Empirical claim #1 — US chip-control event study
Generated 2026-05-23 01:57 UTC by `scripts/py/intelligence/event_study_chip_controls.py`. Re-run anytime; pure-Python, no network.
Pre-registered cohort
All IPTM actions with issuer_country = US, action_type = export-control, severity >= 4, target_countries containing CN, announced_date in [2018-01-01, 2026-12-31], with a full T+60 trading-day window available in the ETF dataset.
Matched: 30 actions in window; 0 excluded for insufficient post-event ETF data (announced after 2026-02-18). Final n = 30.
The claim
US BIS export-control actions targeting China at severity 4+ are historically followed, over a 60-trading-day window, by:
- MCHI (China, the target) trailing the ACWI global benchmark; AND
- EWY (Korea) / EWT (Taiwan) / EWN (Netherlands) — the home jurisdictions of the supplier majors — outperforming the same benchmark.
Effect sizes, sample, and p-values follow. This is a historical association, not a forecast.
Results — abnormal returns vs. ACWI
Median log-return on each ETF minus log-return on ACWI over the same window. Two-sided sign test against H0: median = 0 (binomial exact). Wilcoxon signed-rank p is the more powerful rank-based test (normal approx; reported when n >= 6 non-zero observations).
Horizon: T+5d
| ETF | n | median (%) | mean (%) | IQR (%) | % neg | sign-test p | Wilcoxon p |
|---|---|---|---|---|---|---|---|
| MCHI | 30 | +0.08 | -0.04 | 3.20 | 47% | 0.856 | 0.877 |
| EWY | 30 | +0.94 | +0.97 | 2.46 | 27% | 0.016 | 0.026 |
| EWT | 30 | +0.70 | +0.39 | 2.68 | 37% | 0.200 | 0.262 |
| EWN | 30 | -0.06 | -0.04 | 1.81 | 53% | 0.856 | 0.750 |
| SPY | 30 | -0.03 | -0.02 | 0.57 | 53% | 0.856 | 0.894 |
Horizon: T+20d
| ETF | n | median (%) | mean (%) | IQR (%) | % neg | sign-test p | Wilcoxon p |
|---|---|---|---|---|---|---|---|
| MCHI | 30 | -0.73 | +1.25 | 12.14 | 57% | 0.585 | 0.704 |
| EWY | 30 | +1.80 | +2.79 | 6.15 | 37% | 0.200 | 0.039 |
| EWT | 30 | +0.76 | +0.97 | 3.77 | 33% | 0.099 | 0.153 |
| EWN | 30 | +1.55 | +1.63 | 2.71 | 23% | 0.005 | 0.001 |
| SPY | 30 | -0.18 | -0.33 | 0.82 | 57% | 0.585 | 0.037 |
Horizon: T+60d
| ETF | n | median (%) | mean (%) | IQR (%) | % neg | sign-test p | Wilcoxon p |
|---|---|---|---|---|---|---|---|
| MCHI | 30 | -3.03 | -1.57 | 17.41 | 67% | 0.099 | 0.417 |
| EWY | 30 | +1.85 | +3.42 | 9.54 | 30% | 0.043 | 0.075 |
| EWT | 30 | +1.18 | +1.52 | 7.13 | 40% | 0.362 | 0.178 |
| EWN | 30 | +2.93 | +2.07 | 5.81 | 27% | 0.016 | 0.049 |
| SPY | 30 | +0.31 | -0.12 | 2.74 | 43% | 0.585 | 0.861 |
Results — raw log-returns (for reference)
| Horizon | MCHI | EWY | EWT | EWN | SPY | ACWI |
|---|---|---|---|---|---|---|
| T+5d | -0.49% | +0.71% | +1.42% | +0.31% | +0.40% | +0.61% |
| T+20d | +1.10% | +3.38% | +4.28% | +3.47% | +2.63% | +2.91% |
| T+60d | +2.73% | +6.59% | +6.27% | +5.46% | +5.37% | +4.84% |
Events used
| # | Date | Severity | Action |
|---|---|---|---|
| 1 | 2020-08-17 | 5 | 2020-08-17-us-bis-huawei-affiliates-entity-list-fdpr-tgl-removal |
| 2 | 2020-12-18 | 4 | 2020-12-22-us-bis-entity-list-smic-77-entities |
| 3 | 2020-12-23 | 4 | 2020-12-23-us-bis-ear-hong-kong-removal |
| 4 | 2020-12-23 | 4 | 2020-12-23-us-bis-ear-meu-list-creation |
| 5 | 2021-01-15 | 4 | 2021-01-15-us-bis-military-intelligence-end-use-controls |
| 6 | 2021-04-09 | 4 | 2021-04-08-us-bis-entity-list-7-china-supercomputing |
| 7 | 2021-11-26 | 4 | 2021-11-26-us-bis-entity-list-27-quantum-pakistan-meu |
| 8 | 2022-08-24 | 4 | 2022-08-24-us-bis-entity-list-7-china-military-aerospace-electronics |
| 9 | 2022-10-07 | 5 | 2022-10-07-us-bis-advanced-ai-chip-controls-china |
| 10 | 2022-10-07 | 4 | 2022-10-07-us-bis-uvl-31-additions-china-ymtc |
| 11 | 2022-12-19 | 4 | 2022-12-19-us-bis-entity-list-36-additions-china-japan |
| 12 | 2023-03-06 | 4 | 2023-03-06-us-bis-entity-list-38-china-burma-pakistan-belarus |
| 13 | 2023-10-11 | 4 | 2023-10-11-us-bis-entity-list-49-entities-russia-diversion |
| 14 | 2023-10-17 | 5 | 2023-10-17-us-bis-advanced-chip-controls-expansion |
| 15 | 2023-10-25 | 4 | 2023-10-25-us-bis-advanced-computing-updates-corrections |
| 16 | 2024-02-27 | 4 | 2024-02-23-us-bis-entity-list-93-additions-russia-china-turkey |
| 17 | 2024-05-09 | 4 | 2024-05-14-us-bis-entity-list-37-additions-china-quantum-russia-balloon |
| 18 | 2024-08-27 | 4 | 2024-08-27-us-bis-entity-list-123-additions-russia-china-iran-diversion |
| 19 | 2024-09-05 | 4 | 2024-09-05-us-bis-quantum-biotech-additive-manufacturing-controls |
| 20 | 2024-12-02 | 5 | 2024-12-02-us-bis-hbm-sme-entity-list-package |
| 21 | 2025-01-06 | 4 | 2025-01-06-us-bis-entity-list-13-additions-china-burma-pakistan |
| 22 | 2025-01-13 | 5 | 2025-01-13-us-bis-ai-diffusion-framework |
| 23 | 2025-01-16 | 4 | 2025-01-16-us-bis-advanced-ic-due-diligence-ifr |
| 24 | 2025-03-28 | 4 | 2025-03-28-us-bis-entity-list-12-additions-supercomputers-china-taiwan |
| 25 | 2025-03-28 | 4 | 2025-03-28-us-bis-entity-list-70-additions-china-iran-pakistan-south-africa |
| 26 | 2025-05-13 | 4 | 2025-05-13-us-bis-ai-chip-guidance-package-huawei-gp10 |
| 27 | 2025-09-02 | 4 | 2025-09-02-us-bis-veu-revocation-china-fabs |
| 28 | 2025-09-16 | 4 | 2025-09-16-us-bis-entity-list-32-additions-china-russia-iran |
| 29 | 2025-09-30 | 4 | 2025-09-30-us-bis-affiliates-rule-entity-list-50-percent |
| 30 | 2026-01-13 | 4 | 2026-01-13-us-bis-advanced-computing-license-review-revision |
Honest caveats
1. Small sample (n=30). Effect-size estimates have wide intervals; sub-cohort analysis (e.g. severity 5 only) would be under-powered. The reported medians are point estimates, not confidence-bounded.
2. Overlapping event windows. Many events cluster within 60 trading days of each other (e.g. the 2022-10-07 and 2024-12-02 packages each spawn follow-up rules within a quarter). Observations are NOT independent, so the reported sign-test and Wilcoxon p-values are anti-conservative — they understate the true probability under the null.
3. No factor-return adjustment. Abnormal returns are computed against ACWI only, not against a multi-factor model (FF3 / FF5 / momentum). Some of the EWN / EWY / EWT outperformance over 2020-2026 is sector + size factor exposure that would attenuate under a richer benchmark.
4. Cohort-period confounding. The 2020-2026 window contains COVID, the 2022 inflation cycle, US elections, the AI capex boom, and the Russia war — all of which independently moved the test ETFs. The event-study claim is conditional on these joint regimes.
5. No out-of-sample validation. Future BIS packages may exhibit different post-event dynamics if (a) the architecture saturates, (b) Chinese onshoring closes the supplier-country opportunity, or (c) the regime shifts to demand-side (compute-export) controls where the supplier-country lever weakens.
6. Selection of the cohort is pre-registered — issuer = US, action_type = export-control, severity >= 4, target includes CN, date window pre-specified — but the ETF list (MCHI/EWY/EWT/EWN/SPY) is post-hoc: chosen because the case-study work identified these as the regime's economic perimeter. Replication on a randomly drawn ETF list would not show the same pattern by construction.