Financial diversification between cryptocurrencies and equities depends on which dependence measure you use because different statistics capture different facets of co-movement. Simple linear co-movement is summarized by Pearson correlation, but that measure assumes linearity and normal-like distributions and therefore can mislead when returns are skewed or contain extreme events. In crypto markets, heavy tails and asymmetry are common, so linear summaries understate joint extreme risk.
Time-varying dependence
For dynamic portfolios, DCC-GARCH models developed by Robert F. Engle at New York University explicitly model time-varying correlations, letting risk managers see how relationships evolve through volatility regimes. This matters because crypto–equity correlations can rise sharply during market stress, eroding diversification exactly when it’s most needed. Using a static correlation matrix can therefore overstate protection.
Nonlinear and tail dependence
Rank-based measures such as Spearman and Kendall tau capture monotonic but nonlinear dependence and are more robust to outliers than Pearson. For extreme co-movements, however, copulas and tail-dependence coefficients are superior. Work by Paul Embrechts at ETH Zurich on copulas and extreme-value theory shows these tools quantify the probability that assets crash together, a critical input for stress testing. Mutual information from information theory, rooted in Claude Shannon at Bell Labs, captures general nonlinear dependence without assuming monotonicity, useful when relationships are complex and driven by structural market differences.
Empirical evidence matters. Early empirical analysis by Yukun Liu and Aleh Tsyvinski at Yale University found cryptocurrencies had low correlations with traditional assets, suggesting diversification benefits. Subsequent studies using time-varying and tail-focused methods, however, documented higher correlations in crisis periods, implying that diversification can be fragile.
Relevance, causes, and consequences converge: measurement choice affects portfolio construction, regulatory capital, and hedging. Causes for shifting dependence include synchronized risk-off sentiment, regulatory announcements, and technical market microstructure like 24/7 trading and concentrated mining operations that link crypto to energy markets. Territorial and cultural factors—differences in national regulation, varying retail adoption across regions, and localized liquidity pools—alter observed dependence and can produce misleading cross-asset inferences if ignored. For robust diversification assessment, combine DCC-type models for time variation with copula/tail-dependence analysis and supplement with nonparametric measures like mutual information or rank correlations to capture a fuller picture of crypto–equity relationships. No single metric suffices; triangulation is essential.