Abstract:

Today, there exist a lot of models for modelling yield curves. However, plenty of these models just like many other statistical models have disadvantages such as non-flexibility, non-variability, and non-automatization. Moreover, it is not possible to make automatic dynamization of the model, the modelling has to be realized from the beginning. Finally, client who does not have sufficient theoretical background in the area of yields curves, has problem in realizing these estimations of yield curves. The goal of this paper is to create application for modelling time series of yield curves on base of theoretical knowledge of selected methods. Authors come out from Nelson Siegel and Svensson model, they create their application in JAVA software and use R software for statistical modelling procedures. In order to connect JAVA and R they use the LPGL R-caller. Customized version of Svensson model is programmed in R (function Sven.est). The result is the software which is able to create yield curves, compare them a produces outputs of these models. Created software minimizes negative impacts of yield curves models. Moreover, authors’ application enables easy importing of data, modelling yield curves. Finally, tests for validating this applications were realized. The application was tested on American state bonds.