Why 'Full Reserve' Stablecoins Are Not Immune to Bank Runs

Full-reserve assets are often held in low-liquidity, high-yield instruments. A mass redemption event forces a fire sale, crystallizing losses and breaking the peg.

• Key Risk: Off-chain collateral (e.g., commercial paper, T-Bills) cannot be liquidated at par under stress.
• Real-World Example: The 2022 UST depeg was preceded by Luna Foundation Guard selling its ~$3B Bitcoin reserve into a falling market.

The on-chain representation of off-chain collateral depends on oracles. A delayed or manipulated price feed can trigger unjustified liquidations or mask insolvency.

• Attack Vector: Oracle latency or failure during market volatility creates a information asymmetry between chain state and real-world reserves.
• Systemic Example: MakerDAO's $4M flash loan attack in 2020 exploited a price oracle delay to drain vaults.

Users must trust the issuer's attestations and the integrity of opaque, regulated custodians (e.g., banks, trust companies). A seizure or freeze of reserves by a single jurisdiction breaks the global peg.

• Centralized Failure Point: The issuer remains a legal entity subject to government action, as seen with Tether's past settlements.
• Contagion Risk: A run on one stablecoin (e.g., USDC post-SVB) creates panic across the sector, draining liquidity from DeFi protocols like Aave and Compound.

The collapse of Silicon Valley Bank (SVB) triggered a run on Circle's $3.3B reserve deposit, causing USDC to de-peg. This proved full-reserve models are vulnerable to concentrated banking risk and regulatory seizure. The contagion spread to DAI and other protocols, demonstrating that off-chain reserve custody is a single point of failure.

• Key Risk: Centralized reserve custodian failure.
• Key Metric: USDC de-pegged to $0.87.
• Contagion Vector: Protocol reliance on a single stablecoin as primary collateral.

MakerDAO's Peg Stability Module (PSM) allowed 1:1 USDC-to-DAI swaps, creating an illusion of perfect liquidity. When USDC de-pegged, the PSM became a one-way exit ramp, draining the 'full reserve' and threatening DAI's solvency. The system's stability depended entirely on the market price of its underlying collateral, not just its existence.

• Key Risk: Secondary market collateral devaluation.
• Key Flaw: No circuit breaker for mass redemptions.
• Outcome: Governance had to freeze the PSM to prevent a bank run.

True resilience requires geographic diversification of custodians, on-chain verification of assets (via RWAs or Treasuries), and circuit breakers for redemption queues. Protocols like Frax Finance (hybrid model) and MakerDAO's RWA vaults are moving towards this, but the base layer risk of fiat banking remains.

• Key Mitigation: Multi-sig, multi-jurisdiction custody.
• Key Tech: On-chain attestations & real-time audits.
• Trade-off: Increased complexity and regulatory surface area.

Even with 1:1 collateral, a stablecoin can trade below $1 on secondary markets like Uniswap. This creates a profitable arbitrage: buy cheap stablecoin, redeem for $1 of collateral, pocket the difference. This redemption pressure is the modern bank run.

• Arbitrageurs, not panicked users, drive the run.
• Liquidity on DEXs like Curve and Uniswap evaporates first.
• The peg is defended by the speed and cost of the redemption mechanism.

Isolate redemption liquidity from trading liquidity. Protocols like MakerDAO (PSM) and Frax Finance (AMO) hold a portion of reserves in highly liquid, low-slippage pools.

• PSM holds USDC directly to absorb immediate sell pressure.
• AMOs programmatically manage stablecoin supply against demand.
• This creates a circuit breaker that doesn't rely on external DEX depth.

'Full-reserve' often means 'full of a single asset' like USDC. This creates a systemic point of failure. A depeg or regulatory seizure of the underlying asset (e.g., USDC blacklist) collapses the entire stablecoin system.

• This is not a bank run, but a collateral failure.
• Tether (USDT) and USDC dominance creates network risk.
• Over-collateralized models like DAI are exposed to the same underlying basket.

Mitigate single-asset risk with a diversified, censorship-resistant collateral portfolio. This is the original MakerDAO ETH-backed DAI vision, now extended to LSTs and real-world assets.

• Increases resilience to any single asset's failure.
• Introduces new complexity: liquidation risk and oracle dependence.
• Requires robust, decentralized oracle networks like Chainlink.

On-chain redemption is not free or instant. During a crisis, network congestion leads to sky-high gas fees and failed transactions. Users and bots engage in a gas auction, where only the highest bidders can exit, trapping others.

• Creates a two-tiered system: whales exit, retail is stuck.
• Turns a liquidity crisis into a settlement layer crisis.
• Highlights the failure of using a volatile gas token for stable system exits.

Architect the stablecoin natively on high-throughput, low-cost Layer 2s like Arbitrum or Optimism. Pair this with intent-based redemption pathways via CowSwap or UniswapX that batch and optimize settlements.

• L2 reduces base-layer congestion risk.
• Intents allow users to specify a desired outcome (e.g., 'exit at $0.995') without managing gas.
• Solvers compete to fulfill the redemption efficiently.