How the magnificent seven reshape market spillovers: TVP-VAR evidence

초록

This study examines the dynamic connectedness of the Magnificent Seven (M7) stocks. Using a time-varying parameter vector autoregression (TVP-VAR) framework with frequency decomposition, we analyze Apple, Microsoft, Alphabet, Amazon, Nvidia, Meta, and Tesla. With 1507 daily observations from January 2020 to December 2025, we document four main findings. First, in raw returns, total connectedness declines from 75.56% to 69.00% between the pre-and post-November 30, 2022 periods. Second, M7 stocks shift toward a stronger net receiver role, with average net spillover moving from-2.31 to-4.53 (t(HAC) = -14.48, p < 0.001). Third, the connectedness decline is concentrated in short-term (1-5 day) frequency dynamics, while longer-horizon (6-20 day) connectedness remains comparatively stable. Fourth, group-average net pairwise directional connectedness from M7 to benchmark proxies (SPY, IWM) increases from 1.16 to 2.26. The raw-return pattern survives an alternative benchmark specification using RSP and IWM, and a market-adjusted specification, although richer factor adjustments attenuate the directional effect. Expanded robustness now also considers volatility connectedness via squared daily log returns: it reinforces the system-wide TCI decline but qualifies the directional interpretation because volatility-based M7 NET and NPDC remain broadly flat. We therefore interpret November 30, 2022 as an economically salient marker for a broader regime shift rather than as a standalone causal breakpoint.

키워드