DOI: 10.12688/f1000research.179434.3 ISSN: 2046-1402

The Concentration-Fragility Nexus: Record Concentration, Elusive Fragility in Three Decades of S&P 500 Evidence

Jorge A. Restrepo Morales, Rosa Ysabel Moreno Rodriguez, Freddy Zea Restrepo, Emerson Andrés Giraldo Betancur
Show PDF Cite
Background Equity-market concentration has reached historic levels, prompting warnings from regulators and asset managers that a market dominated by a handful of mega-capitalization stocks is structurally fragile. Whether concentration actually predicts market fragility, however, has rarely been tested over a sample long enough to contain multiple concentration cycles. Methods We hand-construct a thirty-year, semiannual series of S&P 500 portfolio concentration (48 observations, December 1995 – March 2026) from the complete portfolio schedules in SEC N-30D and NPORT-P filings of the SPDR S&P 500 Trust, cross-validated against independent NPORT-P parses to a mean absolute difference of 0.2%. We then run three independent tests: (i) Newey-West predictive regressions of four next-semester fragility measures (realized volatility, maximum drawdown, empirical VaR 1%, crash-day frequency) on concentration; (ii) a panel of 47 equity index funds (1,260 fund-quarters, 2019Q3–2026Q1) with fund and time fixed effects; and (iii) a walk-forward, out-of-sample evaluation of a non-circular composite financial-stress index against the VIX for crash prediction at 5- and 21-day horizons. Results Concentration is at an unprecedented level: the S&P 500 Herfindahl-Hirschman Index reached 0.0222 in September 2025, 2.1 times the dot-com peak of 0.0107, and the top-10 weight reached 38.9% versus 24.3% in 2000. Yet none of the three tests links concentration to subsequent fragility. Across all long-run specifications the HHI, CR10 and ΔHHI coefficients are statistically indistinguishable from zero (p = 0.28’0.95); the panel coefficient with two-way fixed effects is 0.27 (wild-cluster bootstrap p = 0.89); and the composite stress index underperforms the VIX alone out-of-sample (AUC 0.740 vs. 0.781 at t+5; 0.665 vs. 0.712 at t+21). Conclusions Portfolio concentration at the index level, however extreme by historical standards, is not by itself an early-warning indicator of market fragility. Volatility-based indicators dominate any concentration signal. Fragility channels associated with concentration, if present, do not operate at the aggregate index level over horizons of days to semesters — a well-identified null result with direct implications for concentration-based macroprudential proposals.

More from our Archive