A network structural analysis of Malaysian stock market with edge density constraint
The main study of this thesis is the topological properties of the Malaysian stock market correlation network. In the first part of this study, two sets of networks are constructed to respectively capture the fuctuation of stock prices during the bearish market period (June 2007 - May 2009) and bul...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/41446/1/IPM%202013%202R.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The main study of this thesis is the topological properties of the Malaysian stock market correlation network. In the first part of this study, two sets of networks are constructed to respectively capture the fuctuation of stock
prices during the bearish market period (June 2007 - May 2009) and bullish market period (May 2009- March 2011). The correlation networks constructed during the bearish market period have higher clustering coe±cient,stronger disassortativity and the stocks in the largest component of the net-work develop more correlation relationships with greater correlation strength among each other. Both network constructed during the bullish or bearish market period belong to the class of disassortative network with disassortative hubs. In the second part of this study, the correlation network is constructed based on the stock prices between the period of May 2002 to May 2012. When the market mode is removed from the correlation matrix, the constructed network has lower clustering coe±cient, and lesser intra connection of edges between the stocks in the largest component. In addition, the network has also become assortatively mixed with assortative hubs in comparison to disassortatively mixed network with the presence of market mode. The stability of the correlation network structure is studied in terms of two diferent types of attack, namely attack based on removal of vertices from the network and attack by perturbation (Gaussian attack on the time series of stocks return). The Malaysian stock market is more vulnerable to preferential removal of vertices when the edge density of the network is decreased. In the context of perturbation, the stocks traded in Malaysian stock market are more sensitive to perturbation and have stronger clustering property in comparison to stocks of NYSE.
In the last part of this study, the time shift 1 day cross-correlation of stock prices between Malaysian stock market and New York Stock Exchange (NYSE) is analyzed. The time-lag behavior of the stocks traded in NYSE (as the mature market) is organized diferently compared to Malaysian stock market (as the emerging market), in particular the in-degree disassortative hubs of the correlation network of NYSE remain as indisassortative which is in contrast with the in-degree hubs in the network of Malaysian stock market that become inassortative as the edge density of the network increases from 0.0014 to 0.0046. |
|---|