Predicting environmental migration

Author Bo R Döös discusses the predictability of large-scale environmental migration. The author is Chairman of Global Environmental Management in Vienna, Austria.

IN MANY PARTS of the world people are being forced to leave their homelands due to various types of environmental problems. According to recent studies the number of such refugees is now at least 10 million, but it may be as much as 25 million. Although the number of these refugees today is alarming, it cannot be excluded that during the next decades forced displacement of people — environmental migration — will increase dramatically.

So far such forced displacement of people has, in most cases, been over comparatively short distances and confined to their home country, or to a neighbouring country. In the future, however, refugees may be forced to move considerably farther away from their country of origin.

Some studies have been concerned with the problem of trying to estimate the future magnitude of large-scale environmental migrations, when and where they are likely to occur and their expected directions. However, practically no attention has been given to evaluate to what extent they can be predicted. Indeed, this is not very surprising due to the complexity of the problem and the limited prospects of reaching firm conclusions.  

Even taking a simplistic approach in developing a method for prediction of environmental migration it is apparent that, as well as the numerous environmental factors, there are many other factors, such as population growth, famine, war, religious intolerance, unemployment and poverty, which come into play.  

In view of the interdependencies of the various "driving forces" behind migrations it can be argued it is not entirely appropriate to use the term "environmental migration." Nevertheless, it will be used in this paper with the understanding that environmental factors are not the only factors contributing to the onset of migrations but that they can play an important role.

Given, then, that all these "driving forces" are difficult to quantify and possess a limited degree of predictability, it cannot be expected that their integrated effect can be predicted with a high degree of accuracy. At the same time it should be recognized that what is required in the first hand is not a detailed and precise prediction of future environmental migrations. Even an approximate prediction, indicating those regions which are particularly vulnerable and the order of magnitude of the number of people who may be forced out of these regions, can be of vital importance for the planning of concrete response measures.  

It should also be pointed out that, in general, governments of more developed nations have not demonstrated any pronounced interest in understanding the migration problem, its nature and possible magnitude for the future. Nor have they seriously addressed the question as to what kind of preventive measures could be taken to reduce the risk of such an involuntary mass displacement of people. Instead, it seems their interest is mainly focused on how to obviate inflows of refugees from developing countries.  

Nevertheless, the problem remains, and in the following discussion an attempt is made to assess the present ability to predict environmental migration. In particular, we will then be concerned with migrations that are  driven basically by insufficient food availability, and migrations forced by a rising sea level as a consequence of a climatic change induced by an enhanced greenhouse effect.

Major causes and linkages

In trying to identify the factors that play a major role in influencing the risk of environmental migration it is often difficult to differentiate between refugees driven by environmental factors and those driven by economical problems. Another point is that any type of migration, regardless of what its major driving force might be, can have consequences leading to political instability and military conflicts, thereby having a harmful impact on the environment which in turn can generate or reinforce environmental migration. In a very schematic way such interdependencies of influencing factors are given in Figure 1.

Figure 1: Schematic illustration of the interconnections between the major factors that can influence or reinforce environmental degradation resulting in an increased risk of environmental migration.

Population pressure

There is no doubt that rapid population growth can represent one of the most serious driving forces behind any type of migration. This is definitely true for environmental migration. Thus an increased density of the population in a certain region may lead to increased environmental stresses and a decline in food production at the same time as the demand for food production is increasing.

Environmental degradation

Due to natural processes and human activities the environment is degraded in numerous ways. However, here we will primarily be concerned with  environmental stresses which, directly or indirectly, can be expected to have a significant influence on the risk for a mass displacement of people, for example, severe reductions of food production or extensive flooding of low-lying coastal zones. A list of such stresses is given in Table 1 in which it is indicated whether these stresses are mainly caused by natural processes or anthropogenic activities, or both.

With regard to the possibilities of predicting to what extent these stresses can have an impact on the environment we are confronted with different types of limitations. For example, we have not sufficient information about the past and present magnitude of these stresses, and our knowledge about the various processes governing these stresses is limited. Another limitation is caused by the difficulties in estimating to what extent governments are willing and able to take decisive actions aimed at reducing environmental degradation.

Natural disasters

According to a recent study more than four million people were killed during the period 1967-1991 by various types of natural disasters, such as tropical and extra-tropical cyclones, droughts, flooding, earthquakes and volcanic eruptions.

The prospects of achieving a definite skill in predicting the occurrence and magnitude of such disasters is indeed limited. The only statement that can be made with some confidence is that due to an increasing population the number of people affected will also increase over time.

Socio-economic-political factors

In many countries, particularly in the developing world,  governments have not been able to cope with the problems facing their country which implies a risk of an acceleration of the chain of processes leading to migration and conflicts as illustrated in Figure 1. There are many reasons for this: for example, the sheer number of socio-economic-political problems; a lack of knowledge and experience in how to manage them; insufficient resources; and poor infrastructure. It has to be recognized that the predictability of these factors is very low.

In almost all countries there exist minorities (for example of ethnic, tribal or religious character) whose existence can often cause problems and conflicts that governments are seldom able to resolve in a completely satisfactory way. The problems can become so severe that they lead to political instability, civil war and large-scale migrations, thereby contributing to degradation of the environment.

There is no doubt it will be possible to continue to increase global food production during the next few decades. However, environmental stresses will become more and more dominant. It is still an open question as to whether global food production can be expected to increase at a greater rate than that of the world population. The situation in many developing countries cannot be viewed optimistically. In 1993 about two thirds of the developing countries recorded stagnant or declining levels of per capita food production.  

Clearly, present predictions of global food production and regional distribution cannot be considered reliable. Nevertheless, they can be judged sufficiently accurate to identify weak points in the global food production system.

It should be emphasized that efforts to increase food production can have negative consequences through different types of feedback processes. For example, in many regions the need for expansion of agricultural land has significantly contributed to the increased rate of tropical deforestation, implying an increased rate of carbon emission, thereby accelerating greenhouse gas-induced climatic change. This in turn may have a negative impact on food production.  

Food shortages — famines

In addition to the impacts of environmental degradation and rapid population growth, there are a number of other factors such as poverty, insufficient distribution systems and international trade factors that can contribute to food shortages in a given region. Other contributing factors are failure of governments to take action and insufficient responses by donor countries and international organizations, as evidenced by the modest response to the famine of 1984 in Ethiopia and Sudan resulting from opposition in the European Parliament.  

From experience gained during the severe famine in 1975-85 in Ethiopia, it is possible to distinguish between five different stages of the response of the peasantry to crop failures:

  • exchange of livestock for food;
  • attempts to borrow food;
  • scavenging and consumption of food that could be found in the wild;
  • stealing and robbing; and,
  • migration accompanied by begging, misery and high mortality.

Migration and conflicts

Only after environmental degradation and its consequences have reached a critical level is it likely that major movements of people will occur. It should be noted that in most cases the migratory flows cover only comparatively short distances, and are of a temporary character. In other cases, however, the migration is definite and long range, and directed to neighbouring or more distant countries. Such migrations are likely to generate problems ranging from political instability to civil wars and international conflicts.

Estimates of future migrations

As a background to the discussion of the prediction of environmental migration it may be appropriate to provide some information about available estimates of the expected total amount of migrations and those which to a large extent are caused by environmental factors.

All types of migration

The available information is a weak base on which to estimate current inter-regional migrations or derive assumptions of future trends. Nevertheless, estimates have been made which indicate that in 2030 the annual migration may be as much as about four million persons. It is expected that future migrants will move for one of three reasons, just as they do now:

  • political, seeking asylum;
  • economic, seeking better standard of living; and,
  • environmental, fleeing environmental degradation.

However, it is emphasized that the intensity and pattern of migrations can change significantly within a very short time.

Environmental migration

Some estimates of the possible magnitude of future environmental migration have been made. It has been estimated that sea-level rise caused by global warming, together with the ongoing subsidence of some coastal areas, may force a permanent displacement of as much as 100 million people living in low-lying coastal areas by the middle of the next century.  

Crude estimates have also been made of migrations caused by insufficient food availability as a consequence of various types of environmental degradation, including water shortages. Large regions of Africa have been identified to be particularly vulnerable due to the prospect of severe and persistent droughts. Also, due to possible dislocation of monsoon systems, the entire Asia-Pacific region has been identified as exceptionally vulnerable. It is estimated that such disruptions of food production may trigger mass migration of people from famine-affected areas, and that the number of refugees may be as high as about 50 million people by 2050.

Predictability of environmental migration

Based on the discussion above we may conclude that the problem of predicting future environmental migration is of an enormous magnitude, involving a wide range of factors interacting with each other. Moreover, the problem is aggravated by two further factors:

  • first, both with regard to causes and effects, environmental migration cannot be dealt with in isolation from other types of migration; and,
  • second, only a very few of the main driving forces behind environmental migration can be predicted with significant accuracy.

Is environmental migration predictable?

It should be recognized that we cannot take for granted that we are dealing with a deterministic problem. Simply to indicate the reason for making this point, it may be of interest to look into what is known about another (closely related) problem, namely the predictability of weather and climate.

Regarding weather, it has been shown that it is at least theoretically possible to perform a deterministic prediction of the detailed motion of the atmosphere. However, it has also been made clear that there exists a limit in time for such predictions. It has been demonstrated that even if we have the correct system of equations for all the dependent variables governing the atmospheric motion, a perfect observing system, and an infinitely powerful computer, it would still not be possible to produce a useful forecast beyond about two to three weeks.  

The problem of climate prediction is less straightforward. Here we are concerned with the problem of predicting the averages of the various variables that specify the climate, and it is not obvious how this problem should be tackled. Neither is it clear whether the climate is predictable, or not.

However, it should also be pointed out that currently available climate models demonstrate a certain ability to simulate the present climate, at least over global and continental scales. This indicates that these models can be used to estimate the response of the climate system  to changes in the external forcing, for example, due to changes in the atmospheric concentration of greenhouse gases. The computation of such a response of the climate system is usually referred to as a climate sensitivity study.

Clearly, the climate prediction problem is very different from the environmental prediction problem. Nevertheless, there exists a certain similarity from the point of view that in both cases we are primarily interested in the changes of the dependent variables caused by changes of external driving forces. With regard to the climate problem, the change in the driving force is the enhanced greenhouse effect due to human activities. In the case of environmental migration, it is basically environmental degradation and the fast growth in population.

Consequently, we can consider the prediction of environmental migration as a prediction problem of the "second kind" or a sensitivity study. This fact may provide an indication of the existence of a certain amount of predictability.  

Acceptance of uncertainties

Without underestimating the possibilities to develop reliable methodologies for predicting environmental migration, it can safely be stated it will never be possible to predict their occurrences in great detail: for example, with regard to where and when they can be expected to take place, their magnitude and their directions.

It can be stated with confidence though that for the planning and implementation of preventive response measures a certain range of uncertainty of the quantities defining the expected migration can be tolerated. Without going into detail in providing definite levels of uncertainties, the following rough estimates might be considered realistic.

Origin and destination of migrations: It may prove sufficient to identify from which country (or countries) a major, permanent migration can be expected. For the destination, the horizontal resolution may not need to be better than the scale of a continent.

Magnitude: The risk of occurrence of migrations involving more than about hundred thousand people needs to be identified. The uncertainty of the predicted size of such a mass displacement should preferably not exceed an order of magnitude.

Timing: The warning of such flows of migrants ought to be known a few years in advance. Expected possible significant changes of the scale of the migrations should be made available at least a year in advance.

Prospects of significant predictability

Should it be possible to assess to what extent environmental migration is predictable, we need to be able to provide realistic answers to the following two questions: To what extent can the individual influencing factors be predicted? What are the relative contributions of these factors to the risk of environmental migration?

Regarding the first question, the following assumptions are made.

Population pressure: It can be stated that this factor can be predicted far more accurately than any of the other influencing factors.

Environmental degradation: An attempt has been made to estimate both the present level of ability to predict  the various types of environmental factors and what their ultimate limit of predictability might be (see Table 2). In some cases, there is already a not insignificant level of accuracy.  

Socio-economic-political factors: In general, such factors are indeed very difficult to predict, and it seems unlikely that prospects exist for substantial improvements in such predictions. About the same low level of accuracy must also be  assigned to predictions of such factors as political instabilities, national and international conflicts, and related feedback processes.

With regard to the second question it can generally be assumed that:

  • although there are a large number of factors that can influence the probability of an environmental migration, there is only a subset of these factors that can be expected to have a significant influence — that is, different types of migration have different groupings of contributing factors; and,
  • for each type of migration, there are usually only a few factors that play a dominant role — if this limited group of influencing factors possesses a comparatively high level of predictability, then it is likely that this type of migration may be predicted with a fair amount of accuracy.

Given the validity of these assumptions, we do have a certain base for taking an optimistic view of the predictability of some types of environmental migration. It has to be recognized, though, that there exist types of migration that are bound to have a very low predictability. This is particularly true with regard to migrations caused by (the threat of) natural disasters such as earthquakes and volcanic eruptions.  

Estimation of predictability of two different types of environmental migration

As an illustration of the approach indicated above in estimating the predictability of environmental migration we will consider two cases with different groupings of influencing factors: one in which shortage of food is the main driving force and one which is caused by rising sea level due to global warming.

For each of these two cases we will attempt to answer the following basic questions: Which are the main influencing factors? With what accuracy can the predictability of these factors be estimated?

If these questions could be answered with a reasonable degree of precision we would be able to evaluate to what extent the ranking of the most influential factors matches the ranking of the predictability of these factors. If these two rankings coincide, this would indicate a comparatively  high level of predictability of the risk of migration.  

Migration caused by insufficient food availability

In this case, an attempt has been made to provide answers to the two basic questions formulated above (see Table 3) by ranking the various influencing factors according to their relative importance together with indications of their expected level of predictability.

With regard to the predictability of the "natural" occurrence of drought, these events appear to occur quasi-periodically on two timescales: one with comparatively short "episodes" which usually last a year or two; and one with long dry periods spanning a decade or two. Even if the exact timing of these droughts cannot be predicted, there exists a certain base for optimism that they can be predicted one to two years in advance.

Concerning the impacts of a greenhouse gas induced-climatic change on food production, recent studies indicate that they can be predicted with a degree of confidence on global and continental scales.  

Given the approximate and subjective estimates of the ranking and predictability of the factors identified in Table 3, it can be concluded as follows.

  • A global change of food availability per capita can be predicted with a fair amount of accuracy over a period of about one decade.
  • For individual regions the predictability must be considered to be very limited. This is  particularly the case in drought-prone regions where a radical change in the food availability can occur from one year to another.
  • Even if food availability in a region falls below the level of sufficiency, it may take some years before the stage of famine occurs. i.e. famines do not strike unexpectedly.
  • In case extensive droughts occur, the question is not so much if, but rather when, famines will occur.
  • It has to be recognized that no reliable formula exists for predicting migration following a famine. Experience suggests that migration is not an early response to a famine. The lag-time may be several years.

It must be acknowledged that past assistance to developing countries suffering from famines has been both reluctant and insufficient.

Migrations caused by a sea-level rise

One of the most serious implications of global warming is that it will cause a rise in global sea level due to expansion of the sea-water and melting of  glacier ice. The present best estimate of the magnitude of this rise is that it will be about half a metre by the year 2100, with a range of uncertainty of 13-94cm.  

Increase of local sea level will affect low-lying coastal areas in several ways: permanent loss of land, temporary flooding  and salt water intrusion. It should also be recognized that a rise sea level will affect the water level of rivers and cause problems tens of kilometres inland.

In trying to estimate the predictability of this type of migration, we are basically concerned with only one major driving force, the rate of increase of sea level. However, even if population pressure is not a direct driving force it needs to be taken into account in predicting the magnitude of the expected migration.  

As a general conclusion it can be stated that even if it is not possible to assign accurately the rate at which sea level will rise, it is beyond doubt that it will rise, and that this eventually will cause extensive migrations from low-lying, densely-populated coastal areas. This in turn may cause a "domino effect,"  resulting in migration to more distant destinations.

More specifically, we may safely make the following conclusions.

  • In many densely populated coastal areas the subsidence of land is currently contributing much more to the change in relative mean sea level than the rise of global sea level, and it can also be predicted with a better accuracy.
  • Thus, it can be said that the rise in relative mean sea level can be predicted with some accuracy permitting fairly realistic predictions of future migrations in coastal areas where the relative mean sea level is also affected by local subsidence.
  • In contrast to the migration caused by extensive famines that can develop over a comparatively short period of time (years), the migration caused by a sea level rise will be spread out over a considerably longer time period (decades). However, it can be expected that in many cases migration from the threatened areas will be accelerated by an increased frequency of temporary flooding caused by storm surges.  
  • Undoubtedly, it is technologically feasible to hold back the sea as water levels rise. However, in some cases it may not be economically or environmentally sound.

It should be noted here that, despite the threat of the expected rise of sea level, few nations are taking this into account in their long-term planning. In many countries there is even a tendency to intensify the development of their coastal areas.


In general, there are a large number of factors that can contribute to the risk of environmental migration, including not only stresses on the environment but also a variety of socio-economic-political factors. In addition, national and international conflicts and wars can, through feedback processes, contribute to degradation of the environment and thereby cause or amplify an involuntary displacement of people.

Moreover, the problem of predicting environmental migration is aggravated by the fact that some of the influencing factors either have very low, or no, predictability. However, with regard to the most serious types of environmental migration, there are compelling reasons to believe they can be predicted with sufficient reliability to motivate mitigating measures.

As an overall conclusion with regard to the entire migration problem (not only migrations caused by degradation of the environment) it can be stated that the more developed countries have so far demonstrated little interest and imagination in its possible occurrence and consequences. It appears they believe the problem can be solved simply by sharpening their immigration policies.

Further information

Bo R Döös, Global Environmental Management, Jordangasse 7/13, A-1010 Vienna, Austria. Fax: 43-1-5350188.

This article is based on a more detailed discussion by the author published in Global Environmental Change, 7(1), 41-61, 1997.