Baseline evidence and impact analysis for resettlement

Design the baseline around the decisions it must support: who is affected, what each person or group loses, which remedies are feasible, and how recovery will be assessed. A long questionnaire provides little protection if the census misses seasonal users or the asset inventory cannot be linked to livelihood dependence.

Coordinate distinct evidence instruments

Use shared identifiers and a common design version so these instruments describe the same affected relationships. The World Bank sourcebook and IFC handbook treat baseline collection as part of the resettlement program's planning and monitoring system. World Bank sourcebook, IFC handbook

Identify people through use as well as ownership

Work in both directions. Identify people connected to the mapped footprint, then ask communities which activities extend beyond it. The first approach finds owners and occupants; the second may reveal herders, fishers, resource collectors, employees and seasonal cultivators.

Define visit times, revisits, absent-person procedures, duplicate checks and claims review before enumeration. A daytime household visit can miss workers. A visit in one season can miss another season's users. Local leaders can help verify facts, but one gatekeeper should not control recognition of every affected interest.

Separate persons, households, structures, enterprises, employees and collective user groups. Two families may share a dwelling while maintaining separate budgets. A household may use several parcels. Physical and economic displacement can overlap: count unique households separately from the two impact categories, and state the overlap. Preserve plan estimates, verified register totals and actual assistance totals as different measures.

The cutoff date needs a disclosed and documented basis under the governing framework. Distinguish a later arrival from a person who held an existing interest but was missed. IFC GN5 discusses seasonal users, outdated dates and changes in eligible households. Project delay also requires attention to restrictions that prevent people continuing ordinary livelihood activities. World Bank ESS5, IFC GN5, GN32–33

For a historic date, review the original census's coverage, communication and subsequent changes. Keep evidence, reasons and an appeal route for each contested decision. Update the affected register when the engineering footprint changes while preserving the earlier version.

Connect a loss to its remedy

A verification field should point to evidence and a reviewer. For contested claims, retain the assertion, supporting and contrary evidence, decision and date. Shared resources need records of their users and governing arrangements as well as household links.

Protect identity documents, payment details, health information and grievance accounts through restricted access. Public explanations of impacts and progress can use aggregates and de-identified examples. Do not publish a household register simply because the planning report is public.

Measure livelihoods through seasons and functions

Record how activities contribute to welfare across the year. Farming may provide sales and food for consumption; fishing may be most important during a lean season; a dry-season pasture may keep a herd viable when other land is available. Annual cash averages can conceal these dependencies.

For agriculture, record production, own consumption, sales, inputs, labor, water, tenure and crop cycles. For businesses, distinguish revenue from net income and examine customers, suppliers, operating costs and restart conditions. For common resources, investigate access, permitted users, seasonality, management and substitutes.

Triangulate interviews with observation, market information and relevant records. An administrative source covering formal transactions may miss a real informal activity. A household account may also need corroboration. Use disagreement to direct further inquiry.

The published abstract of the Three Gorges study by Wilmsen, Webber and Duan describes households in two villages whose resources diminished during resettlement and who borrowed largely to fund new housing. Their responses reflected constraints on their choices. The transferable question is whether the proposed alternative actually replaces the lost contribution under the conditions facing the intended participants. Wilmsen and colleagues

Use samples for estimates and a census for recognition

A probability sample can estimate population characteristics. It cannot determine that an unsampled household is ineligible. Maintain the census and claims process separately from the statistical survey design.

For larger populations, select strata suited to the impact profile, such as relocation, land loss, business displacement and tenure categories. Rare groups may need complete coverage or additional qualitative investigation. Apply appropriate weights when oversampling groups and reporting population estimates. Define precision, variation, design effects and nonresponse assumptions rather than choosing a convenient percentage of households. Impact Evaluation in Practice

Plan follow-up while people can still be contacted. Obtain safe recontact arrangements, link household splits and moves, and record attrition. People who leave an unsuitable relocation site remain relevant to the outcome assessment.

The baseline survey methods develop questionnaire design, sampling, measurement and fieldwork in more detail.

Use spatial evidence to test a causal account

Maintain dated layers for permanent works, temporary occupation, easements, access roads, utilities and construction stages. Check coordinates, resolution and positional accuracy before overlaying parcels, buildings, resources and travel networks. Then verify use, dependence and rights through fieldwork.

A partial acquisition illustrates the difference between exposure and consequence. Suppose a one-hectare farm loses 0.1 hectare containing its irrigation intake. Ten percent land loss can interrupt production on the remainder. Assess the feasibility and timing of restoring irrigation, residual viability and transition support. An area percentage does not determine the remedy.

ADB's ESS5 guidance addresses residual parcels that become unviable, unsafe or inaccessible under its specified provisions. The broader inquiry into function is useful across frameworks. ADB ESS5 guidance

Use routes and travel times for access questions where suitable data exist. Straight-line proximity does not establish safe access to a school, water source or market. Satellite imagery can support evidence about structures and cultivation, but cannot establish tenancy, customary claims or who controls income. See the spatial methods discussion.

Make the baseline useful when evidence is incomplete

Keep unknown, zero, not applicable and refusal distinct. Record unvisited areas, omitted groups, uncertain quantities and the decisions these gaps affect. Budget provision can manage uncertainty while investigation continues; it cannot replace a defensible eligibility decision.

Where displacement preceded the baseline, reconstruct the earlier conditions using dated imagery, production and business records, agreements, earlier assessments and accounts from multiple parties. State which variables can be recovered credibly. A later survey describes current conditions; naming it a baseline does not recreate the earlier situation.

Before accepting the evidence, trace several real affected interests through identification, loss, remedy and outcome measure. If those links cannot be followed, resolve the missing evidence while the relevant activity and relationships can still be observed.

Previous: Safeguard decisions · Overview · Next: Entitlements and livelihood restoration