This article attempts to translate my understanding of the macroeconomics of cryptoasset mining into a deductive, simple, but quantifiable model with clear feedback loops.
The Why
There is currently no united framework, which could describe the economics of cryptoasset mining. Mining economics can get hard to conceptualize due to hidden variables, which become backbreaking if ignored.
The What
I believe there are four main variables in play, which influence each other in a direct, circular fashion. There are indirect or weaker interactions between the variables in on-circular fashion, which we can explore later, however, the base model is sufficient without their consideration.
The indirect relation between hashrate and asset price is explored in this series of articles: https://medium.com/@ambroidcrypto/the-hashrate-deep-dive-1-2-85b681f0d945
I propose these main variables all tied together by Margin:
- Asset Price (AP) — The price per one unit of a cryptoasset, denoted in USD.
- Average Hardware Efficiency (HW) — The amount of hashpower an average piece of hardware in the network produces compared to the electricity it consumes.
- Global Hashrate (H)— The total amount of hashpower pointed at a given network.
- Average Network Electricity Rate (ER) — The average rate at which a theoretical miner is mining. The main expenditure for a mining farm.
- Margin (M)— The main profit variable. Positively impacted by higher HW efficiency and higher asset price. Negatively impacted by higher electricity rate and higher hashrate.
The How
The margin considerations are:
Higher asset price increases margin.
Higher hardware efficiency increases margin.
Higher electricity rate lowers individual miners margin.
Higher global hashrate lowers individual miners margin.
Let us explore the circular feedback loop behind the variables:
AP →HW
Development pressure — with higher underlying asset price, or with the expectation of an eventual higher asset price, more prospective miners want to join the network. This means more hardware. With more HW, Wright’s Law* implies higher efficiency in producing these mining accessories, similar to Moore’s law.
An example of lack of growth of price stalling innovation is the case of Litecoin. Litecoin network is still mostly stuck with Antminer L3+, machines from 2017.
HW → H
Growth pressure — more (and more efficient) hardware causes growth pressure on terms of hashrate. Difficulty of mining rises proportionally with more hashrate joining the network (to eliminate higher emission).
H → ER
Efficiency pressure — higher hashrate means more machines. This constant expansion causes miners to look for localities with cheap(er) electric rates. As mentioned above, these are places with abundant (too abundant) local sources of usually renewable energy. Said differently, higher hashrate is the reason for search of higher margin in form of lower electricity rate.
ER → AP
Miners not only want to, they need to get paid. This is caused by monthly electricity bills, creating sell pressure on the market of the underlying asset being mined. This is especially noticeable in high inflation markets, where miners generate high relative amounts of supply compared to the current circulating supply (eg. GRIN, ZEC).
The takeaway is that the economy of mining cryptoassets is circular and full of feedback loops.
On top of this, we completely ignore the lower issuance over time (halving), which sometimes suddenly cuts the margin in half.
Quantification for Bitcoin
The model is completely deductive and doesn’t have to be enumerated to make sense intuitively, however, it can be done.
Profit Margin is simply the cost of the existence of the network compared to revenue from block rewards (coinbase and transaction fees).
Profit Margin will be the ratio (or percentage) of our mined coins that are purely profit and do not have to go towards paying for our electricity bill.
The costs of the network C:
C = ER * HW * H
Revenue of the Bitcoin network R:
R = 13.1 * 6 * AP/565 s
Where 13.1 is our average block reward (coinbase + tx fees) and 565 s is the actual block time (every 9 and a half minutes).
Where
AP = 8000 $
ER= 5 c/kWh = 0.05 $/kWh
H =102 EH/s = 1.02 * 10⁶ TH/s
HW = 0.08 J/GH = 80 J/TH = 80 * 2.78*10^-7 kWh/TH
Considerations:
HW, or the efficiency of the average miner is the hardest variable to get. It’s going to be a mix between the largest bulk of miners in the network (Antminer S9 @ 0.098 J/GH) and the newest generation which is being produced (Antminer S17 @ 0.04 J/GH), leaning towards the older gen.
ER is similarly hard to estimate and it might be even higher than 5 cents and still be profitable, as we will see lower.
C = ER * HW * H
C = 0.05 $/kWh * 80 * 27.78 * 10^-6 kWh/TH * 1.02 * 10⁶ TH/s
C = 0.05 * 80 * 27.78 * 1.02 $/s
C =113.34 $/s
R = AP * 13.1/565 s = 8000 * 13.1/565 $/s
R = 185.48 $/s
Raw profit (R - C) of the miners in the Bitcoin network is therefore in the whereabouts of 60 dollars per second.
M = (1 - C/R) * 100 % = (1 - 113.34/185.48 ) * 100% = 38.9%.
Profit Margin of the average Bitcoin miner is in the neighbourhood of 35–40%.
Discussion
Please keep in mind that for this discussion, we ignore the (often significant) first time investment (CAPEX) into building a mining farm and consider only already developed farms, which have ROI’d.
If we increase ER to 0.08 $/kWh, we notice the profit margin becomes negative (C>174.7 $). This means that any miner with an average market mix of ASIC miners is in the negatives if his electricity costs are above 8 cents per kilowatt hour.
Similarly, we can easily calculate when it will again be profitable to mine on an Antminer S5 (efficiency 0.511 J/GH). The answer is that if your electricity rate is below 1.2 cents you’re profitable even on an ASIC from 2015.
Consider what would happen if the entire network instead switched to Antminer S17’s: the profit margin would shoot up to 70%. This means that the price of Bitcoin could drop to 2700$ and the network would still be profitable at the current hashrate.
Or instead — how high can the hashrate of the network grow at current price and the miners will still be profitable? The answer is about 300 EH/s.
Notes
This model is of course imperfect, as all economic models are. Anyone who claims to build a perfect economic model and/or says the world should conform to their model is a charlatan.
If you can think of any improvements that do not overly complicate the model, please let me know. If anything seems outright wrong, again, please let me know.
As a bonus so you can play around, I have made a spreadsheet with the model above applied to different cryptos so you can play around with it:
Source for Wright’s model:
*http://www.uvm.edu/pdodds/research/papers/others/1936/wright1936a.pdf
