Athens University Economics and Business (AUEB)
Affine models: change point detection and applications
Abstract
dc:descriptionThe purpose of the present thesis is the application of statistical process control (SPC) techniques, specifically control charts in Gaussian affine term structure models (ATSM). In recent years SPC methods have been widely popular not only in industrial applications but also in several non-industrial scientific areas such as in finance. Gaussian ATSMs under no-arbitrage conditions have been a very important research tool in the area of term structure of interest rates. Financial time series are usually subject to changes at unknown time points and the detection of these change points are of great importance.In our work we propose several control chart procedures that develop the ATSMs from the change point perspective. First, we construct control chart procedures for monitoring the parameters of an ATSM and examine their ability of detecting changes for various types of shifts in the yield curve. Also, we propose a technique for reestimating the target process of the control chart procedure in case of a detection of a change. The results show that there is no single chart that performs well in all types of shifts but a combination of control chart is needed. Second, we extent the class of term structure models estimated using the minimum chi-square estimation (MCSE) method by constructing fixed-income government bond portfolios. The proposed bond portfolio strategies from the ATSM in most of the cases perform better than traditional bond portfolio strategies. Next, the control charts are applied for monitoring the optimal portfolio weights. Third, we propose and construct control charts for monitoring shifts in the autoregressive and moving average matrix of a VARMA ATSM. The estimation procedure of the model is a two-step process where in the first step among standard estimation procedures we apply a minimum distance estimation method based on the impulse responses and define its advantages in the forecasting of the yield curve. In the secondstep, we estimate the market prices of risk given the estimates of the first step by minimizing the sum of squared fitting errors of bond yields.
Degree
thesis:*- Grantor dc:publisher
- Athens University Economics and Business (AUEB)
- Year dc:date
- 2021
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Bisiotis, Konstantinos
- Μπισιώτης, Κωνσταντίνος
Subjects
dc:subject × 14Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/51494