SMU SOE Seminar Series (May 13, 2025): Stock Co-jump Network Models

TOPIC:

ABSTRACT

Stock prices often exhibit co-jumps, even without market jumps. To capture such phenomena, we introduce a class of network models based on site-percolation, contrasting with usual network models that rely on edge-based connections to represent dependencies. We discuss the fundamental differences between these two modeling approaches, develop community detection methods tailored to our proposed framework, and demonstrate their economic significance through empirical applications.

Keywords: Network, Community Detection, Jumps, Co-Jumps, Stock Dependence, High-Frequency Data.

JEL: C14, C38, C58, G17.

Click here to view the CV.
Click here to view the paper.