KTH Matematik |

In this study, issuer specific term structures of interest rates are estimated in the presence of missing data. A smoothing spline estimation is applied on a set of bootstrapped yield notations for a number of Emerging Market issuers. Based on estimated term structures, in which a significant fraction of data is missing, univariate GARCH(1,1) filters are applied on constant maturity time series, after which a PCA based algorithm fills missing observations. Term structures are thereafter recursively reconstructed. In the center of the study lies the designed backfilling routine. Its application onto a number of synthetic data sets shows that its performance is satisfactory. Although, it is dependent on the fraction of missing values, the complexity of the underlying factor structure and the amount of noise in the data set. Applied onto the real set of data, the routine produces results of varying quality. Obtained estimates of missing term structures appear credible but their correctness remains to be assessed. Keywords: Yield Curve Estimation, Backfilling, GARCH(1,1), Principal Component Analysis. |

Sidansvarig: Filip Lindskog Uppdaterad: 25/02-2009 |