Pre-registration
China's dynamic-zero-COVID policy 2020-2022, peaking with the Shanghai April-2022 lockdown and Beijing/Wuhan late-2022 closures, produced sharper-than-counterfactual demand contraction — retail-sales YoY turning negative, manufacturing PMI sub-50, and youth unemployment spiking — and the December-2022 abrupt reopening produced an incomplete and short-lived recovery rather than the V-shaped rebound expected from pent-up demand and excess household savings.
Falsification criterion — what would disprove this
This hypothesis is considered falsified if:
Multi-metric checklist on China 2022Q1-2023Q4. SUPPORTED if >=4 of 6 metrics meet thresholds; REFUTED if <=2 do. Each metric independently sourced (NBS retail-sales, Caixin/NBS PMI, NBS labour-force survey, customs trade data, OWID stringency) and probes a different layer of the lockdown demand-shock and reopening response.
formal test & threshold
test: china_zero_covid_demand_collapse_2022_2023_multi_metric threshold: >=4/6 supports; <=2/6 refutes
Method
- Template
multi_metric_checklist- Clustering
none- Sample
- 1 countries · 2018 – 2024
- Evidence type
- canonical_case_multi_metric
Single-country canonical case (China zero-COVID 2020-2022 and abrupt 2022-12 reversal). Pattern-match across stringency, retail, PMI, youth unemployment, and post-reopening growth. NBS data-quality concerns addressed by cross-source independence (OWID, ILOSTAT, IMF, WDI).
Data
| Variable | Source | Transform |
|---|---|---|
real_gdp_growth_yoy outcome | imf:NGDP_RPCHtier 2 world_bank_wdi:NY.GDP.MKTP.KD.ZGtier 2 | pct_change_yoy |
retail_sales_growth outcome | world_bank_wdi:NE.CON.PRVT.KD.ZGtier 2 | pct_change_yoy |
manufacturing_pmi outcome | owid:china_manufacturing_pmitier 2 | level |
youth_unemployment_rate outcome | ilostat:UNE_2EAP_SEX_AGE_RT_NBtier 2 | level |
industrial_production outcome | owid:industrial_production_chinatier 2 | pct_change_yoy |
zero_covid_stringency treatment | owid:government_response_stringency_indextier 2 | level |
shanghai_lockdown_indicator treatment | constructed:1 from 2022-03-28 to 2022-06-01tier 5 | indicator |
reopening_indicator treatment | constructed:1 from 2022-12-07 (10-point reopening) onwardtier 5 | indicator |
oil_price_brent control | imf_pcps:POILBREtier 1 | log_diff |
global_manufacturing_pmi control | owid:global_manufacturing_pmitier 2 | level |
● ready · ● pending · ● reconstruct-needed
Detailed result card
Result card — china_zero_covid_2022_2023_demand_collapse_recovery
Verdict: inconclusive (data gaps)
Reason: 0 metrics met, 5 pending; 4 more need resolution
Pre-registered rule: SUPPORT if >= 4 of 6 metrics met; REFUTE if <= 2 met (impossible to hit support).
Counts: 0 MET · 1 NOT_MET · 3 PENDING_DATA · 2 PENDING_EVAL
Primary country: CHN
Metric-by-metric
| # | Metric | Status | Observed | Threshold | Notes |
|---|---|:---:|---:|---|---|
| 1 | shanghai_lockdown_industrial_production_drop_2022q2 | NOT_MET | 0 (2022) [peak_to_trough_pct_decline] | >25% MoM decline in Shanghai industrial production 2022-04 vs 2022-03 | |
| 2 | retail_sales_negative_2022 | PENDING_EVAL | 1.49 (2022) [max_in_window] | retail-sales YoY <=-7% at 2022 trough | threshold expression unparseable by regex |
| 3 | manufacturing_pmi_sub50_persistent | PENDING_DATA | | >=6 months of 2022 with manufacturing PMI <50 | No usable vintage for: owid:china_manufacturing_pmi |
| 4 | youth_unemployment_spike_2023 | PENDING_DATA | | >=20% peak | No usable vintage for: ilostat:UNE_2EAP_SEX_AGE_RT_NB |
| 5 | incomplete_v_recovery_2023 | PENDING_EVAL | 0 (2023) [pct_increase_from_baseline] | actual 2023 full-year real GDP growth <5.5% (vs IMF 2019 pre-COVID forecast ~6.0%) | threshold expression unparseable by regex |
| 6 | stringency_index_extreme_2022 | PENDING_DATA | | >=70 mean stringency 2022 (vs OECD ~30) | No usable vintage for: owid:government_response_stringency_index |
Claim
China's dynamic-zero-COVID policy 2020-2022, peaking with the Shanghai April-2022 lockdown and Beijing/Wuhan late-2022 closures, produced sharper-than-counterfactual demand contraction — retail-sales YoY turning negative, manufacturing PMI sub-50, and youth unemployment spiking — and the December-2022 abrupt reopening produced an incomplete and short-lived recovery rather than the V-shaped rebound expected from pent-up demand and excess household savings.
Interpretation
Verdict is inconclusive (data gaps) — 3 metric(s) cannot be evaluated because the underlying data source is not yet in the vintages pipeline, and 2 metric(s) have data but a threshold expression the auto-evaluator does not recognise (complex conditions, discrete event counts, cross-country gaps). Close these gaps then re-run.
Steelman live concerns
See hypotheses/steelman/china_zero_covid_2022_2023_demand_collapse_recovery.md for the strongest opposing arguments. Canonical-case multi-metric evidence is a pattern match, not a causal identification — the result card should be read as 'outcome trajectory matches the predicted pattern to degree X' rather than 'policy P caused the outcome'.
Provenance
Vintages pinned in manifest.yaml. Full per-metric diagnostics in diagnostics.json. Machine-readable results in metric_results.parquet.
Strongest opposing argument
Every hypothesis ships with its charitable opposing argument. The framework earns credibility by handling objections at their strongest, not weakest.
Notes
2020s natural experiment of an extreme NPI regime, then a sharp policy reversal. Pattern-match on demand-side metrics; growth-side outcome.