Issue #47: Consensus Mechanisms in Traditional Finance and Decentralized Finance
In Issue #36 of Monetary Mechanics, I introduced the idea that blockchain-based finance (i.e. decentralized finance a.k.a. DeFi) is a fundamental improvement in the accounting of economic transactions that facilitates the expression of a whole series of novel financial functions. This expansion in financial functions will undoubtedly be commensurate with the historic explosions in the volume and variety of economic activity that are associated with the Renaissance, the Industrial Revolution, and the post-WWII economic boom.
In this issue of Monetary Mechanics, I am going to focus on the common problems that both traditional finance and decentralized finance try to solve, as well as the different methods that each financial system uses to try to solve these common problems, in order to further flesh out the aforementioned idea. On the surface, it seems that the two financial systems have their own unique concerns, challenges, and jargons. At heart, however, the two financial systems actually attempt to solve the same problem – establishing and maintaining trust in economic systems.
This is accomplished in traditional finance through bureaucratic backstops such as central banks, sovereign debt immunity, insurance from the Federal Deposit Insurance Corporation (FDIC), accountants and auditors, legislations and regulations, and mutually agreed upon systems of standards and methods of measuring assets and risks (e.g. value-at-risk).
This is accomplished in decentralized finance through the development of Byzantine Fault Tolerance (BFT), the development of robust consensus mechanisms, and the establishment of public/private key encryption, as well as other constantly evolving techniques and technologies.
What is a “Consensus Mechanism”?
A proper explanation of my overarching philosophical argument probably necessitates an explanation of what exactly a “consensus mechanism” is and why this concept, despite being from the decentralized financial system, is directly applicable to the traditional financial system.
An excerpt from the founders of Ethereum:
Consensus mechanisms (also known as consensus protocols or consensus algorithms) allow distributed systems (networks of computers) to work together and stay secure.
For decades, these mechanisms have been used to establish consensus among database nodes, application servers, and other enterprise infrastructure. In recent years, new consensus mechanisms have been invented to allow cryptoeconomic systems, such as Ethereum, to agree on the state of the network.1
Another excerpt from Distributed Systems: Concepts and Design:
A distributed system is one in which components located at networked computers communicate and coordinate their actions only by passing messages. Examples of modern distributed systems include web search, massively multiplayer online games, and financial trading systems.
A fundamental problem in distributed computing and multi-agent systems is to achieve overall system reliability in the presence of a number of faulty processes. This often requires coordinating processes to reach consensus, or agree on some data value that is needed during computation. Examples of consensus include agreeing on what transactions to commit to a database in which order, state machine replication, and atomic broadcasts.2
Basically, a consensus mechanism can be considered as a shared system of standards and steps that allows a distributed system (a group of individuals or computers) to accurately, precisely, and reliably coordinate on shared problems. Where have we heard something like this?
The “new networks of interbank relations” 60 years ago were attempting to solve problems that were strikingly similar in their essence to the problems that consensus mechanisms in distributed computing systems today are attempting to solve.3 The Eurodollar banking system was essentially nothing more than an analog consensus mechanism, a consensus mechanism that used human beings and primitive telecommunications such as telex networks to come to consensus over shared facts (e.g. funds transfers, financial transactions, deposits, loans, etc.).
According to Chris Berg, “the key invention of blockchains and cryptocurrencies is the consensus mechanism. Coming to consensus over shared facts has been the overriding concern of economists and political philosophers for centuries. […] In a very real sense, the consensus mechanism, the Byzantine Generals Problem, is the basic problem of all human activity, not just blockchains and cryptocurrencies, but also the economy and the society [in general].”4
What is “Ledger Money”?
This section might seem a bit abstract.
What is “ledger money” and a “master ledger”? As I have explained in various previous issues of Monetary Mechanics (Issue #9, Issue #14, and Issue #20), the modern financial system depends on a few large central intermediaries to provide liquidity and process transactions. Whenever anyone wants to purchase any good or any service or transfer funds from one place to another, banks (and sometimes also nonbank financial institutions) must come together to facilitate that purchase or transfer of funds across all their books. The involved transactions are cleared over one of several payment systems and are eventually settled on the books of a correspondent bank or the books of the Federal Reserve Bank of New York (FRBNY).
Let’s use Venmo as an example. You may believe that your payment is instantly processed because you sent money that your recipient immediately received through the app. However, what Venmo is doing is taking on the risk that your payment does not properly clear in the next 3-5 days, which is how long it typically takes for a payment to clear the entire payment processing network. This happens with credit cards too. You can purchase goods and services for a specified period of time with your credit card until ultimately you must settle your purchases with the credit card company at the end of the specified period of time, which is usually a month.
Hence, these types of money are referred to as “ledger money” and “inside money” because they do not exist in reality as physical paper money – they exist only as a recorded entry on somebody’s balance sheet. There are, strictly speaking, some differences between what I would consider to be “ledger money” and what most people would consider to be “inside money” (e.g. I would not consider physical bank coins and notes to be “ledger money,” but they are technically “inside money,” because they are a liability of the Federal Reserve). However, for our own purposes here, we should be able to use “ledger money” and “inside money” interchangeably.
The reason why this is important is because the majority of purchases and transfers of funds are settled in this manner – on a net basis, as opposed to on a gross basis. Consequently, when transactions are settled on a net basis, only the net difference needs to be transferred between two counterparties in order for the transaction to be considered settled.
Let’s imagine two banks, Bank A and Bank B, engaging in a series of transactions, as an example. Bank A owes Bank B $1 million, and Bank B owes Bank A $2 million. If the two banks are allowed to use net settlement, the only transfer that needs to occur is for Bank B to transfer $1 million to Bank A. If the two banks are forced to use gross settlement, then Bank A must transfer $1 million to Bank B, and then Bank B must transfer $2 million to Bank A.
The difference in efficiency between net settlement and gross settlement is negligible for a small number of transactions in the presence of ample liquidity. However, on a realistic level, the difference in efficiency between net settlement and gross settlement is cripplingly massive.
Hence, from an economic and financial perspective, “ledger money,” “master ledger,” and “consensus mechanisms” matter because these concepts determine the rules for recording and settling economic and financial transactions. At the simplest level, these rules might be:
Anyone can conduct transactions
Everyone must settle at the end of the day
Everyone must settle in physical US dollar banknotes
Obviously, much more complex rules would have to be put in place, as the volume and variety of economic and financial transactions continue to increase. Consequently, as these complexities complicate the machinery of exchange, it would be not only important, but imperative, for everyone to have an efficient, equitable, transparent, and secure system to settle transactions.
Centralized Systems vs. Decentralized Systems
On the one hand, centralized systems can be effective and efficient in some situations, especially where alternative systems are too costly to be reasonably considered. This is a fact that remains true in many more fields than just economics and finance – it is a fact that is perhaps better exemplified by a political perspective. For example, an authoritarian governmental structure can be much more effective and efficient in some ways, as it is much easier to make swift, sweeping, and definitive decisions if disagreement with a central authority figure is impossible, as is the case in China. On average, such a governmental structure can act faster and much more uniformly than decentralized (i.e. democratic) forms of government.
On the other hand, centralized systems possess their own problems too. Firstly, central authority figures are not always honest and competent, resulting in a single point of failure risk. If the central authority figure is dishonest and incompetent, the performance of the entire system suffers. Furthermore, centralized systems tend to work in a top-down manner, from the core out to the periphery, oftentimes resulting in imperfect information and suboptimal outcomes.
The diagram above, borrowed from complexity science, demonstrates this conceptual trade-off between centralization and decentralization (efficiency vs. resiliency). Centralized systems are more efficient, but systems that are too centralized can tend to be brittle and have low sustainability. Decentralized systems are more resilient, but systems that are too decentralized can tend to be stagnant and have low sustainability. The optimal balance, observed in ecological ecosystems in real life, tends to be roughly ⅓ efficiency and ⅔ resiliency. Obviously, not every system will be optimized by these exact parameters, but it remains a good rule of thumb.
The main takeaway here is that there is always a trade-off between centralization and decentralization, or between efficiency and resiliency. For some purposes, it may be perfectly acceptable to use a completely one-sided and private ledger. For other purposes, that may be unacceptable. Generally, any given centralized system tends to be more efficient than any given decentralized system at the same level of “technological advancement.” For example, it is easier to broadcast the updates of a master ledger to 15-20 large banks than to 350 million citizens.
However, it is also important to remember that what is efficient in the short run is not necessarily always what is efficient in the long run. Decentralized systems support greater experimentation and innovation and can even surpass the capabilities of centralized systems over enough time.
Let’s imagine the political model of a democratic republic like the United States. The federal government has some centralized authority, but it is certainly not complete. In addition, states and cities also have some authorities of their own, and can act independently from one another. Each state and city might implement its own version of a larger plan, some being more efficient, some being less efficient. In the short run, bureaucratic inefficiencies can make it difficult to pass policies, particularly controversial policies, resulting in the obstruction of progress. In the long run, however, assuming an ideal political environment, each state and city can learn from each other’s implementation, evaluating and choosing the best combination of practices, resulting in an optimal outcome over time.
Centralized systems typically do not support such bottom-up experimentation and innovation and make it much more difficult, if not totally impossible, to do so. Hence, centralized systems actually create their own inflexibility and fragility in the long run.
Now, let’s bring our focus back to the modern monetary system. On the surface, it might make more sense to allow only a few privileged global systemically important banks (G-SIBs) and primary dealers access to the “master ledger.” This means not only direct access to Federal Reserve master accounts but also more broadly things like direct access to the Clearing House Interbank Payments System (CHIPS), direct access to tri-party repo, being a clearing member of the Fixed Income Clearing Corporation (FICC), and other more informal trading arrangements.
One might make the argument that, by allowing only a few large entities access to the vital vessels of the modern monetary system, and by doing our best to ensure the resilience of these entities via legislations, regulations, and bureaucratic backstops, we are able to get the best of both worlds. However, most financial market observers and participants would probably disagree with that argument. Over the years, it has become painfully obvious that the perennial backstopping of “too big to fail” entities has precipitated a buildup of inefficiencies and an overall ossification in what is perhaps one of the world’s most important and most centralized systems.
Robert V. Roosa, “Exchange Rate Arrangements in the Eighties,” Federal Reserve Bank of Boston, Economic Research Conference Series, Conference Series 28: The International Monetary System: Forty Years After Bretton Woods, 1984.