Crisis regimes, tariff uncertainty, and cross-asset ETF efficiency: A rolling multifractal analysis

This paper examines the time-varying multifractal efficiency of nine U.S.-listed exchange-traded funds (ETFs) representing equities, bonds, real estate, commodities, gold, crude oil, and the U.S. dollar. To this end, multifractal detrended fluctuation analysis (MFDFA) is applied to daily ETF logarithmic returns from March 2012 to February 2026, and full-sample estimates are complemented by rolling 500-trading-day windows advanced by 21 observations. The analysis distinguishes the pre-COVID period, COVID-19, the Russia-Ukraine war, the Israel-Hamas conflict, and the U.S. tariff period associated with the second Trump administration. Efficiency is measured as the absolute deviation of the singularity-spectrum peak from the random benchmark 0.5. The full-sample ranking identifies the U.S. dollar, gold, U.S. government bonds, and commodities as the closest ETF exposures to the benchmark, whereas investment-grade bonds, real estate, high-yield bonds, and crude oil are more distant. Rolling estimates show that average inefficiency rises during COVID-19 and Russia-Ukraine, while the shorter Israel-Hamas endpoint window records lower average inefficiency in this ETF universe. Two-way clustered OLS regression indicates that mean event effects are not statistically significant after conservative dependence corrections, whereas bootstrap quantile regressions show upper-tail amplification: COVID-19, Russia-Ukraine, and U.S. tariffs are most visible when ETF markets are already in high-inefficiency states. The findings support an adaptive view of ETF efficiency in which crisis and political-policy regimes affect the distributional tail of multifractal inefficiency more strongly than its conditional mean.