Designing the socio-economic baseline: the instrument that constrains everything after it
Abstract
The socio-economic baseline is the binding constraint on a resettlement programme's entire evidentiary position: an impact not measured at baseline cannot be shown to have been mitigated at completion. This paper examines baseline design against the uses the data must later serve — eligibility, valuation, vulnerability identification, livelihood monitoring and grievance adjudication — and identifies the design errors that are structurally uncorrectable once enumeration is complete. It argues that baselines fail most often not through poor execution but through being scoped as a census of assets when they are required to function as a measurement of livelihoods.
1. Five uses, one instrument
The baseline is usually commissioned to establish who is affected and what they will lose. It is then required to serve four further purposes: identifying vulnerability, providing the pre-displacement measurement against which livelihood restoration is assessed, supplying the factual record against which grievances are adjudicated, and evidencing the affected population at the cut-off date.
An instrument scoped only for the first purpose will fail the others, and the failure surfaces years later when it cannot be remedied. This is the central design problem: the survey is designed by people focused on enumeration and consumed by people conducting monitoring, adjudication and audit.
2. Assets versus livelihoods
The most consequential scoping error is to enumerate assets without measuring income and its sources. An asset inventory supports valuation and nothing else. It cannot establish whether livelihoods were restored, because it never measured them.
Measuring income in smallholder economies is genuinely difficult — seasonal, partly non-monetary, spread across activities, and reported unreliably where the respondent believes the answer affects compensation. The difficulty is real and is not a reason to omit the attempt, since the alternative is an unfalsifiable restoration obligation. Proxy approaches, consumption measures and activity inventories all have weaknesses, and all are better than nothing measured at all.
3. Unit of enumeration
Surveying the household as an undifferentiated unit is administratively convenient and analytically destructive. It aggregates the interests of members whose losses differ, and once aggregated they cannot be separated. Women's cultivation of specific plots, tenants' use of land they do not own, and junior members' derived rights all disappear at this step.
The correction is cheap at design and impossible afterwards: record individuals within the household and attribute uses and interests to them. Every later analysis disaggregated by sex, age or status depends on it having been done.
4. Errors that cannot be corrected
Certain baseline defects are structurally uncorrectable. Anything not asked cannot be recovered, because the pre-displacement state no longer exists to be observed. Records without unique, stable identifiers cannot be reliably linked to later monitoring rounds, so a longitudinal comparison becomes impossible. Enumeration without timestamps cannot evidence presence at the cut-off date. And records without location data cannot be recomputed when a project footprint changes.
Each is trivial to build in and terminal to omit. A baseline reviewed against the five uses before enumeration begins is the cheapest risk mitigation available to a resettlement programme, and it is skipped more often than any other.
References
- [1]Performance Standard 5: Land Acquisition and Involuntary Resettlement. International Finance Corporation, 2012.
- [2]Guidance Note 5: Land Acquisition and Involuntary Resettlement. International Finance Corporation, 2012.
- [3]Good Practice Handbook: Land Acquisition and Involuntary Resettlement. International Finance Corporation, 2023.
- [4]Handbook Module 5: Livelihood Restoration and Improvement. International Finance Corporation, 2023.
- [5]Environmental and Social Standards (ESS). World Bank, 2018.