Ethereum and Arbitrum gas demand near-inelastic, new study finds

A new econometric paper from Pranay Anchuri and Akaki Mamageishvili tackles a fundamental question in blockchain economics: how much does gas demand actually respond to fee changes? Using a two-way fixed effects panel regression instrumented by each wallet's own lagged base fee, the authors remove the congestion-driven endogeneity that causes naive regressions to underestimate demand sensitivity. On Ethereum mainnet (L1) for the full year 2025, the causal price elasticity of gas demand is a mere -0.006, meaning a 10% fee increase reduces total gas demand by only 0.06%. On Arbitrum One (L2), the pooled elasticity is -0.036—still inelastic, but six times more responsive than L1. These aggregate numbers mask significant heterogeneity: a per-resource decomposition on L2 reveals that refunds (-0.27) and storage growth (-0.15) are the most elastic categories, while computation (-0.027) remains highly inelastic.

Behavioral clustering further illuminates the results. Always-on protocol wallets (e.g., automated bots, smart contract infrastructure) are effectively unresponsive to fee changes, while high-volume operators show cluster-level elasticities up to roughly 6x the pooled estimate. For L1, the near-zero elasticity confirms that Ethereum's user base faces little price-driven demand reduction at current usage levels, a finding with direct implications for EIP-1559 tuning and future fee market reforms. For L2, the higher responsiveness suggests that rollup users are more cost-sensitive and could shift activity in response to L2 fee structures. The study provides essential empirical calibration for fee simulation models and resource pricing debates on both layers.