Title: Monitoring structural breaks in vector autoregressive models Authors:  Masoud Abdollahi - Institute for Advanced Studies (Austria) [presenting]
Abstract: A closed-end monitoring approach is proposed for the online detection of structural breaks in non-stationary cases. The framework is based on an error correction model that accommodates potential breaks in both the cointegrating relationships and adjustment vectors, with the possibility of a changing cointegration rank. To this end, we develop Lagrange-multiplier statistics tailored for sequential monitoring. Following a calibration period used to estimate the model parameters, monitoring begins and terminates once the test statistic crosses its corresponding critical value. Through an extensive simulation study, we investigate the performance of the monitoring procedure, including assessments of the observed size and power of the test.