Choosing the Weighting Coefficients for Estimating the Term Structure from Sovereign Bonds

The term structure of interest rates depends heavily on the quality of the market data on which it is built. Data quality is an actual topic at the moment, as the received rates can be incorrect due to errors and omissions in the data. In this paper, we study how artificially modeled errors in bond prices affect the results of constructing the term structure of interest rates by data of government bonds of several countries in the Eurozone and Russia. In addition, ways of improving the construction of the yield curve are examined. According to the results of the study, it can be concluded that data errors can greatly distort the built-up yield curve. At the same time, the use of different weighs or other ways of recording errors does not always improve the result, and sometimes only worsen the situation. This concerns not only robustness, but also the predictive power of the model.