IESET.
Hypotheses·energy·africa_south_africa_load_shedding_manufacturing_2007_2024

South Africa's persistent electricity-supply crisis (Eskom load-shedding episodes 2007, 2014-2015, 2018-2024) caused a measurable and cumulative manufacturing-sector contraction and total-factor-productivity drag, with effects intensifying in the post-2018 phase as load-shedding hours per year escalated to record levels.

The pre-registered claim is that, in a synthetic-control design with upper-middle-income emerging-market peers (Brazil, Mexico, Turkey, Malaysia, Thailand, Indonesia), South Africa's manufacturing share of GDP declines by at least 4 percentage points more than the synthetic counterfactual 2007-2024 AND real-GDP-pc cumulative growth falls at least 15 log-points below counterfactual. The null counter-claim is that South Africa's manufacturing decline is part of a broader Brazil/Latin-American-style premature deindustrialisation pattern and the load-shedding signal cannot be cleanly separated from commodity-cycle and competition-from-China effects.

PARTIALengine/runs/africa_south_africa_load_shedding_manufacturing_2007_2024

PARTIAL — mean_gap=+2348, |gap|/pre_sd=6.9, p_perm=0.571; claim direction ambiguous

confidence cueThe result is useful, but not decisive. Treat it as a clue, not a settled conclusion.

policy briefMixed or noisy

In ordinary language

Over a long period, do more market-oriented institutions translate into higher income or productivity, once the comparison looks beyond a single success story?

plain answer

The evidence is suggestive but not decisive. mean_gap=+2348, |gap|/pre_sd=6.9, p_perm=0.571; claim direction ambiguous

why it matters

This matters because energy claims should change belief only when they survive a pre-declared empirical test.

how the test works

It compares 7 country or place units from 2000 to 2024, using a synth did design.

what was measured
What changed
  • Load shedding indicator
  • Load shedding hours year
What we checked
  • Manufacturing share income
  • Real income pc
  • Industrial va
what this does not prove

A single test is not the whole truth. It narrows the claim under a specific sample, time period, and method. Strong policy conclusions need the pattern to survive nearby tests, alternative data, and serious objections.

verification

No evidence packet has been generated yet.

Results

engine/runs/africa_south_africa_load_shedding_manufacturing_2007_2024
1007550250200020122024ZAFBRAMEXTURMYSTHAIDN
illustrative sketch · run pending
No coefficients yet. When the model fires, this chart will show manufacturing_share_gdp across 7 sampled countries over 2000–2024.
The shapes above are stylised — none of the lines are real data.
Placeholder for africa_south_africa_load_shedding_manufacturing_2007_2024. Published chart will be generated from engine/runs/africa_south_africa_load_shedding_manufacturing_2007_2024/chart_data.json.

Pre-registration

registration ordering unverified
first-spec commit 4c8ce8e · 2026-07-18T22:11:21Z
run generated · 2026-04-30T10:15:29Z
Run timestamp predates this path's first git-add commit (rebase, rename, or pre-git local run). Spec hash is still the path's first-add commit — not repository HEAD — but ordering is not a clean pre-registration proof.

South Africa's persistent electricity-supply crisis (Eskom load-shedding episodes 2007, 2014-2015, 2018-2024) caused a measurable and cumulative manufacturing-sector contraction and total-factor-productivity drag, with effects intensifying in the post-2018 phase as load-shedding hours per year escalated to record levels. The pre-registered claim is that, in a synthetic-control design with upper-middle-income emerging-market peers (Brazil, Mexico, Turkey, Malaysia, Thailand, Indonesia), South Africa's manufacturing share of GDP declines by at least 4 percentage points more than the synthetic counterfactual 2007-2024 AND real-GDP-pc cumulative growth falls at least 15 log-points below counterfactual. The null counter-claim is that South Africa's manufacturing decline is part of a broader Brazil/Latin-American-style premature deindustrialisation pattern and the load-shedding signal cannot be cleanly separated from commodity-cycle and competition-from-China effects.

Falsification criterion — what would disprove this

set before the run · honoured after

This hypothesis is considered falsified if:

Not supported if EITHER (a) the manufacturing-share gap vs synthetic counterfactual over 2007-2024 is less than -4 percentage points at p_perm < 0.10, OR (b) the real- GDP-pc cumulative growth gap is less than -15 log-points, OR (c) the loadshedding- hours coefficient on log(industrial_va) in the panel-FE specification is not significantly negative at p < 0.05 (indicating intensity does not predict damage).

formal test & threshold
test:      synth_did_with_intensity_panel_validation
threshold: manufacturing_share_gap_2024 - manufacturing_share_gap_2007 <= -4 ppts AND cumulative_log_gdp_pc_gap_2007_2024 <= -0.15 AND intensity_panel_coef_negative at p < 0.05

Method

Template
synth_did
Clustering
country
Sample
7 countries · 2000 – 2024
Evidence type
causal

Primary: synth_did with ZAF treated from 2007 and EM-UMI peer donor pool. Secondary: panel FE with loadshedding-hours intensity treatment (continuous). Tertiary: event study around the 2018 intensification break. Robustness drops Turkey (parallel-FX- crisis confound) and adds an interaction with the China-import-share control.

Data

VariableSourceTransform
manufacturing_share_gdp
outcome
world_bank_wdi:NV.IND.MANF.ZStier 2
level
real_gdp_pc
outcome
world_bank_wdi:NY.GDP.PCAP.KDtier 2
pwt:rgdpetier 3
log_level
industrial_va
outcome
world_bank_wdi:NV.IND.TOTL.KDtier 2
log_level
total_factor_productivity_index
outcome
pwt:rtfpnatier 3
level
load_shedding_indicator
treatment
constructed:1 for ZAF from 2007 onward (treatment); intensified post-2018tier 5
graded_indicator
load_shedding_hours_year
treatment
constructed:scraped CSIR / Eskom annual loadshedding hours; 0 for non-treated countriestier 5
log_level
terms_of_trade
control
world_bank_wdi:TT.PRI.MRCH.XD.WDtier 2
level
china_imports_share
control
world_bank_wdi:TM.VAL.MRCH.WL.CDtier 2
log_level
us_policy_rate
control
fred:FEDFUNDStier 1
level
oil_price
control
fred:DCOILBRENTEUtier 1
log_level
pwt_capital_stock
control
pwt:rnnatier 3
log_level

● ready  · ● pending  · ● reconstruct-needed

Detailed result card

Result card — africa_south_africa_load_shedding_manufacturing_2007_2024

Verdict: PARTIAL — mean_gap=+2348, |gap|/pre_sd=6.9, p_perm=0.571; claim direction ambiguous

Pre-registration

  • Claim: South Africa's persistent electricity-supply crisis (Eskom load-shedding episodes 2007, 2014-2015, 2018-2024) caused a measurable and cumulative manufacturing-sector contraction and total-factor-productivity drag, with effects intensifying in the post-2018 phase as load-shedding hours per year escalated to record levels. The pre-registered claim is that, in a synthetic-control design with upper-middle-income emerging-market peers (Brazil, Mexico, Turkey, Malaysia, Thailand, Indonesia), South Africa's manufacturing share of GDP declines by at least 4 percentage points more than the synthetic counterfactual 2007-2024 AND real-GDP-pc cumulative growth falls at least 15 log-points below counterfactual. The null counter-claim is that South Africa's manufacturing decline is part of a broader Brazil/Latin-American-style premature deindustrialisation pattern and the load-shedding signal cannot be cleanly separated from commodity-cycle and competition-from-China effects.
  • Falsification rule: Not supported if EITHER (a) the manufacturing-share gap vs synthetic counterfactual over 2007-2024 is less than -4 percentage points at p_perm < 0.10, OR (b) the real- GDP-pc cumulative growth gap is less than -15 log-points, OR (c) the loadshedding- hours coefficient on log(industrial_va) in the panel-FE specification is not significantly negative at p < 0.05 (indicating intensity does not predict damage).

Synthetic-control estimate

  • shape: synth_did
  • treated_country: ZAF
  • event_year: 2007
  • n_donors: 6
  • donor_weights (top): {'IDN': 0.9565, 'MEX': 0.0323, 'BRA': 0.0083, 'TUR': 0.0029, 'MYS': 0.0}
  • pre_rmse: 2791.8359053626177
  • pre_period_sd: 342.4511149408722
  • mean_post_gap: 2348.3155383821313
  • end_period_gap: 1080.1325407566137
  • post_period_years: [2007, 2024]
  • placebo_p_value: 0.5714285714285714
  • n_placebos: 6
  • method: synthetic-control via NNLS, permutation inference

Variables resolved

  • world_bank_wdi:NY.GDP.PCAP.KD; pwt:rgdpe → real_gdp_pc (outcome, n=14131)

Generated by scripts/run_synth_did.py at 2026-04-30T10:15:29+00:00

Strongest opposing argument

Every hypothesis ships with its charitable opposing argument. The framework earns credibility by handling objections at their strongest, not weakest.

Notes

Data-gated on Stats SA quarterly manufacturing series, CSIR loadshedding-hours dataset (publicly released annually), and PWT TFP series. The 2018-2024 intensification phase is the cleanest window for the test; pre-2014 loadshedding was episodic.

Authored framework. Read the transparency note.