The Regime Switch of Capital Flows: How the China-U.S. Yield Spread Drives Global Capital Reallocation in a Tiered Manner— Exhaustive Evidence Based on TIC Panel Data, ETF Flows, and China's Cross-Border Data

The international capital flow literature has long treated the interest rate differential as a linear, stable factor. This paper proposes instead the Regime Switch Theory of Capital Flows: the China-U.S. yield spread is a regime variable determining which side's logic dominates. When positive, U.S.-side factors (VIX, DXY) dominate; when negative, China-side factors (internal yield spread, RMB exchange rate) gain independent pricing power. Institutional interfaces (Stock Connect, Bond Connect) are the activation conditions. From the fact that central banks independently set interest rates under fiat money, we deduce the global capital flow equation must have a Unified Field structure, yielding four testable predictions. Using the TIC 68-country panel (1978–2023), Northbound/Southbound flows, ETF flows, and China's four asset categories, twelve methods—exhaustive testing under VIF<5, interaction effects, Chow tests, threshold regressions—systematically exclude old-theory alternatives. Principal findings: (1) China factors systematically enter optimal pricing combinations—private Treasury flows are spread-dominated; Northbound and Southbound optimal combinations are nearly identical, proving the spread prices bidirectional flows. (2) The spread prices indirectly by modulating market factor weights—spread×VIX and spread×DXY interactions are significant (p=0.032, p=0.0001), while the direct effect is entirely insignificant after Newey-West correction (p=0.969); interaction incremental R² is 1.78 times the direct effect. This pattern replicates across six disaggregated flow categories and six volatility measures, proving cross-asset, cross-market universality. (3) The flow equation breaks structurally near spread≈0—Northbound Chow F=33.15 (p=0.000), all six disaggregated Chow tests significant, four core thresholds near spread≈0 (Bootstrap p<0.05). Foreign bond holdings are the exception (Chow F=0.22), establishing the boundary: bond flows react far more slowly than equity flows. (4) Official and private capital respond oppositely—official embodies strategic de-dollarization, private embodies tactical risk aversion; China is the only top-10 TIC holder with a negative spread coefficient (-153.55). (5) The MCHI/SPY ratio is not spread-driven (p=0.875), a negative verification of the Institutional Interface Theory complementing the DID activation verification. (6) Reserve currency adjustments show a two-year lag (8-quarter coefficient on USD share=24.96, p=0.035). Tang (2026e) found that in the price dimension, regime factors exhibit "direct effect≈0, interaction effect significant." This paper replicates this across aggregate flows, disaggregated flows, and volatility—proving it an inevitable cross-dimensional deduction from the Unified Field Formula. Theoretically, this paper reduces the Mundell-Fleming model, the Triffin Dilemma, and traditional capital flow literature into boundary-specific special cases. With Tang (2026a–2026e), it completes the Dual-Anchor Regime theory from price to quantity. Research Paradigm Statement: The core methodology, research direction, and final decisions were independently directed by the author. DeepSeek assisted with code implementation, data presentation, and text drafting. The author takes full academic responsibility for the final content.