资源描述
Journal of Financial Economics 134 (2019) 91109 Contents lists available at ScienceDirect Journal of Financial Economics journal homepage: elsevier From mining to markets: The evolution fees David Easley a , b , c , Maureen OHara c , d , e , , Soumya Basu a Department of Economics, Cornell University, United States b Department of Information Sciences, Cornell University, United States c University of Technology Sydney, Australia d Cornell University, Johnson College of Business, United States e Cornell University, Initiative for Crypto Currencies and Contracts, United States f Department of Computer Science, Cornell University, United States a r t i c l e i n f o Article history: Received 2 October 2017 Revised 31 May 2018 Accepted 27 June 2018 Available online 11 March 2019 JEL classification: G10 G18 D47 D6 Keywords: Bitcoin Blockchain Market microstructure Transaction fees a b s t r a c t We investigate the role that a mining-based structure to explain the factors leading the strategic behavior of miners rewards, transaction fees, how microstructure featu ics of user participation on predictions and discuss im “In a few decades when the reward gets too small, the transaction fee will become the main compen- sation for nodes. Im sure that in 20 years there We are grateful to the editor (Bill Schwert), two anonymous referees, Levon Barseghyan, Campbell Harvey, Andrew Karolyi, Francesca Molinari, Talis Putnins, Gun Sirer, Clara Vega, David Yermack, Yao Zeng, and semi- nar participants at the Board of Governors of the Federal Reserve System, Cornell University, HEC Montreal, Johns Hopkins University, Queens Uni- versity, the RFS FinTech Conference, and the University of Technology Syd- ney/ University of Sydney Market Microstructure Conference for helpful comments and discussion. Soumya Basu thanks the Initiative for Crypto- currencies and Contracts (IC3) and National Science Foundation Graduate Research Fellowship Program under Grant No. DGE-1650441 for support. Corresponding author at: University of Technology Sydney, Australia; Cornell University, United States E-mail address: mo19cornell.edu (M. OHara). doi/10.1016/j.jfineco.2019.03.004 0304-405X/ 2019 Elsevier B.V. All rights reserved. /locate/jfec of bitcoin transaction e , f transaction fees play in the bitcoin blockchains evolution from to a market-based ecology. We develop a game-theoretic model to the emergence of transactions fees, as well as to explain and users. Our model highlights the role played by mining price, and waiting time, discusses welfare issues, and examines res such as exogenous structural constraints influence the dynam- the blockchain. We provide empirical evidence on the models plications for bitcoins evolution. 2019 Elsevier B.V. All rights reserved. will either be very large transaction volume or no volume.” Satoshi Nakamoto 1 1. Introduction By a variety of metrics, bitcoin is no longer a financial curiosity. From its genesis transaction of 50 bitcoins in Jan- uary 2009, bitcoins in circulation now number more than 17 million. An estimated 35 million Bitcoin wallets are held worldwide with 10 0,0 0 0 companies accepting payments in bitcoins, some via the newly issued bitcoin debit card. 1 Cited in is- the- bitcoin- block- reward/ . 92 D. Easley, M. OHara and S. Basu / Journal of Financial Economics 134 (2019) 91109 Daily trading volume on November 7, 2017 at major bitcoin exchanges first exceeded a record $5.0 billion, with volume across all cryptocurrencies on Dec. 20, 2017 exceeding $50 billion. 2 While volume has fallen from that record level, it is still substantial, with Nov. 28, 2018 volume of $20.4 bil- lion. A recent study (see Hileman and Rauchs, 2017 ) esti- mates that 10 million people now hold a material amount of bitcoins as a financial asset. Approximately 10,467 active global bitcoin nodes exist, each containing a complete copy of the Bitcoin blockchain. 3 Envisioned in 2008 as a decentralized, trustless digital currency and payment system, the bitcoin blockchain oper- ates on a worldwide basis via a complex set of rules orig- inally proposed in Nakamoto (2008) . Fundamental to the bitcoin ecology are miners, who play a crucial role both in creating new bitcoins and in verifying transactions on the blockchain. Mining involves using specialized computer hardware to find a particular mathematical hash function, with the reward for success being payment in new bit- coins. The amount of such payments, as well as a variety of parameters such as the diculty of the underlying com- putational problem and even the total amount of bitcoins that can ever be mined are specified exogenously. The bit- coin protocol also exogenously specifies a dynamic adjust- ment process for these payment and diculty parameters. As Fig. 1 illustrates, an additional form of compensation for miners has emerged in the form of transactions fees. These transaction fees are voluntarily appended to bitcoin transactions by buyers and sellers wanting to ensure that their transactions are included in the block of transactions that the miner attaches to the blockchain. While such fees are still not the primary component of miners total rev- enues, the endogenous development of transactions fees reflects an important step in the evolution of the bitcoin blockchain from being a mining-based set of rules toward being a market-based system capable of adapting to chang- ing economic conditions. What is less clear is how success- ful this transition will be. In this paper, we investigate the evolution of transac- tions fees in bitcoin. We build a framework for understand- ing why such fees developed and how they influence the dynamics of the bitcoin blockchain. Examining the bitcoin ecology is complex as it involves strategic behavior on the part of both miners and users, all packaged within a set of exogenous rules. We develop a game-theoretic model to explain the factors leading to the emergence of transac- tion fees, as well as the interactions between miners and users. Our model also highlights the roles played by the bitcoin price, waiting times, and mining-based revenues known as block rewards. We then provide empirical evi- dence supporting our theoretical conclusions about the im- pact of waiting time on fees and we offer estimates of the magnitudes of these effects. Our research has a number of results on the role and behavior of transaction fees in the bitcoin blockchain. Our 2 See update- bitcoins- daily- trade- volume- surpasses- 5b/ and cryptocurrency- volumes- vs- stock- market- volumes- 2017- 12 . 3 This is the number of reachable, or active global nodes see . model confirms that eventually without transaction fees the blockchain would not be viable as miners revenues from block rewards are deterministically programmed by the bitcoin protocol to reach zero. Given that this zero block reward level is not projected to occur until the year 2140, our model shows the important role played by the bitcoin price level in sustaining the viability of the bit- coin blockchain. Our model shows why, with increasing bitcoin price levels, transaction fees can long play only a secondary role in explaining the willingness of miners to participate. Where transaction fees play a bigger role is in affecting the participation of users. We show that, even with trans- action fees, there are limits on the size of the blockchain imposed by waiting times confronting users. We show how these waiting times arise in equilibrium and how they are influenced both by endogenous transactions fees and by exogenous dynamic constraints imposed by the bitcoin protocol. More intriguing, we show that waiting times (or for that matter, transaction fees) are not influenced by the block reward; provided mining is viable, our model pre- dicts that the reward level is irrelevant for determining transaction fees in equilibrium. Our model also demonstrates how equilibrium trans- action fees evolve in the bitcoin ecosystem. If the arrival rate of potential transactions is low, transactions without fees attached are written to the blockchain but, as the ar- rival rate of potential transactions increases, the equilib- rium shifts and only transactions with fees attached are posted to the blockchain. These equilibria exhibit strate- gic complementarity, meaning that while some equilibria are stable, others are not. Moreover, these equilibria can be Pareto-ranked, and we demonstrate that transactions fees are not welfare-improving. We also show that, in these equilibria, transaction costs can induce user nonparticipa- tion. The fees directly induce some users to drop out, while increasing wait times cause other fee-paying users to de- part as well. Our model suggests that these welfare and user participation effects are reasons why transaction fees alone are not a panacea for the dynamic challenges facing the evolving bitcoin blockchain. Our empirical work supports these findings. We show that higher transactions fees are being driven by queu- ing problems facing users, not by reductions in bitcoin- denominated block rewards. As predicted, in at least most empirical specifications, we find no statistically significant effects for the block reward level. As users battle to get transactions posted on the blockchain, transaction fees rise to levels that discourage bitcoin usage, highlighting an im- portant structural issue confronting the blockchain. While bitcoin can continue to develop as a financial asset, longer waiting times and higher transactions fees could impede its development as a transactional medium. Overall, our re- sults delineate the complex role that transaction fees play in bitcoins evolution from a mining-based structure to a market-based ecology. Finally, our general results apply more broadly to the many other systems (for example, Ethereum, Litecoin, and Dogecoin) that use the bitcoin blockchain management protocols. In these and other similar systems, transac- tion fees also rose dramatically over our sample period, D. Easley, M. OHara and S. Basu / Journal of Financial Economics 134 (2019) 91109 93 Fig. 1. Total bitcoin transaction fees. This figure gives yearly data on the level of on the relative importance of transaction fees for overall miner revenue. reflecting the interaction of increasing demand and fixed blockchain protocols. For such cryptocurrencies, a funda- mental challenge is whether a static rules-based protocol can remain a single entity or whether the disparate needs of users result in fragmentation into multiple coin-based currencies. We argue that such fragmentation can be a nat- ural part of the evolution to a market-based system, un- derscoring Satoshis concern about whether volumes will be large or none at all. This paper is organized as follows. Section 2 provides a literature review. Section 3 gives a brief overview of the bitcoin blockchain, explaining the various exogenous components of the system. Section 4 develops game- theoretic models of the games played by the miners and transaction fees measured both in US dollars and in bitcoins (BTC) and by the users. We find the Nash equilibria in these games and demonstrate their stability properties and welfare im- plications. We then characterize the factors influencing these equilibria, with a particular focus on the important role played by waiting times. We examine the equilibrium transaction fee structure and how it influences user wait- ing times. Section 5 then provides empirical analyses based on predictions from our model. Section 6 concludes by dis- cussing the current structural challenges confronting bit- coin and the role played by transaction fees. 2. Literature review Our research joins a growing literature examining bit- coin, digital currencies, and the broader applications of 94 D. Easley, M. OHara and S. Basu / Journal of Financial Economics 134 (2019) 91109 blockchains. A variety of authors (see, for example, Eyal and Sirer, 2014; Gans and Halaburda, 2015 ; Gandel and Halaburda, 2016 ) analyze design issues of the bitcoin pro- tocol, as well as the dynamic interactions between cryp- tocurrencies. Other recent research analyzes aspects of the bitcoin ecosystem specifically as they relate to finance and and the block is added to the blockchain. If a block is pro- posed with an invalid transaction, then the block proposal is ignored. The blockchain grows as each validated block is added. This validation process of posting transactions to the blockchain can also provide another source of revenue to the winning miner via transaction fees embedded in pend- the financial markets (see Boehm et al., 2015 ; Harvey, 2016; Malinova and Park, 2016; Raskin and Yermack, 2017, 2018; Aune et al., 2017 ). Still other research (see Cong and He, 2017; FINRA, 2017 ) examines smart securities and more general uses of the blockchain. A literature also looks at the game played by miners on the bitcoin blockchain. Huoy (2014) , for example, analyzes the mining game but his analysis includes neither users nor transactions fees. Kroll et al. (2013) do include transaction fees but argue that they have little importance. Huberman et al. (2017) , written contemporaneously with our paper, analyze a congestion queuing game that includes miners and fees. The basic ingredients of our model and their model are similar: free entry into min- ing yields zero expected profit, and waiting time for users lead to fees. But, the focus of our analyses is different. Huberman et al. (2017) are concerned about the viability of mining when mining rewards are zero and fees are nec- essary to incentivize miners. Their concern is that equilib- rium fees could be too low for the blockchain to be viable, and they propose changes to the protocol to address this potential issue. We have the opposite concern, i.e., waiting times and equilibrium fees could be high enough to dis- courage user participation. Also, our model examines the evolution of these mining rewards and transaction fees in equilibrium, while their analysis looks only at the long- run steady state where mining rewards have disappeared and price effects of bitcoin are assumed irrelevant. This difference is particularly important as whether bitcoin can reach such a steady state depends critically on this evolu- tion from mining to markets. As we demonstrate, the com- plex interactions in the bitcoin ecology make this evolution far from a sure thing. 3. A brief overview of the bitcoin blockchain protocol The bitcoin blockchain set out in Nakamoto (2008) in- volves a decentralized trustless network composed of nodes, with each node containing a complete copy of the blockchain. Miners run nodes, validate transactions, and provide the requisite security for the blockchain. Mining uses computers with dedicated hardware and software to find a specific hash function or string of numbers. 4 The first miner to do so is compensated with
展开阅读全文