Research On The Pricing Of Treasury Bond Futures With Hull-White Model And Dynamic Programing Method

Treasury bond futures market has grown significantly for the last 3 years, and it is playing a more and more important role in hedging the interest risks and reflecting the expectation of the market. As the marketization of interest rate continues to proceed, more effort will be put in the studying on treasury bond futures. The fact that treasury bond futures allows the short position to choose the cheapest to delivery bond from a variety of qualified treasury bond and to choose exactly when to operate the transaction procedure during the delivery month creates a lot of difficulties in pricing the treasury bond futures, thus a lot of efforts has been put in the study of pricing of both the futures and the delivery options, and a lot of attentions has been paid by the investors.After studying the former literatures, we found that most studies simply split the future prices into 2 parts: the forward price and the option prices. Scholars price the 2 parts separately with different models and different assumptions. This may cause a theoretical dilemma. But the dynamic program method allow us to price the future with 1 model and under 1 assumption, which gives the method a great advantage in pricing. After thoroughly studying the transaction and trade rules of treasury bond futures traded in China Financial Futures Exchange, we build a model based on the dynamic programming method proposed by the former research to describe these rules and use the data of Chinese market to estimate the Hull-White model and price the treasury bond futures and the embedded options under no-arbitrage assumption. What's more, we improve the traditional method by using the treasury bond future market prices to correct the parameters produced by the dynamic programing model, thus further improve the accuracy of pricing. The result shows that the improved dynamic programing method can price the treasury bond futures and its embedded options precisely. In the end we use two different term structure models, Bjork and Christensen model and NSM model to ensure the robustness.