PREreview of Demand Under a Mechanical Price: Elasticity and Welfare in Ethereum's Data-Availability Market

1. Summary of the Work

This manuscript offers a valuable and timely empirical analysis of the data-availability (blob) fee mechanism introduced in Ethereum by EIP-4844. By examining the Pectra hard fork (7 May 2025) as an exogenous supply shock that doubled capacity and brought base fees down to 1 wei, the author estimates the price response of major rollups using Poisson pseudo-maximum likelihood (PPML). The paper provides a thoughtful decomposition of fiscal transfers versus deadweight loss, supported by an extensive specification curve and mempool transaction records from Xatu.

2. Key Merits & Strengths

• Empirical Transparency: The author deserves high praise for open methodology, including public pre-registration records (commit 811ab9b), transparent discussion of pre-analysis plan adjustments, and reporting across 450 alternative model specifications.
• Appropriate Single-Event Inference: Using 516 Conley-Taber placebo break dates and Kline-Santos wild score bootstrap methods is a sound and rigorous approach to handling single-event date uncertainty and unequal cluster sizes.
• Granular Mempool Evidence: The inclusion of 5.4 million Xatu mempool transaction records provides clear empirical evidence that the 1-wei floor reflects genuine operational slack rather than unobserved queue rationing.
• Clear Exposition: The institutional details of Ethereum’s blob pricing rules and hard-fork milestones are presented clearly and accessibly.

3. Constructive Suggestions & Questions for the Author

1. Batch Packing Density vs. Raw Data Demand: As noted in Section 12, the analysis observes capacity purchased in discrete 128 KiB units rather than raw payload bytes. When fees dropped to 1 wei, rollups had no financial penalty for posting partially filled blobs in order to provide faster confirmation times to their users. It would be helpful to clarify whether rollups posted more frequently with fewer transactions per blob post-Pectra. If possible, discussing whether transaction packing density changed would help confirm how much of the 24% increase represents faster batching frequency versus an increase in underlying user payload demand.
2. First-Stage Sampling Uncertainty: The elasticity bound (-0.31) is calculated by dividing the upper bound of the post-fork coefficient (0.337) by the estimated first-stage price drop (π = -0.90 mETH). Because the first stage is an estimated parameter (SE = 0.33), incorporating its standard error into the bound (for example, using a delta-method or Fieller-type ratio interval) would provide an even more robust confidence set for the elasticity estimate.
3. Subgroup Architectural Differences: The primary sample pools Optimistic rollups (where calldata scales directly with user execution) and ZK-rollups (where state diffs scale with unique storage slot modifications). A brief comment or breakdown on whether Optimistic and ZK-rollups showed different responsiveness to the price drop would add valuable insight for protocol researchers.

4. Minor Corrections & Polish

• Fusaka Duration Clarification: In Table 4, Fusaka is listed as 7 days, whereas Table 1 lists 40,084 blocks (about 5.6 days), and the text on page 9 mentions 5 days. Harmonizing these definitions (calendar days vs. block intervals) would improve consistency.
• Total Calendar Days: In Figure 1, the sample is reported as N = 930 UTC days, while the sum of days across periods in Table 4 is 934 days. A brief note clarifying boundary day definitions would resolve this minor discrepancy.
• Appendix Cross-Reference: Section 3 concludes with a reference to "update rule (Appendix A)", but no appendix is included in the preprint. Citing the external repository or specification document directly would help readers locate the update code.
• Table 2, Column 3 Note: In Column 3 (intensity margin), a brief footnote explaining that the upper bound of +0.110 corresponds to the elastic bound of -0.10 due to the offset in the denominator would help readers interpret the sign.

5. Conclusion

This is a well-crafted, rigorous, and highly informative paper that significantly advances our empirical understanding of blockchain data-availability pricing. The author’s transparency and careful econometric execution are commendable. Addressing the points above on batch packing density and first-stage variance will further strengthen this excellent contribution.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.