摘要
本篇論文主要針對十二個國家長期公債的預期報酬做研究。這十二個國家分別為:美洲區的美國、加拿大；亞洲區的日本；澳洲區的澳大利亞與歐洲區的德國、法國、英國、比利時、義大利、西班牙、丹麥和荷蘭。這十二個國家的公債市值佔全全球公債市值的98%。研究期間從1991年的一月到2003年的二月。主要探討的問題為(1)是否可以用某些因素來解釋長期公債的預期報酬?(2)長期公債市場是否附合資產訂價理論? 在本篇論文中，作者用了五個因素來解釋長期公債的預期報酬。這五個因素分別為:投資人前期財富與當期財富的比率、實質債券殖利率、長短利差、債券β值與匯率波動率。發現長短期利差與實債債券殖利率在這十二個國家中，最有一致性的解釋能力。另外，匯率的報酬也對長債公債的預期報酬有負向的影響。 本篇論文用了二個訂價模型來檢測長期公債市場是否附合資產訂價理論。第一個訂價模型為Latent Variables Model。在Two Latent Variables Model中，作者無法拒絕長期公債市場附合資產訂價理論這個假設。第二個訂價模型為Bivariate GARCH-in-Mean Model。在這個模型中，作者發現有些國家的公債市場受全球市場的影響。 本篇論文的架構如下:I緒論。II訂價模型的檢測。III實證結果與探討。IV 結論。
並列摘要
Abstract： In this article, I want to address two questions: (a) Can I predict bond returns by using some instruments? (b) Are expected bond returns in these twelve countries consistent with some asset pricing models? I study twelve countries’ long-term government bonds: United States, Canada, Japan, Australia, Germany, France, United Kingdom, Belgium, Italy, Spain, Denmark and Netherlands from January 1991 to February 2003. I use five factors to explain expected returns on government bonds. These five factors are: inverse relative wealth (the ratio of exponentially weighted past wealth to current wealth), the bond beta (a bond market’s exposure to a stock market index), the term spread, the real bond yield, and the change of exchange rate. I find the term spread and the real bond yield has consistent effect on bond returns positively. And the returns on currency have negatively impact on the returns on government bond returns. I use two pricing model to test whether government bond market is consistent with asset pricing model. The first model is latent variables model. In this model, I can’t reject asset pricing model by using two latent variables model. The second model is bivariate GARCH-in-mean model. In this model, I find the conditional covariance between country government bond returns and world portfolio returns have significant effect on government bond returns. The reminder of the paper is divided into four sections. Section II introduces asset pricing models used. In this section, there are two parts. The first part presents the latent variables model, and the second part discusses a bivariate GARCH-in-mean model. Section III defines five instruments and explains the possible ways these instruments will affect bond returns. Section IV shows the results of empirical test. There are two parts in this section. The first part takes every country’s Treasury bill rate or short-term deposit rate to be the short-term rate and risk-free rate. The second part uses U.S. three-month Treasury bill rate to be every country’s short-term rate and risk-free rate. Section V is conclusion.