lowest price during the last two years of around 2,000 US-Dollars, with the inclusion of transaction fees, there is a mining reward of 5 million US-Dollar every block (i.e., every ten minutes) (Blockchain.com, 2019h). This led to an arms race by the miners, who kept investing in their hashing capability in order to increase their chance of finding the nonce in the next block and therefore receiving the block award. As mentioned in the previous section, the difficulty of finding the nonce adapts dynamically to the hashing power of the network. Figure6 shows the development of this difficulty and the corresponding network hashing power from mid 2016 to mid 2019. With the rising investment occurred another phenomenon. In order to decrease their risk, miner begun to work together in finding a nonce. This collaboration is called a Mining Pool. These pools make up to around 90.4% of the total network as of today (Blockchain.com, 2019d).
The raise of hashing power was accompanied by a rise of energy consumption. Dur-ing its peak time in August 2018, the total Bitcoin network had an estimated energy consumption of 73 tera watt hours per year (Digiconomist.net, 2019). For comparison:
That is more than the whole country of Austria consumed in 2016 (CIA, 2016). Due to the surge in demand since 2017, the network also witnessed a huge congestion prob-lem. Since the Bitcoin protocol has a limited block size capacity, which was discussed in the previous section, the transaction fee rose substantially. The mean transaction fee increased to 1.89 US-Dollar. However, for short periods of time, the transaction fee did increase to 37.49 US-Dollars per transaction (Bitcoinfees.info, 2019).
2.2 Economic Literature
2.2.1 Economic Analysis of Bitcoin
With raising popularity of cryptocurrencies and Bitcoin, the amount of economic lit-erature on Bitcoin also grew. This subsection will introduce several topics. First, it introduces literature on Bitcoin adoption. The second part will introduce the literature on mining games including selfish mining. The third part of literature will cover trans-actions and their fees. The fourth part introduces the literature on security concerns in Bitcoin and the last part will cover the literature of design issues in Bitcoin.
This first part will begin discussing the literature on Bitcoin adoption as a currency.
Luther (2016) writes that due to switching costs, the widely acceptance of Bitcoin is
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Figure 6: Bitcoin Hashing Power and Difficulty since 2016 Data from Blockchain.com (2019c) & Blockchain.com (2019a)
unlikely, unless there exists governmental support or the occurrence of hyperinflation.
This is due to network effects and a possible lack of coordination. Since his writing, an example for a broader adoption due to hyperinflation actually has been witnessed in the case of Venezuela, where inflation rates of over 300,000% (Di Salvo, 2019) drove people towards the adoption of Bitcoin. Athey et al. (2016) analyzed the geo-graphic distribution of Bitcoin users and they were able to differentiate them by their behavior patterns and times of usage. Catalini & Gans (2016) believed that Bitcoin and other cryptocurrencies theoretically can reduce the costs of verification and networking, hence the blockchain is challenging existing business models and opens up new ones at the same time.
A larger amount of research is committed to Bitcoin mining and its games. Altman et al. (2018) analyzed the mining procedure from a congestion game approach. They believed, that more miners increase the security of the overall network of a blockchain, since a higher hashing capability makes it harder for a single malicious miner to take over the blockchain. This higher robustness could increase the real value of the under-lying cryptocurrency. On the other hand, the more miners there are on a given chain,
‧
the higher the negative mining externalities for all miners, because everyone’s chance of mining a block (finding the correct nonce) decreases with the amount of miners in the network. This rises the risk of the investment of miners, find Cong et al. (2018). They write that as a consequence miners are inclined to collaborate in mining pools in order to decrease their overall risk. Since a bigger pool leads to a lower risk, mining pools tend to centralize. Nonetheless, they self-regulate themselves in not becoming too big, as in fear of a majority attack, the whole currency might collapse, self damaging the pools.
Another aspect of mining falls under the term of selfish mining, where the assumption, that every miner always mines the newest block and broadcasts his findings immedi-ately, is dropped. Carlsten et al. (2016) say that only relying on transaction fees, as it is currently planned in the future of Bitcoin, would harm the overall security of the Bit-coin network, as miners would only mine a block when the transaction fees in prospect would outvalue the mining cost. They state that their hypothesis only holds in a world where the block size remains vastly larger than the demand for transactions. Given the developments of recent years, it can be said though that this assumption is not always true. Goren & Spiegelman (2019) confirm that selfish mining, or smarter mining as they call it, will lead to a reduction of the network security, but also has the potential to decrease the energy consumption as a consequence. Sapirshtein et al. (2016) find that selfish mining is a real threat to Bitcoin. Given that the total amount of miners and their hashing power is not too large, it becomes profitable and feasible for miners to mine selfishly. The selfish mining strategy outlaid by the authors includes an inten-tional forking of the blockchain in order to force other miners to abandon their currently mined block.
The following literature focuses more on the transactions and their fees inself. Kroll et al. (2013) say that they do not believe that transaction fees will play any important role in the future of the Bitcoin ecosystem, given the current protocol, since miners will always underbid their competitors and therefore decreasing the fees.1 Easley et al.
(2019) included the waiting time for a transaction to be picked up by a miner in their analysis and concluded that even if there is a high amount of transaction fees that this
1It appears, that the authors did not foresee the increase of demand of transactions on such a big magnitude as one could observe during recent years.
‧
does not guarantee a long term equilibrium for the network security. If the waiting time for a transaction of users becomes too long, they might leave the blockchain all together in order to optimize their personal utility in a similar fashion as a miner which leaves a blockchain to optimize his revenue. Therefore, fees are also a way to balance the demand for transactions and decrease the waiting time for users, which gives the trans-action fee a crucial role to stabilize the network security. Pappalardo et al. (2018) point out that the Bitcoin protocol in its current form is not efficient. Miner tend to ignore transactions with smaller fees, leaving 42% of all transactions still unprocessed after the first hour of issuing and discover that 20% of transactions still have not been processed after 30 days. The majority of these transactions have a small volume, therefore the nominal efficiency of the Bitcoin blockchain is still on an ”acceptable” level. However, the authors questions how well Bitcoin is suited as a time stamping system for small transactions with a condition as described.
There is also a focus on security concerns in the literature. Pagnotta (2018) writes that the Bitcoin price and the network security are positively correlated and jointly deter-mined. Users value a higher security of the network and are willing to pay a higher price for it, while miners respond positively to a higher Bitcoin price and invest in more hashing power, therefore increasing the network security. Budish (2018) believed that the continuous flow of payments for miners must be large enough to prevent a majority one-off attack of the system. Buterin (2016) states that the costs for security in form of the verificators or miners are inevitable. He argues that there should be an optimal combination of inflation and transaction fees which minimizes the dead weight loss for the transaction market. Furthermore, he discusses other ways for the redistribution of the transaction fees. He proposes the idea to either distribute the fees among all miners no matter who mined a block or to burn the fees all together (therefore appreciating the currency). He concludes that eventually it is a tradeoff. If the fees are distributed, there is a high level of certainty of Bitcoin supply, but the level of security is less certain and if the fees are burned the Bitcoin supply is less certain, but the level of security is.
The last section of economic research focuses on the design questions in regard to scal-ability and economic performance of Bitcoin and blockchain. According to Ma et al.
‧
(2018), free entrance to the Bitcoin mining market is the main cause for the high energy consumption. The authors suggest two possibilities to reduce the energy waste. Firstly, restrict miners from entering the blockchain freely.2 Alternatively, removing the infla-tion (i.e., the new Bitcoin generainfla-tion) and relying solely on transacinfla-tion fees. Croman et al. (2016) find that a reparameterization of the current Bitcoin protocol should only be the beginning and is not sufficient to solve the scalability problem of Bitcoin in the long term. They propose a variety of technical improvements and additions to the pro-tocol that might be better suited to address Bitcoins problems. Huberman et al. (2019) also find that Bitcoin does not scale enough with its current protocol. They argue for a flexible block size limit, which reacts to the demand of transactions. Therefore, bal-ancing the delay costs for users and the revenue for the miners. Lastly, Chiu & Koeppl (2017) analyze how well Bitcoin is serving as means of payment. Their analysis finds that with peak demand, Bitcoin is 500 times more costly than other traditional curren-cies. Given the welfare as the overall surplus of a modeled market, using Bitcoin as a currency, the welfare loss totaled with 1.41%. The authors then proceed to establish an equation-based model to calculate the parameters for optimal welfare and conclude that a Bitcoin protocol with zero transaction fees and a small inflation would minimize the mining costs and maximize the total welfare.
2.2.2 Agent-Based Computational Economics in Blockchains
The literature on agent-based computation of blockchains or cryptocurrencies is com-paratively small. Terna et al. (2016) simulated the adoption of Bitcoin wallets. The authors adapted an epidemic model lent from health care research to analyze the spread of trust and adoption of Bitcoin. One of their findings is the lack of trust in cryptocur-rency which is correlated with the amount of transactions and the amount of agents in an agent’s proximity that uses a cryptocurrency as well. Due to a lack of confidence in cryptocurrencies, agents do not like to hold a cryptocurrency for a long time, selling it therefore quickly. This leads to a higher amount of transactions, which then allows for the generation of confidence in cryptocurrencies in the model.
Cocco et al. (2017) modeled the Bitcoin blockchain with a trading system using two kinds of agents. The first one called random trader, trades Bitcoin for diversification or
2This would imply a smaller degree of decentralization of the blockchain.
‧
liquidity, similar to the zero-intelligence agent by Gode & Sunder (1993) without any real constraints besides their own budget. The other category is called chartist, which used different strategies involving genetic algorithms to optimize its revenue. They could confirm that the daily trading returns of absolute prices are indeed autocorrelated and follow a negative exponetialy shaped cumulative distribution function with a fat tail. In their next work, Cocco & Marchesi (2016) modeled the Bitcoin blockchain in-cluding the mining system. It added the new agent type miner. This category could acquire mining hardware and generate new Bitcoins. Its goal is also to maximize its revenue. The authors were able to reproduce stylized facts, such as the unit root and volatility clustering. This thesis is based on this model. In their newest publication, Cocco et al. (2019) solely focus on the mining aspect of Bitcoin and ignores the trading aspect. It compares the profitability of mining Bitcoin to that of mining gold and comes to the finding that mining Bitcoin is superior compared to mining gold. The authors state that if everything included, the costs of using Bitcoin might actually be smaller than those of a traditional currency.
Zhou et al. (2017) use a simple agent-based model to try to capture the effect of the amount of agents. Their findings, while limited, say that the amount of agents directly affects the Bitcoin price and the deal rate of transactions. A high number of agents decreases the price fluctuation and the overall deal rate is higher. Finally, Lee et al.
(2018) first use inversed reinforced machine learning to deduct trading rules from the real Bitcoin transactions. In a second step they applied these rules to an agent-based model in order to forecast the price and performance of Bitcoin. They were able to establish an 80% directional accuracy of their price prediction within the first six days of forecasting.