Contagion: a panel analysis. Preliminary version
Common tests for contagion are often based on the bi-variate correlation coefficient. These tests can be biased through heteroscedasity and be inappropriate as a measure of non-linear market association. More crucially, these tests only provide results for pairs of markets and do not reveal the linkages for all markets simultaneously. Furthermore, these tests hide the direction of common movements, i.e. positive joint return movements or negative joint return movements. We propose the use of a panel model that provides results for all markets simultaneously and also reveals changes of common return movements in a crisis period by the estimation of time-specific effects. The power to detect such movements is evaluated in a simulation study. We analyze eleven Asian stock markets and show that there are simultaneous upward and downward movements in crisis periods that are stronger than in normal times. The model also provides results conditioned on different common factors.
Bibliographic Reference: EUR 20826 EN (2003),16pp. Free of charge
Availability: Available from European Commission, JRC Knowledge Management Unit, Ispra (IT) Tel: +39 033278 9843 or +39 033278 9864 Fax: +39 033278 9623 E-mail: email@example.com
Record Number: 200317138 / Last updated on: 2003-11-18
Original language: en
Available languages: en