A01Same envelopeThe withheld resource is about R253.5 billion in constant 2026 purchasing power, the same base as the industrial-fund paper.
The Welfare Cost of Not Paying the Grant
The poverty, consumption and social cost omitted from the industrialisation counterfactual.
PDF · 27 pages · 1.44 MB · 03.02 · GSX:SRD-002
Study record · GSX:SRD-002
What poverty, consumption, and social costs does the industrialisation counterfactual leave out when the grant is not paid?
Evidence status: Modelled. Counterfactual. Version 1.0. Peer review: none. The paragraph under the title is the published summary.
Conditional answer · GSX:SRD-002:C01
Withholding the SRD envelope has a central explicit welfare cost of R426.9 billion in welfare-equivalent 2026 rand, about R1.68 of social cost for every R1 withheld. The scenario range is R237 billion to R743 billion.
Evidence status: Modelled. Sensitivity: high. Depends on GSX:SRD-002:A01, GSX:SRD-002:A02, GSX:SRD-002:A03. This is not a Treasury invoice, a forecast or a confidence interval. Crime, unrest and household dissolution are left outside the priced total. The central case is an analytical benchmark.
C01
Related questions
- What is the welfare cost of withholding the SRD grant?
- Which social costs sit outside a productive-capital account of the grant budget?
3
Assumptions
Assumptions
These are the registered model assumptions. A result stays conditional on the ones a claim names.
A02Distributional weightsIn the conservative, central and severe cases, R1 of consumption at the recipient margin is valued at R1.10, R1.50 and R2.00. The weights are normative assumptions.1.50 welfare rand per consumption rand, central case
A03No double-counting of local spendingThe local spending multiplier is reported separately and is not added into the welfare cost.qualitative not a scalar
1
Model
Model
The registered model statement. It is a scenario model, and the manuscript remains the full specification.
M01Welfare-cost ledgerThe model prices the explicit welfare cost of withholding the same SRD envelope valued in the industrial-fund paper. It combines consumption, costly credit, scarring and labour effects and keeps the local spending multiplier out of the welfare total.GSX:SRD-002:M01
2
Results
Results
Modelled scenario results. They are not forecasts.
R01426.9 billion welfare-equivalent 2026 rand, central estimateThe explicit welfare-equivalent cost of withholding the envelope is between R237 billion and R743 billion. The central estimate is R426.9 billion, about R1.68 of social cost for every R1 withheld.Central welfare case
R022.5 million person-years, low end of the scenario rangeAdditional poverty exposure across the scenarios is 2.5 million to 11.0 million person-years.Seven-year poverty exposure range
Sensitivity
- High · 2
- Moderate · 0
- Low · 0
Limitations
This is not a Treasury invoice, a forecast or a confidence interval. Crime, unrest and household dissolution are left outside the priced total. The central case is an analytical benchmark.
The comparison is between a public asset and a social cost. It does not say that cash transfers are always better than investment.
2
Claims
Claims
Each claim names the assumptions it uses and the limitation that travels with it.
C01GSX:SRD-002:C01Withholding the SRD envelope has a central explicit welfare cost of R426.9 billion in welfare-equivalent 2026 rand, about R1.68 of social cost for every R1 withheld. The scenario range is R237 billion to R743 billion.Modelled · high sensitivity
C02GSX:SRD-002:C02The average industrial outcome in the companion paper does not clear this central welfare hurdle. Only an exceptionally well-executed industrial programme clearly beats it, and the later beneficiaries are not automatically the households who gave up the grant.Modelled · high sensitivity
Sources and reproducibility
The PDF and complete Markdown source are both published.
Full manuscript: 6 829 words across 53 indexed headings. Source checksum: fb1ca8cc364a…. Code recorded: yes. Data recorded: yes. Environment recorded: no. Published files: 3. the published statement does not rely on a recorded random seed A public code repository is not published.
Community record · version 1.0
Published community record: 0 critiques, 0 forks, and 0 revisions.
Pakang Senosha is the author on this version. External review: none. A later critique or revision appears here after a person publishes it.
Share this article
A permanent link to the work.
Use this page when citing or sharing the study. The PDF contains the complete analysis.
greyscienx.com/research/r350-counterfactual/the-welfare-cost-of-not-paying-the-grant
03
More in this series
The R350 Counterfactual
What South Africa might have built with the SRD budget—and what society would have lost by withholding it.