Tanzania SGR: the RAP that wrote down its own gap with the standard

Most of the case studies in this series have to reconstruct the gap between national law and a lender standard from a valuation dispute, a court judgment, or a complaint filed after something went wrong. The RAP for Lot 6 of Tanzania's Standard Gauge Railway — the 55 km section from Tabora to Kigoma, km195 to km250 — does something rarer. It states the gap itself, in a table, before anyone has been paid.

Section 4.5 of the disclosed RAP sets Tanzanian law against the African Development Bank's Operational Safeguard 2 (OS2), issue by issue, and names the population each regime leaves out.[1]

"PAPs under category C such as the encroachers to the land are not entitled to compensation under the Tanzanian laws."

The three categories, and the one Tanzanian law drops

OS2 classifies displaced people into three groups: those with formal legal rights, those with customary and traditional rights, and those with no legal right who can demonstrate occupation before a cut-off date — the same three-category structure PS-5 uses.[3] The RAP records that the third category is entitled to assistance for developments and relocation under OS2 — and that "the lost assets under OS are much wider than land and include loss of access to livelihoods and standard of living," seeking to restore them to pre-displacement levels.[1]

Tanzanian law, the same table records, compensates "unexhausted improvements" — land and what is built or grown on it — and nothing else. Category C simply is not a compensable class under the domestic statute. The RAP does not soften this into "the frameworks differ in emphasis." It states plainly that the encroacher population is uncompensated under national law and picked up only because OS2 requires it.[1]

The same table addresses tenants and squatters: Tanzanian law does not recognise them as entitled to compensation at all, treating squatters the same way. Under OS2 they fall into category C and are owed compensation for their developments plus relocation assistance.[1]

This is what a gap analysis is supposed to look like: not a paragraph asserting that lender standards "go further," but a named class of person, a named exclusion, and a named remedy. Every case study in this series that has had to infer this gap from a court record or a CAO finding is looking at the consequence of a RAP that did not do what this one did.

The numbers

The section covers six villages — Nguruka, Nyangabo, Bweru, Malagarasi, Mpeta and Chakulu, all in Uvinza District, Kigoma Region — and reports 1,082 project-affected persons identified and interviewed, of whom 457 were recorded as vulnerable.[1] RAP implementation was budgeted at USD 4,922,937.13. The project is Category 1 — the AfDB's highest risk tier — reflecting its scale and the sensitivity of its footprint.[1][2]

Two cut-off dates, and a sequencing question worth asking

As with Dar es Salaam BRT, this RAP runs two cut-off regimes: one for land parcels, set at intervals between 31 October 2023 and 27 January 2024 across the six villages, and a separate one for the railway right of way, running from March to April 2025.[1]

The RAP is explicit about how the date is fixed: "the cut-off date for the approved 55 KMs of the RoW and land parcels were announced after the valuation."[1] That follows Tanzania's Valuation and Valuers Registration Act, whose section 53(1) states plainly that the cut-off date is the date valuation commences.[1]

Compare that with Rusumo Falls, where the cut-off was fixed at the date of each community's own self-validation meeting and signed by the affected people themselves. Both approaches satisfy the basic requirement that a cut-off exist and be disclosed. But a date announced once valuation is already underway carries a specific risk this RAP does not discuss: anyone who improves or occupies the corridor in the window between the valuation team's arrival and the formal announcement is in a grey zone the document does not resolve. Where the community itself signs the date, that ambiguity has no room to exist.

The grievance mechanism has two tiers, on paper

The RAP describes a project-level Grievance Redress Mechanism plus a separate route to "the Bank's grievance and redress mechanism"[1] — meaning the AfDB's own accountability channel sits above the project mechanism, structurally independent of TRC. That is the tiered design this site's paper on grievance escalation argues for: a project-level route with real authority, and an external one that does not depend on the party complained about.

What the document cannot show, because it is a plan rather than a monitoring report, is whether that second tier is reachable in practice by a household in Nguruka or Chakulu, or whether it functions only as a line in a disclosure document. That is the same distinction Bujagali's CAO history draws sharply: a grievance mechanism described in a RAP and a grievance mechanism a person can actually use are not verified by the same evidence.

What the register had to carry

• Category C as a tracked class, not an omission. Where a RAP has already conceded, in writing, that encroachers and tenants are uncompensated under national law, the register needs a field that identifies every person in that category, distinct from category A and B claimants, so the OS2 top-up cannot be quietly forgotten between the RAP's approval and its implementation.
• The valuation-to-cut-off interval, dated, per village — because the RAP's own account shows the date is set after valuation begins, and that interval is exactly where the ambiguity this article flags would surface.
• Vulnerability status linked to a defined assistance measure. 457 of 1,082 PAPs are recorded vulnerable; Guidance Note 5 treats this as the standard's central identification requirement, not a discretionary extra,[4] and the register needs to show what, specifically, each received because of that status, not merely that the flag was set.
• Grievances tagged by tier — project mechanism or Bank mechanism — with the date each was reached, so the second tier's actual accessibility becomes measurable rather than assumed.

What this article does not establish

This is a reading of a disclosed planning document dated September 2025, covering only Lot 6 of a multi-lot project; other sections of the SGR have their own instruments and are not addressed here. The RAP is a plan, not a monitoring or completion report, and nothing here evidences what was actually paid, to whom, or whether the described grievance mechanism has handled a real case. Figures are the RAP's own; readers should open the document before relying on any of them.[1] This is not legal advice on Tanzanian compulsory acquisition law or on AfDB safeguard compliance.