Directional Decoupling and the Volatility-Ratio Channel: A Beta-Decomposition Framework for Exchange-Native Tokens — Evidence from the BNB-ETH Relationship, 2022-2026

Abstract

Standard factor models in cryptocurrency asset pricing [Liu & Tsyvinski, 2021; Liu et al., 2024] treat beta as a sufficient statistic for systematic exposure. The identity β = ρ · (σᵢ / σₘ) implies, however, that beta can change for two distinct reasons: a change in directional co-movement (the correlation channel) or a change in relative volatility (the volatility-ratio channel). The economic interpretations differ sharply, but the existing literature does not systematically separate them.

This paper proposes a variance-decomposition framework that isolates these two channels and applies it to a striking empirical pattern in cryptocurrency markets that we term directional decoupling: a decline in beta accompanied by an unchanged correlation. Using daily price data from Coin Metrics covering May 2022 to April 2026 (n = 1,460 trading days), we document that BNB’s static beta to ETH declines by 16.9% (from 0.643 to 0.534), while the Pearson correlation remains essentially unchanged (0.731 → 0.731).

The decomposition reveals that 100% of the static beta change is attributable to a decline in the volatility ratio σ_BNB / σ_ETH (from 0.879 to 0.730), with no contribution from correlation change. A further numerator/denominator decomposition shows that BNB-side volatility compression accounts for approximately 57% of the ratio decline, while ETH-side volatility expansion accounts for approximately 43%, indicating that the apparent BNB pattern is jointly produced by BNB-specific and ETH-side dynamics. Robert F. Engle’s DCC-GARCH framework [Engle, 2002] confirms the qualitative pattern under dynamic conditional moments: dynamic correlation rises (0.698 → 0.747, p < 0.001), while dynamic beta declines (0.575 → 0.541, −5.84%).

Cross-sectional analysis of 16 altcoins—including exchange tokens, Layer-1 platforms, DeFi/oracle tokens, and other major altcoins—reveals substantial heterogeneity in beta dynamics across and within categories, with BNB occupying the moderate-to-strong decoupling segment but not an extreme position (BNB rank 6/17 in the cross-sectional pool, t = −2.22, p = 0.043 against the control mean; synthetic-control rank 12/17, p ≈ 0.71). The cross-sectional evidence supports interpreting BNB’s pattern as part of a broader altcoin beta recompression coinciding with ETH’s post-2024 volatility expansion, rather than as an asset-specific regime change.

We discuss implications for cryptocurrency factor models, arguing that beta-based models should be augmented with explicit decomposition into correlation and volatility-ratio components when interpreting changes in systematic exposure. We also identify on-chain stablecoin flow data and staggered-treatment designs around individual exchange holder-reward launches as priority directions for follow-up causal-mechanism studies. The April 2024 launch of the Binance HODLer Airdrops program serves as a natural timing benchmark for partitioning the sample but is not the basis for any causal claim in this paper.

Keywords: cryptocurrency, beta decomposition, directional decoupling, volatility-ratio channel, factor models, asymmetric volatility, BNB, exchange-native tokens

JEL Classification: G12, G14, G15, C58