Writing terms of reference for a RAP consultancy

Most of what is wrong with a resettlement action plan was decided in the terms of reference that procured it. The scope, the schedule, the team composition and the deliverable definitions determine what the consultant can produce; a good team working to a weak TOR produces a document that satisfies the TOR.

Sequence the work rather than parallelising it

The most consequential clause in a resettlement TOR is the one nobody thinks of as a resettlement clause: the milestone structure. A RAP cannot enumerate until the footprint is fixed, and the footprint is an output of the impact assessment's alternatives analysis.[1]

Where the TOR requires census delivery on a date that precedes footprint confirmation, it has commissioned enumeration on assumptions. Make the census milestone conditional on written footprint confirmation, and price the consequences of a change after that point as a variation rather than as the consultant's risk — otherwise the consultant prices the risk into the bid and the project pays for it either way.

Specify the instrument, not just the report

A TOR that asks for a RAP in accordance with the applicable standards leaves every methodological decision to the bidder, and bidders differ. Specify:

• The census instrument and that it must support multiple claims per parcel, disaggregation by individual rather than household head, and the fields needed to assess severity of loss.
• The valuation approach, including an explicit requirement to benchmark statutory rate schedules against replacement cost and to document the gap.[2]
• The socioeconomic baseline, with the indicators that livelihood restoration will later be measured against named in the TOR, so the baseline is designed to support them.
• Common property and access losses as a required survey component, seasonally timed. If it is not required it will not be done.
• Data deliverables in a specified structure with the audit trail intact, and ownership vesting in the client.

Everything a monitoring framework will later need has to be collected at baseline. The TOR is the last point at which that can be required.

The team, stated as time rather than titles

Key expert lists specify titles and qualifications and rarely specify person-days. Since the sociologist is frequently the input that gets compressed when the schedule slips, specify minimum field time for each role and require the enumerator-to-household ratio and the supervision arrangement.

Two roles are worth naming explicitly because they are routinely under-provided: a valuer with local practising standing, since a valuation that cannot be lodged under national procedure is of limited use; and a data specialist responsible for the register as a system rather than as a spreadsheet deliverable.

Deliverables that are testable

Define acceptance criteria that someone can apply. A draft RAP is accepted when it contains an entitlement matrix covering every loss category identified in the census, a rate benchmark for each category, a cut-off date with evidence of disclosure, a grievance procedure with response times, a budget reconciling to the entitlement calculations, and a monitoring framework whose indicators are computable from the baseline.[3]

Where acceptance is defined loosely, review becomes a matter of opinion and the argument happens after the money is spent.

Provide for the implementation phase

The TOR should say what happens after approval. Options are to extend the same consultant into implementation support, to require a documented handover to the implementing agency with a defined transition period, or to procure implementation support separately with a specified overlap.

What should not happen is the default: the plan is accepted, the team disperses, and the agency begins implementation with a document and no institutional memory.[4]

Budget realism

Two lines are routinely under-budgeted in TORs and both are predictable. Enumeration cost scales with the number of parcels and with terrain, not with corridor length, and TORs that estimate from length underestimate in fragmented smallholder areas. Consultation cost scales with the number of settlements and the number of separate sessions required to reach women, tenants and absentees.

A TOR that under-provides for either will receive compliant bids that have solved the problem by doing less of both, and the deficiency will surface as an inadequate baseline two years later, when it cannot be fixed.