Independent monitoring: what an external monitor is for, and when to appoint one

Internal monitoring reports the programme's progress to the people running the programme. It is necessary, it is generally honest, and it is structurally incapable of saying the one thing that matters most: that the approach is not working.

Two different functions

Internal monitoring tracks delivery — payments made, houses built, grievances logged, training delivered — at high frequency, for management use. Its purpose is operational control.

External monitoring tests outcomes and compliance independently: whether households actually received what the records say, whether livelihoods are recovering, and whether the commitments in the plan are being met. Its purpose is assurance.[1]

Conflating them produces a monitoring function that reports activity frequently and outcomes never, which describes most resettlement programmes.

The value of an external monitor is precisely that it can report something the project management does not want to read. Everything about its terms of reference should protect that capability.

What independence requires

• Appointment and reporting outside the implementing unit — to a board, a steering committee or the lender, not to the manager whose programme is being assessed.
• Unrestricted access to the register, the payment records, the grievance log and the affected communities, without escorts who determine who is interviewed.
• Sampling chosen by the monitor, not supplied by the project. A list of satisfied households provided by the resettlement unit is not a sample.
• Reports issued without editorial control by the implementing agency, with management response published alongside rather than incorporated into the findings.
• A contract that cannot be terminated for reporting adverse findings.[3]

Appoint early, not at completion

An external monitor engaged at the end can only document what happened. Engaged before implementation begins, it can verify that the baseline is adequate for the monitoring framework, that the register is capable of supporting reconciliation, and that the grievance mechanism is operating before offers go out — all of which are correctable at that point and not later.[2]

The most useful single deliverable in most programmes is an early verification of the baseline, because everything the completion audit will ask depends on data collected once, at the start.

What a midterm review should actually do

Not repeat the progress report. A midterm review earns its cost by answering four questions:

1. Are the entitlements as designed producing restoration, or are households compensated and worse off?
2. Which categories of affected person are falling behind — and are they the ones flagged as vulnerable at baseline?
3. Where is the process failing operationally: the payment tail, the grievance backlog, the livelihood component nobody has started?
4. What has to change now, while there is still time and budget to change it?

The fourth is the point. A midterm review that produces observations rather than a corrective action plan with owners and dates has documented a problem instead of fixing it.[5]

Methods that catch what records do not

Household visits against a random sample from the register, not from a list. Reconciliation of confirmed receipt against assessed entitlement for that sample. Interviews with people who lodged grievances, about whether they were answered. Interviews with households that moved away from the resettlement site, which requires actually finding them. And conversations conducted without project staff present, which is the condition under which people say what happened.[4]