Kanis Saengchote, a researcher from Thailand's Chulalongkorn University, has developed a framework aimed at identifying and quantifying systemic risk within decentralized finance (DeFi) institutions.
Known as the Global Systemically Important Protocol (G-SIP), this approach draws inspiration from similar initiatives in the traditional banking sector. Following the 2008 global financial crisis, conventional financial institutions collaborated to create a protocol for identifying critical banking structures, allowing for strategies To prevent future collapses. This led to the identification of "global systemically important banks" (G-SIBs), enabling the implementation of measures to protect against losses.
Saengchote's research paper outlines a method for applying a comparable standard to "blockchain banks," referring to DeFi protocols operating on blockchains. The paper underscores the necessity of identifying systemic risks in DeFi due to the self-amplifying nature of deleveraging, which can be triggered by financial interactions and fire sales. Given the algorithmic nature of DeFi, deleveraging can occur rapidly, leading to unstable cycles and potential protocol breakdowns.
The framework, G-SIP, assesses interactions between various DeFi protocols and identifies nodes with significant influence in the network. To establish protocol parameters, Saengchote analyzed four distinct protocols—Aave, Compound, Liquity, and MakerDAO—representing 88% of Ethereum's "blockchain banks." Among these, MakerDAO scored the highest in the G-SIP category due to its complexity and interconnectedness. It earned 37 points on the G-SIP rating scale, surpassing Aave (31.56), Compound (28), and Liquidity (4.57 ).
This analysis implies that MakerDAO could potentially possess a higher risk profile compared to the other protocols, necessitating more substantial capital requirements to effectively mitigate those risks. The G-SIP framework aims to provide a structured approach to understanding and managing systemic risk within the DeFi space .
