Bubble Characteristics and Tail Risk in Global REIT Markets: Duration, Volatility, and the Anatomy of Severe Drawdowns
摘要
This paper examines whether BSADF-identified explosive regimes in global REIT markets predict subsequent tail risk, and whether economically interpretable upward explosive episodes are followed by more severe drawdowns than comparable non-explosive market peaks. Using daily price data for fourteen FTSE REIT indices between 2011 and 2026, we identify explosive regimes with the Backward Supremum Augmented Dickey-Fuller (BSADF) methodology and explicitly distinguish upward from downward explosive episodes. This distinction is important because BSADF detects statistical explosiveness relative to a unit-root benchmark and does not, by itself, establish speculative overvaluation. Post-regime outcomes are evaluated using maximum drawdowns, recovery times, a nearest-neighbor matched control sample of normal peaks based on pre-event return, volatility, and trend conditions, and daily forward-looking tail-risk measures with block-bootstrap inference. The results show substantial heterogeneity across markets and regime types. Upward explosive regimes are not followed by systematically larger drawdowns than matched normal peaks: in the matched sample, mean subsequent drawdown is 12.05% after upward regimes and 13.50% after matched normal peaks, with no statistically significant paired difference. Downward explosive regimes display a different profile, consistent with the interpretation that part of the adjustment occurs during, rather than after, the BSADF episode. Daily-level evidence is also sensitive to dependence and index composition; after accounting for overlapping forward windows and excluding aggregate indices, broad BSADF regime indicators do not provide robust evidence of universal tail-risk amplification. Overall, the findings indicate that BSADF-identified regimes in REIT markets capture heterogeneous price dynamics rather than a single speculative mechanism with predictable crash consequences. Effective tail-risk assessment therefore requires conditioning on regime direction, market structure, and local pre-event conditions rather than relying on pooled bubble signals alone.