Chainalysis Archive

Ethereum upgrades and bitcoin on-chain activity falls as market liquidity moves from exchanges

06 August 2021
    • The Ethereum price has climbed with much interest around the London Hard Fork and EIP-1559, which decreases supply as demand to use Ethereum increases, but there is a large amount of illiquid supply that could activate in replacement. 
    • Bitcoin sent via the blockchain has declined 32% since early March, primarily due to a decline in bitcoin sent by entities holding less than 0.1 bitcoin for less than 2 weeks, which is typically retail and not so important to the market, and a decline in bitcoin sent from crypto-to-fiat exchanges, which is important.
    • Bitcoin liquidity on exchanges is declining but the overall liquidity of the supply is not, suggesting that bitcoin, more of which is now in the hands of institutional investors, often held by custodians, is being traded bilaterally.

The bitcoin price is slightly up over the week, closing at $40k on Thursday 29 July and $40.9k yesterday, Thursday 5 August; although it moved in a range between $42.6k and $37.7k. The Ethereum price continued rising, increasing over the week from $2.4k to $2.8k.

Last week I said that the bitcoin market was at least in a sideways phase and unlikely to enter an extreme price fall phase, with a price floor at $37k. It is nice when an opinion survives at least a week! The Ethereum price has been climbing with much interest around the London Hard Fork, which occurred yesterday.

I probably should give a hot take on how the Hard Fork and EIP-1559 – an Ethereum Improvement Proposal being implemented in the Hard Fork – may affect the price. EIP-1559 changes the way transaction fees are calculated, which will involve ‘burning’ some of the fees, thereby reducing the supply. The more transactions there are – so the more demand there is to use Ethereum – the greater fees will be. As some fees will now be burned, this means that the higher the demand for Ethereum, the lower the supply will be.

I regularly say that crypto prices are determined by a simple interaction of demand and supply, it is just that measuring demand and supply is very hard. This applies to EIP-1559. Any mechanism that decreases supply while demand is increasing should increase the price. But we will have to wait and see how much supply actually gets burned, and whether more of the existing Ethereum supply becomes active to make up for the burned supply.

The supply of Ethereum is one of the most active amongst cryptocurrencies. But there is still a lot of illiquid supply. 53 million Ether, 45% of the supply, is held in self-hosted wallets that hold more than 10 thousand Ether and, based on historical behaviour, have an 8% probability of sending Ether in a given week. The question I have is whether EIP-1559 encourages people to hold rather than use Ether. If they expect the price to rise because supply will decrease then they will hold. But if more people hold then there will be fewer transfers so supply will not decrease as much. As always, the outcome will be a mix of effects, but given the transparency of on-chain data, we’ll be able to measure it! So I’ll skip the hot take for now in favour of a post-event analysis in the future.

This week I’m going deeper into a trend that people have been talking about: the decline in bitcoin on-chain activity. This decline is clear in the first chart of the report, which shows the amount of bitcoin sent each week by different types of users, presented as a 12 week moving average.

The main groups that send bitcoin are entities that hold bitcoin for less than two weeks, in orange, and services, such as exchanges, in gray. In early March they sent 753k and 725k bitcoin respectively, together accounting for 86% of the bitcoin sent on-chain. By the end of July, entities holding bitcoin for less than two weeks were sending 473k bitcoin and services were sending 551k bitcoin, declines of 37% and 24% respectively. All other groups except for traders experienced similar or greater percentage declines but less in absolute terms.

The decline in bitcoin sent by entities holding for less than two weeks came almost entirely from entities that hold less than 0.1 bitcoin, as the second chart in the report shows. Entities that quickly spend small amounts of bitcoin are typically either retail, or bitcoin addresses that are part of businesses Chainalysis has not yet confirmed are controlled by a business. These are typically the addresses used to process withdrawals, and there is often a time lag before we can confirm control. So this decline suggests two things: less retail activity and fewer smaller withdrawals from exchanges, which is also typical of a decline in retail activity. There is a caveat to this: exchanges may have just changed how they process withdrawals but I don’t think this is a major factor.

Overall, I don’t think this decline in on-chain retail activity is significant. Much of this flow is noise, as bitcoin often moves from one quickly-spent wallet to another, then another, then another. So the level of activity represented in this data is an upper bound and the amount of bitcoin actually changing hands is likely much lower. This is important, as it means that half of the recent decline in on-chain activity likely overstates the actual decline in activity, and this actual decline is likely concentrated in retail activity.

The decline in bitcoin sent by services is more significant to the market. The majority, 65%, of this decline from services since early March is due to a 33% decrease in bitcoin sent from crypto-to-fiat exchanges, as shown in the third chart of the report. Other data, not shown here, shows that most of the decline in bitcoin sent from crypto-to-fiat exchanges is due to a decline in flows between crypto-to-fiat exchanges. This also includes bitcoin moved off a crypto-to-fiat exchange by traders and then sent back to the same exchange at a later date. So, since early March there appears to have been a decline in arbitrage and market-making activity between crypto-to-fiat exchanges.

This decline in activity coincides with an increase in bitcoin held by institutions that entered in Q4 of 2020 and Q1 of 2021, who have largely held on to their bitcoin. This appears to have taken liquidity out of crypto-to-fiat exchanges. As the fourth chart in the report shows, median bitcoin trade intensity, the ratio of bitcoin traded for every bitcoin received by exchanges via the blockchain, has been rising since September 2020, from 3.5 then to 7.6 today. This means that the supply of the underlying asset has been declining relative to demand to trade the asset. Even as trade volumes fell in June, trade intensity kept rising, suggesting supply of the underlying remained lower than demand even in a quiet month.

However, this does not mean that liquidity is entirely out of the market. Much depends on how institutional investors trade bitcoin. Many of them will be holding bitcoin via custodians, who can arrange bilateral trades that do not go via crypto-to-fiat exchanges. This may be happening, as the liquidity of the entire bitcoin supply has stayed broadly the same since the end of March. So there has been a change in the venues that provide liquidity, but not in the overall liquidity of the market.

The bitcoin market has always had this dual-structure, of retail on exchanges and a bilateral market for larger players. But historically that was because there wasn’t the market depth on exchanges to accommodate large orders without significant slippage. That is less of a problem today, so if this dual-structure is emerging again it is more likely due to advances in the offerings of custodians and the preferences of institutional investors. That at least is my hypothesis from the data, let me know if you see this in your day-to-day market activity via!

Philip Gradwell, Chief Economist

Privacy Policy
© 2022 Chainalysis Inc. All right reserved.