Multi-chain DeFi lending risk exposed in new study of 53 bridges

A new academic paper by Hasret Ozan Sevim, published on arXiv, dives into how cross-chain interoperability affects DeFi lending risk. The study analyzed 15 decentralized lending protocols, 53 cross-chain bridges, and 9 EVM-compatible blockchains (Ethereum, alternative layer-1s, and Ethereum layer-2s) from October 2022 to January 2025. Using panel regression fixed effects and OLS models, the research used total value locked (TVL) and total revenue as performance proxies. The results reveal that cross-chain activity significantly impacts protocol performance, with bridge volume emerging as a critical but heterogeneous driver. Surprisingly, increased bridge integrations are associated with decreased TVL and revenue, indicating that liquidity tends to escape from lending ecosystems — users bridge assets out rather than keeping them locked. Conversely, bridge hacks showed a significant positive relationship with TVL, likely due to temporary liquidity injections from exploited pools or insurance payouts.

Liquidations produced mixed effects across chain categories, while new network launches had no significant relationship with TVL or revenue. The study also found behavioral differences: Ethereum attracts large depositors (whales), while layer-2s skew toward retail participants. With high R-squared values confirming strong explanatory power, the paper concludes that traditional DeFi risk models must be updated to include cross-chain metrics and adopt a layer-aware approach. As DeFi becomes increasingly multi-chain, ignoring interoperability effects could lead to mispriced risk and systemic vulnerabilities. This research provides a data-driven foundation for protocol designers, auditors, and risk managers to better understand the complex dynamics of cross-chain lending.