On Libra and DC/EP（上）

关注我们，关注南湖基金小镇

【作者介绍：邹传伟，南湖金服合伙人、南湖基金小镇产业发展顾问、万向区块链首席经济学家。先后就读于北京大学、清华大学和哈佛大学，曾长期供职于中央汇金公司和中国投资公司。邹传伟博士曾荣获首届“孙冶方金融创新奖”和第五届中国软科学奖（前沿探索奖）。邹传伟博士在区块链方面的代表性研究是《区块链能做什么、不能做什么》（中国人民银行工作论文2018年第4号）。】

RECOMMEND
原文阅读

Recently[1], both Libra and DC/EP (Digital Currency/Electronic Payment) of the People’s Bank of China (PBoC) have drawn lots of attention worldwide. There have been lots of discussion from various angels. This article studies Libra and DC/EP from a new perspective, i.e. blockchain as a financial infrastructure (BaaFI). It consists of 4 parts. Part Ⅰ introduces BaaFI and discusses two models of financial infrastructures, i.e. the account model and the token model. Part Ⅱ briefly evaluates the risk and regulation of Libra. Part Ⅲ studies DC/EP in great details. Part Ⅳ is a brief summary.

Ⅰ. Blockchain as a Financial Infrastructure (BaaFI)

There exist two models of financial infrastructures. The first is the account model, which is represented by the 2-tier bank account system. The second is the token model, which is made possible by blockchain and related technologies. Those two models are quite different from each other but both can record financial assets and transactions. Under lots of circumstances, they complement as well as substitute for each other.

Ⅰ.1. The Account Model

The best-known and most studied example of the account model is the 2-tier bank account system, which consists of a central bank and multiple commercial banks. Individuals, enterprises, and government departments open deposit accounts in commercial banks. Commercial banks open deposit reserve accounts in the central bank. Fiat money, no matter it is M0 or M2, exists on the liability side of the central bank and commercial banks. M0 is the central bank’s liabilities, among which cash is its liability to the public, and deposit reserve is its liability to commercial banks. Deposits are commercial banks’ liabilities to individuals, enterprises, and government departments. In a modern economy, cash is only a very small portion of total money supply. All layers of money except cash, especially deposit reserve and deposits, exist in a digital form.

Cash is anonymous and accessible to all. Cash transactions also have a unique feature. As long as both the sender and the receiver are sure of the authenticity of cash, cash transactions can be settled without any third-party trusted institution. This is similar to token transactions which will be discussed later.

Bank transfers and remittances are conducted through bank accounts. If both the sender and the receiver have deposit accounts with the same bank, transfers between them only require the bank to credit the receiver’s account and debit the sender’s account. If the sender and the receiver have deposit accounts with different banks, transfers between them, besides requiring the two banks to adjust the sender’s and the receiver’s accounts respectively, will also lead to transfers between the two banks, which are settled through their deposit reserve accounts with the central bank.

Figure 1: Cross-border Payments[2]

Cross-border payments under the account model are more complicated (Figure 1).  Let’s suppose 2 countries, A and B, with their respective currencies. Both countries have established payment systems for their currencies[3]. There are two people, Alice and Bob. Alice is a resident of country A. She has a deposit account denominated in currency A with bank A1 of country A. Bob is a resident of country B. He has a deposit account denominated in currency B with bank B2 of country B. Suppose Alice wants to send Bob a certain amount of money. However, bank A1 and bank B2 have no direct business relationship. That is where correspondent banks become necessary.

Suppose bank A2 of country A and bank B1 of country B have direct business relationship with each other. As correspondent banks, they will act as bridges between bank A1 and bank B2. Cross-border payments are performed as follows. First, in country A, Alice transfers money from bank A1 to bank A2, which is supported by the payment system of currency A. Next, bank A2 transfer money to bank B1. Correspondent banks open deposit accounts with each other. For example, from bank A2’s perspective, its account with bank B1 is called the nostro account and denominated in currency B. Bank B1’s account with bank A2 is called the vostro account and denominated in currency A. Transactions between correspondent banks are settled through the nostro and vostro accounts. Finally, in country B, bank B1 transfers money to Bob’s account with bank B2.

I want to emphasize two points in cross-border payments. First, correspondent banks use SWIFT (Society for Worldwide Interbank Financial Telecommunication) to communicate with each other. SWIFT is an interbank messaging system that processes information flow in cross-border transactions. However, flow of funds in cross-border payments is processed through bank accounts, especially those of correspondent banks. Therefore, information flow and flow of funds are separate. Second, many people think that SWIFT is main cause of high cost and low efficiency of cross-border payments. However, that is not necessarily true. The network of correspondent banks and compliance checks in every step of money movement do make cross-border payments less efficient.  But according to a study of the McKinsey & Company in 2016, the main contributor of cost in cross-border payments is liquidity locked in the nostro and vostro accounts of correspondent banks. Correspondent banks have to keep extra money in those accounts, which could have been used to generate higher returns. As a result, they have to charge users higher fees to cover the opportunity cost.

Ⅰ.2. The Token Model

Mainstream blockchain systems, no matter they are public or consortium chains, and no matter their script languages have Turing completeness and support smart contracts or not, can all be summarized by the token model (Figure 2).

Figure 2: The Token Model

Tokens, smart contracts, and consensus algorithms all exist inside the boundary of consensus. There exists an inseparable relationship between tokens and smart contracts. Consensus algorithms ensure the trustless environment inside the boundary of consensus. There are two types of interactions across the boundary of blockchain. First, outside information is written into blockchain by Oracle mechanisms. Second, the representative relationship between tokens and off-chain assets, as well as trade between them. Essentially, blockchain as a financial infrastructure (BaaFI) is to use tokens to represent off-chain assets. The following features of blockchain and tokens are key to understand BaaFI (Table 1).

Table 1: Blockchain as a Financial Infrastructure (BaaFI)

First, different addresses inside blockchain correspond to different users, which is similar to bank accounts. Cryptographic technologies such as ECDSA (Elliptic Curve Digital Signature Algorithm) and Hash functions ensure the anonymity of addresses. For any address, only the person with the right private key can own and spend tokens in that address, which is similar to the passcode of a bank account.

Second, Tokens are essentially state variables defined inside blockchain. Tokens defined by the same rule are homogeneous and can be divided into smaller units. The number of tokens in an address is similar to the balance of a bank account. Although blockchain’s distributed ledger technology (DLT) is stored and updated by a group of miners or verification nodes, DLT maintains consistent in this decentralized arrangement. That is why blockchain is used in post-trade settlement to simplify custodian network and streamline transaction process.

Third, tokens can be transferred between different addresses, which is similar to bank transfers. Token transactions do not change the total number of tokens. That is, while the transferred-out address loses 1 unit of token, the transferred-in address gains 1 unit of token at the same time. Confirmation of token transactions and update of DLT are completed simultaneously, which means there is no settlement risk. Besides, DLT and confirmed transactions are public and tamper resistant.

Forth, consensus algorithms, such as the Nakamoto consensus, and blockchain’s tamper resistant feature make “double spending” impossible.

Fifth, since blockchain runs on the Internet, token transactions are cross-border by nature. That is why blockchain is used in cross-border payments.

Tokens exist as a piece of computer code and have no intrinsic value. Tokens derive their value from off-chain assets they are linked to. Linking tokens to off-chain assets is a legal as well as an economic problem. A centralized trusted institution (thereafter “issuance institution”) is indispensable in this process (Table 2).

Table 2: Linking Tokens to Off-chain Assets

The issuance institution follows 3 rules in linking tokens to off-chain assets. The first rule is “1: 1 issuance”: Tokens are backed 1: 1 by the underlying assets. The second rule is “1:1 two-way exchange”: The issuance institution guarantees a 1:1 two-way exchange between tokens and underlying assets. Whenever a user gives the issuance institution 1 unit of asset, the issuance institution will issue 1 token to the user, and vice versa. The third rule is credibility: The issuance institution must be regularly audited by a third party and fully disclose information to demonstrate the authenticity and adequacy of the underlying assets.

Under those 3 rules, 1 unit of token represents 1 unit of underlying asset. If there is a secondary market for tokens, token price may deviate from asset value, but arbitrage will make them converge. If those 3 rules  are not strictly followed, arbitrage will not have perfect effect and token price will be decoupled from asset value.

Ⅰ.3. A Comparison between the Account Model and the Token Model

Ⅰ.3.1. Difference in Transaction Performance

Token transactions do not rely on any centralized trusted institution. Transaction confirmation and DLT update occur simultaneously so there is no settlement risk. However, because of the existence of hard forks, many types of blockchain can only ensure settlement finality in a probabilistic sense. Blockchain is also constrained by the “impossibility trinity”: No blockchain system can achieve accuracy, efficiency, and decentralization at the same time. In particular, the higher its degree of decentralization, the lower its efficiency.

In contrast, in the account model, centralized and trusted institutions such as banks play an indispensable role in confirming transactions and updating ledgers. It is hard to eliminate counterparty risk and settlement risk. But the account model can guarantee settlement finality. Being not subject to the “impossibility trinity”, the account model can achieve very high efficiency.

Ⅰ.3.2. Difference in Openness and Privacy Protection

Data management and privacy protection have increasingly become important policy issues and can significantly impact the future of online business models.

Theoretically speaking, blockchain is open to all. Anyone who can generate a pair of public and private keys can own an address in blockchain. Addresses are anonymous. It is almost impossible to discover the true identity of address owners only from information contained in addresses. The number of tokens in addresses and confirmed transactions in blockchain are open to all and tamper resistant. The anonymity feature protects the privacy of address owners but increases the difficulty in enforcing regulation such as KYC (“know your customers”), AML (anti-money laundering), and CFT (counter-terrorism financing).

In contrast, accounts generally require prior approval to open and therefore can be highly selective. Particularly, financial accounts such as deposit accounts and third-party payment accounts must follow strict KYC requirements. It is relatively easy to infer information about account owners from the names of financial accounts. But the balance of financial accounts is only visible to those with authorization.

If we expand our horizons from financial accounts to other types of accounts such as social network accounts and e-commerce accounts, it is straightforward to see that accounts are inevitable associated with user identification. Accounts record the behaviors of their owners in various scenarios and make consumer profiling possible. Consumer profiling then helps technical companies to engage in targeted advertising and online lending. However, in the account model, the collection and use of users’ behavioral information can easily evolve into privacy infringement. Personal information is difficult to protect but easy to use without prior authorization. Information collected from one line of business can be used in another line of business. Organizations holding personal information may be attacked, leading to information leakage. Those problems faced by the account model do not apply to the token model.

注释
[1]The author is grateful for HAO Kai’s help in preparing the English version of this article.

[2]This figure is adopted from Bank of Canada, Bank of England, Monetary Authority of Singapore, 2018, "Cross-Border Interbank Payments and Settlements".

[3]For example, HVPS (High Value Payment System), BEPS (Bulk Electronic Payment System), and CIPS (Cross-Border Interbank Payment System) in China, and Fedwire (Federal Electronic Funds Transfer System) and CHIPS (Clearing House Interbank Payment System) in US.