Interconnected DeFi: Ripple Effects from the Terra Collapse

TLDR

Federal Reserve publication quantifying the marginal impact of a bridge on the utility of chains integrated with failed blockchains. Paper finds that for every bridge a blockchain had to Terra decreased the probability of increasing TVs over the next months by 40%.

Bridging risk is real and quantifiable, especially with large chains with deep DeFi integration. This is equally true for large lending protocol integrations - Concrete must monitor this risk.
Blocking for the obvious impact of the failed chain’s native token outflow, the programmable of bridged chains quantifiably decreases

The impact analysis conducted on contagion effects from bridging can be applied to integrations with lending protocols for Concrete. This methodology provides some useful insight in selecting future partners/integrations.
Having a diverse set of lending protocols is important in derisking Concrete’s risk exposure. Having integrations with lending protocols of different sizes, collateral types, and on different chains is vital for future risk management.

The impact of riding on contagion/interconnectedness is measured via positive share change. This is utilized rather than simply TVL as it captures not just decrease in utilization but also changes in a bridge-connected blockchain’s programmable use value.
All data was sourced from DeFi Llama. Number of bridges, share change, shared dApps, shared bridges, shared non bridges, and TVL are all analyzed between Terra and 44 public blockchains.
The paper finds that not all blockchains were equally affected by Terra’s collapse. This finding, in and of itself, is not novel but the finding that smaller chains were more resilient than larger chains even when smaller chains had a larger percentage of TVL locked in Terra ecosystem-related tokens.
The size of the blockchain is blocked for and shared bridges, dApps, and TVL share are used to analyze the effect of bridging on the non-Terra blockchain. The paper finds that per shared bridge, the likelihood of the chain’s TVL increasing over the next month after the Terra collapse went down 40%.
Analysis on the reduction in TVL in specific dApps show that outflow of TVL unrelated to short-term volatility and price instability occurred roughly a month after the collapse. This shows that the effect of bridging on TVL is not simply an effect of less liquidity but rather lower utility of the blockchain as it is less connected to the broader DeFi ecosystem.

It is impossible to block for the infinite number of confounding variables when analyzing the impact of Terra on other chains but this paper does a less-than-stellar job and makes bold assertions on causal relationships.