Institutional capacity and the implementation gap in resettlement programmes

1. Two different documents

A resettlement action plan is written by a consultancy team assembled for the purpose: a resettlement specialist, a valuer, a sociologist, enumerators, a GIS technician. It is delivered, approved, and the team disperses. Implementation then falls to a unit inside the implementing agency, often two or three people with other responsibilities, whose authority to commit funds, hire staff or vary procedures is limited by rules the plan did not consider.

The plan and the implementing unit are therefore matched only by assumption. The plan specifies a grievance mechanism with defined response times, a livelihood programme running for several years past construction, quarterly monitoring against outcome indicators, and a records system capable of surviving audit. Each of these is a standing operational capability. The unit was sized to process payments.

2. Where the constraint actually binds

Staffing is the visible constraint and rarely the binding one. The binding constraints are more often procedural: an agency that cannot pay a household without a national identity document in a population where a fifth of adults have none; a procurement rule that makes engaging a livelihoods NGO a nine-month exercise; a treasury release cycle that funds compensation annually while the schedule requires payment before possession; a delegation limit that sends every above-threshold valuation to a committee that meets monthly.

None of these appear in a risk register phrased in terms of capacity, and each of them can stop a programme for a year. They are visible only to someone who has asked how the agency actually executes a payment, which is a different question from whether the agency has a resettlement unit.

The second underrated constraint is coordination. Resettlement implementation typically requires a land authority, a valuation office, a local government, a utility, a police function for possession and sometimes a court. The implementing unit usually has authority over none of them. Where the plan assumes their cooperation on a schedule, it has assumed the thing most likely to fail.

3. Why capacity assessments do not predict

Standards require an assessment of the borrower's or client's capacity to implement, and the assessments exist. They are typically structured as an inventory: does the agency have a resettlement policy, a unit, prior experience, a grievance procedure. Answered honestly, this produces a description of institutional form and almost no information about throughput.

The predictive questions are quantitative and awkward to ask. How many compensation payments has this agency completed in the last twelve months, and how long did the median one take from valuation approval to funds received by the claimant? How many grievances were logged in its last project, and what proportion were closed within the stated period? What was the variance between its last resettlement budget and its actual expenditure, and where did the variance arise? A unit that has never processed more than a few hundred payments cannot be assumed to process several thousand because the plan says it will.

The reason these questions are not asked routinely is that the answers are institutionally sensitive, arrive during appraisal when the project is committed in all but signature, and would imply either a longer schedule or a larger budget. The assessment therefore records form, and form is not a constraint.

4. The consultant substitution problem

Where capacity gaps are recognised, the standard remedy is to contract them out: an implementation support consultant, a monitoring consultant, a livelihoods contractor. This works, in the narrow sense that the activity happens. It also has a systematic side effect — the institutional knowledge accrues to the contractor and leaves when the contract ends, which is generally before the obligations do.

The pattern is most damaging in records. Where the register of affected persons, the valuation files and the payment reconciliation live in a consultant's systems and are handed over as a final deliverable, the agency receives a dataset it did not build, in a format it cannot query, without the people who know its exceptions. Completion audit then tests a record nobody in the room maintained. This is one of the more common routes by which a substantively adequate programme fails on evidence.

5. What would make the assessment useful

Three changes would move capacity assessment from documentation to prediction. First, measure throughput rather than structure, using the agency's own recent history as the base rate and treating any planned rate materially above it as a risk requiring a specific mitigation rather than an assertion of commitment. Second, trace one payment end to end during appraisal — from valuation approval through internal authorisation to funds actually received — and record every step, since the resulting map is where the schedule will fail. Third, treat the records system as a capability the agency must hold rather than a deliverable it receives, and fund it as such.

The corresponding budget consequence is not marginal, and it is the reason these recommendations are made more often than they are adopted. It remains cheaper than the alternative, which is a programme that spends its full compensation budget and cannot demonstrate at completion that the obligations were met.

6. Conclusion

The distribution of resettlement failures is not random across the project cycle. Planning is done to a professional standard by people who do it repeatedly; implementation is done by institutions doing it for the first or second time, under procedural constraints that the plan did not model. Analysis that continues to treat the plan as the locus of quality is examining the part of the process that is least often the problem.