Focus Article

Linking climate change,

environmental degradation,
and migration: a methodological
overview
Etienne Piguet∗

Empirical research focusing on the links between climate change, environmental

degradation, and forced migration has risen significantly in recent years and uses an
impressive variety of methods. The present article suggests a typology identifying
six research method families: ecological inference based on area characteristics,
individual sample surveys, time series, multilevel analysis, agent-based modeling
(ABM), and qualitative/ethnographic studies. The main technical features and
empirical results of each family of methods are presented and critically discussed.
We conclude by calling for a coordinated international effort to improve the
quality and variety of data that could be used with existing research methods
and significantly improve our understanding of the migration-environment
nexus.  2010 John Wiley & Sons, Ltd. WIREs Clim Change 2010 1 517–524

A lthough a wide spectrum of estimates, ranging

anywhere up to 1 billion ‘climate refugees’ by
2050, have been put forward by NGOs and advocacy
agency in deciding to migrate or to choose other
adaptation strategies.11 For these reasons, migration
induced by climate change cannot be modeled in the
organizations, the numbers ‘circulating’ in the media same ‘hardware modeling’ way as the climate itself,
are nothing but rule of the thumb—more or less and the idea of producing quantitative predictions of
informed. It is clear among scholars that there migration, assorted with probabilities of occurrence,
are no established methods of providing overall is little more than a dream.12–14
quantitative predictions concerning the additional Despite this major limitation, at least two
human migration that might be caused by climate research strategies appear scientifically relevant with
change, and that there is truly no such thing respect to migration and climate change. The first
as a climate or environmental migrant in the is mainly descriptive and prospective. It focuses on
narrow sense of a migrant exclusively moving for the identification of the main regions threatened by
environmental reasons. Except in extreme cases, environmental degradation (the so-called hotspots)
population displacements are always the result of and on integrated assessments of the vulnerability and
a multicausal relationship between environmental, resilience of their inhabitants, which provide insights
political, economic, social, and cultural dimensions.1–8 into possible future migrations.15,13,16 In several
Environmental stressors do not impact equally on cases, environmental degradation scenarios could be
all individuals, households, and communities, and incorporated in economic models that are then used
information related to climate change is not perceived to forecast migration, but these are as yet largely
in the same way everywhere and by everyone.9,10 unbeaten tracks.17 The second research strategy is ana-
Even when confronted with severe environmental lytical and attempts to disentangle the environmental
degradations, human beings and communities are impact from other migration drivers. Empirically,
resilient and have at least some minimum level of it questions the role and weight of environmental
factors in already occurring human migration.
∗ Correspondence to: [email protected] The present article deals exclusively with this last
Etienne Piguet Institut de géographie IGG, Université de Neuchâtel, question and presents a critical assessment of the dif-
Espace L-Agassiz 1, 2000 Neuchâtel, Switzerland ferent methods used in its response. Such an attempt
DOI: 10.1002/wcc.54 at a systematic methodological inventory is missing

Vo lu me 1, Ju ly /Au gu s t 2010  2010 Jo h n Wiley & So n s, L td. 517

 Focus Article wires.wiley.com/climatechange

so far in the literature dealing with the environment– cyclones, etc., most studies that apply ecological infer-
migration nexus, as well as in more general syntheses ence deduce a significant impact of the environment
of methods related to the population–environment on emigration (Saldana-Zorilla between Mexican
relationship.18 It is, however, rendered necessary municipalities,25 Munshi between Mexican provinces
by the growing scientific interest in the topic, the and the United States,23 Naudé in sub-Saharan
resultant recent upsurge in empirical research—in Africa,26 Van der Geest in Ghana,27 Henry et al.
postdisaster situations for example—and the variety in Burkina Faso,28 Chopra and Gulati in India,29
of methods used. We suggest a six-group typology Barrios30 and Reuveny31 among developing countries,
of empirical studies and outline the variables used and Afifi and Warner across 172 countries of the
to capture environmental change and migration. We world32 ), but the level of correlation varies greatly
discuss the principal results and the pros and cons of across these works, and environmental variables
the six methods, before briefly outlining future direc- always appear as only one driving force of migration
tions for data collection and empirical research. We among others. In the case of interprovincial migration
do not consider here the inverse relationship, namely in Burkina Faso, for example, they add only 5% to the
the impact of migration on the environment.19–21 explanation of migration measured by a coefficient
of determination (R2 ).28 No correlation at all has
been found when the dependent migration variable
TYPE 1: ECOLOGICAL INFERENCE is limited to asylum requests lodged in Europe. This
specific kind of migration is explained, to a significant
BASED ON AREA CHARACTERISTICS
extent, by other factors such as the political situation
The central idea of ecological inference is to in zones of departure.24
reconstruct individual behavior from group-level data. As those results testify, ecological inference is a
The word ‘ecological’ indicates that the unit of very fruitful approach. Ecological variables are often
analysis is not an individual but a group of people, much easier to collect than individual data and allow
usually corresponding to a geographical area. The for a good level of comparability between studies.
hypothesis here is that if the environment plays Two limitations should be mentioned. The first lies in
a role in migration decisions, the environmental the paucity of the environmental variables used: most
characteristics of a specific geographic area should indicators are very basic and concern either rainfall
be correlated with the migratory characteristics of or natural disasters, leaving aside more elaborate
that same area during the same period of time (or indicators of climate change or environmental degra-
after a certain time lag). To give an example: there dation. This limitation could be overcome by a more
should be a correlation between the intensity of systematic exploitation of existing data collected by
natural disasters in the municipalities of a country organizations such as the World Meteorological Orga-
and the emigration rates of these municipalities. One nization, the Centre for Research on the Epidemiology
can trace back to Durkheim’s work on suicide22 this of Disasters, the International Earth Science Informa-
general approach to inferring causal relationships. tion Network, etc., and by the identification of key
Considering that many factors impinge on migration, environmental variables related to migration which
contemporary researchers nowadays use multivariate could be better collected at a world or regional level.
methods to control the effect of socioeconomic or A second limitation is more directly related to
political confounding variables so as to isolate the the method itself and the fact that exposures and
specific impact of the environment. Whereas, many responses are only measured for spatial aggregates
existing studies explicitly target the environmental rather than for individuals. This alludes to the well-
dimension of migration in a direct way, interesting known problem of ‘ecological fallacy’: correlations
results also stem from studies intended at analyzing measured at the aggregated level might not hold true
migration determinants in a more general context, at the individual level.33 In other words, nothing
albeit one that includes environmental explanatory guarantees that the very people who emigrated and
variables.23,24 Numerous studies enhance their sample contributed to a negative migration balance in an
by using pooled data at different time periods: for one area under environmental stress, for example, are the
single variable such as, for example, pluviometry, the same individuals who experienced that environmental
characteristics observed in one area during period 1 stress and took a decision to migrate accordingly.
constitute one observation while the characteristics One could argue that the problem of ecological
observed during period 2 constitute another, etc. fallacy remains less severe in environmental/migration
Based on different types of environmental studies than in studies linking two purely individual
indicators including rain, drought, floods, tropical characteristics (such as literacy and origin) because

518  2010 John Wiley & Sons, L td. Volume 1, July/Augu st 2010
 WIREs Climate Change Linking climate change, environmental degradation, and migration

the environmental variables themselves are not based against disaster-induced migration’.34 Halliday37
on aggregated individual characteristics. One must utilizes panel data among 739 rural households in
nevertheless bear in mind that in interpreting the El Salvador. He shows in a multivariate model that
above empirical results, any consideration regarding adverse agricultural conditions did increase migration
the mechanisms at stake should be done at the level to the United States during the nineties but that the
of the area and not at that of the individuals. In the 2001 earthquakes—in accordance with Paul concern-
same vein, ecological inference makes it very difficult ing the counterintuitive impact of sudden disaster—ac-
to differentiate the impact of environmental variables tually reduced net migration to the United States.
between population subgroups, in relation to gender Finally, an impressive 108-month long panel study
or socioeconomic status for example, unless one can conducted between 1997 and 2006 in 151 Nepalese
use specific migration data for these groups. All of neighborhoods of the Chitwan valley35,38 shows that
these difficulties are reinforced, especially in Low whereas the quality of drinking water has no impact on
Income Countries, by the varying size and shape of the population displacements, deforestation, population
spatial units employed and by the fact that they might
pressure, and agricultural decline do indeed produce
cut across meaningful cultural, economic, social, and
elevated rates of local population mobility, but no
environmental boundaries.
significant increases in interregional or international
migration. These results partly contradict a previous
TYPE 2: INDIVIDUAL SAMPLE SURVEYS study using the same method in the same area but
with a smaller sample and over a shorter time span.39
This second approach differs from the first precisely One main weakness of the aforementioned stud-
because it is aimed at considering processes at ies is that environmental change is only very incom-
the level of individuals or households. Data on pletely captured. In certain cases, the information
environmental pressure and socioeconomic context on environmental evolution is limited to one single
are collected through relatively large surveys (from a documented event (hurricane, drought, etc.) and the
few hundred to several thousand cases). The surveys analysis compares ‘before’ and ‘after’ situations.36,34
either inquire about past migrations (reconstitution of In other instances, the environmental situation at the
biographies)34 or take the form of a panel in which beginning of the period is used as a predictor of all
households are contacted several times and questioned future migrations.35,38 Halliday does ask questions
about the migration of one, or several, of their about agricultural shocks in all three waves of his
members during the intervals.35–37 This information panel survey, but the level of detail is limited to
is then used as a dependent variable in regression
‘harvest loss’ and ‘livestock loss’.37 On the whole,
models35,37 or to compute simple cross-tabulations.36
none of the studies based on sample surveys draws
Environmental variables are captured either by
on detailed environmental evolutions captured along
asking direct questions in the survey or by collect-
the whole period under review, and it thus remains
ing information at the local level. One of the most
difficult to disentangle environmental variables from
cited studies using this approach is based on two
other contextual effects. Just as ecological inference
surveys (1982 and 1989) conducted in rural Mali
with a pool of 7079 individuals and 309 households can be subject to ecological fallacy (cf. section Type 1:
before and after a series of droughts that affected Ecological Inference Based on Area Characteristics),
the country.36 The results document no increase in we see that analyses strictly centered on individual
international emigration but shorter-cycle migration data are symmetrically subject to the so-called atom-
from food-short to food-surplus zones. Most of the istic fallacy of missing the context in which behavior
other studies using similar methods also emphasize takes place.40 Although the studies just mentioned do
the complexity and indirect linkages between migra- lead to some very valuable results, this is an impor-
tion and environmental variables: Paul questions 291 tant shortcoming that is not intrinsically linked to the
respondents from eight tornado-affected villages in method itself but to its implication on data collection.
Bangladesh and discovers that none of their house- It could be overcome by designing large panel ques-
hold members had migrated because of the 2004 tionnaires including, over a sufficiently long period
tornado, that respondents were unaware of any out- of time, a broader array of environmental questions,
migration within their localities, and that one-third of or by combining local information on environmental
respondents even suspected that outsiders had been evolutions with repeated waves of questionnaires.41
flocking into the tornado-affected areas in the hope of Such research strategies are costly but, as demon-
benefiting from disaster relief schemes. These results strated by the Nepalese case mentioned above, they
led him to provocatively entitle his paper ‘Evidence may lead to very valuable new insights.

Vo lu me 1, Ju ly /Au gu s t 2010  2010 Jo h n Wiley & So n s, L td. 519

 Focus Article wires.wiley.com/climatechange

Several more sophisticated methods have been appeared quite recently in numerous disciplines of the
tested to overcome the failures of the two families of social sciences with the aim of analyzing explanatory
methods we have just referred to; we group them in factors at various levels of aggregation (individual,
three families. household, classroom, geographical area, etc.) or,
more generally, disentangling the specific impact of
different contextual scales.42 These methods appear
TYPE 3, 4, AND 5: TIME SERIES, well suited to the study of human–environment inter-
MULTILEVEL ANALYSIS, actions in geography as they allow for a significant
AND AGENT-BASED MODELING expansion of the range of variables analyzed and
Time series, multilevel analysis, and agent based thus enhance the precision of the analysis.43 They
modeling (ABM) are three very different approaches, have, as yet, been applied to migration by only a
yet all seek to bridge the gap between individual and very limited number of authors. Henry et al.12 are
ecological data or, in other words, to avoid both actually the only ones to fully apply a multilevel
ecological and atomistic fallacies. approach to migration in the case of Burkina Faso.
Although similar to type 1 methods, time series They collected migration histories among 3911
analyses substitute data on temporal evolutions in a individuals and environmental data at the community
given area for data on spatial units. The measure level in about 600 places of origin mentioned by the
of the degree of correlation over time between, migrants. The environmental indicator consists of
for example, monthly pluviometry and migration, rainfall data covering the 1960–1998 period and the
allows this approach to establish if, and to what dependent variable is the risk of the first departure
extent, migration patterns are explained by the of a migrant from his village. Findings suggest that
evolution of environmental parameters, controlling people from drier regions are more likely to engage
for other factors that might evolve during the period. in both temporary and permanent migration to other
Unfortunately, the two studies at our disposal to rural areas, but that short-term moves to distant des-
have taken this route both used a limited number tinations decrease with rainfall deficits. Three other
of variables and periods that make it difficult to draw studies can be mentioned as they introduce contextual
significant results: Van der Geest27 simply compares characteristics of the region of departure as additional
time series of north–south internal migration and information related to households in a ‘type 2’
average annual rainfall in Ghana and obtains the survey study (one can also mention an older paper
counterintuitive result that migration is reduced at on Mexican–US migration44 ). The first study, in
times of most pronounced environmental stress. Nicaragua, shows that a household highly exposed to
Kniveton et al. analyze the relationship between Hurricane Mitch has a higher probability of having a
climate variability in Mexico and migration to the member abroad than a household with similar adap-
United States in the drought prone states of Zacatecas tive capacity but living in a nonexposed area.45 The
and Durango for the 40-year period between 1951 second, in Ethiopia, shows a positive impact between
and 1991.5 They show in the case of Durango the (perceived) local vulnerability to a food crisis
(Zacatecas presenting no significant correlations) that and emigration among 2000 households of 40 village
the greater the rainfall, the larger the emigration. This communities.46 The third, in Ecuador, collects events
result contradicts the conclusions reached by Munshi, histories, including migration information at local,
also for Mexico, with type 1 method.23 Although internal, and international level, from 1995 to 2006
these two studies clearly pave the way for promising among 279 households. This person–year dataset
developments, their conclusions should be handled is completed by time-varying contextual variables
carefully: no real control variables have been used, including, among other control variables and predic-
and migrant numbers are low and statistically not tors of migration, environmental information about
very significant. As noted by Kniveton et al., one precipitations and ‘unusual harvests’.47 Overall, the
major limitation of this approach can also be found results show that environmental conditions play a
in the absence, for much of the world, of monthly role for all three types of migration, but are most
or quarterly migration flow data time series, which significant for local and internal mobility.
would enable us to link changes in the environment at Although the multilevel approach appears very
time ‘t’ with migration at subsequent periods.5 promising, one drawback of the method is the use
Another approach, that we shall call ‘multilevel’, of a predefined hierarchy of spatial units (usually
combines ecological data (including, e.g., satellite the administrative units at which level the data is
imagery), individual data from household surveys and, collected) that might not reflect the spatial distribution
in certain cases, time series. Multilevel methodologies of the phenomenon at stake. To give an example,

520  2010 John Wiley & Sons, L td. Volume 1, July/Augu st 2010
 WIREs Climate Change Linking climate change, environmental degradation, and migration

one half of a unit—say a district—might be exposed not be so common in the field of environmentally
to landslides whilst the other is not. This weakness induced migration, where many stimuli consist of
could only be overcome by defining small enough sudden events with which populations have never had
statistical units to capture the spatial variation of to cope before. These two points render the situation
the environmental degradation. This is the case, for of environmental migration quite different from the
example, in the aforementioned study on Ecuador, classical fields of application of ABM.50 ABM retains
where precipitations were measured with a 1 km clear potential nevertheless. It forces researchers to
resolution. Another drawback is the difficulty to explicitly formalize their hypotheses about the mech-
collect reliable contextual information apart from anisms at stake and could be fruitfully combined with
the most basic climatic data such as pluviometry. participative methods involving local populations in
Another—not yet completed—Ecuador case study the process of building the model (as a game play
seems extremely promising in overcoming these process), as was the case, without an explicit link to
limitations41 . migration, in a recent experiment in Kiribati.50
ABM has recently been advocated by several We have just examined five families of methods
researchers in the field of environment and migration. specifically geared at answering the question of the
According to Kniveton et al. who use ABM in a case exact weight of the environment on migration by
study on Burkina Faso48 : ‘A solution to the complex- using a variety of statistical tools and data. Let us now
ity of climate-migration linkages is to use agent-based briefly introduce one last approach, which mainly uses
models to simulate the behavioral responses of indi- qualitative method.
viduals and households to climate signals, as well as
relevant interactions between these social actors’.5,47
The central idea is to identify or hypothesize the TYPE 6: QUALITATIVE/
rules of behavior that lead to migration decisions in ETHNOGRAPHIC METHODS
a context of multiple stimuli. A computer simulation Qualitative methods have been by far the most
then allows researchers to observe the outcome on widely used research design in recent years. The
a population of agents over time and to modify the number of existing ethnographic local field studies
contextual parameters. One of the great strengths performed since 2005 can be estimated at around
of ABM versus other methods is that it can easily 50 worldwide,51 nearly one half in the context
take into account heterogeneities of behavior between of the EACH-FOR program.52 These studies use
agents (e.g., according to gender) or bounded ratio- either interviews or small sample questionnaires
nalities (the fact that the rationality of individuals among inhabitants of threatened areas, contacts with
is limited by their level of information, cognitive privileged informants, or, in some cases, literature
abilities, and amount of time available to make sources on historical analogs.53–55
decisions), and that interactions between agents and Providing an exhaustive list and summary of all
retroaction loops can be dealt with (e.g., if a certain these case studies is beyond the scope of the present
number of agents decide to emigrate, the remaining paper. One can note that these qualitative approaches
agents face an increased incentive to leave too). are well established and raise far fewer methodological
ABM is not a new method in migration studies. and data difficulties than the quantitative methods
It has been famously used in the past to analyze described in the previous chapters (which does not
segregation processes (e.g., the preference of an ethnic mean they are easier to use!). As a general observation,
group to live in a homogeneous neighborhood along one can note that most case studies strongly support
co-ethnic lines) leading to intraurban migration.49 the multicausality hypothesis regarding migration.
Until now, however, one can note that only very Whereas numerous authors simply confirm that the
tentative studies have used ABM in the field of environment plays a role in migration in many parts of
environment-migration relations, and that no con- the world, others are challenging the idea that climate
vincing results have been published so far. One can change is already a central driver of migration, even in
wonder if the method will really fulfill its promises for areas such as Tuvalu that are considered to be at the
two reasons. First, preexisting knowledge about the forefront in terms of environmental deterioration.56
ways in which people react to environmental stress Although, by definition, they are not in a
and, more specifically, about the reactions of specific position to provide a quantitative measure of the
subgroups is very limited and makes it difficult to cre- weight of environmental factors on migration, such
ate the rules of behavior necessary for ABM. Second, studies offer invaluable insights into people’s attitude
the routine behaviors themselves (i.e., rules and regu- toward, and their perception and representation of,
larities developed over a certain period of time) might climate change in general and the migration option in

Vo lu me 1, Ju ly /Au gu s t 2010  2010 Jo h n Wiley & So n s, L td. 521

 Focus Article wires.wiley.com/climatechange

particular; a central dimension if one wants to gather a collection of studies are as yet impossible. This is
a coherent and complete theory of migration related largely due to the lack of data available to measure
to environmental change.57 migration behavior and environmental evolutions at
temporally and spatially comparable scales.
The collation of results and the combination
CONCLUSION: WAYS FORWARD of methods applied on more relevant datasets thus
appear promising avenues in order to overcome the
It is only through a better understanding of the ways limitations of single approaches. The most illuminat-
in which migration intervenes as a coping strategy ing and original studies that we have referred to make
responding to environmental degradation that local use of data especially developed through time consum-
and regional scenarios of the migratory consequences ing collection processes involving qualitative as well as
of climate change will be conceivable and might, even- quantitative methods. Comparable efforts shall hope-
tually, be aggregated to deliver overall predictions. In fully be intensified. A complementary strategy would
doing so, it should be kept in mind that migration is be to add relevant questions about environmental
only but one of a range of responses to environmental change and migration to existing censuses or large
degradation. It can be a last resort solution but can sample surveys.41 With this objective, official pro-
also be a complementary, efficient individual choice ducers of statistics at the national and international
to promote in situ adaptation at the household or levels should be actively involved in order to lay
community level. Studies should thus not treat migra- the foundations for a large coordinated international
tion outcomes in isolation but connect them with data production effort. This effort will need to be
nonmigration responses. informed by the methodological discussions that have
This methodological review has identified six long taken place in migration studies around questions
families of methods, some in their infancy, others well such as longitudinal versus cross-sectional studies,
established, that can all contribute to the aim of better studies in the origin or the destination, etc.3,58,59
understanding the migration–environment nexus. The In any case, a climate/migration module, inspired
inventory also underlines that the empirical research by methodological discussions from both main-
has often been pursued in isolation by a fairly limited stream migration studies and environmental studies,
number of authors. Meta-studies that could assess the should definitely be introduced in future international
migratory impact of different factors on the basis of research efforts on climate change.

ACKNOWLEDGEMENTS
I would like to thank Raoul Kaenzig, Laurence Crot, and Fabrice Lehmann for their help in preparing this article.

REFERENCES
1. Adamo S. Addressing Environmentally Induced Pop- 5. Kniveton D, Schmidt-Verkerk K, Smith C, Black R.
ulation Displacements: A Delicate Task. Back- Climate Change and Migration: Improving Method-
ground Paper for the Environment Research Network ologies to Estimate Flows. Migration Research Series.
Cyberseminar on Environmentally Induced Popula- Vol. 33. Geneva: International Organization for Migra-
tion Displacements; 2008. Available at: http://www. tion; 2008.
populationenvironmentresearchorg. Accessed March
24, 2010. 6. Perch-Nielsen S, Bättig MB, Imboden D. Exploring
the link between climate change and migration. Clim
2. Barnett J, Webber M. Accommodating Migration to Change 2008, 91:375–393.
Promote Adaptation to Climate Change. Stockholm:
Commission on Climate Change and Development; 7. Piguet E. Climate change and forced migration. New
2009. Issues in Refugee Research. United Nations High
3. Brettell CB, Hollifield JF. Migration Theory—Talking Commissioner for Refugees Research Paper; 2008,
Across Disciplines. London: Routledge; 2007. 153.

4. Hugo G. Migration, Development and Environment. 8. Tacoli C. Crisis or adaptation? Migration and climate
Geneva: IOM International Organization for Migra- change in a context of high mobility. Environ Urban
tion; 2008. 2009, 21:513–525.

522  2010 John Wiley & Sons, L td. Volume 1, July/Augu st 2010
 WIREs Climate Change Linking climate change, environmental degradation, and migration

9. Dessai S, Adger WN, Hulme M, Turnpenny J, 23. Munshi K. Networks in the modern economy: Mexi-
Kohler J, et al. Defining and experiencing dangerous can migrants in the U.S. labor market. Q J Econ 2003,
climate change—an editorial essay. Clim Change 2004, 118:549–599.
64:11–25. 24. Neumayer E. Bogus Refugees? The determinants of asy-
10. Marx SM, Weber EU, Orlove BS, Leiserowitz A, Krantz lum migration to Western Europe. Int Stud Q 2005,
DH, et al. Communication and mental processes: expe- 49:389–410.
riential and analytic processing of uncertain climate 25. Saldaña-Zorrilla S, Sandberg K. Impact of climate-
information. Glob Environ Change 2007, 17:47–58. related disasters on human migration in Mexico: a
11. Grothmann T, Patt A. Adaptive capacity and spatial model. Clim Change 2009, 96:97–118.
human cognition: the process of individual adapta- 26. Naudé W. Conflict, Disasters and No Jobs—Reasons
tion to climate change. Glob Environ Change 2005, for International Migration from Sub-Saharan Africa.
15:199–213. United Nations University, WIDER, Research Paper;
12. Henry S, Schoumaker B, Beauchemin C. The impact of 2008, 85.
rainfall on the first out-migration: a multi-level event- 27. Van der Geest K. North–South migration in Ghana:
history analysis in Burkina Faso. Popul Environ 2004, what role for the environment? Paper presented at
25:423–460. the International Conference on Environment, Forced
13. Kniveton D, Smith C, Black R, Schmidt-Verkerk K. Migration and Social Vulnerability, Bonn, 9–11 Octo-
Challenges and approaches to measuring the migration- ber 2008.
environment nexus. In: Laczko F, Aghazarm C, eds. 28. Henry S, Boyle P, Lambin EF. Modelling inter-
Migration, Environment and Climate Change: Assess- provincial migration in Burkina Faso: the role of
ing the Evidence. Geneva: International Organization socio-demographic and environmental factors. Appl
for Migration (IOM); 2009, 41–111. Geogr 2003, 23:115–136.
14. Perch-Nielsen S. Understanding the Effect of Climate 29. Chopra K, Gulati SC. Migration, Common Prop-
Change on Human Migration - The Contribution erty Resources and Environmental Degradation.
of Mathematical and Conceptual Models. Diploma New Delhi/Thousand Oaks/London: Sage; 2001.
Thesis, Department of Environmental Sciences ETH,
30. Barrios S, Bertinelli L, Strobl E. Climatic change and
Zurich; 2004.
rural–urban migration: the case of sub-Saharan Africa.
15. Black R, Kniveton D, Skeldon R, Coppard D, Murata J Urban Econ 2006, 60:357–371.
A, et al. Demographics and climate change: future 31. Reuveny R, Moore WH. Does environmental degra-
trends and their policy implications for migration. dation influence migration? Emigration to developed
Working Paper, Brighton, Development Centre on countries in the late 1980s and 1990s. Soc Sci Q 2009,
Migration, Globalisation and Poverty, University of 90:461–479.
Sussex; 2008, T-27.
32. Afifi T, Warner K. The Impact of Environmental Degra-
16. Warner K, Ehrhart C, de Sherbinin A, Adamo S, dation on Migrations Flows across Countries. United
Chai-Onn T. In Search of Shelter—Mapping the Effects Nations University, EHS, Working Paper; 2008, 5.
of Climate Change on Human Migration and Dis-
placement. CARE/CIESIN/UNHCR/UNU-EHS/World 33. Robinson WS. Ecological correlations and the behavior
Bank; 2009. of individuals. Am Sociol Rev 1950, 15:351–357.

17. Barbieri AF, Confalonieri UEC. Climate change, migra- 34. Paul BK. Evidence against disaster-induced migration:
tion and health: exploring potential scenarios of pop- the 2004 tornado in north-central Bangladesh. Disas-
ulation vulnerability in Brazil. In: Piguet E, Pécoud A, ters 2005, 29:370–385.
de Guchteneire P. eds., Migration and Climate Change. 35. Bohra-Mishra P, Massey DS. Environmental Degrada-
Paris; Editions de l’UNESCO; 2010, (In preparation). tion and Out-Migration: New Evidence from Nepal. In:
Piguet E, Pécoud A, de Guchteneire P. eds., Migration
18. Lutz W, Prskawetz A, Sanderson WC. Population
and Climate Change. Paris; Editions de l’UNESCO;
and Environment. Methods of Analysis in Supple-
2010, (In preparation).
ment to Population and Development Review. Vol. 28.
New York: The Population Council; 2002. 36. Findley SE. Does drought increase migration? A study
of migration from rural Mali during the 1983–85
19. Black R. Refugees, Environment and Development.
drought. Int Migr Rev 1994, 28:539–553.
London: Longman; 1998.
37. Halliday TJ. Migration, risk and liquidity constraints
20. Carr D. Population and deforestation: why rural migra-
in El Salvador. Econ Dev Cultural Change 2006,
tion matters. Prog Hum Geogr 2009, 33:355–378.
54:893–925.
21. Hugo G. Environmental concerns and international 38. Massey DS, Axinn WG, Ghimire DJ. Environmental
migration. Int Migr Rev 1996, 30:105–131. change and out-migration: evidence from Nepal. Popu-
22. Durkheim E. Le suicide (1897). Paris: PUF; 1995. lation Studies Center Research Report; 2007, 07–615.

Vo lu me 1, Ju ly /Au gu s t 2010  2010 Jo h n Wiley & So n s, L td. 523

 Focus Article wires.wiley.com/climatechange

39. Shrestha S, Bhandari P. Environmental security and Migration in Burkina Faso; 2008. Manuscript avail-
labor migration in Nepal. Popul Environ 2007, able at: http://wwwinformaticssussexacuk/users/cds21/
29:25–38. publications/ (Accessed December 21, 2009).
40. Gray PS, Williamson JB, Karp DA. The Research Imag- 49. Schelling TC. Micromotives and Macrobehavior.
ination: An Introduction to Qualitative and Quantita- New-York: Norton; 1978.
tive Methods. Cambridge: Cambridge University Press; 50. Gilbert N. Agent-Based Models. London: Sage; 2007.
2007. 51. Piguet E, Pécoud A. Introduction. In: Piguet E, Pécoud
41. Bilsborrow RE. Collecting data on the migration- A, de Guchteneire P. eds., Migration and Climate
environment nexus. In: Laczko F, Aghazarm C, eds. Change. Paris; Editions de l’UNESCO; 2010, (In prepa-
Migration, Environment and Climate Change: Assess- ration).
ing the Evidence. Geneva: International Organization 52. Jäger J, Frühmann J, Grünberger S, Vag A. EACH-
for Migration (IOM); 2009, 115–196. FOR—Environmental Change and Forced Migration
Scenarios: Synthesis Report; 2009.
42. Kreft IGG. Introduction: hierarchical linear models:
problems and prospects. J Educ Behav Stat 1995, 53. Arenstam Gibbons SJ, Nicholls RJ. Island abandon-
20:109–113. ment and sea-level rise: an historical analog from the
Chesapeake Bay, USA. Glob Environ Change 2006,
43. Zolnik EJ. Context in human geography: a multilevel 16:40–47.
approach to study human–environment interactions.
54. McLeman R, Mayo D, Strebeck E, Smit B. Drought
Prof Geogr 2009, 61:336–349.
adaptation in rural eastern Oklahoma in the 1930s:
44. Janvry Ad, Sadoulet E, Davis B, Seidel K, Winters lessons for climate change adaptation research. Mitig
P. Determinants of Mexico–US migration: the role of Adapt Strat Glob Change 2008, 13:379–400.
household assets and environmental factors. Depart- 55. McLeman R, Smit B. Migration as an adaptation to
ment of Agricultural and Resource Economics, UC climate change. Clim Change 2006, 76:31–53.
Berkeley, Working Paper Series; 1997, 853. 56. Mortreux C, Barnett J. Climate change, migration and
45. Carvajal L, Pereira I. Evidence on the link between adaptation in Funafuti, Tuvalu. Glob Environ Change
migration, climate disasters, and human development. 2009, 19:105–112.
In: Paper presented at the International Conference on 57. Piguet E. Re-embedding the environment into migration
Environment, Forced Migration and Social Vulnerabil- theory. Paper presented at the Conference Remaking
ity, Bonn, 9–11 October, Munich; 2008. migration theory: intersections and cross-fertilisations
of the Population Geography Research Group of the
46. Ezra M, Kiros G-E. Rural Out-migration in the drought
RGS-IBG, University of Brighton and University of Sus-
prone areas of Ethiopia: a multilevel analysis. Int Migr
sex 2009, 13th–14th May, submitted for publication.
Rev 2001, 35:749–771.
58. Bilsborrow RE. International Migration Statistics:
47. Gray CL. Environment, land, and rural out-migration Guidelines for Improving Data Collection Systems.
in the southern Ecuadorian Andes. World Dev 2009, Geneva: International Labour Office; 1997.
37:457–468.
59. Bilsborrow RE, Oberai AS, Standing G. Migration
48. Smith C, Kniveton D, Black R, Wood S. Climate Surveys in Low Income Countries: Guidelines for Sur-
Change, Migration and Agent-Based Modelling: Mod- vey and Questionnaire Design. London: Croom Helm;
elling the Impact of Climate Change on Forced 1984.

FURTHER READING
Laczko F, Aghazarm C. eds., Migration, Environment and Climate Change: Assessing the Evidence. Geneva: International
Organization for Migration (IOM); 2009.
Piguet E, Pécoud A, de Guchteneire P. Migration and Climate Change. Paris: Editions de l’UNESCO; 2010, in preparation.

524  2010 John Wiley & Sons, L td. Volume 1, July/Augu st 2010