Conditional Correlation Network Data from the Financial Sector

Published: 26 August 2021| Version 1 | DOI: 10.17632/pns8jrrc7h.1
Contributor:
Stephen Rush

Description

This data set contains rolling conditional correlation networks estimated from stock returns and the volume synchronized probability of informed trading. Only the largest 104 financial firms are included for the period of 1996 through 2012. The data was used to analyze banking sector systemic risk.

Files

Steps to reproduce

I match the NYSE TAQ data set to CRSP common equity. I retain firms with Standard Industrial Classification (SIC) Industry codes from 6000 to 6800 before merging with OptionMetrics. I eliminate firms with prices less than $5 per share, daily trading volume less than 1,000 shares, and market capitalization of less than $100mm. I create an hourly panel of VPIN and stock returns using python. I then estimate daily conditional correlation networks using the bnlearn software package in R.

Institutions

University of Connecticut, Bowling Green State University

Categories

Financial Crisis, Financial Institution, Bayesian Network

Licence