HashKey Cao Yixin: In-depth Analysis of Ethereum Fee Attribution Model and Rising Logic

This article was published by HashKey Capital, author: Cao Yixin.

One of the reasons for the diversion of the DeFi ecosystem to other blockchains is the increasing transaction fees on the Ethereum network. This article conducts an in-depth analysis of the existing fee model, provides a causation model for fee growth, and attempts to deduce the logic behind the recent rise in fees based on this model. Finally, some conclusions and predictions are presented based on this logic.

Ethereum's Fee Model

A brief review of the analysis of blockchain usage costs in the 21st research report of 2021, "The Development of the DeFi Ecosystem on Different Public Chains," shows that the transaction fee for each transaction on Ethereum and exchange chains is calculated using the following fee model:

Where GasUsed is the actual amount of Gas consumed by each transaction, which is only related to the complexity of the transaction; GasPrice is the cost the transaction sender is willing to pay per unit of Gas, also known as "gas fee," and this cost is measured in the functional token of the blockchain. Therefore, to obtain the fee priced in USD, it must be multiplied by the market price of the functional token UtilityTokenPrice.

As shown in Table 1, the differences in the average "gas fees" and the USD prices of functional tokens among the three chains lead to significant variations in the fees users must pay for a blockchain transaction that consumes the same blockchain resources. The surge in transaction fees on the Ethereum network is also one of the direct reasons for the diversion of the DeFi ecosystem.

Table 1. Comparison of Average Transaction Fees for Simple Transactions on Three Chains over the Last 30 Days (February 14 to March 15)

Fee Causation Model

The above fee model corresponds to an auction bidding model: under the incentive mechanism, miners generally prioritize packaging transactions with higher GasPrice into blocks. Users compete for the right to be packaged into blocks by bidding (setting GasPrice). However, the simple transaction fees on the three chains show three orders of magnitude of difference, which can be attributed to the scarcity value of obtaining block accounting services and the value addition of functional tokens (as shown in Figure 1).

Figure 1. Analysis of the Causes of Transaction Fee Growth

We can break down formula (1) into:

The first factor affecting fee fluctuations is the scarcity value of block dimensions. Under constant TPS, a surge in transaction volume can lead to block congestion, and the "scarcity" characteristic of block dimensions drives the bidding price to continuously rise (①). This part of the cost represents the cost of obtaining block accounting services, which is generally directly priced by traders participating in market bidding and is more related to the urgency of transaction time. For example, in extreme market conditions that trigger a series of urgent liquidation operations or popular applications like CryptoKitties, it can lead to a sudden spike in GasPrice, which is essentially attributed to market heterogeneity. Of course, as the application ecosystem on the blockchain expands, it will also lead to transaction congestion and raise the scarcity value of block dimensions.

When making a horizontal comparison between multiple blockchains, the difference in transaction fees also comes from the difference in GasPriceBase (②), which can be attributed to the scarcity value of the blockchain network dimension, representing the entry cost into that blockchain network ecosystem. This cost is generally set by ecosystem builders as a reference value and then dynamically adjusted with the development of the ecosystem (relying on the market). From the average GasPrice data in Table 1, we can determine that Ethereum has the highest GasPriceBase, followed by BSC, and Heco has the lowest. This value is mainly related to user scale, and the growth of user scale depends on the expansion of the application ecosystem. When the user scale exceeds the limit that the blockchain network can bear, the market will drive GasPriceBase to rise, indirectly increasing the entry threshold of the network to control user scale. This is also one of the reasons for the current rise in Ethereum network fees, which will be explained in detail later.

At the same time, the market price of functional tokens, which mainly captures value through network development (③), will also rise with the expansion of the application ecosystem. In addition, the design of mechanisms that reduce the inflation rate of functional tokens or even lead to deflation will provide upward momentum for their market prices. The rise in functional token prices drives transaction fees higher.

These three factors collectively influence the fluctuations in transaction fees. If we believe that an ideal DeFi infrastructure should have a stable, predictable recurring cost, then the current fee model does not meet this requirement. This fee model mixes the recurring costs that users need to pay with the value-added costs obtained after participating in a specific blockchain network. In the early stages of blockchain network development, if it is promising, the value addition of its functional tokens has a continuous upward expectation, leading to an increasing transaction fee. If we exclude the fluctuations of value-added factors, GasPriceBase and UtilityTokenPrice in formula (2) can be regarded as constants, then the fees will fluctuate around a constant base, and △GasPrice can be used to represent the magnitude of fluctuations in recurring costs.

Value-Added Costs

As mentioned earlier, the time series fluctuations of the scarcity value of block dimensions can be seen as the fluctuations of GasPrice above GasPriceBase, while the horizontal comparison of the scarcity value of blockchain network dimensions (entry costs) determines the size of GasPriceBase. The relative appreciation of this base reflects a part of the value-added costs of the blockchain network, which is the second factor that raises transaction fees (as shown in Figure 1). The measurement of network scarcity value mainly depends on user scale and TPS. Unfortunately, this value is proportional to user scale, resulting in a反规模效应 of value-added costs, meaning that the larger the user scale, the higher the additional costs brought by scarcity. The fundamental reason is that the fee model, which raises GasPrice to compete for scarcity value, essentially sells the blockchain as an exclusive service commodity rather than a public infrastructure; however, the decentralized characteristics of the blockchain rely on such auction competition mechanisms to incentivize participants to act in favor of the public chain's operation.

Another part of the value-added costs corresponds to the favorable conditions that can be obtained by participating in a specific blockchain ecosystem or, conversely, the potential risks faced, such as higher liquidity and security, which will push up the value-added costs. This part of the cost fluctuates according to the development of the blockchain network ecosystem and is also related to the issuance and circulation mechanisms of functional tokens, ultimately reflected in the price of functional tokens.

Logic Behind the Rise in Ethereum Transaction Fees

The transaction fees on the Ethereum network have experienced several significant fluctuations over the past five years, spanning five orders of magnitude, with the fee for a single simple transaction (consuming 21KGas) exceeding $10, as shown in Figure 2.

Next, we will attempt to provide the logic behind the rise in fees based on the aforementioned fee growth causation model. From Figure 4, we can clearly identify four intervals:

1. Before mid-March 2016, GasPriceBase was approximately 50 GWei;
2. From mid-March 2016 to mid-October 2017, GasPriceBase was approximately 22 GWei;
3. From mid-October 2017 to the end of April 2020, GasPrice showed more frequent daily fluctuations, with GasPriceBase around 11 GWei;
4. From early May 2020 to mid-March 2021, GasPrice showed a clear upward trend.

Figure 2. Time Series of Factors Dependence in Ethereum Fee Model and Relationship Between Simple Transaction Fees and Ether Price

This indicates that starting from early May 2020, GasPriceBase has a rising relationship over time. This can also be reflected in the logarithmic regression of fees and Ether prices at the bottom of Figure 4. Assuming that GasPrice fluctuates around a constant in the fee model expressed in formula (1), then ignoring fluctuation noise, fees only have a linear relationship with Ether prices. The first three stages confirm this: fees and Ether prices show a linear distribution on a logarithmic scale, with a slope of 1±0.1, indicating that the fluctuations in fees are mainly determined by Ether prices, while GasPrice in formula (1) fluctuates around a constant (GasPriceBase) in each stage. However, in the fourth stage, the linear relationship between fees and Ether prices on a logarithmic scale has a slope of about 1.9, indicating that GasPriceBase in formula (4) can no longer be regarded as a constant.

Next, we can make some interesting deductions. If we boldly assume that the price of Ether is mainly represented by the square of the network scale n, before May 2020, the fees required to pay for each unit of Gas on the Ethereum network can be approximately expressed as:

After May 2020, the effect of GasPriceBase rising due to the expansion of network scale exceeding the carrying limit begins to manifest in the aforementioned fee causation model:

Based on the above very rough estimates and deductions, we can roughly estimate that k is approximately 1.8. The network utilization chart shown in Figure 3 indicates that since early May 2020, the Ethereum blockchain has entered a state of full load for more than 10 months, which just confirms the assumption in the previous deduction that the network scale has exceeded the network's carrying capacity in the fourth stage.

Figure 3. Time Series of Ethereum Network Utilization

Methods to Reduce Fee Fluctuations

In theory, separating value-added costs from the recurring cost calculation model can avoid significant fluctuations in transaction fees, but there are many considerations in reality. The reasons may include two points:

1. Currently, functional tokens still need to rely on network development to capture value, and the cost users are willing to pay for transactions can indirectly reflect the urgency of using that blockchain, thus measuring the value of the blockchain network to users. Therefore, linking the fee model with functional tokens is the most straightforward way to achieve market pricing for functional tokens and avoid human intervention. After all, it is currently difficult to measure value-added costs using a fixed model.
2. If the part of functional tokens attributed to network value addition is separated, it may raise regulatory issues, posing a risk of being classified as a security token.

A potential consequence of reason 1 is that since both GasPriceBase and functional token prices in formula (4) represent the network's value-added costs and are not completely orthogonally decomposed, and the representation method is purely achieved through market adjustment mechanisms, there may be repeated or excessive measurement of the scarcity value of the network dimension under the inertia of the market.

EIP-1559

To address the issue of significant fluctuations in fees, the Ethereum community is planning to adopt EIP-1559 in the London upgrade tentatively scheduled for July. EIP-1559 attempts to stabilize fee fluctuations by establishing a dynamic adjustment mechanism for GasPriceBase (as block utilization increases, GasPriceBase increases, and vice versa).

However, in fact, under the current circumstances where user scale exceeds the carrying limit of the blockchain network (supply does not meet demand), a foreseeable outcome is that the rising GasPriceBase with increasing block utilization may not necessarily suppress traders from reducing transaction demand but may continue to fill the expanded blocks with higher fees (GasLimit increased from 12.5 million to 25 million), further raising the entry cost of the Ethereum network.

At the same time, the destruction mechanism introduced by EIP-1559 (all GasPriceBase will be destroyed, and miners only receive user-defined tips) reduces the inflation rate of Ether, which will further strengthen the proportion of value-added costs in each transaction fee.

Conclusion and Reflection

Through the above analysis, we can draw the following conclusions:

1. Ethereum transaction fees can be decomposed into the costs of block accounting services, network entry costs, and the value-added costs of functional tokens, attributed to the scarcity value of block dimensions, the scarcity value of network dimensions, and ecological value addition. Among them, the network entry cost behaves as a stage constant before the user scale reaches the carrying limit, and after the user scale exceeds the carrying limit, it will rise with the expansion of the user scale.

2. Since May 2020, the Ethereum network has entered a state of full load, and the network entry cost shows a unilateral upward trend, indicating that the block accounting services provided by the Ethereum network are actually in short supply, and the current fees are still within the acceptable range for mainstream users.

3. In the absence of separating value-added factors, solving the problem of significant fee increases still returns to enhancing network TPS, which requires increasing TPS to accommodate a larger user scale and reduce network entry costs.

4. Although BSC and Heco chains have not experienced congestion or excessively high fees in the short term, this is mainly due to their small user scale. If the user scale expands, they will still face the same issues as Ethereum, so migrating to these two alternative chains is not a long-term solution.

Stimulated by DeFi applications, the Ethereum network has established a value threshold built on years of proven decentralization and security. However, as an ideal DeFi ecosystem infrastructure, its usage costs should exhibit economies of scale, while the current Ethereum public chain resembles a luxury item from the late 1980s and early 1990s for ordinary users. The Ethereum 2.0 plan aims to introduce PoS consensus to increase TPS, which will further expand the network scale it can accommodate, but there will still be an upper limit. Another idea is to migrate DeFi and other businesses to layer two networks, using the Ethereum public chain as a dedicated settlement chain for large transactions, primarily targeting Dapp business parties and high-net-worth individuals, thereby reducing the fee rate by increasing the transaction amount per transaction, and Dapp business parties can then spread this cost to users on the layer two network. The ultimate solution may require revolutionary upgrades to computing power, storage, bandwidth, and other resources through new energy and new technologies.