The PAP census and socioeconomic survey: getting the baseline right

The census is the foundation everything else in a resettlement action plan rests on. It fixes who is eligible, records what each person loses, and establishes the baseline that livelihood restoration will eventually be measured against. Errors made here do not stay here — they propagate into entitlements, into payments, and into the grievances that arrive eighteen months later.

It is also the step under the most schedule pressure, which is why it is the step most often compressed.

Two exercises, not one

The census and the socioeconomic survey are frequently run together and serve different purposes. Conflating them is a common source of gaps.

• The census and asset inventory enumerates every affected person and every affected asset. It is a complete count, not a sample, and it determines eligibility.
• The socioeconomic survey establishes conditions — income sources and levels, livelihood strategies, vulnerability, access to services. It may be sampled, and it is the benchmark restoration is measured against.[2]

A project that runs a thorough asset inventory and a thin socioeconomic survey has counted what it will pay for and failed to record what it will later have to restore. That gap only becomes visible at completion audit, when there is no baseline to compare against and no way to create one.

Who gets missed

Undercounting is systematic rather than random, and it follows the same pattern across projects. Enumeration tends to find the household head, on the plot, during the day, in the dry season. Everyone whose situation differs from that is at risk:

• Tenants. Enumerators interview owners; tenants have separate entitlements under both standards and are frequently never recorded.[1]
• Employees of affected businesses. A distinct eligible group from the business owner, and almost never enumerated as such.
• Seasonal and migrant users. Pastoralists, seasonal cultivators and transhumant herders may be absent during enumeration and hold genuine use rights.
• Women with independent claims. Where enumeration records one head per household, women's separate crop plots, trees and income sources disappear into a single male-headed record.
• Absent household members working elsewhere but dependent on, or contributing to, the affected asset.
• Users of common property — grazing land, water points, collection areas — who lose access without losing any titled asset.

Every category above is eligible under PS-5 and ESS-5, and every one of them is invisible to an enumeration method built around titled household heads.

What has to be captured per person

Working backwards from what the project must later evidence, each record needs enough to support an entitlement determination, a payment, and a defence of both:

• identity, with a durable identifier and a photograph;
• household composition, including dependants and absent members;
• tenure status and the evidence for it — which drives land compensation and nothing else;
• every affected asset, itemised and measured, geo-referenced to the parcel;
• income sources and their relative contribution;
• vulnerability screening against the project's written criteria;
• consultation record — what the person was told, when, and in what language;
• enumerator identity and timestamp, so an entry can be traced back to who made it.

That last item is the one most often omitted and the one that matters most later. When a valuation is disputed, the question is who measured this, when, and what did they record at the time. An unattributed row cannot answer it.

Why paper and spreadsheets fail here

The standard enumeration stack — paper forms in the field, keyed into a spreadsheet later — introduces failure modes that are structural rather than careless:

• Transcription loss. Every value is written once and typed once, with no validation between.
• No attribution. The spreadsheet row does not record who collected it or when.
• No change history. A revised figure overwrites the original, and the fact of revision disappears with it.
• No spatial anchor. Assets recorded without coordinates cannot be re-found for verification.
• Version divergence. Field teams working offline produce copies that are reconciled by hand, and reconciliation decisions are not recorded either.

Offline-first digital collection addresses these directly — validation at the point of entry, automatic attribution and timestamping, geo-referencing, and an append-only change history. It is not a matter of modernity. It is that the completion audit asks questions the paper method structurally cannot answer.[3]

Verification and disclosure

A census is not finished when enumeration ends. Affected people should be able to see and check what was recorded about them — their assets, their measurements, their entitlement category — and to correct it before it hardens into a payment schedule.[4]

Projects that skip this step pay for it twice: once in grievances that could have been corrections, and once in the credibility cost of arguing with people about data they were never shown. Disclosure at this stage is the cheapest grievance prevention available, and the record of it is a useful thing to hold when a determination is challenged later.