In-depth analysis of the potential impact of EIP-1559 through on-chain data

Written by: William, works at Huobi Research Institute

summary

Judging from the on-chain data, EIP-1559 currently has four major phenomena that go against market intuition.

First, the transaction fees of the Ethereum network have not dropped significantly. The reason is that the "tax distortion" model commonly used in the market is not applicable to the Gas transaction market, and the motivation of EIP-1559 is to improve efficiency rather than reduce prices.

Secondly, EIP-1559 can tactically alleviate the problem of transaction congestion. The block capacity has increased significantly after the London upgrade, which can alleviate the problem of transaction congestion to a certain extent. The main reason for the increase in block capacity comes from the particularity of the BaseFee price formula, which increases the average block utilization rate from 50% to 51.7%.

Again, EIP-1559 did not cause deflation in Ethereum, but it successfully slowed down inflation. Judging from the daily output of ETH and the daily destruction of ETH, at least for now, EIP-1559 alone cannot achieve deflation, but it has successfully slowed down inflation.

Finally, after the London upgrade, miners’ income has not dropped significantly. Even if the ETH price increase is removed, miners’ income has not dropped significantly. The reason is that more than 80% of the transactions in the current blocks are generated in the original transaction (Legacy) mode, while the EIP-1559 transaction mode only accounts for about 20%.

It has been more than ten days since EIP-1559 was launched. Frankly speaking, even though it has been successfully upgraded, EIP-1559 is still full of controversy. These controversies include the impact of EIP-1559 on handling fees, miners' income, ETH prices, etc. So what is the real situation? Let's explore these controversies through on-chain data.

On-chain data 1: Transaction fees have not dropped significantly

One of the major controversies of EIP-1559 is whether the proposal can reduce Ethereum transaction fees. From the on-chain data since the successful London upgrade on August 5, if denominated in ETH, the transaction fee has not dropped significantly; even when denominated in US dollars, due to the rise in ETH prices, the transaction fee has increased to a certain extent.

There may be different opinions: EIP-1559 mainly reforms the Gas auction mechanism, which affects the Gas auction price rather than the simple handling fee. Similarly, we can see from the on-chain data that the Gas auction price did not change significantly before and after the London fork.

The above results are consistent with our previous research predictions. The reason is that the main purpose of the EIP-1559 proposal is to improve efficiency rather than reduce the transaction fee price. From the motivation of the proposal, EIP-1559 mainly improves the inefficiency of current Gas transactions from the following four aspects:

1. The mismatch between the volatility of transaction fee levels and the social costs of transactions;

2. Unnecessary delays for users;

3. The inefficiency of first-price auctions

4. Instability of blockchains without block rewards

It is worth mentioning that some users in the Chinese community currently like to use the tax distortion effect theory to analyze the issue when discussing the impact of EIP-1559, and thus infer that EIP-1995 may increase Gas prices, reduce Gas supply, and cause unnecessary losses.

However, the problem with the above analysis method is that it ignores the particularity of Ethereum Gas transactions. In the traditional analysis of tax distortion effects, the implicit assumption is that all manufacturers are competitive manufacturers. At this time, the market supply curve is derived from the marginal cost curve of the manufacturer. The marginal cost curve of an ordinary manufacturer is a right-sloping curve, so the supply curve in tax theory is a right-sloping curve (see the figure below).

The special thing about Gas transactions is that after miners win the accounting competition in each block, they obtain the monopoly right of Gas auction, and monopoly manufacturers do not have a supply curve. The way a monopoly manufacturer determines the price or output is that the marginal revenue (MR) is equal to the marginal cost (MC); the special thing about miners is that the marginal cost of obtaining transaction fees is almost 0. Because the main source of income for miners is block rewards, the marginal cost in the short term mainly comes from electricity costs, and the impact of packing one more or one less transaction on the marginal cost of miners is negligible.

In addition, due to the existence of the auction mechanism, miners can achieve first-level price discrimination against users, that is, miners can set different prices for each unit of Gas based on the user's willingness to pay. Therefore, as long as each user's Gas quote (P) is higher than the miner's marginal cost (MC=0), the miner will package the transaction until the block Gas limit is reached. Even if miners can only charge a small amount of TIP or even no TIP in EIP-1559, miners are still willing to package transactions (TIP≥MC=0). This also explains why miners are still willing to package transactions even if the Base Fee is destroyed after the implementation of EIP-1559.

On the other hand, it is biased to regard the Base Fee as a "tax". Taxes are an additional cost (P+t) added to the original price, while EIP-1559 splits the original price into two parts (P = Base Fee + TIP). Therefore, the Base Fee under EIP-1559 cannot be considered a tax because it does not bring cost burden to both supply and demand. The only change brought by EIP-1559 is that the maximum price that users are willing to pay to miners has changed from the previous Gas Price to TIP.

As the on-chain data shows, several important conclusions drawn from the tax distortion theory: the increase in transaction fees, the reduction in gas supply and the emergence of unnecessary losses, have never actually occurred.

On-chain data 2: EIP-1559 can tactically alleviate transaction congestion

Another interesting data since the London upgrade is that Ethereum's average daily Gas consumption has increased significantly, from 92 billion to the current 100 billion. This also suggests that the block capacity has increased significantly after the London upgrade, which can alleviate the transaction congestion problem to a certain extent.

So, what is the source of the nearly 10% increase in Ethereum block capacity? --- The block relaxation mechanism design of EIP-1559.

As we all know, EIP-1559 uses two parameters to control the block size: Gas limit and Gas target. Gas target is an ideal block space that Ethereum wants to maintain for a long time. It is numerically 50% of the Gas limit, and the average block utilization is controlled at 50%. In order to achieve this goal, EIP-1559 uses Base Fee for adjustment:

• When the usage rate of the previous block exceeds 50%, the Base Fee increases proportionally (the upper limit for each block is +12.5%)

• When the usage rate of the previous block exceeds 50%, the Base Fee increases proportionally (the upper limit of each block is -12.5%)

Furthermore, the specific adjustment formula of Base Fee is as follows:

Here, b represents Base Fee, t represents block height, T represents block size, the target block load is T/2, g represents the number of transactions included in the block, and since g depends on the base fee b, we denote gt | bt as the number of transactions included in Bt when the base fee is equal to bt. D is the adjustment factor, currently set to d=0.125.

However, the above formula design will lead to an interesting mathematical phenomenon: Assuming there are four blocks now, the utilization rate of the first block is 50%, the utilization rate of the second block is 0%, and the utilization rate of the third block is 100%. Assuming the initial Base Fee is 1, what is the Base Fee of the fourth block? The details are shown in Table 1.

As can be seen from the table, although the average utilization rate of the first three blocks is 50%, under the algorithm adjustment, the Base Fee of the fourth block is 63/64 instead of 1. So what is the equilibrium condition that keeps the Base Fee unchanged under the fluctuation state? Roughly speaking, this requires the ratio of full blocks to empty blocks to be -ln(7/8)/ln(9/8)=1.134, that is, the proportion of full blocks to all blocks is 53.13%, and the proportion of empty blocks to all blocks is 46.87%, or the average utilization rate of each block is 53.13%, that is, the average Gas consumption of each block will increase by about 6%.

Of course, the above estimates are only rough calculations. According to the actual on-chain data, after the London fork, the average Gas consumption per block rose to 15.5 million, that is, the average block utilization rate was 51.7%, and the actual increase rate was 3.3%.

The above are the main reasons for the increase in block capacity after the EIP-1559 upgrade. It should be noted that this is not the only reason for the increase in Ethereum's daily average gas consumption. Some current data studies have found that the London upgrade delayed the arrival of the Ice Age and shortened the block time, which is also an important reason.

It should be emphasized that EIP-1559 cannot fundamentally solve the current congestion of the Ethereum network. After all, EIP-1559 only expands each block by 3%. The fundamental solution to the congestion problem lies in Ethereum Layer 2, sharding technology and other expansion solutions.

On-chain data 3: EIP-1559 did not cause deflation, but successfully slowed inflation

Another focus of EIP-1559 that has attracted market attention is that the destruction of Base Fee may bring deflationary effect to Ethereum. However, from the actual data, EIP-1559 did not cause deflation of Ethereum, but successfully reduced its inflation, making the actual circulation of ETH significantly slower than before the "London Upgrade".

There are three main sources of Ethereum supply: one is the block reward (Block Rewards), which currently provides 2 ETH for each block; the second is the uncle reward (Uncle Rewards); and the third is the reward for including uncle blocks in confirmed blocks (Uncle Incl Rewards). Currently, the daily output of ETH, including these three parts, is around 13,000-13,500 ETH, while the amount of ETH destroyed by EIP-1559 is only around 5,000 ETH per day. Therefore, at least from the current data, EIP-1559 cannot cause deflation in Ethereum.

On-chain data 4: Miners’ income has not yet shown a significant decline

Before the London upgrade, many people predicted that EIP-1559 would reduce miners’ income. However, surprisingly, from the on-chain data, when denominated in USD or USDT, miners’ income increased significantly after the London upgrade due to the increase in ETH prices; but even if the influence of ETH prices is removed, miners’ income (or handling fees) denominated in ETH has only slightly declined, and income has not decreased significantly.

However, a new question arises: why is there only a "slight" decline in revenue? After all, the Base Fee, which accounts for most of the transaction fees, has been destroyed. In theory, it is reasonable for the transaction fees to decline significantly.

From the analysis of the income structure of Ethereum miners, it can be seen that after the London upgrade, most of the fees are still paid in the original mode. This is why the income of miners has only declined "slightly" even if it is denominated in ETH.

EtherScan’s data can further confirm the above reasons: after the EIP-1559 upgrade, transactions in each block retain three modes. The first is the EIP-1559 mode we are familiar with. The second is the Access List mode. This transaction mode is mainly derived from EIP-2930. It is mainly to solve the main network transactions blocked by EIP-292; the third is the original transaction (Legacy) mode.

Refer to "1559 Cheatsheet for Implementers", the original transaction mode here is not the original first-price auction mode, but interprets the gas price into three parts: maxFeePerGas, baseFeePERgas and maxPriorityFeePerGas, where maxFeePerGas represents the maximum amount the user is willing to pay, baseFeePerGas will be destroyed, and maxPriorityFeePerGas belongs to the miner. For example, assuming the Base Fee is 100 gwei:

• EIP-1559 transaction mode: The user sends a transaction with a GAS Price of 250 gwei and a TIP of 5 gwei. The user will pay 100+5=105 gwei, of which 100 gwei will be burned, 5 gwei will go to the miners, and the remaining 145 gwei will be returned to the user.

• Legacy mode: A user sends a transaction with a GAS Price of 250 gwei. At this time, maxPriorityFeePerGas will be sent out as 250 gwei, of which 100 gwei will be burned, and 250-100=150 gwei will belong to the miner.

According to EtherScan data, more than 80% of the transactions in the current blocks are generated in the original transaction (Legacy) mode, while the EIP-1559 transaction mode accounts for only about 20%. This is the main reason why the miners' fee income has only decreased "slightly".