The recessionary impact of bitcoin energy consumption
Much commentary has been devoted to bitcoin’s unsustainable proof-of-work algorithm that requires the consumption of a vast quantity of electricity. By some estimates, by February 2020 Bitcoin will use as much electricity as the entire world does today. This post explores the energy intensity of Bitcoin and what it might mean.
One of the benefits of cryptocurrencies is the theoretical reduction in transaction costs due to the elimination of centralised trust, and therefore the disruption of the conventional finance industry and the enormous costs of running and profiting the industry.
The idea of crypto was to piggyback onto the existing internet infrastructure, and run the system on a network of servers to perform the bitcoin ‘mining’, verification and updating of the transaction ledger (or blockchain). The trade-off is that investment is required to finance and run the server infrastructure, and also pay for the electricity for running and cooling the server farms. Investors are compensated by payment in bitcoin.
The sting in the tail is that the greater the processing power that is engaged in mining, the greater the mining ‘difficulty’, and therefore the greater the hash rate of the Bitcoin network. This comes about because the average time between blocks is fixed at 10 minutes.
The staggering energy statistic is that the electricity per bitcoin transaction is given as around 852 kWh, equivalent to 29 US households, based on this estimate. Since the processing effort is the same regardless of the value of each transaction, the absurd outcome is that a transaction worth a few dollars required the electricity equivalent of 29 US households. In practice, the average value of each transaction is USD3,010 using data from here. Based on bitcoin electricity consumption data from here and transaction data from here, and converting to primary energy equivalence, I calculate 2.1 MJ of primary energy per 1 USD of bitcoin transaction. This is the critical number to explore the energetic viability of bitcoin.
To begin with, I calculated the energy intensity of the US economy at 4.5 MJ/USD based on GDP of USD18.6 trillion and primary energy of 84 quadrillion Btu from here. So as a starting point, the bitcoin network runs at about 47% (2.1 / 4.5) of the energy intensity of the entire economy. Is this high or low? What does this number mean?
In general, the energy intensity of primary and secondary industries are much higher than the national average, and for service industries much lower. Primary and secondary sectors include agriculture, mining, construction and manufacturing. From a study I conducted for Australia (assuming it is comparable with the US), I estimated the primary energy intensity of the ‘Auxiliary Finance and Insurance Services’ industry at around 0.08 MJ/USD of value-added (1 AUD = 0.76 USD). A low figure is what we would expect for a service industry. Energy consumption in finance consist of the direct consumption of electricity (e.g. for office buildings) and petroleum (e.g. to run vehicles), along with the indirect energy costs of those industries respectively. Significant costs for the finance industry include computer systems design, internet services, employment services, and travel agency services. The services of finance consists of more than simply transactions, but providing the entire suite of financial services.
Returning to the US economy, the energy intensity of bitcoin transactions is 47% of the national economy. Therefore, if the entire economy was transacted in bitcoin, primary energy consumption would increase by 47% just to energetically fund transactions. This is obviously absurd. Depending on the cost of energy, and the prevailing price elasticity, an increase in energy consumption of as little as 5% can easily be recessionary.
If bitcoin was to run at a similar energy intensity to finance, it would probably need to reduce its energy intensity by 97% using the figures in this post. Since bitcoin represents an infinitesimally small proportion of economic transactions, energy consumption estimates that are well outside the bounds of normal are possible. However, it seems obvious that bitcoin is not scalable as long as it relies on the proof-of-work algorithm. Even if common sense didn’t prevail, the iron law of energy would ensure that economic collapse preceded large scale implementation of bitcoin.