Below 1,000 kWh per person, concessions work only when subsidy, tariff and losses are engineered together — 19 of 35 studied operators were not viable
Asked:
“We do not yet know the economics of concessions in countries below 1,000 kWh per person. Can you help figure this out? (These are the questions a feasibility phase must answer with real numbers but we want to get as close as possible to the truth)”
The evidence base is 35 operator and program cases — national utilities, distribution concessions, rural concessions and mini-grids — drawn from 28 distinct source documents (heavily World Bank), spanning roughly 1960–2025 across Africa and Asia. About 43 of 218 economies sat below 1,000 kWh/person in a recent 2024 listing, but other datasets and years give counts from 43 to 77 — the cutoff is a screening threshold, not a verified universe, and national per-capita consumption is not concession-area customer consumption.
The bankability line: realized tariff vs cost per kWh
Not viable — tariff below costConditional — at or near costViable — margin above cost10 cases with comparable US$/kWh pairs; hover a point for detail
Every utility below the diagonal sells each kWh for less than it costs. The wedge is then compounded by losses (Chad 40%, Guinea 30%, Sierra Leone >45%) and noncollection (Chad ~50%, Guinea 42%, Gaza 41%). Cases without a clean US$/kWh pair (Umeme, CIE, mini-grids) appear in the table below.
What the wedge chart cannot say alone
Scale thresholds are explicit in the record: Madagascar’s HIER said projects need at least 400–500 subscribers plus a 75% capital subsidy on generation and 100% on distribution to be profitable; Uganda’s KRECS needed nearly 20× its electricity sales to break even. documents1.worldbank.org
The strongest turnaround is operational, not tariff magic: Uganda’s Umeme cut losses from 38% to 21% and lifted collections from 80% to 99.1%, with a guaranteed 20%/yr return on capex — and stopped needing subsidies. documents.worldbank.org
Risk allocation can substitute for sector health: Côte d’Ivoire’s CIE stayed profitable (~5% net income/revenue, losses 29%→15%) because its remuneration insulated the operator from broader sector risk. openknowledge.worldbank.org
Priced-in viability exists at both ends: Senegal’s first rural concession reached a 12.22% FIRR with ~US$286 subsidy per connection and ~60% private financing, while Cambodia’s spontaneous mini-grids ran with no subsidy at US$0.40–1.25/kWh. openknowledge.worldbank.org
CIE (Côte d’Ivoire): profitable through the concession, ~5% net margin, losses 29%→15%; operator remuneration ring-fenced from sector deficits. openknowledge.worldbank.org
Cash cost = (kWh billed ÷ (1 − losses)) × energy purchase or generation cost + fixed O&M + debt service
Viability gap = cash cost − cash revenue, to be closed by capital subsidy per connection, operating subsidy or availability payment, tariff indexation, or load growth
Observed anchors: household consumption of newly connected rural customers runs 3–30 kWh/month in the cases (Madagascar 3–15, Uganda KRECS 30), Eastern Africa household end-use averages 55 kWh/person/year, global mini-grid funding averages US$411 per connection, and Senegal’s bankable concession needed ~US$286 subsidy per connection. These are orientation figures, not underwriting assumptions.
Sensitivity: monthly cash margin per customer
Inferred design implication, not an observed figure: margin = kWh × tariff × 95% collection − (kWh ÷ 0.8 to gross up for 20% losses) × US$0.25 energy cost × 60% energy-cost share − US$1.75 fixed O&M per customer per month (the observed African isolated-grid average). Green cells clear cash costs; red cells lose money on every customer. Hover a cell for the exact figure.
Thirteen feasibility questions with the thresholds the record supports
Addressable connections and ramp-up — below ~400–500 subscribers per site, Madagascar’s operators were structurally unprofitable even with 75% generation subsidy; Senegal’s winner bid 21,800 vs an 8,500 minimum.
kWh per customer per month — 3–15 kWh/month (Madagascar) does not carry a concession; ~30 kWh (KRECS) still fell ~20× short; Umeme’s weighted 315 kWh/month did. Measure household and productive load separately.
Tariff by segment, indexation, affordability — viable cases sat at or above cost (Guatemala 16.7 vs 13.5 US¢); mini-grid customers demonstrably paid US$0.40–1.25/kWh; affordability ceiling for unelectrified households is about a US$10 monthly bill.
Energy purchase or generation cost — the killer variable: US$0.22–0.63/kWh in failing cases; fuel was >80% of opex for Mali’s KAMA mini-grids.
Loss trajectory — winners hold or drive losses to 15–21% (CIE, Umeme); >30% (Guinea, Gaza, Chad, Sierra Leone >45%) predicts failure.
Collection and prepaid share — winners collect 93–100%; prepaid metering took HIER from ~85% to 100%. Below ~75% (Chad, Guinea, Gaza) no tariff fixes it.
Connection and network capex — observed range US$112 (Uganda) to US$1,171 (Lao) per connection; Senegal rural averaged US$725.
Capital subsidy per connection and % of capex — bankable Senegal needed ~US$286/connection (~40% of capex); Madagascar design point was 70–75% of capex and 100% of distribution; Tanzania socializes 85–90% of connection cost.
Operating subsidy / availability payment — quantify it up front: Chad ran ~US$1,000 per customer per year; Rwanda’s EUCL needed ~1.5% of GDP in subsidies through its turnaround.
Operator remuneration and allowed return — Umeme's 20%/yr on capex and CIE’s cost-plus margin (~6% of eligible costs) are the observed prices of private capital here.
Working capital, FX, fuel/hydrology, government payment risk — Lao EdL carried 18 months of government receivables; Mozambique EdM needed a 35% tariff rise just to cover O&M; diesel price moves flipped Madagascar’s BETC into loss.
Renewal capex and handback — Madagascar plants need rebuilding after ~10 years, and first-investment-only subsidies leave no funding for it.
Downside DSCR / equity IRR / breakeven — reference points: Senegal FIRR 12.22% vs a 12% discount rate; Lao DSCR 1.9; Gambia DSCR −0.5 at the trough.
Go / no-go gates for the feasibility phase
Gate 1 — demand: verified addressable base ≥400–500 paying connections per service area and measured willingness to pay at a bill ≤US$10/month; anchor productive loads identified. Fail → standalone solar territory, stop.
Gate 2 — price vs cost: realized tariff × collection ≥ energy cost grossed for losses on a 5-year view, or a committed, indexed subsidy/availability payment closing the gap in the contract. Fail without a funded gap → no-go.
Gate 3 — capital structure: capital subsidy per connection sized (reference US$286–411 observed; up to 75–100% of capex where loads are thinnest) and disbursement-credible; renewal capex funded, not just first investment.
Gate 4 — risk allocation: operator remuneration insulated from sector cashflow (CIE model) or regulated return on capex (Umeme model); FX, fuel and government-payment risk explicitly assigned; downside DSCR ≥1 with equity IRR above the local discount rate.
Proposed feasibility data room
Customer census of the concession area: counts by segment, current spend on kerosene/diesel/phone charging, appliance ownership.
12+ months of metered consumption from comparable sites (not national per-capita statistics), split household vs productive.
Full tariff schedule, indexation formula, history of political tariff interventions, and regulator track record on adjustments.
Energy purchase contracts or generation cost build-up with fuel/hydrology sensitivity; losses by technical vs commercial with meter audit.
Billing and collection ledgers, prepaid penetration, arrears aging including government accounts.
Connection cost bill of quantities; subsidy program rules, disbursement history and arrears.
A tariff–cost wedge plus losses >30% plus collection <75% is fatal; no observed case survives all three.
Household-only rural demand of 3–30 kWh/month cannot fund a full-cost concession; capital subsidy of roughly 40–100% of capex is the observed bankable range.
Operational contracts with regulated returns and ring-fenced remuneration (Umeme, CIE) are the repeatable success pattern.
People do pay US$0.40+/kWh where service is reliable (Cambodia) — affordability is about the bill, not the per-kWh price.
Only field work can answer
Actual concession-area demand and ramp-up — national per-capita figures are the wrong denominator, and site demand varies 100× (Zambia: <0.5 to >50 MWh/yr).
Real willingness to pay and collection under prepaid metering in the target area.
Achievable loss trajectory given the existing network and theft environment.
Credibility of the specific government as subsidy payer and offtaker — receivables and indexation in practice.
Method: 35 operator/program cases from a curated result set (28 distinct source URLs across 3 independent web hosts; no URL backs more than 5 rows), each row carrying tariffs, costs, losses, collections, subsidies and a viability assessment as stated in its source, mostly World Bank documents spanning roughly 1960–2025 across different technologies, scopes and price years — figures are as reported, not inflation-adjusted, and not directly comparable without qualification. The scatter shows the 10 cases with a comparable US$/kWh tariff–cost pair; Chad is plotted at its tariff-band midpoint. The country-universe framing draws on 30 public per-capita consumption tables (counts below 1,000 kWh/person range 43–77 by source and year); benchmark figures (US$411/connection, 55 kWh/person/yr) come from a 100-row benchmark set and are orientation only. The sensitivity grid is a design calculation, not observed data. Long source text is truncated in table cells for space; each row links its full document.