解码 DeFi：抵押限制了信用风险，但网络威胁依然存在

This collateralisation system is designed to limit the credit risk for the lender, but it does not eliminate it entirely. The lender is still exposed to the risk of a sharp devaluation of the collateral (asset B in the diagram), for example in the event of a contagion between the crypto-assets used as collateral (see article by Tovanich, Kassoul, Weidenholzer and Prat, 2023), or a cyberattack on the platform. On this issue, see also the rapport by the ACPR.

这种抵押系统旨在限制贷方的信用风险，但并不能完全消除风险。贷方仍然面临抵押品（图中的资产 B）急剧贬值的风险，例如用作抵押品的加密资产之间发生传染时（参见 Tovanich、Kassoul、Weidenholzer 和 Prat 的文章） ，2023），或平台上的网络攻击。关于这个问题，另请参阅 ACPR 的融洽关系。

Blockchain, an unrivalled source of data

This blog post studies the data directly available on the Ethereum blockchain, which accounts for 80% of DeFI activity. We analyse two DeFI platforms: Aave and Compound. By way of illustration, according to Defillama, on 1 September 2023, Aave had USD 4.5 billion locked in its smart contracts, of which 85% was on the Ethereum blockchain.

区块链，无与伦比的数据来源这篇博文研究了以太坊区块链上直接可用的数据，该数据占 DeFi 活动的 80%。我们分析了两个 DeFi 平台：Aave 和Compound。例如，根据 Defillama 的说法，2023 年 9 月 1 日，Aave 的智能合约锁定了 45 亿美元，其中 85% 位于以太坊区块链上。

To extract the data, the Banque de France has deployed an archive node in its cloud, in other words an archive of all data generated since the blockchain was created, including historical states, which is synchronised in real time with the global Ethereum network. This node allows us to collect historical data without relying on a third-party data provider. 

为了提取数据，法兰西银行在其云端部署了一个存档节点，换句话说，存档了自区块链创建以来生成的所有数据，包括历史状态，并与全球以太坊网络实时同步。该节点允许我们在不依赖第三方数据提供商的情况下收集历史数据。

We then collected more than a million lending and borrowing transactions from the Aave and Compound platforms for three stablecoins: USDC, USDT and DAI, all of which aim to keep their tokens at parity with the US dollar.  

然后，我们从 Aave 和 Compound 平台收集了超过 100 万笔借贷交易，涉及三种稳定币：USDC、USDT 和 DAI，所有这些稳定币都旨在保持其代币与美元平价。

The setting of interest rates in DeFI 

On the main DeFI platforms, interest rates are set automatically using an algorithm coded into each platform’s smart contract. These algorithms are based on the amount deposited and borrowed for each crypto-asset, and are known to all users. The formula is simple: the borrow rate is an increasing function of the utilisation rate – in other words, the ratio between the amount borrowed and the amount deposited (Chart 3 shows the function applied by Aave for the token USDC).

DeFi 中的利率设置在主要 DeFi 平台上，利率是使用编码到每个平台的智能合约中的算法自动设置的。这些算法基于每种加密资产的存入和借入金额，并且为所有用户所知。公式很简单：借款利率是利用率的递增函数，换句话说，就是借入金额与存入金额之间的比率（图 3 显示了 Aave 对 USDC 代币应用的函数）。

The interest rates cannot be negative. They can be as high as 80% if there is an imbalance between borrowers and lenders. They rise to a limited extent when utilisation rates are low and very sharply beyond an inflection point decided by the platform and programmed into its smart contract. This inflection point is intended to keep the equilibrium utilisation rate close to a “target rate” (of 80% or 90% depending on the platform), as a high rate will attract lenders to the platform (see Chart 3). The parameters differ across platforms and across tokens, and depend in particular on the risk associated with each crypto-asset. The deposit rate is generally equal to the borrow rate multiplied by the utilisation rate.

利率不能为负。如果借款人和贷款人之间存在不平衡，比例可能高达 80%。当利用率较低且急剧超出平台决定并编程到其智能合约中的拐点时，它们的上升幅度有限。该拐点旨在使均衡利用率保持接近“目标利率”（80%或90%，具体取决于平台），因为高利率将吸引贷款人进入平台（见图3）。这些参数在不同平台和代币之间有所不同，并且特别取决于与每种加密资产相关的风险。存款利率一般等于借款利率乘以利用率。

Chart 3: Interest rates in DeFI are an increasing function of the utilisation rate

图 3：DeFI 的利率是利用率的递增函数