Heterogeneity and clustering of defaults

Karlis, Alexandros; Galanis, Giorgos; Terovitis, Spyros and Turner, Matthew. 2021. Heterogeneity and clustering of defaults. Quantitative Finance, 21(9), pp. 1533-1549. ISSN 1469-7688 [Article]

[img]
Preview
Text
Heterogeneity and clustering of defaults.pdf - Published Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (1MB) | Preview

Abstract or Description

This paper studies how the degree of heterogeneity among hedge funds' demand orders for a risky asset affects the possibility of their defaults being clustered. We find that fire-sales caused by margin calls is a necessary, yet not a sufficient condition for defaults to be clustered. We show that when the degree of heterogeneity is sufficiently high, poorly performing HFs are able to obtain a higher than usual market share, which leads to an improvement of their performance. Consequently, their survival time is prolonged, increasing the probability of them remaining in operation until the downturn of the next leverage cycle. This leads to an increase in the probability of poorly and high-performing hedge funds to default in sync at a later time, and thus also in the probability of collective defaults. Our analytical results establish a connection between the nontrivial aggregate statistics and the presence of infinite memory in the process governing the hedge funds' defaults.

Item Type:

Article

Identification Number (DOI):

https://doi.org/10.1080/14697688.2021.1882686

Additional Information:

JEL Classification: C22, G01, G23, G32, G33

Keywords:

Long memory; Financial crises; Hedge funds; Survival statistics; Bankruptcy

Departments, Centres and Research Units:

Institute of Management Studies

Dates:

DateEvent
8 January 2021Accepted
30 March 2021Published Online
2021Published

Item ID:

29904

Date Deposited:

31 Mar 2021 12:59

Last Modified:

24 Sep 2021 11:00

Peer Reviewed:

Yes, this version has been peer-reviewed.

URI:

https://research.gold.ac.uk/id/eprint/29904

View statistics for this item...

Edit Record Edit Record (login required)