Abstract
This thesis tests financial contagion from the US to ten African markets during the 2007-09 financial crisis. For comparative purposes, testing procedures are also extended to cover a number of developed-economy markets.There is considerable debate within the literature as to how to measure contagion. A central focus of my research is therefore to compare alternative econometric methodologies. VAR based constant-correlation based techniques are examined alongside dynamic conditional correlation (DCC) based techniques. I find that the DCC approach is superior in respect to my dataset.
The 2007-09 crisis was unique from a contagion perspective in that its impact was truly global. This provided a unique opportunity to examine the subject across different continents and market types. African markets were found to have lower levels of integration (correlation) with the US than developed-economy markets and this resulted in considerable differences in the way that the contagion event spread across these two groups.
As well as being truly global, the 2007-09 crisis was a contagion event that lasted more than a year. I use the volatility index (VIX) to identify both a long crisis period and a series of sub-events. The former ran from 15 September 2008 to 15 October 2009. The four sub-events were 15/09/2008-10/10/2008, 15/09/2008-17/10/2008, 15/09/2008-27/10/2008 and 15/09/2008-20/11/2008.
Correlations (and contagion) changed significantly as sub-events unfolded. At the onset of the crisis, correlations with all African markets increased relatively quickly. I suggest that this can possibly be considered as being consistent with fast herding behaviour. The impact on developed markets was very different in that contagion spread slowly. I suggest that this can possibly be considered as being consistent with slow herding.
I argue that differences in contagion found between African and developed markets reflect differences in social network effects in investor communities. I apply behavioural finance theory to more fully explore this issue and identify the channels through which contagion events developed.
| Date of Award | 2014 |
|---|---|
| Original language | English |
| Awarding Institution |
|