Fitting And Predicting Bond Excess Return In China

Bond have always been considered as an investment tool of low risks. Thebond market is the frst choice for investors to hedge diferent economic impact-s. Using the general additive models,this paper focus on ftting and forecast thebond excess return based on macroeconomic variables and term structure of in-terest rates.Many researches have found the bond return relates to some macroe-conomic,but the relationships between them are complexity, nonlinearity and un-certainty.The general addictive model is suitable to analysis the non-monotone ornon-linear relationship between independent variables and dependent variables.Inthis paper, we take the term spread, infation, stock market, exchange rate, theCRB index, investor confdence, etc.as predictors into account. When analysisingthe bond data,the paper take the bond excess return and microeconomic monthlybetween2008and2013into account,and the data is devided into two parts,data insample and data out of the sample.The paper uses the GAM ftting based on datain sample and forecasting on data out of sample.Otherwise, we fnd the ability offtting and forecasting of GAM is better by comparing the linear model and thegeneral linear model.