Policy Research Working Paper 10082
Understanding the Climate Change-Migration
Nexus through the Lens of Household Surveys
An Empirical Review to Assess Data Gaps
Marco Letta
Pierluigi Montalbano
Adriana Paolantonio
Development Economics
Development Data Group
June 2022
Policy Research Working Paper 10082
Abstract
Over the past two decades, the causal relationship between and contextual factors. Then, it discusses open issues and
climate change and migration has gained increasing prom- assesses the main data gaps that currently prevent more
inence on the international political agenda. Despite recent robust quantifications. Finally, the paper highlights oppor-
advances in both conceptual frameworks and applied tech- tunities for exploring these research questions, exploiting
niques, the empirical evidence does not provide clear-cut the potential of the existing multi-topic and multi-purpose
conclusions, mainly due to the intrinsic complexity of the household survey data sets, such as those produced by the
phenomena of interest, the irreducible heterogeneity of World Bank’s Living Standards Measurement Study. The
the transmission mechanisms, some common misconcep- paper focuses on the Living Standards Measurement Study–
tions, and, in particular, the paucity of adequate data. This Integrated Surveys on Agriculture program to discuss
data-oriented review first summarizes the findings of the potential improvements for integrating standard household
most recent empirical literature and identifies the main surveys with additional modules and data sources.
insights as well as the most important mediating channels
This paper is a product of the Development Data Group, Development Economics. It is part of a larger effort by the
World Bank to provide open access to its research and make a contribution to development policy discussions around the
world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may
be contacted at apaolantonio@worldbank.org.
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.
Produced by the Research Support Team
Understanding the Climate Change-Migration Nexus through the Lens of Household
Surveys: An Empirical Review to Assess Data Gaps*
Marco Letta
Sapienza University of Rome and Global Labor Organization (GLO)
marco.letta@uniroma1.it
Pierluigi Montalbano1
Sapienza University of Rome and University of Sussex
pierluigi.montalbano@uniroma1.it
Adriana Paolantonio
Development Data Group, The World Bank
apaolantonio@worldbank.org
Keywords: migration, climate change, household surveys, microdata, data gaps
JEL classification: C80; O12; O15; Q54
*We are grateful to Calogero Carletto and Alberto Zezza from the LSMS team for their continued support and invaluable
advice throughout the development of this work. We thank Mariapia Mendola for helpful comments and Giulia Altomare and
Carlotta Vicario for excellent research assistance in collecting background materials.
1
Corresponding author. E-mail: pierluigi.montalbano@uniroma1.it
1. Introduction
In the last couple of decades, the causal relationship between climate change and migration has emerged
as a central issue for both scholars and policy makers, receiving growing attention in the media and public
debate and gaining an important place among the policy priorities of the global agenda. Environment and
climate change have been formally recognized as key drivers of migration in the UN Global Compact
for Safe, Orderly and Regular Migration (GCM) endorsed by the UN General Assembly in 2018. The
Sustainable Development Goals (SDGs) explicitly include a specific target (10.7) to “facilitate orderly,
safe, and responsible migration and mobility of people, including through implementation of planned
and well-managed migration policies”. As a consequence of this increased prominence at the
international level, climate-induced human migration has become one of the main channels of interest
for the quantification of the socioeconomic impacts of future climate change. The common expectation
is that an increase in average global temperature of 2°C or more above pre-industrial levels would result
in substantially higher migration flows in the coming decades (Myers, 2002; Biermann and Boas, 2010),
a view which is endorsed by the recent Groundswell reports from the World Bank (Rigaud et al., 2018;
Clement et al., 2021).
Concerns about potentially dramatic increases in migratory flows triggered by large-scale climate-
induced environmental phenomena have spurred the birth of an empirical research field dedicated to
shedding light on the human mobility consequences of climate-related hazards. As a result, a growing
body of research has started to investigate the causal link between the two phenomena to satisfy the need
for reliable projections and provide robust empirical evidence on the issue. While common wisdom
assumes that climate change will progressively become one of the main push factors shaping migratory
flows in the coming decades, experts warn that a direct link between environmental factors and migration
is not easy to identify, and the empirical evidence does not provide clear-cut conclusions. The empirical
results differ depending on the environmental factors considered, the data and scale of the analysis, the
methodology employed, and the geographical contexts covered. Even within the same studies, estimates
of the size and direction of climate-related migration differ substantially (Hoffmann et al., 2020).
In this respect, lack of data has been the most severe constraint for a long time. Notwithstanding the
increasing availability of panel micro data sets, coupled with refinements of conceptual frameworks and
advances in econometric techniques, there are still many key questions to tackle and technical hurdles to
overcome: given the need to differentiate across types of migration (displacements, rural-urban flows,
2
international migrations), what are the information needs? And how to prioritize? What are the
opportunities offered by the household survey data collection? What would it be possible to achieve with
slight modifications to the current data collection instruments utilized by popular, long-established
household survey programs, such as the World Bank’s Living Standards Measurement Study (LSMS)?
This critical review aims to shed light on the issues above. Given the vastness of the topic and the huge
body of work in the field, a few preliminary clarifications on the scope of this review are in order: first,
this review has an explicit orientation toward the empirical and data perspective on the climate change-
migration links. Hence, we look at retrospective studies using historical data, we assume as given the
state of the art of the theoretical underpinnings of the current empirical investigation, and we do not
consider studies on projections or predictions of future migratory flows; second, since it is now widely
recognized that developing countries, which are both hotter and poorer, will be disproportionately
affected by climate change (Auffhammer, 2018; Dell, Jones and Olken, 2014; Nordhaus and Moffat,
2017; Tol, 2018), the focus of this work will be on migration originating within and/or from these
countries; third, we will focus on the more recent empirical advances. This means we limit our
exploration to studies and works published in the last decade (specifically, from 2010 on);2 fourth, in our
assessment of data gaps in household survey data, we will look specifically at the Living Standards
Measurement Study (LSMS) collection by the World Bank. The LSMS has been the World Bank’s
flagship household survey program since 1980, and it is currently a global leader in the methodological
development of rich and extensive multi-topic questionnaires designed to study various aspects of
household welfare. Within the broader LSMS program, we restrict our attention to the LSMS-ISA
household survey program, the explicit aim of which is to improve the availability, quality, and relevance
of agricultural data in multi-topic, multi-purpose household surveys (Carletto and Gourlay, 2019). The
reason is that the LSMS-ISA has a distinct advantage with respect to standard LSMS surveys when it
comes to studying issues specifically related to climate change, as its strong focus on agriculture allows
one to study the impacts of weather- and climate-related events on several household welfare outcomes,
that might either trigger or prevent migratory flows, as well as exploring farmers’ adaptation responses,
especially in situ ones, that complement migration as coping strategies in the face of climatic shocks.
Nonetheless, most of the reflections and recommendations we propose throughout this paper are also
relevant to LSMS surveys more broadly and to multi-topic and multi-purpose household surveys in
general.
2
For reference to the literature before 2010 see Piguet et al. (2011).
3
This work is organized as follows. Section 2 reviews the most recent empirical literature on the causal
relationship between climate change and migration. This section also clarifies important conceptual
distinctions, summarizes the key open research questions, and takes a closer look at Sub-Saharan Africa,
the area where the LSMS-ISA survey program currently operates. Section 3 presents the empirical and
methodological challenges and highlights the related data gaps. Section 4 takes a closer look at the
LSMS-ISA data collection and discusses its main limitations and possible use for empirical work. Section
5 suggests potential improvements and add-on modules to be integrated into future LSMS-ISA surveys.
Section 6 sets up an agenda for future work and concludes.
2. Review and current state of the literature
2.1 A synthesis of recent empirical findings
In this subsection, we carry out an updated review of the literature by focusing on the period 2010-2022.3
Overall, even the most recent literature still provides mixed and inconclusive evidence about not just the
magnitude, but even the sign of the climate-migration relationship. Heterogeneity, depending on several
context-specific features including, inter alia, the type and frequency of climatic shocks, peoples’
resources and adaptation strategies, seems to be an irreducible aspect of this delicate matter, that rules
out blanket generalizations on the nexus between climate change and (im)mobility. Such complexity
calls for caution in issuing predictions that climate change will force tens of millions of people to move
within and/or out of their own country (Rigaud et al., 2018; Clement et al., 2021) and to pay more
attention to the potential of climate change in preventing voluntary migration and trapping more
vulnerable and poorer populations in immobility.
By analyzing this flourishing strand of the literature, we flesh out some thematic issues that we see as
pivotal for the subsequent analysis of the empirical issues and data gaps. Specifically, we look at recent
insights into five key facets of the climate–migration relationship as they have emerged in the literature:
slow- vs fast-onset events (United Nations Framework Convention on Climate Change, 2012; Bohra-
Mishra, Oppenheimer, and Hsiang, 2014); direct and indirect links (Bardsley and Hugo, 2010); internal
3
Our survey is a data-oriented synthesis of selected recent findings relevant for the study of the potential of household survey
data. For general and more comprehensive reviews and meta-analyses of the literature on the climate-migration relationship,
we recommend readers to consult Cattaneo et al. (2019), Kaczan and Orgill-Meyer (2020), and Hoffmann et al. (2020).
4
vs international migrants; liquidity constraints; and migration as adaptation.4
A. Fast-onset events versus slow-onset changes
The migration impacts of fast-onset extreme weather events (such as hurricanes, heavy rains, floods, and
landslides) related to climate change are usually sudden and direct, resulting mainly in temporary
movements over short distances (McLeman and Gemenne, 2018). A specific strand of the literature
investigated the role of floods on migration, with mixed results: while Gray and Mueller (2012a) and
Bohra-Mishra, Oppenheimer, and Hsiang (2014) document a lack of significant effects of floods on
migration in, respectively, Ethiopia and the Philippines, Mueller et al. (2014) find that floods reduce the
probability of migrating in Pakistan. Koubi et al. (2016) show that in Vietnam fast-onset shocks are more
likely to trigger migration, whereas long-term environmental changes (such as salinization) reduce the
likelihood of migration. Robalino, Jimenez, and Chacón (2015) provide evidence that flooding and other
hydro-environmental emergencies increase migratory flows in Costa Rica, but also find that more severe
emergencies decrease migration. Kaczan and Orgill-Meyer (2020) argue that this may happen as floods
and other fast-onset shocks rapidly deplete household assets and resources, leaving households unable to
migrate. Using a global data set which combines climatic, census, and night-light data, Castells-Quintana,
McDermot and Krause (2021) study the relationship between changes in weather patterns and the spatial
distribution of population and economic activity, and find that worse climatic conditions are associated
with higher urbanization. Their results suggest that while slow-onset changes in climate may lead to more
permanent, and more long-distance, movements, sudden-onset events are primarily associated with
temporary displacements and short-term migration to nearby urban areas. Koubi et al. (2022) use survey
data from Cambodia, Nicaragua, Peru, Uganda, and Vietnam and document that less educated and lower-
income people are less likely to migrate after exposure to fast-onset shocks, compared to people endowed
with higher education and more economic resources. Finally, even in the absence of sudden resource
depletion, in poor countries the populations affected may not have enough monetary resources for long-
distance migration (Zickgraf and Perrin, 2016). Cattaneo et al. (2019) emphasize that the main current
insight of this strand of the literature is that the potential for fast-onset events to cause long-term, long-
distance migrations appears limited, especially in the case of costlier international migration.
In contrast, slow-onset changes are more likely to induce migration than rapid-onset ones, but the
4
These five thematic subsections have been selected among many topics, due to their relevance and prominence for the
assessment of data gaps we conduct.
5
literature has paid less attention to the migration outcomes of slow-onset changes compared to sudden
disasters (Kaczan and Orgill-Meyer, 2020). Part of the reason is that slow-onset changes are not regarded
as sufficiently extreme to trigger migration, since they have less of an immediate impact on people
(Koubi et al., 2022). The effect of events such as drought, desertification, and warming on migration is
less sudden than floods, landslides, hurricanes and similar, because they tend to emerge gradually, and
attribution is intrinsically more difficult for departures in response to gradual changes. Moreover,
migratory flows can be staggered, more difficult to capture and more susceptible to measurement error.
There is a body of works documenting that slow-onset changes and rising temperatures increase
migration (Bohra-Mishra et al., 2017; Cai et al., 2016; Dallmann and Millock, 2017; Dillon, Mueller,
and Salau, 2011; Feng, Krueger, and Oppenheimer, 2010; Gray and Mueller, 2012a, 2012b; Hunter,
Murray, and Riosmena 2013; Jessoe, Manning, and Taylor 2018; Mastrorillo et al. 2016; Mueller et al.,
2014). Many of these studies are discussed in other sections of the paper. However, there are also some
notable exceptions in this literature showing that slow-onset changes may also have the opposite effect,
resulting in a reduction in migration. For example, Cattaneo and Peri (2016) find that, consistent with
the presence of severe liquidity constraints, higher temperatures reduce the probability of migration from
poor countries. The results of Cottier and Salehyan (2021) and Martinez Flores, Milusheva and Reichert
(2021) on the role of droughts in influencing international migration support this conclusion. From a
within-country perspective, Hirvonen (2016) shows that temperature anomalies decrease (male)
migration in rural Tanzania because of tightened liquidity constraints. Liu et al. (2022) study responses
to slow-onset temperature changes in Indian districts, and conclude that progressive warming and rising
temperatures inhibit structural transformation and limit rural-urban migration for households living in
isolated areas. Other authors emphasize that, in most cases. migration depends much more on political
and economic factors and is only minimally associated with slow-onset changes. For example, Selby et
al. (2017) find that environmental variables have a marginal role in explaining migration flows to the
Syrian Arab Republic in the period before the outbreak of the civil war. Similarly, Niva et al. (2021)
performed a geospatial analysis of a gridded global net migration data set for the decade 1990-2000, and
found that slow-onset changes, such as droughts and water scarcity, were the dominant environmental
events in explaining net-migration. Yet, they also emphasize that income levels and adaptive capacity
crucially mediate environmental variables in determining migration outcomes.
In short, while there seems to be a prevailing consensus that slow-onset shocks do increase migratory
flows, there are cases in which the opposite happens, and affected populations are instead trapped in
6
immobility because of the negative consequences of such shocks on their liquidity. This should come as
no surprise: since the determinants of migration, and especially climate-related migration, are complex
(IPCC, 2014), it is expected that the sign of the relationship cannot be known a priori, as the direction
will depend on local circumstances and is ultimately an empirical question.
B. Direct versus indirect effects
Climate change could exert both direct and indirect effects on migration. In the latter case, one must
understand how climatic events affect other drivers of migration, via demographic, socio-economic and
political channels. Among the main channels of the indirect effects of weather- and climate-related
hazards on migration, the two most important ones are the economic and socio-political drivers. To date,
however, there is only limited and partial evidence about these links. Feng, Krueger, and Oppenheimer
(2010) provide empirical support about the links between extreme temperatures, crop yields, farmers’
income, and migration from Mexico to the United States. In an important study, Marchiori, Maystadt,
and Schumacher (2012) leverage annual panel data from 1960-2010 and find that in Sub-Saharan Africa
weather shocks boost rural-urban migration through a decrease in rural wages. In turn, this weather-
driven internal migration into cities brings about downward pressure on urban wages, causing urban
workers to move internationally, a mechanism they label the ‘economic geography channel’. Similarly,
Dallmann and Millock (2017) find that drought effects on inter-state migration in India are stronger in
agricultural states. On top of these microeconomic, household-specific income channels, there is also a
macro-level income driver: developing countries are particularly vulnerable to climate-related hazards,
because they have a large share of their income in agriculture (the most weather-dependent sector), tend
to be hotter and closer to biophysical limits, and lack adaptive capacity to cope with the negative impacts
of climate change. For example, using bilateral migration data from 1980 to 2010, Cai et al. (2016)
demonstrate that temperature shocks induce international migration only from agriculture-dependent
countries and increase migration to OECD countries. Investigating the role of climatic factors in
engendering international migration, Beine and Parsons (2015) find evidence of an indirect channel
operating through wages. Niva et al. (2021) find that income is a key determinant in explaining both net-
negative and net-positive migration, and conclude that it is the difference between income-levels of the
origin and destination areas that matters, rather than income level per se. Other than affluence, the main
socio-political factor investigated in the specialized literature is the well-known (and often prominent in
the media) conflict channel. For example, Kelley et al. (2015) suggest that a prolonged and unprecedented
drought in parts of Syria exacerbated the (pre-existing) vulnerability of the affected population,
7
prompting them to migrate. According to their view, migration, in turn, increased tensions and
contributed to the outbreak of the civil war. Note, however, that there are sharp disagreements about this
possible causal role in the Syrian civil war played by climate change via migration (Fröhlich, 2016; Selby
et al., 2017) and that, more generally, the relationship between migration, climate change, and conflict is
particularly complex and context-specific.
C. International versus internal migration
Previous research paid more attention to weather- and climate-related international migration due to the
paucity of internal migration data in developing contexts (Laczko and Aghazarm, 2009). However, the
picture has changed in the last decade, as we are witnessing an increasing number of studies focusing on
the relationship between weather shocks and short- or long-distance within-country movements,
sometimes even comparing migration outcomes across multiple types of destinations. Gray and Mueller
(2012b) study the effects of natural disasters in Bangladesh and find that these are stronger and more
significant for local movements than long-distance outmigration. Jessoe, Manning, and Taylor (2018)
show that extreme heat events in Mexico boost rural-urban internal migration as well as cross-border
migration to the United States. Hirvonen (2016) finds that, in rural Tanzania, temperature increases
reduce internal migration via increased liquidity constraints implied from the estimated negative
consumption shock, but detects this effect only for men. Peri and Sasahara (2019), using a gridded global
data set covering the period 1970-2000, find that progressive warming reduces rural-urban migration in
poorer countries but increases it in middle-income countries. Gray and Bilsborrow (2013) find that
adverse rainfall conditions reduce local, short-distance migration (i.e., moves within the same canton5)
and international migration in Ecuador, but increase internal long-distance (between-canton) migration,
pointing to highly heterogeneous impacts concerning the type of migration. Nawrotzki, Riosmena, and
Hunter (2013) find a positive and statistically significant relationship between weather anomalies and
international migration from Mexico to the United States. Liu et al.’s (2022) work on India also shows
that increasing temperature can reduce internal rural-urban migration. Cottier and Salehyan (2021)
employ temporally disaggregated data on the detection of unauthorized migrants at the EU’s external
borders and report that droughts in origin countries do not increase international migration towards the
European Union but, if anything, they reduce it, in particular for countries dependent on agriculture,
whereas more rainfall increases migration. Their interpretation of the results is that international
5
Cantons are the second-level subdivisions of Ecuador, below the provinces.
8
migration is cost-prohibitive and that adverse weather shocks amplify liquidity constraints. Similarly, the
study by Schutte et al. (2021) on the association between climatic conditions and asylum migration
reveals that temperature anomalies are weak predictors of bilateral asylum migration to the European
Union, and concludes that future asylum migration will mainly be driven by political changes rather than
by climate change. This is in line with the findings of Martinez Flores, Milusheva and Reichert (2021)
for West and Central Africa, who estimate that a standard deviation decrease in soil moisture leads to a
25-percent reduction in the number of international migrants, likely due to liquidity constraints. Bekaert,
Ruyssen and Salomone (2021) leverage individual-level data from Gallup World Polls conducted in 90
countries to show that exposure to environmental stressors increases the probability of intending to
migrate both domestically and internationally (but more so intra-regionally), especially in rural and less
developed regions.
In short, there seems to be wide heterogeneity in the type and destination of migratory flows triggered
by climatic shocks, depending on a variety of factors including, among others, the severity and frequency
of the shock, the gender and resources of the affected individuals, and the context of the case study.
Despite such heterogeneity, however, meta-analyses of the existing studies have found evidence that the
case for weather-related migration is much more compelling for internal, within-country movements than
for international cross-border flows (Hoffmann et al., 2020), a finding which contradicts the
‘conventional wisdom’ about large-scale international migration triggered by climate change. Lastly, an
important missing piece, indeed also due to data gaps, is the paucity of studies reconstructing a potential
‘climate migration chain’ that might be triggered, directly or indirectly, by climatic changes in the
original affected areas. It is not implausible to imagine a scenario in which a future increase in the
frequency or intensity of weather shocks in a rural developing context will determine rural-urban internal
migration which, in turn, gives rise to international, cross-border movements of urban workers due to
changes in local labor markets. Marchiori, Maystadt and Schumacher (2012) show that a similar
‘economic geography’ mechanism had already taken place in Sub-Saharan African countries in past
decades. If and how much this kind of climate migration chain will become more relevant as climate
change intensifies, for the moment remains speculation.
D. Heterogeneous strategies and the role of liquidity constraints
A crucial insight from the most recent literature is that climate-related hazards can cause or worsen
liquidity constraints (Bryan, Chowdhury, and Mobarak, 2014; Cattaneo and Peri, 2016; Cottier and
9
Salehyan, 2021; Hirvonen, 2016). It is precisely for this reason that the capacity for migration in response
to climatic shocks is much more limited than commonly assumed: poor people, who are
disproportionately affected by climate change, have more incentives to migrate but often cannot leave
because they lack the necessary resources (Cattaneo et al., 2019). In this perspective, migration is a costly
investment in risk diversification that only richer households can undertake, while poorer households are
“forced to stay” rather than “forced to move” (Kaczan and Orgill-Meyer, 2020). Depending on the
interaction between the severity of the climatic event and household-specific characteristics such as
wealth, number of income sources, assets and resources, we should expect a wide heterogeneity of
outcomes depending on whether liquidity constraints or migratory responses ultimately prevail. Indeed,
the empirical evidence is mixed, with some studies showing that the poorest households are more prone
to migrate in response to weather shocks (Gray and Mueller, 2012b; Mueller, Gray, and Kosec, 2014;
Mastrorillo et al., 2016), while others find that liquidity constraints trap poor people in immobility
(Cattaneo and Peri, 2016; Hirvonen, 2016; Bazzi, 2017). Cattaneo et al. (2019) argue that this
contradictory evidence can be explained by the different types of migration involved: poor families
respond to negative shocks through low-return or even “survival” migration, taking the form of
temporary movements across short distances, whereas wealthier families engage in risk management
migration, which is typically costlier, semi-permanent and longer-distance migration. Koubi et al. (2022)
suggest that (im)mobility depends on both the type of the climate shock that individuals experience and
their adaptive capacity in terms of endowments and resources. The simulations by Choquette-Levy et al.
(2021), who parameterize an agent-based model on household survey data from Nepal, support this
perspective, but also highlight that cash transfers and risk transfer mechanisms may prevent climate-
induced immobility of farmers.
In short, the final outcome is ultimately household- and context-specific. In such a complex picture,
migration and immobility are only two of the possible outcomes, and the decision to leave can be
interpreted as one of many potential adaptation strategies that an individual or household can adopt. As
such, there is a need to improve our understanding of the causes and consequences of the immobility of
populations ‘trapped’ by environmental disasters, because too often the policy focus is on ‘those who
leave’ rather than to ‘those who cannot’ (Findlay, 2012).
E. Migration as adaptation
One of the most interesting advances in the scientific literature is the progressive integration of the issues
10
of migration and adaptation into a single, unified conceptual framework. In such a framework, the
decision to migrate constitutes one of the possible strategies for adapting to climate change. There is
some empirical support for the notion that migration is a subset of decision options within the broader
issue of adaptation and coping strategies in response to shocks (Black et al., 2011; Alam, Alam, and
Mushtaq, 2016; Kattumuri, Ravindranath, and Esteves, 2017; McNamara et al., 2018). On the other hand,
migration could be a last-resort solution for households, because it is perceived as costlier than other in
situ adaptation strategies (Wodon et al., 2014). While some studies suggest that migration and on-farm
adaptation can indeed be substitutes, there is a dearth of sequential analyses assessing whether the
migration decision happens before or after the implementation of alternative adaptation options (Cattaneo
et al., 2019). The issues of immobility and trapped populations can also be viewed as inability to adapt,
and should be examined through an adaptation lens within an integrated framework, but research to date
remains scarce and fragmented. Among the few exceptions, Martinez Flores, Milusheva and Reichert
(2021), who find that only people living in middle-income areas are less likely to migrate abroad after a
drought (but not people living in wealthy or poor areas), argue that this is evidence that people, who
under normal climatic conditions would be able to migrate, are not able to invest in adaptation
mechanisms such as migration, thus sinking into poverty. Migration is just one of the many options
among potential coping strategies farmers can employ to mitigate the effects of climate change. The
literature has thus ignored for too long that migration is only one of many potential responses to
environmental stress and that, consequently, has to be analyzed against the background of other adaptive
options, which can either complement or substitute migration (Hoffman et al., 2020). Therefore,
understanding how migration fits into this larger pool of coping strategies, and the temporal and causal
dynamics of the migration-adaptation nexus, should be considered as research priorities.
A key take-home message that emerges from this overview is that, despite a growing body of research,
there are still substantial gaps in our understanding of the complex and multi-faceted relationship
between climate change and migration, and thus areas where further research is needed. Nevertheless,
the nuances and differentiations revealed by this body of recent research question the conventional and
simplified narrative that climate change will bring about permanent mass migration (Findlay, 2012).
2.2. Other open conceptual issues
Although not considered as part of the core analysis, it is worth recalling here a set of further areas of
research that are also important to shed light on the global picture. First, the issue of future projections.
11
We should recall here that there is still a huge uncertainty surrounding future projections, both related to
the severity of climate change and to the magnitude of future international and domestic migration flows,
and more research is needed to improve the existing forecasting models. Although important, this issue
remains beyond the scope of the current review that assumes as given the current state of the art on
economic modeling of migration flows. Second, weather vs climate. A too often neglected fact in this
literature is that short-term responses to climatic drivers differ from long-term responses. Weather shocks
and climate change are not equivalent: the first are short-run fluctuations, the latter refers to permanent
and long-run changes in weather patterns over time (Auffhammer et al., 2013). In turn, responses to
weather shocks can differ from responses to climate change for two reasons: intensification effects and
adaptation (Dell, Jones and Olken, 2014). Unmitigated levels of climate change (such as, for instance,
4°C of global warming above pre-industrial levels by 2100) would bring weather shocks well beyond
those experienced in historical records. Such extremes would greatly limit in situ adaptation options
(Gemenne, 2011). There is thus an urgent need for studies that try to fill the gap between adaptation
responses to weather anomalies, which to date have been the predominant focus of the empirical
literature, and responses to long-run and permanent changes in climate. The issue is further complicated
by the fact that observed changes in short-term weather patterns are themselves a manifestation of gradual
climate change. In this respect, a potential bridge is offered by case studies analyzing the impact of
increases in the frequency of natural disasters and the way populations respond to gradual warming as
well as to the risk of cumulative shocks (Cattaneo et al., 2019). To date, in fact, there have been very few
such works because of a dearth of data (a point to which we return below). But this kind of medium/long
run analyses, drawing on longitudinal data, is essential because it allows one to study how people respond
to progressive warming and permanent shifts in climatic conditions, thus reducing the external validity
gap with respect to climate change. Key issues in the more general debate on the impacts of climate
change, such as non-linear effects, tipping points and critical thresholds, are currently not addressed in
the existing literature, where studies usually focus on locally linear approximations of the underlying
non-linear relationship (Hoffmann et al., 2020).
Third, and related, research is scant in other key channels of the climate change-migration link such as
sea level rise (SLR), a phenomenon which is often prominent in the media and popular debate but still
scarcely studied, even though it will certainly be a key driver of climate-related migration, with
projections of 0.26–0.98-m mean sea level rise by 2100 (IPCC, 2014). Africa, and Sub-Saharan Africa
in particular, is considered to be particularly at risk of coastal flooding, due to the combination of
population growth and accelerating urbanization in coastal zones (Neumann et al., 2015a), and because
12
the likelihood of protection being successfully implemented is low (McMichael et al., 2020). We did not
include sea level rise in our review because much analysis on this topic makes use of modeled projections
of exposure rather than retrospective empirical analyses using historical data (Neumann et al., 2015b;
Davis et al., 2018).6 Fourth, future research should look at alternative outcomes, such as survival or risk
management migration, voluntary vs involuntary migration, jointly rather than separately, to improve
our understanding of the response heterogeneity with respect to wealth and income. More generally,
gaining a more systematic understanding of the irreducible heterogeneity of the climate-migration nexus
should be considered as a primary task for future research (Hoffmann et al., 2020). Fifth, more research
is needed on the role played by institutions and policies in ‘interfering’ with the decisions to migrate in
response to climatic stress. Development policies can affect migration outcomes in a way which is
difficult to know a priori, as they could either facilitate or inhibit migration depending on the type of
intervention and the related welfare outcome. For instance, while local investments in climate-resilient
infrastructures or in the development of early-warning systems may reduce the need to migrate and
improve in-situ adaptation, social protection interventions or emergency responses alleviating weather-
induced liquidity constraints may make voluntary migration possible. This is especially important in
order to provide evidence-based recommendations on national and international climate migration
policies.
2.3. A closer look at Sub-Saharan Africa
Finally, bearing in mind the major insights from the main review, we take a closer look at recent studies
with an exclusive focus on Sub-Saharan Africa (SSA), one of the parts of the world indisputably more
vulnerable and exposed to climate change (IPCC, 2014) and the area where LSMS-ISA surveys (which
we use as benchmark in the assessment of data gaps in household survey data to understand the climate-
migration nexus) are currently implemented.
A work by Lilleør and Van den Broeck (2011) in the northern highlands of Ethiopia revealed that
environmental stress shapes migration primarily through impacts on household production. Di Falco,
Veronesi and Yesuf (2011) carry out a study based on a survey conducted on 1,000 farm households
located within the Nile Basin of Ethiopia in 2005. They find that about 58% and 42% of farm households
had used no adaptation strategies in response to long-term changes in temperature and rainfall,
respectively, and that migration is one among many adaptation strategies, adopted by less than 5% of
6
For a recent review of the literature on population exposure to sea level rise and migration, see McMichael et al. (2020).
13
those surveyed. Through a longitudinal household survey, Gray and Mueller (2012a) studied the
consequences of drought on population mobility in Ethiopia’s rural highlands, providing evidence that
drought increases long-distance and work-related relocation of men, especially in land-poor households.
However, severe drought reduces women’s short-distance and predominantly marriage-related mobility.
Another study by Karanja Ng’ang’a et al. (2016) shows that climate change influences the livelihoods of
shepherds in arid and semi-arid lands in Kenya. By analyzing data from a survey of 500 rural households
in northern Kenya that relates adaptive family behavior to family migration, their analysis suggests that
migration and local innovation are complementary mechanisms to ensure resilience to adverse shocks.
In addition, families with at least one migrant member can employ high-cost agricultural innovations
through remittances, thus improving their self-protection against weather shocks. Mueller, Gray and
Hopping (2020) use census data on the migrations of 4 million individuals over 22 years to estimate the
climate effects on migration in Botswana, Kenya, and Zambia. Their results for Kenya show that
temperature had limited effects on migration, whereas a one standard deviation increase in precipitation
caused a 10% reduction in migration. In Botswana, mobility decreases by 19% with a one standard
deviation increase in temperature, and an equivalent change in rainfall causes an 11% decrease in
migration. The effects of temperature appear more severe among poorly educated individuals. Rainfall
shocks increase mobility in Zambia, while an increase in temperature does not affect mobility in the
region. Decreases in inactivity and unemployment coincide with increases in migration, which suggest
that the perspective of new job opportunities may act as a driver of climate-induced migration.
Mastrorillo et al. (2016) also argue that agriculture may function as a primary channel through which
adverse weather conditions influence migration. They combine South African census data with climate
data on spatiotemporal weather variability to examine South African bilateral inter-district migration
flow patterns and determinants during the periods 1997-2001 and 2007-2011. The results reveal that
precipitation scarcity and higher temperatures act as push effects for migration. However, the importance
of the effect of climate on migration varies greatly depending on migrant characteristics, including
ethnicity. In particular, the flows of black and low-income migrants in South Africa are strongly
influenced by climate variables, while white and high-income migrants are weakly or not affected.
Focusing on Uganda, Agamile, Ralitza and Golan (2021) assess gender-differentiated reactions of
smallholder farmers to droughts, finding that adverse weather conditions are an occasion for women to
enter the commercial crop market by exploiting land from subsistence for income-generating crops, while
relatively wealthier and better-educated people, especially men, are among those who benefit most from
14
the migration alternative. Beegle, Joachim and Stefan (2011) document that precipitation anomalies
increased both the probability of people leaving the village and the distance moved in Northern Tanzania.
Mueller et al. (2020) combined NASA’s high-resolution climate data with longitudinal microdata on
migration, labor participation, and LSMS-ISA data (see also Section 4), to test whether climate variability
affects temporary migration to rural and urban East Africa and whether climate-induced migration
coincides with a lack of local job opportunities. The data included surveys conducted in Ethiopia,
Malawi, Tanzania, and Uganda over six years (2009–2014). They found that climate variability
significantly affects temporary migration decisions in eastern Africa, specifically that temperature and
rainfall shocks cause a reduction in temporary urban out-migration. Mueller et al.’s (2020) findings are
consistent with the results of Hirvonen (2016) for rural Tanzania and challenge the narrative that
temporary migration acts as a safety valve in response to climatic push factors.
Grace et al. (2018) found that rainfall did not affect temporary migration rates in two Malian villages.
The authors combine unique data from highly detailed stories of migration collected over 25 years in two
rural communities in Mali, and document that a poor rainy season is not correlated with extreme or even
above-average emigration rates. Even accounting for some known sources of variability (age, gender,
etc.), a decrease in rainfall does not directly lead to a higher emigration rate. Instead, the results suggest
that during low-rainfall years outmigration is lower. Henderson et al. (2017) estimate the effects of
climate variability and change on African urbanization patterns over two different temporal and spatial
scales: i) local, within-district urbanization for an unbalanced 50-year panel of census data from 359
districts in 29 countries; ii) urbanization patterns from 1992 to 2008 in 1,158 cities. Their estimates show
that climatic conditions do affect urbanization rates, with better conditions delaying urbanization and
adverse conditions leading to faster urban population growth, but that these effects are confined to a
subset of about 20%-25% of Sub-Saharan African districts.
In a study already mentioned above, Martinez Flores, Milusheva and Reichert (2021) leverage high-
frequency (daily) migration data on the place of origin of migrants and the time of migration, collected
from the International Organization for Migration in 17 West and Central African countries over the
period 2018-2019, and estimate that droughts, as measured by soil moisture anomalies during the
growing season, strongly reduce the number of international migrants. As they detect these effects only
in middle-income areas but neither in rich nor poor ones, they conclude that climate-induced liquidity
constraints and income losses are the key mediating channels.
Finally, in a recent working paper, Di Falco, Kis, and Viarengo (2022) exploit LSMS-ISA panel data
15
from Ethiopia, Malawi, Niger, Nigeria, and Uganda (see also Section 4) combined with high-resolution
precipitation data to study the effects of cumulative climate shocks on long-term migratory flows in Sub-
Saharan Africa. Overall they find evidence of a persistent impact of droughts on rural households in these
countries, which translates into a much larger effect on migration compared to the period in the aftermath
of the shock as the impacts accumulate over time. The authors also detect the existence of a relationship
between rainfall shortage and accelerating urbanization trends in four of the five countries considered in
their analysis. At the country level, their findings contrast with previous studies that use similar multi-
country micro-level data sets to examine the effect of climate shocks on rural out-migration, which found
no significant or consistent migration-inducing effect of droughts in the short- and long-run. Conversely,
the authors notice that their results are in line with those of macro-level cross-country studies that
corroborate the contribution of rainfall deficits to faster urban development.
Overall, this review confirms that the relationship between migration and climate change in Sub-Saharan
Africa is far from univocal, with some studies considering migration as a direct consequence of weather
shocks and climatic changes, and others that do not find that these factors exert a clear or significant
impact on people’s mobility. The majority of the SSA literature focuses on slow-onset rainfall and
temperature events, whereas only a few studies specify the type of migration. As an aside, we notice that
there is unequal country coverage in this literature, with repercussions in terms of external validity. For
instance, the Sahel region is particularly underrepresented, despite being one of the areas identified
among the hotspots of climate change (IPCC, 2014).
3. Open empirical issues and data gaps
3.1 Open empirical issues
From the conceptual discussion above, a number of key open empirical issues stem:
▪ How to disentangle, from an empirical point of view, short-run elasticities of weather shocks on
migration from the compounding effects of slow-onset, long-run, eventually permanent changes
in climate and progressive warming? Most of the current works only investigate the short-run
weather effects of migration and then extrapolate with respect to climate change. But this poses
the problem of the already mentioned external validity gap between weather shocks and climatic
change. To empirically investigate the latter, one needs to look either at longer time series in a
longitudinal setting, using several lags of the weather parameters or long-run (e.g., 30-year,
16
which corresponds to the agreed definition of ‘climate’) averages, or to the cumulative effect of
many repeated weather events driven by an increase in frequency linked to climate change.
There are some promising approaches in the literature in this respect. For example, Cattaneo
and Peri employ a technique which is now quite common in the new weather-economy literature
(Dell, Jones and Olken, 2014; Burke and Emerick, 2016; Liu et al., 2022) called ‘long-
differences’. This approach consists of replacing annual averages of both the dependent and
independent (climatic) variables of interest with decadal or multi-decadal averages of the same.
This allows one to test whether the short-run relationships retrieved using annual measures also
hold in the medium- and long-run, thus directly testing the external validity of the empirical
findings in a climate change perspective. Cattaneo and Peri (2016) cleverly use long-differences
to confirm their short-run result that warming in poor countries reduces migration (consistently
with the presence of severe liquidity constraints) and find evidence in support of the persistence
of this type of effect.
▪ As explained in the previous subsection, it is now well established that the causal link between
the two phenomena is not as simple and straightforward as it once seemed to be, and there are
many intervening factors (liquidity constraints, assets, in situ adaptation strategies such as
irrigation or other on-farm investments, etc.) that have the power to alter not just the magnitude,
but even the sign of the relationship. Once the analyst retrieves a statistically significant
relationship between a climate shock and the decision to migrate, how can the effect be explained
in light of the above-mentioned mechanisms? Recent advances in empirical micro-econometrics,
such as, in particular, mediation analysis to investigate the mediating role of a variable of interest
(the so-called ‘mediator’) in explaining a causal relationship of interest appear promising,
although not yet picked up by scholars in the migration field.7 The analysis of migration-as-
adaptation, i.e., of how the decision (not) to migrate fits within a broader analytical framework
on the full pool of adaptation options available to farmers in response to climate change, is also
still underexplored. In this respect, while the ‘rare event’ nature of migratory flows is a drawback
for household surveys, the potential to carry out empirical analyses of the full range of intervening
factors, mediating channels, and adaptation options available to households in response to
7
Recent examples of works leveraging the potential of this methodology to unpack the black box of causal relationships in
the development field are Azzarri et al. (2022), Pace et al. (2022), and Prifti, Daidone, and Davis (2019). For a comprehensive
review of the use of mediation analysis in economics, see Celli (2021).
17
climatic stressors using the extensive data embedded in multi-topic and multi-purpose
representative surveys should not be understated.
▪ We also highlighted that migration is not the only possible outcome of adverse weather events.
From an empirical perspective, identifying people trapped in immobility in the aftermath of a
shock is even more difficult than tracing induced mobility, especially in the absence of specific
information on this provided by the affected household or individual.8 While one can separately
investigate effects on a welfare measure and, for example, indirectly argue that a reduction in
consumption or assets, or an increase in liquidity constraints caused by a climatic factor may have
prevented migration (see Hirvonen (2016) for an example of this approach), this may not
necessarily have been the case, because the mediating role of the welfare measure is usually only
indirectly and separately investigated. More generally, the empirical framework on the
determinants of the “migration decision” could greatly benefit from the insights and models of
the well-established literature on geographical and asset-based poverty traps (Barrett and Carter,
2013; Carter and Barrett, 2006; Carter et al., 2008; Jalan and Ravallion, 2002).
▪ Migration is not a random event, and weather- or climate-related migration is not an exception to
this rule. Self-selection comes into play when it comes to the decision to migrate or not (Carletto
and De Brauw, 2012), be this decision due to climatic factors or not. Empirically, there are no
clear prescriptions yet on how to address the self-selection-based endogeneity of migration,9 with
the consequence that many research designs are neither robust nor reliable. Furthermore, since
migration is intrinsically a selective process, any causal inference analysis needs to check for the
determinants of migration and have access to the necessary information for the identification of
8
Incidentally, we also note that the immobility issue is not specific to climate-induced migration. It has long been known that
immobility often masks the inability of people willing to move to do so, due to liquidity constraints, lack of information,
absence of networks, etc. When these obstacles are removed, migration increases. A recent example of international migration
triggered by a program aimed at fostering migration through improved connection and information about employment
opportunities (carried out as part of an experiment implemented in Mizoram, India) is provided in Gaikwad, Hanson and Tóth
(2022).
9
Although there is a wide set of technical alternatives that could be employed to address this key issue, including, but not
limited to, two-phase sampling, Heckman selection, Instrumental Variables (IV) and data-driven statistical methods, there is
often limited credibility of the exclusion restriction of the proposed instruments. New data mining techniques (e.g., LASSO
variable and instrument selection) can help in addressing the identification challenges connected with migration by drawing
on the vast potential and richness of information of multi-topic household surveys (see, inter alia, Belloni et al., 2014).
18
both migrant (treatment) and non-migrant (control) individuals. In addition, data on pre-migration
conditions are also needed (Carletto et al., 2014). While the shift from older cross-sectional
studies, which were vulnerable to a wide range of potential confounders and sources of omitted
variable bias, to more robust panel methods exploiting longitudinal information on household and
individual movements, indeed represents a huge empirical step forward, the use of panel settings
is not in itself a panacea. In fact, the issues highlighted above of self-selection, endogeneity,
transmission channels, longer-run effects and persistence, reverse causality and other key
empirical concerns still need to be addressed through clever research design (De Brauw and
Carletto, 2012). Researchers in the field have traditionally exploited lotteries (e.g., the popular
study on Saudi Arabia to Hajj Visa, Clingingsmith et al., 2009, but also the works by Gibson,
McKenzie and Stillman (2011) and Gibson et al. (2018)) and other possible sets of “exogenous
variations” or “natural experiments” able to mimic the hypothetical situation of a random
selection of migrants. However, what they get is an estimate of the impact of migration only for
the sub-group of beneficiaries (the Local Average Treatment Effect – LATE in the jargon of
policy impact assessments) without “external validity”. A valid alternative, in this case, could be
the use of governmental policy experiments to learn about the effectiveness of alternative policy
initiatives, but the latter goes beyond the scope of this analysis.
3.2 Data gaps and needs
Given the conceptual and empirical limitations of existing studies and the research priorities set out
above, we have identified the main data gaps that currently prevent carrying out the proposed research
agenda and possible solutions to address these data needs that can come from improvements to
longitudinal, multi-topic household surveys such as the LSMS-ISA. We here start by outlining the main
data gaps and needs separately for migration and weather data.
Migration data. Until recently, there was a widespread lack of basic migration data, especially in
developing countries which are more vulnerable to climate change (Laczko and Aghazarm, 2009). While
the situation has improved, and macro-level and international migration data are now available for a wide
range of countries, disaggregated and detailed data on internal migration remain unavailable or
incomplete for many low-income countries (Beine and Jeusette, 2021). This is a paradox because we
know from the literature discussed above that the migration effects of climate change will primarily
concern poor people living in developing countries. The paradox can only be solved by scaling up
19
migration data collection efforts in low- and middle-income origin countries. Despite some recent
progress, there is a specific scarcity of longitudinal and long-term data from migration surveys in
developing contexts. Identification issues due to the lack of panel data have long hampered empirical
progress, so data collection efforts should be primarily focused on tracking individuals over time, either
using self-reported data or information from proxy respondents, also to address non-response issues.
From the discussion on the tight nexus between migration and adaptation emphasized in the previous
section, there is a strong case for integration, in multi-topic household surveys, between questions on
migration and those on risk management, mitigation and coping strategies adopted in response to shock,
such as an explicit distinction between voluntary vs involuntary migration, survival vs risk management
migration, immobility due to in-situ adaptation vs immobility due to liquidity constraints and inability to
move.10 These nuances and amendments to the existing surveys and modules would greatly enhance the
scope for empirical applications in this active area of research, while being relatively low-cost and easy
to implement and collect.
As far as short-term, temporary and seasonal migratory flows are concerned, not only should data be
longitudinal, but they should also be high-frequency, i.e., they should be collected annually. As emerged
from the meta-analysis of Beine and Jeusette (2021), the frequency of the data employed plays a
significant role in determining the findings of econometric analyses: data sampled at higher frequencies
tend to support the case of an effect on mobility more, since migration measures spread over several
years or longer periods are less able to capture short-term migratory flows driven by climatic hazards.
The use of direct measures of mobility, rather than indirect proxies, has often also been stressed as
important but remains the exception rather than the rule. Following Beine and Jeusette (2021), an
example of direct measure of mobility can be found in survey data where people are directly asked about
their migration history, whereas an indirect measurement means that migration is inferred rather than
observed, as in the case of differences in migration stocks reconstructed from censuses. Importantly,
econometric analyses using measures of mobility that are computed or derived from proxies tend to find
10
A caveat is in order here to remind readers of the potential differences and inconsistencies between ‘stated’ and ‘revealed’
migration preferences, which compound over the already complex and multi-faceted nature of the migration phenomenon,
that can take many different forms both across space and over time. The issue is also related to the heterogeneity of the climate
risk perceptions of the individuals interviewed.
20
less empirical evidence in favor of a causal effect of environmental shocks on migration (Beine and
Jeusette, 2021). The use of migration flows as the dependent variable in the regressions increases the
probability of finding an impact, and data on direct measures of migration should be collected
accordingly. Ex post counterfactual policy evaluation of development and social policy interventions
aimed at either favoring or reducing weather-induced migration flows is also scant. There is a need for
data observed on such programs to improve our understanding about the role of policy in mediating the
causal relationship between climatic shocks and human mobility. As a complement to that, the collection
of detailed community-level data to supplement household survey data on the roll-out and
implementation of such programs is also a key element to be considered.
Last but not least, the growing availability of big data and citizen-generated information has spurred a
debate on their potential integration or complementarity with more traditional data collection methods,
such as census and surveys, to fill migration data gaps. In principle, these new data sources could
massively improve the quantity of data available to study climate change and migration. Entities such as
the European Commission and the International Organization for Migration have already started to assess
the potential of sources of big data. Among the most prominent examples of these non-traditional data
sources: mobile phone call detail records (CDR); Internet activity such as Google searches; online media
content; geo-referenced social media activity, which can be obtained via advertising platforms offered
by social media (IOM, 2018). There are also some first recent scientific works based directly on these
sources. Lai et al. (2019) employed a massive data set of 72 billion anonymized CDRs in Namibia from
October 2010 to April 2014, to explore how internal migration estimates can be derived and modeled
from CDRs at subnational and annual scales. As for social networks, Spyratos et al. (2019) used
anonymized and publicly available data provided by Facebook’s advertising platform to estimate the
number of Facebook Network (FN) “migrants” in 119 countries of residence and concluded that these
estimates could be used for trend analysis and early-warning purposes. Specifically concerning climate
change, some have highlighted that in combination with field-level data derived from household surveys
and key-informant networks, big data could be used to detect how sudden-onset natural disasters and
progressive environmental change impact migration patterns (Franklinos et al., 2020). Along these lines,
for example, Lu et al. (2016) used anonymized CDR from a mobile network provider (Grameenphone)
to retrieve the geographical position and movements of users, so as to be able to examine the human
mobility effects of the 2013 Mahasen cyclone in Bangladesh. Finally, for an interesting comparison
between mobile phone and census data, see Wesolowski et al. (2013) and Kirchberger (2022) for a
21
discussion of their potential for research on internal migration. More generally, given their increasing
availability and huge potential, there is a clear need to invest more in research aimed at developing
methods for improving integration and interoperability of household surveys with these new data sources.
Weather data. To investigate climate-related migration outcomes, accurate and georeferenced weather
information is needed. However, most household surveys include, at best, self-reported measures of
weather shocks based on individuals’ recalls, which can hardly be reliable or comparable given their
subjective nature. For this reason, household data from multi-topic surveys are almost always
autonomously integrated with external weather information. The main public domain sources of weather
data include NASA Modern-Era Retrospective analysis for Research and Applications (MERRA-2), the
Terrestrial Air Temperature and Precipitation database from the Center for Climatic Research at the
University of Delaware, and the High-Resolution Gridded Datasets from the Climatic Research Unit of
the University of East Anglia. Remote sensing weather data sets, which can take the form of gridded,
satellite, or reanalysis data, have been used in many studies leveraging LSMS-ISA data to address
weather- and climate-related research questions in Sub-Saharan African contexts (see, among many,
d’Errico et al., 2019; Letta et al., 2018; Mueller et al., 2020; Di Falco et al., 2022).
While this integration is typically carried out directly by the research team, this practice can be
suboptimal for several reasons. First, the intrinsic diversity of weather data products. For instance,
satellite data provide less accurate data than ground stations in most locations and do not extend as far
back in time. Gridded data, on the other hand, aggregate data from ground stations via interpolation and
across a given space. This works well in developed countries, where there is wide and uniform coverage
of weather stations across the entire territory, but not so much in developing contexts, where often
gridded data aggregate weather information from a few old stations spread across the country. Sparse
coverage is a serious issue given the interpolation method adopted by gridded products. Finally, entry
and exit of stations (quite common, especially in poorer countries) can be endogenous and represents an
additional source of measurement error of true weather conditions experienced by people.11 Such
diversity in weather data products can affect econometric estimates of the relationship between climatic
events and a given socioeconomic outcome of interest. Second, the need for spatial anonymization for
privacy protection in household surveys, usually implemented through a random offset of true household
11
See Auffhammer et al. (2013) and Dell, Jones and Olken (2014) for further discussion. For a comprehensive overview of
the availability and quality of climate data in the context of Africa see, instead, Dinku (2019).
22
geocoordinates, can introduce mismeasurement when integrating them with remote sensing weather data.
In a new study based on a pre-analysis plan, Michler et al. (2022) employ 90 linked weather-household
data sets that vary by the spatial anonymization method and show that, as the spatial resolution of most
weather data produce is too coarse, spatial anonymization techniques have an overall small effect on the
estimates of the weather-agricultural productivity relationship and do not introduce substantial
mismeasurement. Depending on the specific type of weather data, however, measurement error can
become significant, especially for higher-resolution data products. Importantly, Michler et al. (2022) also
find that estimates of weather’s impact on agricultural productivity vary substantially in sign,
significance, and magnitude, across different weather data sets for the same spatial anonymization
technique. For these reasons, caution is in order when integrating household surveys such as the LSMS-
ISA with external weather data, and the first-best would be to have high-resolution weather data already
embedded in the survey data set.
4. LSMS-ISA data assessment
To draw concrete operational implications from the review above, we start by outlining what the
implications would be for one of the international survey programs that has been at the forefront of the
methodological debate on data collection in low- and middle-income countries in the past 15 years or so.
The LSMS-ISA program was launched in 2009 with funding from the Bill and Melinda Gates Foundation
and the explicit aim is to fill the gaps in agricultural data through close collaboration with the national
statistical offices (NSOs) of partner countries. The program is based on the implementation of multi-
topic, nationally-representative household longitudinal surveys and, to date, has been carried out in eight
Sub-Saharan African countries, namely Burkina Faso, Ethiopia, Mali, Malawi, Niger, Nigeria, Tanzania,
and Uganda.
LSMS-ISA survey panels are administered approximately every 1 to 3 years. In the LSMS-ISA program,
not only original households are revisited each wave, even if they relocate within the country, but also
individual household members who split off from previously selected households are tracked and
included in subsequent waves. This intertemporal aspect of LSMS-ISA surveys, therefore, unlocks the
potential for the analysis of internal and rural–urban migration patterns, among other things (Carletto and
Gourlay, 2019).
23
4.1 Key limitations of the LSMS-ISA data sets
Given the features of the LSMS-ISA program and based on our previous critical overview, we identify
the following as the main limitation of LSMS-ISA data collection to empirically disentangle the climate-
migration nexus: questionnaire design; sample size and statistical challenges; and respondents’ issues. In
the rest of this section, we focus on each of these areas in turn.
Questionnaire design. Although the LSMS-ISA questionnaires are usually highly standardized across
surveys and countries, the information is sparse, and there are inconsistencies not just across countries,
but also across different surveys within the same country. The lack of consistency in terms of the set of
questions included to detect migration and/or define migrants is certainly the first element to be
considered for a revision of the LSMS-ISA collection aimed at improving data collection on migration.
It also explains why there are only a couple of cross-country investigations on the climate-migration links
which uses LSMS-ISA, the analysis by Mueller et al. (2020) and the one by Di Falco et al. (2022). In
both cases, due to comparable and precise information about migration not being available for all the
countries, the studies have to rely on proxies to define different types of migrations. Specifically, Mueller
and colleagues exploit questions regarding the absence of individuals during the follow-up survey to
define temporary migration as whether an individual present at baseline reported migrating for at least
one month in the previous twelve. Of course, without knowing, in most cases, either the destination or
the reasons for migrating, they had to make a strong assumption on the equivalence between temporary
absence from the household and outmigration. See below (subsection 4.2) for further discussion on this
point. The study by Di Falco and colleagues focuses on long-term migration instead, and thus individuals
are proxied as migrants if households report them to have left between two visits or waves of the survey
and who were not observed to return to their household during the time of analysis. In this specific case,
the assumptions made are even stronger than those implied by Mueller et al. (2020), as the definition of
migrants includes individuals that left the household because they married.
Sample and statistical issues. Migration is, statistically speaking, a rare event. As reported below, the
only five climate-migration studies using LSMS-ISA data have very small samples. This is unsurprising,
as in a normal clustered sample design typical of multi-topic surveys, the expected number of households
associated with emigration may be very low (Carletto et al., 2014). To better identify rare events, two
potential approaches are disproportionate sampling and two-phase sampling. However, both sample
designs require some prior knowledge of migration in the population and are not easy to implement as
24
part of a household survey such as those of the LSMS-ISA collection, which are meant to be, by their
very nature, multi-topic and nationally-representative, and not exclusively targeted to the study of
migration. In this respect, the relatively small sample size of most LSMS-ISA surveys often makes them
unsuitable for the study of migration, as the standard LSMS-ISA multi-stage cluster design is unlikely to
sample a sufficiently large number of households with migrants (De Brauw and Carletto, 2012).
Respondents’ issues. More broadly related to data collection, Lucas (2021) notes that collecting
migration data is essentially limited to two approaches: asking individuals about their migration
experiences or asking remaining household members about those who left, and both present limitations.
The former approach, in fact, provides little information about the household that the migrant left; the
latter assumes that the respondent knows the current whereabouts and activities of the migrated
household members, and memory about the list of those who departed may prove selective.
In addition, as noted by Kirchberger (2022) in her recent review, even when panel household surveys
aim to track respondents, household surveys can still suffer from high levels of attrition. Other general
shortcomings are: i) questions about migrants need to be answered by a proxy, generally a family
member, which may introduce many imperfections and substantial bias (Carletto et al., 2014); ii) the
double-counting of migrants, especially those who can be claimed as members in other households’
rosters; iii) the difficulty in classifying the type of migration: temporary (short-term) migration and (long-
term) permanent migration are usually distinguished by an arbitrary threshold or time criterion set by the
analyst. Return migration, seasonal migration, and circular migration can also be difficult to distinguish
from one another (De Brauw and Carletto, 2012).12
Finally, specifically concerning climate-related hazards, it is unlikely that migration caused by a fast-
onset climatic disaster would be captured in an LSMS-ISA survey, given the localized nature of these
types of events and unless the survey takes place soon after the shock (De Brauw and Carletto, 2012).
12
Seasonal migrants are those who leave for a specified period of time each year and should be identified through questions
about repeated, short migration spells. Return migrants migrated at some time in the past and have returned to the country or
household somewhat permanently. Circular migrants are those who have returned but plan to leave again for a significant
period of time, or repeatedly migrate for long spells.
25
4.2 Limited empirical research uses LSMS-ISA data sets to explore the climate-migration nexus
Our literature search on the number of works on migration and climate change which used LSMS-ISA
data returned only five papers, four of which published in peer-reviewed journals: Ocello et al. (2015);
Kubik and Morel (2016), Mueller at al. (2020), Becerra-Valbuena and Millock (2021), and the recent
working paper by Di Falco et al. (2022).13
Ocello et al. (2015) and Kubik and Morel (2016) both focus on Tanzania, and employ the LSMS-ISA
National Panel Surveys. While the first looks at both the 2008-2009 and the 2010-2011 waves (although
the latter is used only to identify migrants), Ocello et al. (2015) only employ the first. Both also share
similar identification strategies (a two-stage setting with an IV probit model for the former, logit
regression the latter) based on a cross-sectional setting which is not invulnerable to identification threats.
Despite the similarities, they arrive at somewhat contrasting results: Kubik and Morel (2016) find that a
reduction in agricultural income caused by a weather shock increases the probability of internal
migration. However, this effect is significant only for middle-income households, whereas it is
insignificant for the poorest and the richest households, confirming that the decision to migrate as an
adaptation strategy depends on liquidity constraints and initial endowments, with the poorest households
that cannot afford migration costs, while the richest ones can afford in-situ adaptation strategies, such as
irrigation or drought-resistant crops. Ocello et al. (2015) document that being exposed to droughts or
floods or crop diseases is associated with an overall decrease in the likelihood of inter-district mobility,
with the exception of low-educated individuals.
Mueller et al. (2020) employ the LSMS-ISA data from four countries, namely Ethiopia, Malawi,
Tanzania and Uganda. They combine these household panel data with climatic data from NASA’s
MERRA to investigate temporary migration responses to weather anomalies in the East African context.
Using a linear probability model, they find that climate impacts tend to decrease outmigration and,
perhaps surprisingly, are most pronounced in urban areas.
Becerra-Valbuena and Millock (2021) combine LSMS-ISA Malawi surveys with satellite weather data
covering the timespan 2000–2016 to estimate the probability of migration for reasons related to work
and marriage separately for men and women. They find that overall droughts inhibit marriage-related
13
We here refer only to studies exclusively focusing on the causal links between climate change and migration, and exclude
works devoted to other research questions that incidentally find climatic impacts on migration decisions.
26
migration for women, but increase migration of children for work, especially for boys. To carry out their
analysis, they use the migration-related questions on where the individual lived before moving to the
current area of residence, when he/she moved, and the stated motive for doing so. Although this allows
one to retrieve the district of origin and destination as well as the time of migration of individuals
interviewed, they notice that the lack of information at origin before moving is a limitation for their study.
Di Falco et al. (2022) use LSMS-ISA data to construct a multi-country panel data set covering Ethiopia,
Malawi, Niger, Nigeria, and Uganda that is merged with high-resolution gridded precipitation historical
records from the Climate Research Unit to analyze the effects of cumulative drought shocks on the
decision to migrate in rural households. While confirming the existence of an immediate, although small,
impact on migration decisions in the aftermath (i.e., the subsequent year) of a severe and extreme drought
shock, they interestingly show that this impact is long-lasting, increasing migration for at least five years
after the shock occurs, and not even fading or diminishing over time. Furthermore, they find that the
effect of multiple recently experienced droughts (past five years) accumulates over time, which results
in a much higher number of migrants than one would expect based on the immediate effect of the shock
only. The authors emphasize that this has relevant implications for the study of climate-induced migration
and make a plea for advancing the research on the cumulative impacts of climate change on determining
migratory flows in the long-run while at the same time improving the availability of detailed data on
migration.
The fact that out of the vast and growing literature reviewed before, only five studies employ LSMS-ISA
data (and even with conflicting findings), is a clear indicator of the currently limited capacity of the
LSMS-ISA data sets to provide a basis for meaningful analysis on climate change and migration. Figure
1 below provides an idea of the type of migration tracking that is possible using the LSMS-ISA data.
Let us focus, as a benchmark (see below for a comparative analysis of all LSMS-ISA surveys), on the
Tanzanian National Panel survey, which was used by three out of the five studies above. All the available
migration information in the questionnaires is essentially limited to a few questions in two sections in
the Household Questionnaire, Modules B and G.14 Module G, named “Children Living Elsewhere –
Migration” contains some information on households responding affirmatively to the question: “Do you
have any children 15 and older who live elsewhere (outside this household)?” such as information on the
14
Cf. https://microdata.worldbank.org/index.php/catalog/76/related-materials. We use questionnaires from the second round
(2008-2009) as the benchmark here.
27
most recent job, education level, and money sent by the absent individual (who is not necessarily a
migrant). No questions are asked about the reasons which prompted her/him to move in the first place.
Figure 1. Household between-wave mobility – Tanzania National Panel Survey
Source: Carletto and Gourlay (2019)
In addition to this module, some basic questions are also asked in Module B of the household
questionnaire, namely:
➢ B.9: “For how many cumulative months during the last 12 months has [NAME] been away from
this household”?
➢ B.10: “What was [NAME]’s main occupation for the past 12 months?”
➢ B.24: “For how many years have you lived in this community?”
➢ B.25: “From which districts did you move?”
➢ B.26: “Why did you move here?”
➢ B.27: “In which district were you born?”
This is why both Ocello et al. (2015) and Kubik and Maurel (2016) had to rely on some assumptions to
identify migrants. In Ocello et al. (2015), a migrant was defined as a person aged 15 or older who had
moved from one to another district in the five years before the interview, while migrants who moved into
or out of the country were excluded from the analysis, given the focus of the study on internal migration.
Specifically, the authors identified origin and destination districts using questions B.24 and B.25 reported
28
above, and considered respondents living in the community for less than five years as migrants. After
this selection process, their sample included 2,883 individuals aged 15 or above, only 6% of whom
migrated from one Tanzanian district to another in the period between 2004 and 2008. Kubik and Maurel
(2016), instead, in the absence of an explicit question on permanent migration in the data set, exploited
the second wave of the survey collection (2010-2011) to identify migrants, by directly comparing the
place of residence of all household members in the first and the second waves of the survey and
identifying residential moves based on the GPS coordinates of the place of origin and destination.
Following this strategy, they defined their outcome variable of interest as a migration dummy equal to
one for households with at least one member who permanently moved out of the original village between
2008/09 and 2010/11, and found that 14 percent of households had at least one migrant between the two
waves. Note also that Kubik and Maurel (2016), by design, observe internal migration only.
In the subsequent waves of the Tanzanian NPS collection, some changes were implemented. From Wave
2 (NPS 2010-2011) on, Module G on children living outside the household was dropped. In Wave 3,
NPS 2012-2013, an amendment was added among the roster of possible replies to the following question
in the Shocks Section R:
R.6: “What did your household do in response to this [SHOCK] to try to regain your former welfare
level?”
Among the possible replies, there is the following choice: “Household member migrated”. This option,
however, was not present in the subsequent, and currently final, wave of the collection, NPS 2014-2015.
The Tanzania example, chosen as its surveys were used by the studies reviewed above, is emblematic of
the limitations and internal inconsistencies involving migration data in most surveys of the LSMS-ISA
collection.
Prompted by the reply featuring migration as a coping strategy in the Tanzania surveys, we carried out a
screening of all the questions and answers potentially related to the climate-migration nexus that are
currently available in the entire LSMS-ISA collection. The detailed outcome of this screening is reported
in Table A.1 in the Appendix. The migration-as-coping-strategy option is actually present in the shock
questionnaires of many LSMS-ISA surveys in other countries. Interestingly, across the whole the LSMS-
ISA project, the most explicit questionnaire reference to the climate-migration nexus can be found in the
Uganda collection, in rounds 2008-2009, 2009-2010, 2010-2011, 2013-14, and 2015-2016 of the panel,
where the following question appears:
29
Q.3.18: “What was the main reason for moving to the current place of residence?”
And, among the options, there is the following possible reply: “Drought, flood or other weather-related
condition”. This is exactly the type of question that would enable more research on the climate-migration
relationship. Unfortunately, however, this question disappeared from the latest rounds of the Uganda
panel collection and is currently not found in any other collection of surveys across other countries (cf.
Table A.1). Our general conclusion, therefore, is that given the current limitations of the LSMS-ISA data
sets, there is limited scope for these data to help shed light on some aspects of the climate-migration
relationship. However, we believe that a change of perspective on some key issues and relatively
straightforward amendments to the questionnaires could greatly enhance the potential of this multi-topic
household survey collection in this research field. This is the focus of the next section.
5. Adapting LSMS-ISA surveys to collect literature-based migration data
In this section, we elaborate on what is the opportunity window for: i) current LSMS-ISA surveys to
enhance the understanding of the climate-migration nexus; ii) adapting future LSMS-ISA surveys to
collect migration data and position itself as a leading data collection program for the field.
In particular, we identify promising areas to which, with minor improvements, LSMS-ISA surveys can
greatly contribute, including the issues of migration as adaptation, climate-induced immobility and
potential migrants, the role of mediating channels and contextual factors. Finally, we provide an
assessment of the potential integration of LSMS-ISA surveys with other non-traditional data sources.
Clearly, given the reliance on longitudinal migration information tracking individuals and households
over time (and across space), all these data-related opportunities crucially depend on the continuation of
existing panels and the launch of new ones, which should go in parallel with efforts towards improving
questionnaire design and interoperability with other data sources.
A. Look at (im)mobility in the broader framework of adaptation to climate stress
While the issue of small sample sizes of households with migrants, stemming from the ‘rare event’ nature
of migration, is an intrinsic shortcoming of general-purpose multi-topic representative surveys, a change
of perspective on the issue can help in thinking about new research directions. In particular, our review
has emphasized that the potential immobility traps of climate-related hazards are at least as important as
the mobility effects triggered by such events. In this respect, household surveys could be employed not
30
only to look at the migratory flows causally linked to weather anomalies and climatic disasters but also,
and perhaps especially so, to the relationship between such events and the presence of potential
immobility traps associated with adaptation failures. A useful tool in this respect would be an ad hoc
section on migration-as-adaptation embedded within a broader Adaptation Module, with particular
emphasis on questions investigating the reasons (e.g., the role of liquidity constraints) that hampered the
implementation of household- or individual-level coping strategies, including those migration-related.
Migration can be seen as one among a pool of household adaptation strategies in the face of
environmental change. The study of in-situ adaptation strategies, therefore, should be seen as
complementary to the adaptive decision to migrate. McCarthy (2011) provided a set of key indicators
and modules to supplement LSMS and LSMS-ISA survey instruments based on a taxonomy of household
agricultural practices and investments that can contribute to adapting to climate change (the so-called
sustainable land management, SLM). These include agroforestry investments, reduced or zero tillage,
use of cover crops, and various soil and water conservation structures. There are often long-term benefits
to households from adopting such activities in terms of increasing yields and reducing the variability of
yields, making the system more resilient to changes in climate.
The implementation of a comprehensive Adaptation module capturing these aspects as well as other
insights from resilience and vulnerability studies (Magrini et al., 2018; D’Errico et al., 2019) would
represent a clear added value for household surveys. It would allow one to distinguish between migration
as adaptation mainly implemented as a strategic risk diversification strategy by richer and better-endowed
households vs survival migration adopted as a last-resort option by the poorer households, and hence
enhance our understanding of the nature of migration as voluntary or involuntary. It would also have
important implications for studying adaptation policies to climate change in rural developing contexts in
Sub-Saharan Africa. For example, the systematic inclusion, in the roster of potential answers to the
question on coping strategies, of the option “Drought, flood or other weather-related condition” as in
question Q.3.18 of the earlier rounds of the Uganda collection, which we mentioned above, would be a
straightforward way to collect more information on climatic push factors. Even if a module should not
be seen as a viable solution, the collection of additional questions and to make up for the current lack of
an integrated framework on adaptation in household surveys such as the LSMS-ISA is in any case
strongly encouraged.
31
B. Investigate the socio-economic channels and contextual factors driving climate-related
(im)mobility
A complementary and parallel solution to the Adaptation module suggested above is the development of
an Intention-to-migrate (ITM) module on potential migrants to investigate the issues of immobility and
climate-induced reductions in migratory flows. Such an ITM module would draw from the available
evidence on climate risk perception (Helbling et al., 2021; Koubi, Stoll and Spilker, 2016; Zander,
Richerzhagen and Garnett, 2019) and, in order to investigate the potential of climatic hazards to trap
people in immobility, its framing should also be inspired by the theoretical insights of the sound literature
on asset-based and geographic or shock-driven poverty traps (Barrett and Carter, 2013; Carter and
Barrett, 2006; Carter et al., 2008; Jalan and Ravallion, 2002; Letta et al., 2018). With ITM information
at hand, it would also be possible to fully unleash the potential of one of the main advantages household
surveys hold over other data sources on migration, the possibility of assessing the role of transmission
mechanisms and contextual factors that affect the magnitude, or even the direction, of the climate-
migration link. The LSMS-ISA collection, given its multi-topic nature, can be particularly informative
on these issues, because the analyst can fully exploit the wealth of information available about household
demographic, economic, and geographic characteristics to investigate a broader range of issues involving
the causal links between poverty, agriculture, climate and the (in)ability to migrate.
C. Other possible data improvements in longitudinal surveys
Migration data. The two Adaptation and ITM modules proposed above imply a change of perspective
to investigate immobility and transmission mechanisms. But we also call for a series of other data
improvements in longitudinal household surveys such as those of the LSMS-ISA. These surveys, in fact,
exhibit a clear potential given their intertemporal nature (Carletto and Gourlay, 2019), which also allows
them to track over time, and include in the subsequent waves, households that relocate within the country,
as well as individual members who split off from previously selected households. A possible additional
tool to enhance this traceability is the design and implementation of an “associate module” meant to keep
track of “people who are not members of the surveyed households but who are somehow associated with
them, either because they contribute to the household standard of living and/or because they moved out
of the household, and/or simply because they are close relative” (Beauchemin, 2020). A complementary
solution could be to proxy this information by collecting histories and retrospective data in panel waves
covering years between surveys. In any case, the data gaps on migration require one to track people over
time and across space, either directly or indirectly. Finally, a third route could consist of establishing a
32
functional link with proximate data collection on national and local patterns of local labor mobility.
Weather data. Thanks to their panel nature, LSMS-ISA surveys can capture repeated weather shocks,
cumulative weather risk, and their potential impact on migration outcomes. From a data collection
perspective, to maximize this potential, it would be necessary to incorporate, within multi-topic surveys,
a built-in and comprehensive set of (spatially and temporally) high-resolution weather data, ideally
collected in situ during the data collection efforts (for instance, through weather sensors), which should:
i) not be limited to temperature and precipitation, but also incorporate other key weather variables such
as humidity, windspeed, etc.; ii) include ad hoc, agriculture-oriented weather information, such as
weather indicators constructed on the basis on crop-specific and local growing season calendars; iii) also
include, in addition to weather data, all the equivalent climate variables, based on the long-run averages
(e.g., 30-year) of the weather time series. Should such weather data collection in situ be unfeasible due
to timing or budget constraints, efforts could be devoted to the integration of existing surveys with
crowdsourced weather data (Minet et al., 2017; Muller et al., 2015; Overeem et al., 2013), to obtain the
maximum possible spatial and temporal granularity of the climate information. The provision of a set of
built-in weather data would prevent risks deriving from independent user integration with external remote
sensing weather data, i.e., measurement error due to spatial anonymization and the sensitivity of
econometric estimates to the specific weather data product chosen by the researcher (Micher et al., 2022),
as well as the use of self-reported weather shocks based on the perceptions of the individuals interviewed
– such as in Ocello et al. (2015) – whose accuracy remains questionable. The inclusion of longer-run
climatic measures and time series would also allow the implementation of more recent and sophisticated
techniques tailored to bridge the external validity gap between weather and climate, such as long
differences or the inclusion of several weather lags in the regression models.
Lastly, while standard nationally-representative household surveys such as the LSMS-ISA are clearly
disadvantaged with respect to the possibility of investigating the migratory outcomes of localized sudden-
onset weather shocks, which require ad hoc post-disaster surveys, the use of ‘mixed-mode’ data
collections solutions, such as the alternation of standard face-to-face surveys with higher-frequency
phone-survey interviews, would allow one to obtain not only more timely and frequent longitudinal
information in general but also, in case a sudden shock occurs, the collection of basic migration data in
the aftermath of the shock itself, and to do so in an easier and less expensive way compared to the
complexity associated with the implementation of a swift ad hoc survey in the area exposed to the
disaster.
33
D. Integration with non-traditional data sources
As for the opportunities offered by the integration with non-traditional data sources, also known as
‘digital trace’ data (Kirchberger, 2022), such as big data and citizen-generated data, our assessment is
that there are both pros and cons, and that while, at the moment, improving interoperability of household
surveys with non-conventional data presents complex issues, this is likely to be the way to go in a not-
so-far future. On the one hand, big data, such as mobile phone data or smartphone app data, have indeed
the potential to complement traditional data and address significant spatial and temporal gaps. As
emphasized by the International Organization for Migration (IOM, 2018), the main strengths of these
new data sources lie especially in their wide and continuous coverage, flexibility, timeliness, frequency,
high spatial, relatively low cost, and their potential to track temporary, circular, and seasonal patterns of
migration, which are difficult to capture through traditional sources. Additionally, big data can be a
promising tool for better ex-ante sample design, such as first-stage sample selection, and allow one to
compute new covariates that would otherwise be costly or difficult to collect at scale, such as labor market
referrals, social networks, or mobility and social contact (Kirchberger, 2022). On the other hand, the use
of big data comes with significant challenges, among which: i) ownership by the private sector; ii)
privacy, data protection and ethical issues in using data automatically generated by users, and human
rights concerns; iii) their volume, complexity and “noisiness”; iv) the fact that big data reflect behavioral
patterns which may not be representative of the population (selection bias); v) reliability of the self-
reported information on social media; difficulties in applying statistical definitions of migration.
Using the case study of a data science challenge involving West African mobile phone data, Taylor
(2016) argues that big data carries with it the dual risk of rendering certain groups invisible, and of
misinterpreting what is visible and, in addition, raises concerns about the lack of information concerning
the context of behaviors and activities ‘observed’ using big data analysis. Others have argued that the
distance between the researcher and the research looms large for remotely generated data, posing
challenges to validity and ground truth, as well as for the reliability and interpretation of research results
(Kraly and Hovy, 2020). Operationally, while the potential for using such data to better understand
migration dynamics has yet to be fully explored, integrating multi-topic household surveys with varied
sources of big data requires not only addressing the significant technical and ethical challenges just
highlighted but also the development of new methodologies that consider complex interactions over
differing geographic and temporal scales (Franklinos et al., 2020). Furthermore, to leverage big data as
a meaningful source of information for migration analysis, national statistical offices would need to work
34
with all relevant stakeholders to manage the process of data production. (IOM, 2018). These are hard
challenges, but non-conventional data sources offer such great potential in complementing (not in
substituting) survey data that, at some point, these challenges will have to be faced. Our assessment,
therefore, is that while these challenges and drawbacks currently make their potential combination with
household surveys challenging in the short-run, these sources appear promising for the future, especially
because of their ability to provide a large amount of more timely and granular data.
6. Conclusions
It is easy to envisage that, as climate change intensifies, the climate change-migration nexus will keep
gaining prominence in the international agenda. In parallel, the empirical literature will keep growing.
Further refinements to the conceptual framework, as well as developments in econometric techniques,
will shed new light on the relationship between these complex, multifaceted phenomena. However, all
of this cannot provide concrete benefits for policy making without closing the existing data gaps.
Based on the empirical review, we have identified the main data gaps on the climate-migration nexus.
Using a household survey program currently at the forefront of methodological research, the LSMS-ISA
project, we then identified the limitations and opportunities for household survey data to enhance our
understanding of the causal relationship. A summary of this assessment is reported in Table 1, which
provides a list of the most relevant conceptual and empirical issues, with the corresponding data gaps
and a set of initial proposals to boost the potential of household surveys such as the LSMS-ISA. We have
stressed that household surveys currently allow limited exploration of the climate-migration nexus. At
the same time, we have also documented, in light of the open issues and research gaps in the literature,
the great potential that LSMS-type surveys hold to help address some of the most policy-relevant research
questions in the field. Our proposals are twofold: conceptual and operational. Conceptually, we call for
researchers to make use of these data to investigate some of the most important but still unclear aspects
of this nexus, such as the role of immobility, the complementarity or substitutability between migration
and other adaptation strategies, the migration potential of cumulative slow-onset events. Operationally,
we propose to integrate multi-topic, multi-purpose, longitudinal household surveys with additional pieces
of information enhancing both the quality and quantity of the information collected. In particular, with
only slight modifications to the current questionnaires, such as short modules on adaptation and intention
to migrate, the collection of migration histories or associate modules, and the integration of face-to-face
surveys with phone surveys to increase the frequency of the longitudinal information, it would be
possible to maximize the potential of these tools in a cost-effective way.
35
Table 1. Summary of the literature-based assessment of data gaps
➢
Conceptual
Empirical issues Data issues LSMS-ISA issues Proposals
issues
‘Rare event’ nature of
Different Small migrant More focus on immobility and
migration in household
migration samples; lack of cumulative slow-onset changes;
Fast- vs slow- surveys; need for high-
outcomes high-frequency ‘mixed-mode’ data collection
onset events frequency data and long
depending on the data and long solutions combining standard and
panels; need of ad hoc
type of event panels phone surveys
surveys for fast-onset shocks
More empirical research using ad
Direct vs indirect Identification of Need for multi-topic and None, it is its main hoc tools (e.g., mediation
effects channels multi-purpose data added value analysis) to leverage this added
value
Inconclusive Collection of migration histories
evidence of the Lack of consistent and more systematic
Internal vs two phenomena Need for longitudinal information retrospective/recall data in panel
international and their microdata on internal and (across countries, waves for in-between years;
migration interconnectedness international mobility within countries, associate module; potential links
(the climate over time) with complementary labor
migration chain) mobility data
Small migrant
Disentangling the Dearth of data on potential
Liquidity samples; indirect
relationship migrants, intention-to-
constraints, proxies/measures ITM module; empirical focus on
between migrate, stated vs revealed
mobility vs for migration; no immobility traps
migration, wealth preferences, reasons for
immobility information on
and resources migrating
potential migrants
Understanding the
linkage between
agricultural Data gaps on a set of Lack of an
Adaptation Module (in which
practices, in situ indicators capturing the integrated and
Migration as migration appears as a strategy
adaptation and interconnectedness between comprehensive
adaptation among a pool of other adaptation
migration; farmers’ adaptation and framework on
options)
voluntary vs decisions to migrate adaptation
involuntary
migration
36
As noted by scholars and practitioners,15 adding modules to existing surveys such as the LSMS-ISA has
a non-negligible advantage in terms of both timing and cost, as it involves only the relatively small
marginal cost (compared to the total survey cost) of adding questions to an ongoing survey program
already funded and underway. In the medium term, solving technical and ethical challenges related to
the interoperability with non-conventional data sources would unleash even more innovations and
opportunities. For the moment, we see clear scope for general multi-purpose household surveys to play
a leading role in the climate-migration field, as these tools hold key advantages over censuses and other
administrative data sources due to their comprehensiveness and flexibility to collect more detailed data.
Our review calls for the exploitation of these advantages.
15
See, for example, here: https://migrationdataportal.org/blog/household-surveys-key-potential-source-data-migration.
37
References*
Agamile, P., Ralitza, D., & Golan, J. (2021). Crop Choice, Drought and Gender: New Insights from
Smallholders’ Response to Weather Shocks in Rural Uganda. Journal of Agricultural Economics. 72(3),
829-856. [1]
Alam, G. M., Alam, K., & Mushtaq, S. (2016). Influence of institutional access and social capital on
adaptation decision: Empirical evidence from hazard-prone rural households in Bangladesh. Ecological
Economics, 130, 243-251. [1]
Auffhammer, M. (2018). Quantifying economic damages from climate change. Journal of Economic
Perspectives, 32(4), 33-52. [3]
Auffhammer, M., Hsiang, S. M., Schlenker, W., & Sobel, A. (2013). Using weather data and climate
model output in economic analyzes of climate change. Review of Environmental Economics and
Policy, 7(2), 181-198. [3]
Azzarri, C., Haile, B., & Letta, M. (2022). Plant different, eat different? Insights from participatory
agricultural research. PloS one, 17(3), e0265947. [4]
Bardsley, D. K., & Hugo, G. J. (2010). Migration and climate change: examining thresholds of change
to guide effective adaptation decision-making. Population and Environment, 32(2-3), 238-262. [1]
Barrett, C. B., & Carter, M. R. (2013). The economics of poverty traps and persistent poverty: Empirical
and policy implications. The Journal of Development Studies, 49(7), 976-990. [4]
Bazzi, S. (2017). Wealth heterogeneity and the income elasticity of migration. American Economic
Journal: Applied Economics, 9(2), 219-55. [1]
Beauchemin, C. (2020). Towards Migration Modules for the 50x2030 Survey: Literature Review and
Recommendations. Mimeo. [2]
Becerra-Valbuena, L. G., & Millock, K. (2021). Gendered migration responses to drought in
*
To help readers, we report, next to each reference, a number in bold and square brackets depending on the specific topic to
which each reference belongs. The legend is as follows: [1] Migration and climate change; [2] Migration data sources, gaps,
and methods; [3] Climate change impacts and weather data; [4] Other topics.
38
Malawi. Journal of Demographic Economics, 87(3), 437-477. [1]
Beegle, K., De Weerdt, J., & Dercon, S. (2011). Migration and economic mobility in Tanzania: Evidence
from a tracking survey. Review of Economics and Statistics, 93(3), 1010-1033. [1]
Beine, M. A., & Jeusette, L. (2021). A meta-analysis of the literature on climate change and migration.
Journal of Demographic Economics, 1-52. [1]
Beine, M., & Parsons, C. (2015). Climatic factors as determinants of international migration. The
Scandinavian Journal of Economics, 117(2), 723-767. [1]
Bekaert, E., Ruyssen, I., & Salomone, S. (2021). Domestic and international migration intentions in
response to environmental stress: A global cross-country analysis. Journal of Demographic
Economics, 87(3), 383-436. [1]
Belloni, A., Chernozhukov, V., & Hansen, C. (2014). High-dimensional methods and inference on
structural and treatment effects. Journal of Economic Perspectives, 28(2), 29-50. [4]
Biermann, F., & Boas, I. (2010). Preparing for a warmer world: Towards a global governance system to
protect climate refugees. Global environmental politics, 10(1), 60-88. [1]
Black, R., Bennett, S. R., Thomas, S. M., & Beddington, J. R. (2011). Migration as
adaptation. Nature, 478(7370), 447-449. [1]
Bohra-Mishra, P., Oppenheimer, M., & Hsiang, S. M. (2014). Nonlinear permanent migration response
to climatic variations but minimal response to disasters. Proceedings of the National Academy of
Sciences, 111(27), 9780-9785. [1]
Bohra-Mishra, P., Oppenheimer, M., Cai, R., Feng, S., & Licker, R. (2017). Climate variability and
migration in the Philippines. Population and environment, 38(3), 286-308. [1]
Bosetti, V., Cattaneo, C., & Peri, G. (2018). Should they stay or should they go? Climate migrants and
local conflicts. Journal of Economic Geography, lbaa002. [1]
Bryan, G., Chowdhury, S., & Mobarak, A. M. (2014). Underinvestment in a profitable technology: The
case of seasonal migration in Bangladesh. Econometrica, 82(5), 1671-1748. [1]
39
Burke, M., & Emerick, K. (2016). Adaptation to climate change: Evidence from US
agriculture. American Economic Journal: Economic Policy, 8(3), 106-40. [3]
Cai, R., Feng, S., Oppenheimer, M., & Pytlikova, M. (2016). Climate variability and international
migration: The importance of the agricultural linkage. Journal of Environmental Economics and
Management, 79, 135-151. [1]
Carletto, C., & Gourlay, S. (2019). A thing of the past? Household surveys in a rapidly evolving
(agricultural) data landscape: Insights from the LSMS‐ISA. Agricultural Economics, 50, 51-62. [2]
Carletto, C., Larrison, J., & Özden, Ç. (2014). Informing migration policies: a data primer.
In International Handbook on Migration and Economic Development. Edward Elgar Publishing. [2]
Carter, M. R., & Barrett, C. B. (2006). The economics of poverty traps and persistent poverty: An asset-
based approach. The Journal of Development Studies, 42(2), 178-199. [4]
Carter, M. R., Little, P. D., Mogues, T., & Negatu, W. (2008). Poverty traps and natural disasters in
Ethiopia and Honduras. In Social Protection for the Poor and Poorest (pp. 85-118). Palgrave Macmillan,
London. [4]
Castells-Quintana, D., Krause, M., & McDermott, T. K. (2021). The urbanising force of global warming:
the role of climate change in the spatial distribution of population. Journal of Economic
Geography, 21(4), 531-556. [1]
Cattaneo, C. (2019). Migrant networks and adaptation. Nature Climate Change, 9(12), 907-908. [1]
Cattaneo, C., & Massetti, E. (2015). Migration and climate change in rural Africa. FEEM Working Paper,
No. 029.2015. [1]
Cattaneo, C., & Peri, G. (2016). The migration response to increasing temperatures. Journal of
Development Economics, 122, 127-146. [1]
Cattaneo, C., Beine, M., Fröhlich, C. J., Kniveton, D., Martinez-Zarzoso, I., Mastrorillo, M., ... &
Schraven, B. (2019). Human migration in the era of climate change. Review of Environmental Economics
and Policy, 13(2), 189-206. [1]
Celli, V. (2022). Causal mediation analysis in economics: Objectives, assumptions, models. Journal of
40
Economic Surveys, 36(1), 214-234. [4]
Choquette-Levy, N., Wildemeersch, M., Oppenheimer, M., & Levin, S. A. (2021). Risk transfer policies
and climate-induced immobility among smallholder farmers. Nature Climate Change, 11(12), 1046-
1054. [1]
Clement, V., Rigaud, K. K., de Sherbinin, A., Jones, B., Adamo, S., Schewe, J., ... & Shabahat, E.
(2021). Groundswell Part 2: Acting on Internal Climate Migration. World Bank. [1]
Clingingsmith, D., Khwaja, A. I., Kremer, M. (2009). Estimating the Impact of The Hajj: Religion and
Tolerance in Islam's Global Gathering, The Quarterly Journal of Economics, Volume 124, Issue 3,
August 2009, Pages 1133–1170. [2]
Cottier, F., & Salehyan, I. (2021). Climate variability and irregular migration to the European
Union. Global Environmental Change, 69, 102275. [1]
Cui, X., & Feng, S. (2020). Climate Change and Migration. Handbook of Labor, Human Resources and
Population Economics, 1-15. [1]
Dallmann, I., & Millock, K. (2017). Climate variability and inter-state migration in india. CESifo
Economic Studies, 63(4), 560-594. [1]
Davis, K. F., Bhattachan, A., D’Odorico, P., & Suweis, S. (2018). A universal model for predicting
human migration under climate change: examining future sea level rise in Bangladesh. Environmental
Research Letters, 13(6), 064030. [1]
De Brauw, A., & Carletto, G. (2012). Improving the measurement and policy relevance of migration
information in multi-topic household surveys. Living Standards Measurement Study Working Paper, 14,
3358986-1199367264546. [2]
De Brauw, A., Mueller, V., & Lee, H. L. (2014). The role of rural–urban migration in the structural
transformation of Sub-Saharan Africa. World Development, 63, 33-42. [2]
Dell, M., Jones, B. F., & Olken, B. A. (2014). What do we learn from the weather? The new climate-
economy literature. Journal of Economic Literature, 52(3), 740-98. [3]
D'Errico, M., Letta, M., Montalbano, P., Pietrelli, R. (2019). Resilience Thresholds to Temperature
41
Anomalies: A long-run Test for Rural Tanzania. Ecological Economics, 164, 106365. [4]
Di Falco, S., Veronesi, M., & Yesuf, M. (2011). Does adaptation to climate change provide food security?
A micro‐perspective from Ethiopia. American Journal of Agricultural Economics, 93.3 , 829-846. [1]
Di Falco, S., Kis, A. B., & Viarengo M. (2022). Cumulative Climate Shocks and Migratory Flows:
Evidence from Sub-Saharan Africa. IZA Discussion Paper No. 15084. [1]
Dillon, A., Mueller, V., & Salau, S. (2011). Migratory responses to agricultural risk in northern
Nigeria. American Journal of Agricultural Economics, 93(4), 1048-1061. [1]
Dinku, T. (2019). Challenges with availability and quality of climate data in Africa. In Extreme
hydrology and climate variability (pp. 71-80). Elsevier. [3]
Feng, S., Krueger, A. B., & Oppenheimer, M. (2010). Linkages among climate change, crop yields and
Mexico–US cross-border migration. Proceedings of the National Academy of Sciences, 107(32), 14257-
14262. [1]
Franklinos, L., Parrish, R., Burns, R. M., Caflisch, A., Mallick, B., Rahman, T., ... & Trigwell, R. (2020).
Key opportunities and challenges for the use of big data inmigration research and policy. UCL Open:
Environment Preprint. [2]
Findlay, A. M. (2012). Flooding and the scale of migration. Nature Climate Change, 2(6), 401-402. [1]
Fröhlich, C. J. (2016). Climate migrants as protestors? Dispelling misconceptions about global
environmental change in pre-revolutionary Syria. Contemporary levant, 1(1), 38-50. [1]
Gaikwad, N., Hanson, K., & Tóth, A. (2022). How overseas opportunities shape political preferences:
A field experiment in international migration. Unpublished manuscript. Available at:
https://nikhargaikwad.com/resources/OverseasOpportunities.pdf [2]
Gemenne, F. (2011). Why the numbers don’t add up: A review of estimates and predictions of people
displaced by environmental changes. Global Environmental Change, 21, S41-S49. [1]
Gibson, J., McKenzie, D., & Stillman, S. (2011). The impacts of international migration on remaining
household members: omnibus results from a migration lottery program. Review of Economics and
Statistics, 93(4), 1297-1318. [2]
42
Gibson, J., McKenzie, D., Rohorua, H., & Stillman, S. (2018). The Long-term impacts of international
migration: Evidence from a lottery. The World Bank Economic Review, 32(1), 127-147. [2]
Grace, K., & al., e. (2018). Examining rural Sahelian out-migration in the context of climate change: An
analysis of the linkages between rainfall and out-migration in two Malian villages from 1981 to 2009.
World Development, 109, 187-196. [1]
Gray, C., & Bilsborrow, R. (2013). Environmental influences on human migration in rural
Ecuador. Demography, 50(4), 1217-1241. [1]
Gray, C., L., & Mueller, V. (2012a). Drought and population mobility in rural Ethiopia. World
Development, 40(1), 134-145. [1]
Gray, C. L., & Mueller, V. (2012b). Natural disasters and population mobility in
Bangladesh. Proceedings of the National Academy of Sciences, 109(16), 6000-6005. [1]
Helbling, M., Auer, D., Meierrieks, D., Mistry, M., & Schaub, M. (2021). Climate change literacy and
migration potential: micro-level evidence from Africa. Climatic Change, 169(1), 1-13. [1]
Henderson, J. V., Storeygard, A., & Deichmann, U. (2017). Has climate change driven urbanization in
Africa? Journal of Development Economics, 124, 60-82. [1]
Hirvonen, K. (2016). Temperature changes, household consumption, and internal migration: Evidence
from Tanzania. American Journal of Agricultural Economics, 98(4), 1230-1249. [1]
Hoffmann, R., Dimitrova, A., Muttarak, R. et al. A meta-analysis of country-level studies on
environmental change and migration. (2020). Nature Climate Change, 10(10), 904-912. [1]
Hunter, L. M., Murray, S., & Riosmena, F. (2013). Rainfall patterns and US migration from rural
Mexico. International Migration Review, 47(4), 874-909. [1]
International Organization for Migration (IOM) (2108). Data Bulletin Series: Informing the
Implementation of the Global Compact for Migration. Available at:
https://publications.iom.int/books/data-bulletin-series-informing-implementation-global-compact-
migration [2]
IPCC (2014). Summary for policymakers. In: Edenhofer O, Pichs-Madruga R, Sokona EFY, Kadner S,
43
Seyboth K, Adler A et al (eds) Climate change 2014: mitigation of climate change. Contribution of
working group III to the fifth assessment report of the intergovernmental panel on climate change.
Cambridge University Press, Cambridge, pp 1–33. Available at:
https://doi.org/10.1017/CBO9781107415324 [3]
Jalan, J., & Ravallion, M. (2002). Geographic poverty traps? A micro model of consumption growth in
rural China. Journal of Applied Econometrics, 17(4), 329-346. [4]
Jessoe, K., Manning, D. T., & Taylor, J. E. (2018). Climate change and labour allocation in rural Mexico:
Evidence from annual fluctuations in weather. The Economic Journal, 128(608), 230-261. [1]
Kaczan, D. J., & Orgill-Meyer, J. (2020). The impact of climate change on migration: a synthesis of
recent empirical insights. Climatic Change, 158(3), 281-300. [1]
Karanja Ng’ang’a, S., & al., e. (2016). Migration and self-protection against climate change: a case study
of Samburu County, Kenya. World Development, 84, 55-68. [1]
Kattumuri, R., Ravindranath, D., & Esteves, T. (2017). Local adaptation strategies in semi-arid regions:
study of two villages in Karnataka, India. Climate and Development, 9(1), 36-49. [1]
Kelley, C. P., Mohtadi, S., Cane, M. A., Seager, R., & Kushnir, Y. (2015). Climate change in the Fertile
Crescent and implications of the recent Syrian drought. Proceedings of the national Academy of
Sciences, 112(11), 3241-3246. [1]
Kirchberger, M. (2021). Measuring internal migration. Regional Science and Urban Economics, 91,
103714. [2]
Kolstad, C. D., & Moore, F. C. (2020). Estimating the economic impacts of climate change using weather
observations. Review of Environmental Economics and Policy, 14(1), 1-24. [3]
Koubi, V., Schaffer, L., Spilker, G., & Böhmelt, T. (2022). Climate events and the role of adaptive
capacity for (im-) mobility. Population and Environment, 1-26. [1]
Koubi, V., Spilker, G., Schaffer, L., & Bernauer, T. (2016). Environmental stressors and migration:
Evidence from Vietnam. World Development, 79, 197-210. [1]
Koubi, V., Stoll, S., & Spilker, G. (2016). Perceptions of environmental change and migration
44
decisions. Climatic change, 138(3), 439-451. [1]
Kraly, E. P., & Hovy, B. (2020). Data and research to inform global policy: the global compact for safe,
orderly and regular migration. Comparative Migration Studies, 8(1), 1-32. [2]
Kubik, Z., & Maurel, M. (2016). Weather shocks, agricultural production and migration: Evidence from
Tanzania. The Journal of Development Studies, 52(5), 665-680. [1]
Laczko, F., & Aghazarm, C. (2009). Migration, Environment and Climate Change: assessing the
evidence. International Organization for Migration (IOM). [1]
Lai, S., zu Erbach-Schoenberg, E., Pezzulo, C., Ruktanonchai, N. W., Sorichetta, A., Steele, J., ... &
Tatem, A. J. (2019). Exploring the use of mobile phone data for national migration statistics. Palgrave
communications, 5(1), 1-10. [2]
Letta, M., Montalbano, P., & Tol, R. S. (2018). Temperature shocks, short-term growth and poverty
thresholds: Evidence from rural Tanzania. World Development, 112, 13-32. [4]
Lilleør, H. B., & Van den Broeck, K. (2011). Economic drivers of migration and climate change in LDCs.
Global Environmental Change, 21, S70-S81. [1]
Liu, M. Y., Shamdasani, Y., & Taraz, V. (2022). Climate change and labor reallocation: Evidence from
six decades of the Indian Census. American Economic Journal: Economic Policy, forthcoming. [1]
Lu, X., Wrathall, D. J., Sundsøy, P. R., Nadiruzzaman, M., Wetter, E., Iqbal, A., ... & Bengtsson, L.
(2016). Detecting climate adaptation with mobile network data in Bangladesh: anomalies in
communication, mobility and consumption patterns during cyclone Mahasen. Climatic Change, 138(3-
4), 505-519. [2]
Lucas, R. E. (2021). Crossing the Divide: Rural to Urban Migration in Developing Countries. Oxford
University Press. [2]
Magrini, E., Montalbano, P., Winters L. A. (2018). Households’ vulnerability from trade in
Vietnam, World Development, 112. pp. 46-58. ISSN 0305-750X. [4]
Marchiori, L., Maystadt, J. F., & Schumacher, I. (2012). The impact of weather anomalies on migration
in Sub-Saharan Africa. Journal of Environmental Economics and Management, 63(3), 355-374. [1]
45
Martin, M., Billah, M., Siddiqui, T., Abrar, C., Black, R., & Kniveton, D. (2014). Climate-related
migration in rural Bangladesh: a behavioral model. Population and Environment, 36(1), 85-110. [1]
Martinez Flores, F., Milusheva, S., & Reichert, A. R. (2021). Climate anomalies and international
migration: A disaggregated analysis for West Africa. World Bank Policy Research Working Paper
n.9664. [1]
Mastrorillo, M., Licker, R., Bohra-Mishra, P., Fagiolo, G., Estes, L. D., & Oppenheimer, M. (2016). The
influence of climate variability on internal migration flows in South Africa. Global Environmental
Change, 39, 155-169. [1]
McCarthy, N. (2011). Understanding agricultural households' adaptation to climate change and
implications for mitigation: land management and investment options. Inc., Washington, DC, USA. [3]
McCarthy, N., Kilic, T., De La Fuente, A., & Brubaker, J. M. (2018). Shelter from the storm? household-
level impacts of, and responses to, the 2015 floods in Malawi. Economics of disasters and climate
change, 2(3), 237-258. [3]
McKenzie, D and Yang D. (2012). Experimental Approaches in Migration Studies. In Carlos Vargas-
Silva (ed.), Handbook of Research Methods in Migration, Edward Elgar Publishing, 249-269. [2]
McLeman, R., & Gemenne, F. (Eds.). (2018). Routledge handbook of environmental displacement and
migration. Routledge. [1]
McMichael, C., Dasgupta, S., Ayeb-Karlsson, S., & Kelman, I. (2020). A review of estimating population
exposure to sea-level rise and the relevance for migration. Environmental Research Letters, 15(12),
123005. [1]
McNamara, K. E., Bronen, R., Fernando, N., & Klepp, S. (2018). The complex decision-making of
climate-induced relocation: Adaptation and loss and damage. Climate Policy, 18(1), 111-117. [1]
Michler, J. D., Josephson, A., Kilic, T., & Murray, S. (2022). Privacy Protection, Measurement Error,
and the Integration of Remote Sensing and Socioeconomic Survey Data. arXiv preprint
arXiv:2202.05220. [3]
Minet, J., Curnel, Y., Gobin, A., Goffart, J. P., Melard, F., Tychon, B., ... & Defourny, P. (2017).
46
Crowdsourcing for agricultural applications: A review of uses and opportunities for a farmsourcing
approach. Computers and Electronics in Agriculture, 142, 126-138. [2]
Missirian, A., & Schlenker, W. (2017). Asylum applications and migration flows. American Economic
Review, 107(5), 436-40. [1]
Morrissey, J. W. (2013). Understanding the relationship between environmental change and migration:
The development of an effects framework based on the case of northern Ethiopia. Global Environmental
Change , 23.6 , 1501-1510. [1]
Mueller, V., Gray, C., & Hopping, D. (2020). Climate-Induced migration and unemployment in middle-
income Africa. Global Environmental Change, 65, 102183. [1]
Mueller, V., Gray, C., & Kosec, K. (2014). Heat stress increases long-term human migration in rural
Pakistan. Nature Climate Change, 4(3), 182-185. [1]
Mueller, V., Schmidt, E., & Lozano-Gracia, N. (2015, July). Household and Spatial Drivers of Migration
Patterns in Africa: Evidence from Five Countries. In Urbanization and Spatial Development of Countries
Research Workshop, World Bank, Washington, DC. [2]
Mueller, V., Sheriff, G., Dou, X., & Gray, C. (2020). Temporary migration and climate variation in
Eastern Africa. World Development, 126, 104704. [1]
Muller, C. L., Chapman, L., Johnston, S., Kidd, C., Illingworth, S., Foody, G., ... & Leigh, R. R. (2015).
Crowdsourcing for climate and atmospheric sciences: current status and future potential. International
Journal of Climatology, 35(11), 3185-3203. [3]
Myers, N. (2002). Environmental refugees: a growing phenomenon of the 21st century. Philosophical
Transactions of the Royal Society of London. Series B: Biological Sciences, 357(1420), 609-613. [1]
Nawrotzki, R. J., Riosmena, F., & Hunter, L. M. (2013). Do rainfall deficits predict US-bound migration
from rural Mexico? Evidence from the Mexican census. Population research and policy review, 32(1),
129-158. [1]
Neumann, B., Vafeidis, A. T., Zimmermann, J., & Nicholls, R. J. (2015a). Future coastal population
growth and exposure to sea-level rise and coastal flooding-a global assessment. PloS one, 10(3),
47
e0118571. [1]
Neumann, K., Sietz, D., Hilderink, H., Janssen, P., Kok, M., & van Dijk, H. (2015b). Environmental
drivers of human migration in drylands–A spatial picture. Applied Geography, 56, 116-126. [1]
Niva, V., Kallio, M., Muttarak, R., Taka, M., Varis, O., & Kummu, M. (2021). Global migration is driven
by the complex interplay between environmental and social factors. Environmental Research
Letters, 16(11), 114019. [1]
Nordhaus, W. D., & Moffat, A. (2017). A survey of global impacts of climate change: replication, survey
methods, and a statistical analysis (No. w23646). National Bureau of Economic Research. [3]
Ocello, C., Petrucci, A., Testa, M. R., & Vignoli, D. (2015). Environmental aspects of internal migration
in Tanzania. Population and Environment, 37(1), 99-108. [1]
Ortiz-Bobea, A. (2013). Is weather really additive in agricultural production. Implications for climate
change impacts Resources For the Future Discussion Papers. [3]
Overeem, A., R. Robinson, J. C., Leijnse, H., Steeneveld, G. J., P. Horn, B. K., & Uijlenhoet, R. (2013).
Crowdsourcing urban air temperatures from smartphone battery temperatures. Geophysical Research
Letters, 40(15), 4081-4085. [3]
Pace, N., Sebastian, A., Daidone, S., Prifti, E., & Davis, B. (2022). Mediation analysis of the impact of
the Zimbabwe Harmonized Social Cash Transfer Programme on food security and nutrition. Food
Policy, 106, 102190. [4]
Parnell, S., & Ruwani, W. (2011). Sub-Saharan African urbanisation and global environmental change.
Global Environmental Change, 21, S12-S20. [1]
Peri, G., & Sasahara, A. (2019). The impact of global warming on rural-urban migrations: evidence from
global big data (No. w25728). National Bureau of Economic Research. [1]
Piguet, E., Pécoud, A., & De Guchteneire, P. (2011). Migration and climate change: An
overview. Refugee Survey Quarterly, 30(3), 1-23. [1]
Prifti, E., Daidone, S., & Davis, B. (2019). Causal pathways of the productive impacts of cash transfers:
Experimental evidence from Lesotho. World Development, 115, 258-268. [4]
48
Renaud, F. G., Dun, O., Warner, K., & Bogardi, J. (2011). A decision framework for environmentally
induced migration. International Migration, 49, e5-e29. [1]
Rigaud, K. K., de Sherbinin, A., Jones, B., Bergmann, J., Clement, V., Ober, K., Schewe, J., Adamo, S.,
McCusker, B., Heuser, S., & Midgley, A. (2018). Groundswell: Preparing for internal climate migration.
The World Bank. [1]
Robalino, J., Jimenez, J., & Chacón, A. (2015). The effect of hydro-meteorological emergencies on
internal migration. World Development, 67, 438-448. [1]
Schutte, S., Vestby, J., Carling, J., & Buhaug, H. (2021). Climatic conditions are weak predictors of
asylum migration. Nature communications, 12(1), 1-10. [1]
Selby, J., Dahi, O. S., Fröhlich, C., & Hulme, M. (2017). Climate change and the Syrian civil war
revisited. Political Geography, 60, 232-244. [1]
Spyratos, S., Vespe, M., Natale, F., Weber, I., Zagheni, E., & Rango, M. (2019). Quantifying
international human mobility patterns using Facebook Network data. PloS one, 14(10), e0224134. [2]
Taylor, L. (2016). No place to hide? The ethics and analytics of tracking mobility using mobile phone
data. Environment and Planning D: Society and Space, 34(2), 319-336. [2]
Tol, R. S. (2018). The economic impacts of climate change. Review of Environmental Economics and
Policy, 12(1), 4-25. [3]
United Nations Framework Convention on Climate Change. 2012. Slow onset events. Technical Paper
FCCC/TP/2012/7. https://unfccc.int/resource/docs/2012/tp/07.pdf [3]
Viswanathan, B., & Kumar, K. K. (2015). Weather, agriculture and rural migration: evidence from state
and district level migration in India. Environment and Development Economics, 20(4), 469. [1]
Wesolowski, A., Buckee, C. O., Pindolia, D. K., Eagle, N., Smith, D. L., Garcia, A. J., & Tatem, A. J.
(2013). The use of census migration data to approximate human movement patterns across temporal
scales. PloS one, 8(1), e52971. [2]
Wodon, Q., Liverani, A., Joseph, G., & Bougnoux, N. (Eds.). (2014). Climate change and migration:
evidence from the Middle East and North Africa. The World Bank. [1]
49
Zander, K. K., Richerzhagen, C., & Garnett, S. T. (2019). Human mobility intentions in response to heat
in urban South East Asia. Global Environmental Change, 56, 18-28. [1]
Zickgraf, C., & Perrin, N. (2016). Immobile and trapped populations. The atlas of environmental
migration. [1]
50
Appendix
Table A.1. Relevant information for the climate-migration nexus
in standard LSMS-ISA collection questionnaires
Questions/answers/sections potentially related
Survey ID Number Survey Name Country
to the climate-migration nexus
Question CS4:
Quelle a été la stratégie adoptée par le ménage après le
Enquête
Burkina [CHOC] pour faire face à la situation?
BFA_2014_EMC_v01_M Multisectorielle
Faso
Continue 2014
Option 11:
Migration d'un ou plusieurs membres du ménage
Question 8.4:
Rural What did your household do in response to this
Socioeconomic [SHOCK] to try to regain your former welfare level?
ETH_2011_ERSS_v02_M Ethiopia
Survey 2011-
2012 Option 8: Household members migrated
Question 1.22:
Why did [NAME] leave this household?
Option 4: Left to find better land
Option 5: Health reasons
Option 6: Security reasons
Question 1.30:
What was the most important reason [NAME]
Socioeconomic migrated away?
ETH_2013_ESS_v03_M Survey 2013- Ethiopia
2014 Option 4: Left to find better or more land
Option 5: Health
Option 7: For security
Question 8.4:
What did your household do in response to this
[SHOCK] to try to regain your former welfare level?
Option 8: Household members migrated
Socioeconomic
ETH_2015_ESS_v03_M Survey 2015- Ethiopia Question 1.22:
2016 Why did [NAME] leave this household?
51
Option 4: Left to find better land
Option 5: Health reasons
Option 6: Security reasons
Question 8.4:
What did your household do in response to this
[SHOCK] to try to regain your former welfare level?
Option 8: Household members migrated
Question 1.23:
Why did [NAME] leave this household?
Option 4: Left to find better land
Option 5: Health reasons
Socioeconomic Option 6: Security reasons
ETH_2018_ESS_v02_M Survey 2018- Ethiopia
2019
Question 9.4:
What did your household do in response to this
[SHOCK] to try to regain your former welfare level?
Option 8: Household members migrated
Question B13:
What was the main reason that [NAME] moved here?
Third Integrated Option 11: Looking for land to farm
Household
MWI_2010_IHS-III_v01_M Malawi
Survey 2010- Question U04:
2011 What did your household do in response to this
[SHOCK] to try to regain your former welfare level?
Option 8: Household members migrated
Question B13:
What was the main reason that [NAME] moved here?
Integrated
Option 11: Looking for land to farm
Household Panel
Survey 2010-
MWI_2010-2013_IHPS_v01_M Malawi
2013 (Short-
Question U04:
Term Panel, 204
What did your household do in response to this
EAs)
[SHOCK] to try to regain your former welfare level?
Option 8: Household members migrated
Fourth Integrated
Question B13:
Household
MWI_2016_IHS-IV_v04_M Malawi What was the main reason that [NAME] moved here?
Survey 2016-
2017
Option 11: Looking for land to farm
52
Question U04:
What did your household do in response to this
[SHOCK] to try to regain your former welfare level?
Option 8: Household members migrated
Question B13:
What was the main reason that [NAME] moved here?
Option 11: Looking for land to farm
Fifth Integrated
Household
MWI_2019_IHS-V_v04_M Malawi
Survey 2019-
Question U04:
2020
What did your household do in response to this
[SHOCK] to try to regain your former welfare level?
Option 8: Household members migrated
Question 11.05 (Questionnaire Part 2):
Quelle a été la stratégie adoptée par le ménage après le
Enquête Agricole
MLI_2014_EACI_v03_M [CHOC] pour faire face à la situation?
de Conjoncture Mali
Intégrée 2014
Option 11:
Migration de membres du ménage
Enquête Agricole Question 5.05 (Questionnaire Part 2):
de Conjoncture Quelle a été la stratégie adoptée par le ménage après le
Intégrée aux [CHOC] pour faire face à la situation?
MLI_2017_EAC-I_v03_M Mali
Conditions de
Vie des Ménages Option 11:
2017 Migration de membres du ménage
Question 11.05 (Questionnaire Part 2):
National Survey
Quelle a été la stratégie adoptée par le ménage après le
on Household
[CHOC] pour faire face à la situation?
NER_2011_ECVMA_v01_M Living Niger
Conditions and
Option 11:
Agriculture 2011
Migration d’un ou plusieurs membres du ménage
National Survey
Question 10.05 (Questionnaire Part 1):
on Household
Quelle a été la stratégie adoptée par le ménage après le
Living
[CHOC] pour faire face à la situation?
NER_2014_ECVMA-II_v02_M Conditions and Niger
Agriculture
Option 11:
2014, Wave 2
Migration d’un ou plusieurs membres du ménage
Panel Data
53
Question 1.33 (Post-planting questionnaire)
Why did [NAME] leave this household?
Options 4: Left to find better land
Options 5: Health reasons
Options 6: Security reasons
General Questions 1.35 (Post-planting questionnaire) & 1.39
Household (Post-harvesting questionnaire):
NGA_2010_GHSP-W1_v03_M Survey, Panel Nigeria What was the most important reason [NAME]
2010-2011, migrated abroad?
Wave 1
Options 2: To find better or more land
Options 3: Health
Options 8: Security
Question 15A.4 (Post-harvesting questionnaire):
Rank the 3 most significant shocks you have
experienced (1) most severe; (2) second most severe;
(3) third
Option 10: Members of the household migrated for
work
Questions 1.29 (Post-planting questionnaire) & 1.28
(Post-harvesting questionnaire)
Why did [NAME] leave this household?
Options 4: Left to find better land
Options 5: Health reasons
General Options 6: Security reasons
Household
NGA_2012_GHSP-W2_v02_M Survey, Panel Nigeria
2012-2013, Questions 1.35 (Post-planting questionnaire) & 1.34
Wave 2 (Post-harvesting questionnaire):
What was the most important reason [NAME]
migrated abroad?
Options 2: To find better or more land
Options 3: Health
Options 8: Security
54
Question 15A.5 (Post-harvesting questionnaire):
What was the most important consequence of the most
recent [SHOCK]?
Option 10: Members of the household migrated for
work
Questions 1.29 (Post-planting questionnaire) & 1.28
(Post-harvesting questionnaire)
Why did [NAME] leave this household?
Options 4: Left to find better land
Options 5: Health reasons
Options 6: Security reasons
Question 1.34 (Post-harvesting questionnaire):
General
What was the most important reason [NAME]
Household
migrated abroad?
NGA_2015_GHSP-W3_v02_M Survey, Panel Nigeria
2015-2016,
Options 2: To find better or more land
Wave 3
Options 3: Health
Options 8: Security
Question 15A.5 (Post-harvesting questionnaire):
What was the most important consequence of the most
recent [SHOCK]?
Option 10: Members of the household migrated for
work
Questions 1.29 (Post-planting questionnaire) & 1.28
(Post-harvesting questionnaire)
Why did [NAME] leave this household?
Options 4: Left to find better land
Options 5: Health reasons
General
Options 6: Security reasons
Household
NGA_2018_GHSP-W4_v02_M Survey, Panel Nigeria
2018-2019,
Wave 4 Question 15A.7 (Post-harvesting questionnaire):
How did your household cope with the most recent
[SHOCK]?
Option 10: Members of the household migrated for
work
National Panel
Question R.6:
TZA_2012_NPS-R3_v01_M Survey 2012- Tanzania
What did your household do in response to this
2013, Wave 3
[SHOCK] to try to regain your former welfare level?
55
Option 8: Household members migrated
Question 16.4:
How did your household cope with this [SHOCK]?
Option 7: Household member(s) migrated
National
Household
UGA_2009_UNHS_v01_M Uganda
Survey 2009- Question 3.18: What was the main reason for moving
2010 to the current place of residence?
Option 3: Drought, flood or other weather related
condition
Question 16.4:
How did your household cope with this [SHOCK]?
Option 7: Household member(s) migrated
National Panel
UGA_2010_UNPS_v01_M Survey 2010- Uganda
2011 Question 3.18: What was the main reason for moving
to the current place of residence?
Option 3: Drought, flood or other weather related
condition
Question 16.4:
How did your household cope with this [SHOCK]?
Option 7: Household member(s) migrated
National Panel
UGA_2011_UNPS_v01_M Survey 2011- Uganda
2012 Question 3.18: What was the main reason for moving
to the current place of residence?
Option 3: Drought, flood or other weather related
condition
Question 16.4:
How did your household cope with this [SHOCK]?
Option 7: Household member(s) migrated
National Panel
UGA_2013_UNPS_v01_M Survey 2013- Uganda
2014 Question 3.18: What was the main reason for moving
to the current place of residence?
Option 3: Drought, flood or other weather related
condition
56
Question 16.4:
How did your household cope with this [SHOCK]?
Option 7: Household member(s) migrated
National Panel
UGA_2015_UNPS_v01_M Survey 2015- Uganda
2016
Question 3.18: What was the main reason for moving
to the current place of residence?
Option 3: Drought, flood or other weather related
condition
Question 16.4:
National Panel
How did your household cope with this [SHOCK]?
UGA_2018_UNPS_v01_M Survey 2018- Uganda
2019
Option 7: Household member(s) migrated
Question 16.4:
National Panel
How did your household cope with this [SHOCK]?
UGA_2019_UNPS_v01_M Survey 2019- Uganda
2020
Option 7: Household member(s) migrated
57