Sansan Tech Blog

Sansanのものづくりを支えるメンバーの技術やデザイン、プロダクトマネジメントの情報を発信

The Economics of Blockchain Energy Consumption: A Short Review

Economics Meets Data Science


Hi there!. I'm Juan, from Sansan R&D's SocSci Group. Today's post is about blockchains. Cryptocurrencies and blockchain technology have been in the spotlight for the last year. About one year ago Bitcoin was recognized as a legal tender by the first nation-state. Cryptocurrency prices have seen historic volatility. Crypto assets have been a subject of debate regarding the topic of intellectual property rights.


And of course, there's the topic of energy consumption. As energy prices surge all over the world, even more attention has been put on the energy consumption of the major blockchains. On a related note, by the time this article gets published, Ethereum will have gone through a massive upgrade: The Merge, with which it will shift from an energy-intensive consensus protocol to one that is based on staking funds. All of this debate about blockchains and energy consumption made me think: what do economists have to say about that topic? And this reminded me of this post by professor Kohei Kawaguchi, which I saw a few months ago:

The paper by Kawaguchi and Noda employs data on cryptocurrency prices, the source code of some major cryptocurrencies, and blockchain data to explore many topics regarding the incentives that cryptocurrency miners face. Of course, the decisions of miners have implications for the global energy consumption, so I think it is a very relevant paper. So I just want to make a brief comment about their research. Hopefully this will motivate other researchers to keep studying the economics of blockchains, and help policy makers and investors make better decisions.

Inside the Blockchain

Blockchains are essentially decentralized databases with a special structure. They do not exist on a single computer, but in networks of unrelated and anonymous participants. Every node in that network holds a copy of the database. When blockchains hold valuable data, any node could have the incentive to alter their copy in a favorable way. So, how do blockchains decide which copy of the blockchain is the legitimate copy? Incentives!

Different blockchains use different mechanisms for agreeing on the legitimate copy. These are called consensus mechanisms. Two popular ones are the Proof-of-Work (PoW) mechanism and the Proof-of-Stake (PoS) mechanism. Early blockchains, including Bitcoin and some of its forks use PoW. This consensus mechanism makes it difficult for anyone to modify the blockchain in their favor by requiring the consumption of lots of energy for adding new blocks.

PoW and Energy Consumption

Let's look at the case of Bitcoin. A block is a batch of transactions. Every block contains a header, which contains information such as a version number, a pointer to the previous block, a timestamp, and a few other components. Some of those components can be adjusted by the miner when they attempt to add a block to the blockchain.

If you apply some hashing to a block header you can generate a fixed-length output, which can be parsed into a number. By changing its contents even a little, you can generate a completely different number. In other words, changing the contents of the header and hashing it equals to generating uniformly distributed random numbers.

A miner can only include a block in the blockchain if it satisfies a certain condition. In particular, its header, interpreted as an integer, must be smaller than some target value. This happens at random with a very low probability. In other words, the job of miners is to generate lots of random numbers until this condition is met. They do it because, by appending a block to the blockchain, they receive a reward in the form of fresh new cryptocurrency. But only the miner who gets to add the block receives the reward, so all miners have the incentive to be as fast as possible. This is the energy-consuming part!

Energy Consumption and Security

The more energy-consuming this process is, the more secure the blockchain becomes. Why? Because if at some point one miner decided to modify the blockchain on their favor by cheating, they would have to start adding blocks until his version of the blockchain is the longest (and therefore, the legitimate one). That would mean that their random number generation power should be larger than everyone else's. That would require a lot of energy. Cheating on the blockchain would have to be profitable enough to pay for the energy bill.

Of course, the difficulty of adding new blocks is something relative. New technology can make the process much easier and energy-efficient. Higher energy costs may discourage miners from supplying their labor to the blockchain. New miners can join, reducing the probability that a single miner will earn the reward.

If adding blocks to the blockchain becomes too difficult, few miners will want to enter the market. The less miners, the less random number generation and the less security for the blockchain. Also getting transactions confirmed would take much longer, making the blockchain very inconvenient for use cases such as payments. Usually, a layer 1 transaction on Bitcoin gets confirmed once after 10 minutes on average. This could be much longer if the hashrate the miners provide goes down.

That is why Bitcoin, as well as other blockchains, includes a difficulty adjustment mechanism. Through this process, the target value gets adjusted every 2,016 blocks. It is adjusted based on the hashrate in such a way that the time to add a block regresses to the mean of 10 minutes. You can find some information about this process here:

www.bitcoinmining.com

The Economics of Cryptocurrency Mining

Increasing energy expenditure does indeed increase the level of security, but how much depends on the behavior of miners. There is therefore a concept of energy-efficiency of a blockchain. Different blockchains parameterize difficulty in a different way. The process I described in the previous section is the one used as of this date by Bitcoin. Its forks use different difficulty adjustment algorithms. So the question is: are some blockchains more efficient than others?

The paper by Kawaguchi and Noda employs counterfactual simulations and studies how the adoption of different adjustment algorithms can affect energy consumption, security, block time and many other metrics. They create a measure of efficiency called Security per Energy Consumption (SpEC), which is basically obtained from dividing the hashrate at the bottom 5th percentile by the average hashrate. The hashrate at the bottom 5th percentile can be considered a measure of security, because the blockchain is the weakest when the hashrate is the lowest. Under this measure, a blockchain is more efficient when it can keep a low (less energy consuming) but stable (less prone to attacks) hashrate.

The counterfactual simulations are employed to compare how adopting different difficulty adjustment algorithms could affect the efficiency of blockchains, after taking into account the elasticities of miner labor supply with respect to theirs and other currencies rewards. Their simulations show how important an inelastic supply of miner labor contributes to the efficiency of energy consumption of blockchains. Bitcoin, having the most inelastic supply, is the most efficient under any of the three difficulty adjustment algorithms used for comparison. Also, small improvements in Bitcoin's efficiency due to a switch in its difficulty adjustment algorithm can lead to large reductions in energy consumption while keeping the security level constant.

The paper by Kawaguchi and Noda is definitely a great read for anyone interested in understanding the incentives behind blockchains. Of course, I have just highlighted a piece of the paper that I considered very relevant and interesting. The paper explores many other topics regarding the incentives of miners for Bitcoin and some of it forks.

Every day there are more companies that begin to offer blockchain-based services. With so many blockchains out there, it may be difficult to choose a solution or even decide whether a centralized database does the job well enough. I think that it is important for decision-makers to understand the fundamentals of blockchains before jumping into a solution.

Finally

If you are interested in blockchains from the perspective of Economics, you may like the following lectures by Tim Roughgarden:

www.youtube.com


Until the next time!

© Sansan, Inc.