Risk Contagion in Surety Networks
Name
lin-vllin-sm-idss-26-thesis.pdf
Size
875.94 KB
Format
Adobe PDF
Checksum (MD5)
b971342d28e6ed93383510023136568d
Author(s)
Lin, Vanessa
Advisor(s)
Jadbabaie, Ali
Date Issued
February 2026
Publisher
Massachusetts Institute of Technology
Abstract
Surety bonds create directed networks of obligations among contractors, where the failure of one firm can propagate disruption and financial loss to others. This article develops a network-based stochastic framework for analyzing default contagion in surety contractor networks, in which defaults arise from both idiosyncratic risk and upstream contractor failure.
The resulting dynamics define a monotone Markov process on network states that captures risk propagation and converges to a stationary distribution. We first study the deterministic mean-field dynamics and show that the marginal default probabilities induce a network risk centrality measure that quantifies how network structure amplifies baseline failure risk, exhibiting close connections to opinion dynamics and eigenvector-based centrality models, e.g., Friedkin-Johnsen and PageRank.
We then analyze the full stochastic process and establish conditions for positive dependence between default events.
For acyclic contractor networks, we derived a closed-form expression for the limiting joint default distribution; for general networks, we establish logarithmic mixing time bounds that enable efficient numerical approximation.
Under a natural ordering condition, we then prove stochastic dominance results showing that network structure increases both marginal default probabilities and the upper tail of the aggregate loss distribution.
Overall, these results provide a tractable analytical and computational framework for quantifying contagion risk, dependence, and loss amplification in obligation networks, with direct implications for surety risk assessment.
MIT Department
Massachusetts Institute of Technology. Institute for Data, Systems, and Society
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
