Download PDF Winter Bridge on Frontiers of Engineering December 18, 2019 Volume 49 Issue 4 The winter issue of The Bridge is focused on the 2019 Frontiers of Engineering symposium. Cryptocurrencies as Marketplaces Wednesday, December 18, 2019 Author: Jacob Leshno Bitcoin was introduced in 2008 as a computer protocol establishing a decentralized system that allows users to hold balances and make transfers to one another (Nakamoto 2008). Computer systems that provided similar services have existed for decades, but required a trusted party to control and operate them. For example, PayPal Holdings Inc. maintains the required computer infrastructure and charges usage fees to fund its activities and make a profit. Bitcoin Infrastructure and Protocol Bitcoin is a decentralized system. Instead of having a company that is responsible for maintaining the system’s infrastructure, it is operated by a decentralized network of computers called miners (the term also refers to the people who operate these computers). Much like Uber and Lyft, which allow anyone with a car to provide transportation services in return for compensation, Bitcoin allows anyone with a computer to provide the payment processing infrastructure in return for compensation. It eliminates the need for a centralized infrastructure by creating an open marketplace. But Bitcoin is unlike Uber and Lyft in that no entity is in control of the marketplace. Uber can change the price paid to drivers, add or remove the option to tip, and charge fees from the participants in its market. In contrast, Bitcoin is governed by its protocol, which no single entity can change (making changes to Bitcoin is akin to changing communication protocols such as TCP/IP). The computer protocol dictates the rules that govern the system and its implied marketplace, determining how miners are compensated and the fees users pay. The viability and success of Bitcoin, and other crypto-currencies that followed (e.g., Litecoin, Ethereum, Dogecoin), require that the protocol establish a functioning marketplace. But cryptocurrencies cannot control the miners who provide the infrastructure, and incentives are required to get miners to follow a desired behavior (Carlsten et al. 2016; Eyal and Sirer 2018). Miners provide their services at will and can withdraw from the system at any time, or try to exploit the system for profit and jeopardize its security (Auer 2019; Budish 2018). Game theory provides tools to understand how miners and users will behave in such an environment and to determine whether the system is secure. Since Bitcoin is not integrated with (and does not wish to rely on) other financial services, payments to miners can be made only with the system’s native coin, bitcoin, whose value is determined by financial markets, raising questions from monetary theory (Schilling and Uhlig 2019). These elements and others differentiate cryptocurrencies from traditional computer systems and make them economic objects, akin to marketplaces. Monopoly without a Monopolist In my work with Gur Huberman and Ciamac Moallemi (Huberman et al. 2019) we study the properties of this marketplace for transaction processing and ask who pays for the costs of operating the platform, how, and how much. We compare the Bitcoin payment system (BPS) with a traditional payment system (e.g., PayPal) and ask whether the decentralized design offers new benefits. (While we focus on Bitcoin’s design, our analysis also applies to other cryptocurrencies with similar design features.) The system processes transactions in batches called blocks. To ensure that a block is propagated throughout the network before the next one is issued, the protocol limits block size and frequency, limiting the system’s transaction processing capacity. Because of stochastic elements in the system, the system can periodically get congested and transactions can be delayed. We observe that the blockchain design of the BPS has the following features, which are key elements of its economics: Miners can enter or leave the system as they see fit. Miners can select the transactions they process and are rewarded with protocol-determined block rewards and transaction fees offered by the users (because Bitcoin’s protocol specifies that block rewards are halved approximately every 4 years, transaction fees will become more important over time, as they will eventually be the only form of payment to miners). The system’s transaction processing capacity (that is, average number of transactions that can be processed per unit time) is independent of the number of miners. Users choose the transaction fees they pay when their transaction is processed; they may even choose to pay only a minor minimal transaction fee (but when the system is sufficiently congested the delay can be so long that payment is required to get through before a transaction is timed out). We offer a simplified economic model of the BPS that allows analysis of the implied marketplace based on the following: (i) some users are willing to pay to expedite the processing of their transactions, (ii) miners are profit maximizers, and (iii) miners can freely enter or exit the system. Transaction Fees Are Determined in Equilibrium We find that the BPS is well described by an equilibrium in which users choose a transaction fee to gain processing priority over other users; miners process the transactions that offer the highest fee, up to capacity. Nobody dictates the equilibrium fee schedule. Transaction fees are set in an implicit auction without any explicit auctioneer. We offer closed-form expressions for the equilibrium fees and waiting times. We find that total transaction fees depend on three parameters: maximal block size, congestion or load (transaction arrival rate divided by system’s capacity), and the distribution of user willingness to pay higher fees to reduce transaction processing delay. When the system is not congested, the fees are low and essentially insensitive to its use—the expected processing delay is similar across transactions. As the system’s use approaches capacity, fees and cross--transaction variation in processing delays rise rapidly. The fee schedule satisfies the Vickrey–Clarke–Groves property: each transaction fee is equal to the externality it imposes by increasing the delay for transactions that offer lower fees. Comparison with a Profit-Maximizing Firm Pricing under the BPS is structurally different from the pricing of a profit-maximizing firm. A firm sets a price and denies service to users who are unwilling to pay that price. When the BPS has sufficient capacity, the system can raise revenue without denying service to anybody; users who are willing to bear delays can have their transaction processed even without paying transaction fees. Because the miners who collectively operate the system compete with each other, they cannot profitably affect the level of fees paid by users. This provides users protection from price increases: even if the system becomes a monopolist (in the sense that users have no alternative payment methods) users will still pay a low competitive transaction fee. In that way, the decentralized nature of the system may provide economic benefits to users. However, the design has several weaknesses. Transaction processing delays are essential to fee generation and therefore to the BPS’s long-run revenue model. The amount of infrastructure consumed by the system is determined in equilibrium, and there is no mechanism that ensures an efficient level of infrastructure. The amount of energy consumed by Bitcoin has received much media attention. It varies depending on the demand for transactions and the bitcoin-to-USD exchange rate, and the system design does not indicate either a desired level or a way of reaching such a level. Design Suggestions We provide a design that can partly address these concerns. It modifies a component of the protocol so that instead of maintaining a constant capacity, the protocol scales capacity according to demand (within a feasible region) to maintain congestion at a moderate level. This ensures that total transaction fees and the level of infrastructure are kept at a constant level. Our analysis also indicates that smaller block sizes allow the system to raise revenue more efficiently: a smaller block size allows the system to raise the same amount of revenue with shorter transaction processing delays. Governing a Decentralized System The limitations of the Bitcoin protocol have motivated much research and the development of other decentralized systems (e.g., Bentov et al. 2016; Chen and Micali 2016; Poon and Buterin 2017). To update such systems, agreement is needed on a new protocol, but without an entity that controls the system such agreement can be difficult to achieve. The implied rigidity of the system can be advantageous to users, who are guaranteed continuation of service at the same terms (with no ratcheting of fees), but it also reduces the system’s ability to react to new circumstances, which is especially important given the early stage of the technology. Game theoretic analysis can shed light on governance issues and help in the design of systems accordingly (Barrera and Hurder 2018). Conclusion Through a combination of cryptographic tools and economic incentives, Bitcoin and its followers have shown that it is feasible to create a global decentralized system controlled by no one. Services that previously could be provided only by a trusted firm can now be provided by a community coordinated only by a protocol. This allows for new economic models for the operation and funding of such services. The interdisciplinary nature of these systems calls for exciting future collaboration. References Auer R. 2019. Beyond the Doomsday Economics of “Proof-of-Work” in Cryptocurrencies. BIS Working Paper No. 765. Basel: Bank for International Settlements. Barrera C, Hurder S. 2018. Blockchain upgrade as a coordination game. Prysm Group. Bentov I, Pass R, Shi E. 2016. Snow White: Provably secure proofs of stake. IACR Cryptology ePrint Archive 2016:919. Budish E. 2018. The Economic Limits of Bitcoin and the Blockchain. NBER Working Paper No. 74717. Cambridge MA: National Bureau of Economic Research. Carlsten M, Kalodner H, Weinberg SM, Narayanan A. 2016. On the instability of bitcoin without the block reward. Proceedings, 2016 ACM SIGSAC Conf Computer and Communications Security, Oct 24–28, Vienna. Chen J, Micali S. 2016. Algorand. arXiv:1607.01341. Eyal I, Sirer EG. 2018. Majority is not enough: Bitcoin mining is vulnerable. Communications of the ACM 61(7):95–102. Huberman G, Leshno JD, Moallemi CC. 2019. An economic analysis of the Bitcoin payment system. Columbia Business School Research Paper No. 17-92. Nakamoto S. 2008. Bitcoin: A peer-to-peer electronic cash system. Available at bitcoin.org. Poon J, Buterin V. 2017. Plasma: Scalable autonomous smart contracts. Available at https://www.plasma.io/plasma.pdf. Schilling L, Uhlig H. 2019. Some Simple Bitcoin Economics. NBER Working Paper No. 24483. Cambridge MA: National Bureau of Economic Research. About the Author:Jacob Leshno is an assistant professor of economics in the University of Chicago Booth School of Business.