Migration Propensity Index (MPI): A scoring tool to anticipate domestic and cross-border migration

Twelve simple questions that predict which households are most likely to migrate, for smarter development decisions

The Migration Propensity Index (MPI) is a household scoring system designed to estimate the probability of migration within the next 12 to 24 months. It comprises two complementary versions: the internal MPI (i-MPI), which anticipates migration within a country, and the external MPI (e-MPI), which anticipates cross-border migration. Each version uses a calibrated scoresheet containing 12 to 15 simple, non-sensitive questions on observable household characteristics, including demographics, dwelling conditions, assets, and previous migration events. Responses are scored and summed to produce a total between 0 and 100, which maps to an estimated migration probability and classifies households into low, medium, or high migration propensity groups. For each country, the questions and their weights are derived from household panel data, so that household conditions are observed before the migration decision takes place, and are selected through a procedure cross-validated across 300 random partitions of the data to guard against overfitting. For geographic analysis, the scoring rules can also be applied to household-level census or large-survey microdata to estimate migration propensity across geographic areas. Adapting the MPI to a new country requires calibration using suitable local household panel data.

This technology is pre-validated.

9•9

Scaling readiness: idea maturity 9/9; level of use 9/9

Positive impacts: 8

Target group	Positive impacts
Young adults in low-income rural farming households with limited employment opportunities	Programme access: MPI-informed targeting could improve access to relevant employment, training and livelihood programmes. Household visibility: Assessment could identify households overlooked by poverty or geographic targeting alone. Less sensitive questioning: Indirect assessment reduces the need to disclose personal migration intentions.
Women managing resource-constrained farms and households with migration experience	Targeting visibility: Household characteristics could help programmes identify families requiring further needs assessment. Livelihood support: Follow-up assessments could connect women to services addressing farming and household responsibilities. Accessible assessment: Paper or interviewer-administered questionnaires allow participation without owning a smartphone.
Low-income farming households exposed to drought and located far from services	Additional targeting information: Migration propensity can complement poverty and climate-exposure assessments. Geographic visibility: Mapping could help prioritize underserved areas where suitable data are available. Programme relevance: Combined assessments could inform livelihood and adaptation support suited to household needs.
Low-income periurban women and youth entrepreneurs in households with internal migration experience	Livelihood opportunities: MPI-informed programmes could connect eligible households to employment or enterprise support. Planning visibility: Geographic analysis could inform where livelihood services are offered. Lower disclosure burden: Household assessment avoids directly asking entrepreneurs about migration plans.

More...

Unintended impacts: 12
Mitigations: 12

Barriers: 8
Mitigations: 8

Climate adaptability: Highly adaptable

The MPI can be applied across different climatic conditions, provided suitable data are available. Climate-related factors, such as drought exposure, can contribute to predicting migration propensity in relevant contexts.

Farmer climate change readiness: Moderate improvement

Drought exposure during the previous cropping season is a predictor in the i-MPI. The index can help identify households with a higher propensity to migrate in drought-affected areas, informing where climate adaptation programmes are targeted.

Problem

Limitations of existing targeting approaches: Public programmes often rely on poverty or geographic criteria that do not necessarily identify households with a high propensity to migrate.
Insufficiently reliable information on migration intentions: Direct questions about migration intentions or attempts may generate refusals and under-reporting, while responses can be difficult to compare over time.
Risk of misallocating public resources: When households with a high propensity to migrate are not identified, public programmes may fail to reach them or adequately tailor interventions to their needs.
Gaps in identifying climate-exposed households at risk of migration: In contexts where drought contributes to migration, poverty-based targeting may overlook highly exposed households that are not among the poorest.

Solution

Identify migration risk more directly: Use observable household characteristics to estimate migration propensity instead of relying on poverty or geographic criteria alone.
Use a standardized assessment method: Apply 12 to 15 questions and a calibrated scoring system to classify households into low, medium, or high migration propensity groups.
Inform geographic prioritization: Apply MPI scoring rules to existing census or large-survey microdata to estimate migration propensity across geographic areas without collecting new data.
Better distinguish climate exposure from poverty: Identify migration risk that poverty-based targeting may overlook, as households most exposed to drought are not necessarily the poorest.

Key points to design your project

The Migration Propensity Index (MPI) can strengthen programmes addressing rural livelihoods, climate adaptation and youth employment by helping identify households with different levels of migration propensity. It complements conventional poverty-based and geographic targeting, providing additional information to guide resource allocation and intervention planning.

To successfully integrate the MPI into a project:

Assess the availability and suitability of existing household survey or census data before planning additional data collection.
Contact IFPRI for technical support with updates, adaptation or country-specific calibration where necessary, particularly when applying the MPI in a new national context.
Budget for data collection according to the selected approach. Adding MPI questions to an existing survey costs less than USD 2 per household, while a dedicated telephone survey costs approximately USD 2 to 6 per household. Using existing census microdata for geographic mapping avoids the need for a new survey.
Plan for continued use by leveraging the freely accessible MPI methodology and questionnaires. Ensure continued access to suitable data and technical support for updates or adaptation where required.

<2 USD per household

Estimated additional cost when MPI questions are integrated into an existing household survey.

2–6 USD per household

Estimated cost of administering the MPI through a standalone phone survey.

IP

Open source / open access

See details ›

Scaling Readiness describes how complete a technology's development is and its ability to be scaled. It produces a score that measures a technology's readiness along two axes: the level of maturity of the idea itself, and the level to which the technology has been used so far.

Each axis goes from 0 to 9 where 9 is the “ready-to-scale” status. For each technology profile in the e-catalogs we have documented the scaling readiness status from evidence given by the technology providers. The e-catalogs only showcase technologies for which the scaling readiness score is at least 8 for maturity of the idea and 7 for the level of use.

The graph below represents visually the scaling readiness status for this technology, you can see the label of each level by hovering your mouse cursor on the number.

Scaling readiness score of this technology

Maturity of the idea 9 out of 9

Uncontrolled environment: validated

Level of use 9 out of 9

Common use by intended users, in the real world

Maturity of the idea										Level of use
9
8
7
6
5
4
3
2
1
	1	2	3	4	5	6	7	8	9

Positive impact 8

Target group	Positive impacts
Young adults in low-income rural farming households with limited employment opportunities	Programme access: MPI-informed targeting could improve access to relevant employment, training and livelihood programmes. Household visibility: Assessment could identify households overlooked by poverty or geographic targeting alone. Less sensitive questioning: Indirect assessment reduces the need to disclose personal migration intentions.
Women managing resource-constrained farms and households with migration experience	Targeting visibility: Household characteristics could help programmes identify families requiring further needs assessment. Livelihood support: Follow-up assessments could connect women to services addressing farming and household responsibilities. Accessible assessment: Paper or interviewer-administered questionnaires allow participation without owning a smartphone.
Low-income farming households exposed to drought and located far from services	Additional targeting information: Migration propensity can complement poverty and climate-exposure assessments. Geographic visibility: Mapping could help prioritize underserved areas where suitable data are available. Programme relevance: Combined assessments could inform livelihood and adaptation support suited to household needs.
Low-income periurban women and youth entrepreneurs in households with internal migration experience	Livelihood opportunities: MPI-informed programmes could connect eligible households to employment or enterprise support. Planning visibility: Geographic analysis could inform where livelihood services are offered. Lower disclosure burden: Household assessment avoids directly asking entrepreneurs about migration plans.

Unintended impact 12

Target group	Unintended impacts	Mitigation measures
Young adults in low-income rural farming households with limited employment opportunities	Stigmatization: A high score could label young people as likely to leave or unreliable. Exclusion: Young people in low-score households could lose access despite substantial needs. Restricted choice: Programmes could treat preventing migration as the preferred outcome regardless of young people’s aspirations.	Confidentiality: Restrict access to scores and avoid public household labels. Needs-based eligibility: Combine MPI results with individual employment and vulnerability assessments. Personal agency: Offer voluntary support that respects informed decisions to stay or migrate.
Women managing resource-constrained farms and households with migration experience	Hidden inequalities: A household score could conceal women’s limited control over income and resources. Additional workload: Programme participation could increase women’s responsibilities without providing support. Privacy exposure: Disclosure of household migration information could create stigma or family tensions.	Individual assessment: Consult women directly about resource access, responsibilities and priorities. Participation support: Provide flexible schedules, accessible venues and childcare where needed. Data protection: Obtain informed consent and limit collection, access and sharing of identifiable information.
Low-income farming households exposed to drought and located far from services	Misclassification: An unsuitable or outdated model could misrepresent local migration propensity. Exclusion of immobile households: Highly vulnerable households unable to finance migration could receive low priority. Misattributed causes: Programmes could interpret high propensity as proof of climate-driven migration.	Local validation: Check predictive performance in the intended setting and reassess it over time. Separate vulnerability criteria: Preserve access based on need, including for households unable to move. Complementary diagnosis: Assess climate exposure and migration drivers separately before selecting interventions.
Low-income periurban women and youth entrepreneurs in households with internal migration experience	Financial discrimination: Lenders could misuse scores to infer poor creditworthiness. Household-level assumptions: Programme staff could assume that an entrepreneur intends to migrate because of the household score. Displaced priorities: Selection based on migration propensity could overlook business viability and actual support needs.	Purpose restrictions: Prohibit use as a substitute for credit assessment or as grounds for denying finance. Individual consultation: Establish the entrepreneur’s circumstances and plans through voluntary discussion. Enterprise assessment: Combine programme targeting with business needs, viability and inclusion criteria.

Barriers 8

Target group	Barriers	Mitigation measures
Young adults in low-income rural farming households with limited employment opportunities	Limited voice: The household respondent may not accurately represent young adults’ circumstances. Distrust: Young people may fear that participation will expose them to migration monitoring. Limited opportunities: Identification may not lead to useful support where employment services are scarce.	Youth consultation: Add a voluntary individual needs assessment alongside household scoring. Clear communication: Explain the assessment’s purpose, voluntary nature and permitted data uses. Service linkage: Establish practical referral pathways before using scores to recruit beneficiaries.
Women managing resource-constrained farms and households with migration experience	Time constraints: Farming, domestic work and care responsibilities can limit participation. Communication barriers: Literacy, language or limited phone access can impede interviews. Gatekeeping: Other household members may control responses or women’s access to programmes.	Flexible delivery: Offer short interviews at convenient times and locations. Inclusive interviewing: Use local-language oral interviews and in-person options without requiring personal devices. Direct participation: Provide safe opportunities for women to respond and access programme information directly.
Low-income farming households exposed to drought and located far from services	Coverage gaps: Remote households may be absent from surveys or poorly represented in existing data. Connectivity constraints: Weak networks and limited phone access can undermine telephone collection. Calibration requirements: Suitable panel data and funding may be unavailable for a locally appropriate index.	Coverage checks: Identify underrepresented locations and include them through appropriate sampling and outreach. Offline options: Use paper or offline digital interviews with field visits where necessary. Calibration planning: Assess panel-data suitability and secure analytical resources before household targeting.
Low-income periurban women and youth entrepreneurs in households with internal migration experience	Residential mobility: Frequent moves can make households difficult to contact or follow up. Interview costs: Interviews during working hours can interrupt sales and income generation. Weak programme linkage: Household scoring may not connect entrepreneurs to relevant business services.	Flexible contact: Offer multiple contact methods and update details with consent. Convenient scheduling: Arrange brief interviews outside peak business hours. Tailored referrals: Link identified needs to suitable enterprise, skills or employment services.

Costs and benefits

Open-source tools: The MPI methodology, questionnaires and tools are freely available.
Application costs: Less than USD 2 per household when integrated into an existing survey, or USD 2 to 6 per household through a standalone telephone survey. Geographic mapping using existing census or large-scale survey data can be conducted without additional household data collection.
Improved targeting: The MPI distinguishes households by their propensity to migrate, providing information that can help guide the targeting of development programmes and the allocation of resources.

Countries with a green colour: Tested & adopted
Countries with a bright green colour: Adopted
Countries with a yellow colour: Tested
Countries with a blue colour: Testing ongoing

Countries where the technology is being tested or has been tested and adopted
Country	Testing ongoing	Tested	Adopted
This technology has not been tested or adopted in any country.

This technology can be used in the colored agro-ecological zones. Any zones shown in white are not suitable for this technology.

Agro-ecological zones where this technology can be used
AEZ	Subtropic - warm	Subtropic - cool	Tropic - warm	Tropic - cool
Arid
Semiarid
Subhumid
Humid

Source: HarvestChoice/IFPRI 2009

The United Nations Sustainable Development Goals that are applicable to this technology.

Goal 1: no poverty

The MPI helps target social and economic programmes to vulnerable households. Household assets and dwelling conditions are among the predictors used to estimate migration propensity.

Goal 2: zero hunger

The MPI can inform the targeting of programmes addressing rural livelihood challenges, where food insecurity and agricultural shocks are associated with distress migration.

Goal 8: decent work and economic growth

The MPI helps direct livelihood and employment programmes by identifying households with different levels of migration propensity, particularly where limited local economic opportunities contribute to migration.

Goal 10: reduced inequalities

The MPI supports better targeting of development programmes and more equitable resource allocation by identifying households with higher migration propensity.

Goal 11: sustainable cities and communities

Migration affects communities of origin, with implications for rural livelihoods and community sustainability. The MPI can help identify households with higher migration propensity, providing information to target development programmes that address local livelihood challenges and support the sustainability of communities of origin.

Goal 13: climate action

The i-MPI uses drought exposure as a predictor, helping identify households with higher migration propensity in areas affected by climate shocks and informing the targeting of relevant interventions.

The MPI can be applied through a short household questionnaire using a calibrated scoresheet. The choice of application method depends on the available data and the purpose of the assessment.

Select the appropriate MPI component : Choose the internal MPI (i-MPI) for domestic migration or the external MPI (e-MPI) for international migration.
Administer the questionnaire: Collect responses to 12 to 15 questions covering household demographics, housing conditions, assets, and migration history. The questionnaire can be administered in person, by phone, or through digital survey tools, either independently or as part of an existing household survey.
Calculate the score: Apply the predefined weights to the responses to obtain a score from 0 to 100. Scoring can be performed manually using a printed scoresheet or automatically using a standard survey application.
Interpret migration propensity: Classify households as having low (green, below 50), medium (yellow, 50 to 70), or high (red, above 70) migration propensity. Use the corresponding conversion table to obtain the estimated probability of migration.
Analyse the results: Examine household scores to identify different levels of migration propensity. Where suitable household survey or census data are available, aggregate or map the scores to identify geographic patterns.