Assessing project-induced in-migration

An in-migration assessment asks how a proposed investment may change movement, settlement, livelihoods, and access to services. It establishes what is already happening, which project decisions could alter those conditions, and where further evidence or preventive action is needed.

The assessment needs to be proportionate to uncertainty as well as scale. A short construction programme in a well-serviced town may require targeted workforce and housing analysis. A remote mine, industrial estate, port, or transport corridor may require repeated demographic surveys, regional planning, and sustained institutional support. The initial screen determines the depth of inquiry; it cannot substitute for that inquiry.

Establish the decision and the project boundary

Start with the decisions still open to change. Obtain the construction schedule, workforce estimates by phase and contractor, procurement assumptions, accommodation alternatives, proposed recruitment arrangements, access roads, logistics facilities, and closure plans. Verify which information is approved and which remains provisional.

Under IFC Performance Standard 1, the area of influence includes, as appropriate, unplanned but predictable developments caused by a project that may occur later or elsewhere. This provides a basis for examining induced settlement effects beyond the immediate footprint (IFC, 2012, PS1, para. 8). It does not make every regional change attributable to the investment.

Use three nested geographies: the project facilities and activities; places where settlement or service pressures may materialise; and the wider labor market and regional development context. Revise these boundaries as evidence develops. State why a town, corridor, watershed, or service catchment is included, rather than adopting an unexplained radius.

Screen pathways, rather than counting workers alone

The following screen is a proposed assessment tool. An affirmative answer opens a line of inquiry; it is not a numerical risk score.

Some high-consequence concerns warrant specialist review even when expected arrivals are modest. Examples include settlement on hazardous land, child-protection concerns, severe water shortages, or an unresolved land conflict. A low total score cannot provide assurance where the mechanism and consequences remain credible.

Establish a baseline that can be revisited

A baseline needs a reference date, geographic units, population definitions, and a record of how the data were obtained. Describe usual residents, temporary residents, camp occupants, commuters, and seasonal users separately where they create different demands. A payroll is a workforce record; a school register is a service-user record. Neither represents the whole population.

The United Nations' 2025 census recommendations distinguish usual residence from presence at the time of enumeration and discuss frequently mobile populations. Align the project survey with the national statistical definition where comparison is intended, while retaining a separate count of temporary presence for service planning (UN Statistics Division, 2025, paras. 2.60–2.71).

Document at least four baseline domains:

1. Population and mobility: household composition, previous residence, arrival period, temporary absence, seasonal movements, and intended duration.
2. Livelihoods and access: employment, income sources, rent, food and transport costs, land and resource use, and barriers to opportunity.
3. Services and environmental conditions: actual water delivery, sanitation, waste collection, school and clinic capacity, housing safety, and ecosystem pressures.
4. Institutions and relationships: public responsibilities, budgets, customary authority, representation, grievances, and existing disputes.

For each dataset, retain the observation period, publication date, method, coverage, and known exclusions. A new publication may describe old conditions. Administrative boundary changes may make apparently comparable totals misleading.

Choose a sample that answers the question

For an area-wide estimate, develop a probability sample from an updated settlement and dwelling frame. Stratify where the research question requires comparisons: potential receiving settlements, areas outside direct project influence, rental neighbourhoods, and recognised temporary accommodation. Record selection probabilities and non-response so estimates can be weighted and their uncertainty assessed.

Purposive interviews are appropriate for learning mechanisms and reaching people absent from the frame. They do not become representative because many respondents were interviewed. Time-location sampling at markets or transport nodes can reveal mobility practices, but people who visit frequently have a higher chance of inclusion. Document that limitation.

Combine survey evidence with interviews and field verification. The SIA guidance treats assessment as iterative learning, including revision of initial assumptions through consultation and new evidence (Vanclay et al., 2015, Box 2 and associated guidance). In this application, disagreements between methods are findings to investigate: a stable census total and a rising rental population may describe different reference periods or populations.

Build scenarios with transparent components

Use scenarios to test decisions when forecasting is uncertain. A useful accounting structure for additional residents at time t is:

Here, W is newly resident project workers; D is accompanying household members; J is other newly resident job seekers; T is traders and service providers not already counted; O is other project-linked arrivals; and X is departures among these groups. Categories must be mutually exclusive. For a multi-period model, carry forward the resident stock and count new entries and exits in each period. Do not sum workforce positions over time as though every position represented a different person.

Construct lower, central, and higher scenarios using explicit assumptions about recruitment, accompaniment, residence duration, and opportunity-seeking. Treat them as plausible planning cases, not confidence intervals. Compare analogue projects only after considering labor intensity, transport, housing, social networks, public capacity, and project phase. There is no universally applicable worker-to-migrant multiplier.

Keep the population expected without the project separate. Regional urbanisation, another major investment, displacement, and natural increase may produce substantial growth in either scenario. A project can worsen a shortage that it did not originally create.

Translate scenarios into impact pathways

An impact statement identifies a trigger, a mechanism, an affected group or resource, a place, and a time. “Increased population causes social problems” provides none of this detail.

A more useful statement is: During peak construction, contractor rental demand may reduce the availability of low-cost rooms in Junction settlement, increasing the risk that existing tenants move to flood-exposed accommodation. Evidence is then needed on rental stock, affordability, displacement behaviour, and alternative locations.

Record likelihood, consequence, duration, reversibility, distribution, and confidence separately. Describe the criteria used. Do not multiply ordinal ratings and present the result as a probability or expected loss. Distinguish the impact before proposed controls from the residual impact after controls that are funded and demonstrably feasible.

Where several investments draw on the same settlement or water system, the assessment must consider cumulative pressure. IFC's cumulative-impact handbook offers a process for identifying valued environmental and social components, examining combined effects, and developing management responses (IFC, 2013). A shared water source or housing market is often a more useful unit of analysis than a separate headcount for each project.

Explain contribution without claiming certainty

An arrival after project announcement is temporally associated with the project. An interview describing a job expectation strengthens the account of a mechanism. A changed recruitment location followed by changed settlement patterns adds another form of evidence. None, alone, identifies the total causal effect.

Test rival explanations and retain an evidence ledger: observation; proposed interpretation; alternative explanation; confidence; and what would change the conclusion. For important effects, seek several independent sources. Gridded population estimates and built-up layers using the same underlying settlement data are not fully independent corroboration.

The final assessment should produce an actionable set of findings: what can change in design; what contractors must implement; what government capacity is needed; what requires specialist review; and what must be monitored. The spatial methods, management plan, and field tools provide the next steps.