<![CDATA[The prevailing assumption that ASEAN equity markets constitute a homogeneous asset class has profound consequences for regional portfolio construction, hedging strategies, and macroprudential coordination. This study challenges that assumption by identifying three structurally distinct risk-transmission regimes across the ASEAN-4 (Indonesia, Malaysia, Singapore, Thailand) over the 2016–2025 period. Employing a heterogeneous multi-task machine-learning framework augmented with Shapley-value decomposition on daily panel data encompassing global volatility (VIX), crude oil (Brent), crude palm oil (CPO), the US dollar index (DXY), US 10-year Treasury yields, local equity indices, bilateral exchange rates, and the BI-7DRR policy rate, we decompose each market's sensitivity profile into interpretable economic channels. Our analysis reveals a novel empirically grounded taxonomy—the Tripartite Spectrum of market integration: (i) Singapore operates as a Monetary-Driven Hub, with 64.0% of total Shapley attribution concentrated in the DXY and US 10-year yield, consistent with the global financial cycle hypothesis; (ii) Thailand functions as a Geopolitical Risk Transmission Channel, exhibiting disproportionate sensitivity to the VIX with a non-linear threshold effect that amplifies volatility by 340% when the VIX exceeds 35; and (iii) Indonesia and Malaysia constitute Commodity-Linked Segmented Markets, where CPO, Brent crude, and domestic monetary-policy frictions dominate price formation, with a superadditive CPO × BI-7DRR interaction effect (+280% amplification) unique to Indonesia. Kolmogorov-Smirnov tests decisively reject the homogeneous-region null (all cross-country p-values < 0.01). These findings carry direct implications for the limits of intra-ASEAN diversification, the design of regime-specific hedging strategies, and the calibration of commodity-price stress scenarios in macroprudential frameworks.]]>
