short-term treasury bonds.This pushed up the price of short-term bonds relative to long-term bonds(Beber et al.,2008;Caballero & Krishnamurthy,2008;Guerrieri & Shimer,2014),resulting in an increase in the expected treasury bond yield.The main obstacle to the empirical research on rare disaster risk and asset price is that it is difficult to observe the rare disaster risk directly.Therefore,this paper uses six disaster risk indicators constructed by Manela & Moreira(2017)to measure the risk of rare disasters.Manela & Moreira(2017)indirectly measured the risk of rare disasters by mining word frequency information in news media data to measure the public’s concern about rare disasters such as financial system crisis,political turmoil and natural disasters.In terms of research methods,this chapter draws lessons from previous studies and uses partial least square method to extract information from manela & Moreira(2017)disaster risk indicators to construct rare disaster risk factors.Then,in the empirical test of rare disaster risk,firstly,this paper investigates the intra sample prediction effect of disaster risk on national debt risk premium.The results show that the constructed rare disaster risk variables have significant intra sample prediction ability,and can explain 9.93% to 10.44% of the bond yield in the next year.Adding disaster risk factor to the traditional prediction model can also significantly improve the prediction accuracy of bond yield.Secondly,the prediction ability is also significant outside the sample.Using DRF as the only,it produces17.39% to 19.66% of the samples for bonds with different maturities,and they are significant at the level of 1%.After controlling for other predictors,the increase in the use of DRF was still statistically significant.Furthermore,rare disaster risk can significantly improve the economic utility of mean variance investors.Using DRF as the only predictor,building a portfolio of 1-5-year zero coupon bonds can produce an annualized utility return of 1.61%.Similarly,when controlling other predictive variables,the marginal economic utility of increasing the risk of rare disasters as a predictive variable is still significant.Finally,the macro coverage test shows that the prediction ability of rare disaster risk variables is not covered by the yield factor.Based on the empirical bootstrap test of Bauer and Hamilton(2018)and the coverage test of Bauer and Rudebusch(2017)based on the dynamic term structure model,both rejected the macro coverage hypothesis,indicating that the rare disaster risk variable is a non covered bond yield prediction variable.Next,this paper examines the impact of macroeconomic uncertainty on bond risk premium.The level of macroeconomic uncertainty is time-varying and usually shows a reverse cycle,that is,it decreases during economic expansion and increases during economic recession.The uncertainty about the future economy has aroused people’s concern about the economic downturn and made investors more risk averse.Thus,economic uncertainty will affect the risk price and increase the risk premium in the financial market(Drechsler,2013;Caballero & Krishnamurthy,2008;Guerrieri &Shimer,2014).Therefore,higher(lower)economic uncertainty should predict higher(lower)future bond returns.Investigating the relationship between uncertainty and future bond risk premium is helpful to understand the change of bond price.Existing studies focus on first-order moment macroeconomic variables,such as real economic growth,profit margin of industrial enterprises,etc.However,second-order moment macroeconomic variables such as macroeconomic volatility and macro uncertainty may have a new economic mechanism,so that such macro risk variables can affect bond yields through this mechanism,At the same time,the forecast information of this factor will not be covered by the current term structure of interest rate.In order to measure macro uncertainty,this paper adopts the factor enhanced vector autoregressive model of Jurado,Ludvigson & ng(2015),recursively constructs the measurement indicators of uncertainty in multiple economic sectors from a large number of economic indicators,and uses the partial least square method to summarize the economic uncertainty indicators and construct the measurement indicators of macro-economic uncertainty by referring to the previous literature.The empirical results of macroeconomic uncertainty and time-varying bond risk premium show that macroeconomic uncertainty has significant prediction ability for bond risk premium inside and outside the sample.Economic utility analysis shows that macroeconomic uncertainty usually brings marginal utility improvement to risk averse investors.Using the bootstrap method of Bauer & Hamilton(2018),this paper finds that macroeconomic uncertainty can effectively predict the future bond risk premium,and this prediction ability is not fully covered by the yield factor and the first-order moment macro state variables.Furthermore,this paper establishes a dynamic term structure model of interest rate including non macro uncertainty,and makes an empirical estimation of the model by using interest rate and macroeconomic data.Then,the stochastic discount factor process contained in the model is statistically tested.It is found that macro uncertainty has a significant impact on the yield factor of the next period,and leads to the countercyclical change of bond risk premium.Compared with previous studies,the main contributions of this paper are as follows: firstly,this paper makes a certain marginal contribution to the study of time-varying bond risk premium.This paper analyzes the relationship between the second moment of economic variables and bond risk premium,which provides new empirical evidence for the existence of non covered risk factors.Secondly,this paper is an earlier empirical study on rare disaster risk and time-varying bond risk premium,widens the research scope of asset pricing model based on rare disaster risk,and expands the research field from traditional stock market to bond market,thus making a certain marginal contribution to the research of rare disaster risk and asset pricing.Thirdly,this study provides direct empirical evidence for the first time that economic uncertainty has an important impact on the risk premium of the bond market.Economic uncertainty has a significant impact on macroeconomic fluctuations and financial asset prices.However,the research on economic uncertainty and financial asset pricing focuses on the cross-section of the stock market and the yield of various financial assets,and less on the time series changes of bond pricing and bond yield.Finally,aiming at the possible impact of the estimation error of rare disaster risk on the macro coverage hypothesis,this paper further extends the bootstrap test method of Bauer & Hamilton(2018),so that the test program considers the possible distortion of the test results caused by the generation error of prediction variables.Therefore,it makes a marginal contribution to the econometric method of testing the macro coverage hypothesis,and has a certain reference value for the empirical research on the term structure of non coverage interest rates in the future. |
PDF Full Text Request