Mombasa–Mariakani: two cut-off dates, eighteen months apart, on one road

The Resettlement Action Plan for dualling the Mombasa–Mariakani road records not one cut-off date but two, eighteen months apart, and the RAP is unusually candid about why: the road's design changed after the first one was set.[1]

The original cut-off was 30 July 2014. A second cut-off, 18 December 2015, applies specifically to PAPs whose land is acquired "as per the gazette notice of this date and for the new PAPs entrants … as a result of incorporation of additional road design features including the interchanges that had not been concluded on during preparation of the original RAP."[1]

A design change after the census is not a paperwork problem. It is a new population the original cut-off never covered.

Why a second cut-off date is the right answer, not a symptom

It would have been simpler, and wrong, to apply the 2014 date to everyone. Interchanges added after that date created a footprint nobody could have been surveyed against — a household living exactly where an interchange was later routed had no way to appear in a 2014 census for a corridor that, at the time, did not include their land. Treating them as late claimants against an old cut-off would have excluded people for a reason that has nothing to do with opportunism.

The RAP's solution — a second, later cut-off tied explicitly to the gazette notice covering the new footprint — is the correct instrument for exactly this failure mode. It is also uncommon. Most projects that add scope after a census either silently fold the new area into the old cut-off (excluding people who were never surveyed) or treat the addition as an undocumented exception (creating a category the register cannot explain later). This RAP names the mechanism and dates it.

The residual risk is narrower and worth stating: anyone who altered their land use in the eighteen-month gap between the two dates, anticipating that the interchange design might eventually reach them, sits in exactly the grey zone a cut-off exists to close. A RAP amendment can fix the boundary problem. It cannot retroactively remove the incentive that existed while the boundary was open.

The numbers, and what "947 PAPs" actually contains

The project takes roughly 70 hectares for the dualling, and the RAP records 947 project-affected property holders — of whom 914 lose structures, 692 lose business, 146 lose trees and 34 lose crops (the categories overlap, since one claimant can appear in several).[1] Of the 947, 855 are individuals and 92 are institutions or companies.[1]

That last figure is easy to miss and worth noting. A resettlement register built around household entitlements can struggle to represent a claimant that is a business, a church, or a school rather than a person — different eligibility logic, different signatory rules, different documentation requirements. Ninety-two such claimants is not a rounding error; it is a tenth of the affected population needing a different data shape from the other nine-tenths.

The RAP also records 26 vulnerable household heads requiring additional livelihood support, distinct from the larger population of small-scale traders on the road reserve who hold no security of tenure at all.[1] Most PAPs along this corridor are traders and Jua Kali artisans rather than farmers — the RAP notes the land has "no plantation or farms affected" — which places this squarely in the economic-displacement-without-land-title pattern this site's papers on urban displacement and eligibility without title both describe.

Three financiers, one reconciliation table

This RAP was prepared for a project financed jointly by the European Investment Bank, KfW and the African Development Bank.[1] Rather than pick one standard and hope the others do not object, the RAP carries its own table of recommendations to reconcile the three institutions' resettlement policies.[1]

That is the correct response to co-financing, and it is rarer than it should be. A project financed by a syndicate does not get to satisfy the least demanding standard in the group; it has to clear the highest bar on every dimension, as this site's paper on comparing international resettlement standards sets out. A RAP that states which standard governs which requirement, in a table, is doing that reconciliation visibly rather than leaving it to be inferred — or disputed — later.

What the register had to carry

• Cut-off date as a field tied to the gazette notice that created it, not a single project constant — this RAP already needed two, and any future scope change would need a third.
• A claimant-type field distinguishing individuals from institutions, since 92 of 947 PAPs here are companies or organisations with different eligibility and signatory requirements entirely.[4]
• Tenure-security status separate from vulnerability status — the RAP treats "traders with no security of tenure" and "26 vulnerable household heads" as distinct populations requiring distinct support, and a register that conflates them will misdirect assistance.[2]
• The applicable-standard reconciliation, at the entitlement level, so that a claimant's file shows which financier's requirement governed their specific entitlement, not just which financier funded the project overall — Guidance Note 5 sets out how to run that reconciliation in practice.[3]

What this article does not establish

This is a reading of the 2017 revised RAP as disclosed; it does not evidence what was actually paid, to whom, or whether the second cut-off was applied consistently in implementation. This article had no access to any project register or to AfDB's Independent Recourse Mechanism caseload, which has handled complaints on other Kenyan coastal road projects; nothing here should be taken as a statement about complaints on this specific project. Readers should open the RAP before relying on any figure quoted from it.[1] This is not legal advice on Kenyan compulsory acquisition.