The Demographic Benefit of International Migration: Hypothesis and Application to the Middle Eastern and North African Contexts

The view that international migration has no impact on the size of world population is a sensible one. But the author argues, migration from developing to more industrial countries during the past decades may have resulted in a smaller world population than the one which would have been attained had no international migration taken place for two reasons: most of recent migration has been from high to low birth-rate countries, and migrants typically adopt and send back to their home countries models and ideas that prevail in host countries. Thus, migrants are potential agents of the diffusion of demographic modernity, that is, the reduction of birth rates among nonmigrant communities left behind in origin countries. This hypothesis is tested with data from Morocco and Turkey where most emigrants are bound for the West, and Egypt where they are bound for the Gulf. The demographic differentials encountered through migration in these three countries offer contrasted situations - host countries are either more (the West) or less (the Gulf) advanced in their demographic transition than the home country. Assuming migration changes the course of demographic transition in origin countries, the author posits that it should work in two opposite directions - speeding it up in Morocco and Turkey and slowing it down in Egypt. Empirical evidence confirms this hypothesis. Time series of birth rates and migrant remittances (reflecting the intensity of the relationship kept by emigrants with their home country) are strongly correlated with each other. Correlation is negative for Morocco and Turkey, and positive for Egypt. This suggests that Moroccan and Turkish emigration to Europe has been accompanied by a fundamental change of attitudes regarding marriage and birth, while Egyptian migration to the Gulf has not brought home innovative attitudes in this domain, but rather material resources for the achievement of traditional family goals. Other data suggest that emigration has fostered education in Morocco and Turkey but not in Egypt. And as has been found in the literature, education is the single most important determinant of demographic transition among nonmigrant populations in migrants' regions of origin. Two broader conclusions are drawn. First, the acceleration of the demographic transition in Morocco and Turkey is correlated with migration to Europe, a region where low birth-rates is the dominant pattern. This suggests that international migration may have produced a global demographic benefit under the form of a relaxation of demographic pressures for the world as a whole. Second, if it turns out that emigrants are conveyors of new ideas in matters related with family and education, then the same may apply to a wider range of civil behavior.


Non-Technical Summary
It seems sensible that international migration has no impact on the size of world population. However, this paper argues that international migration may have resulted in a smaller world population than in the non migration scenario. The author claims that most of recent migration has been from high to low birth-rate countries, and since migrants typically adopt and send back ideas that prevail in host countries, they are potential agents of the diffusion of demographic modernity to their country of origin.
The author uses data from three major origin countries: Morocco and Turkey (where emigration is bound for the West), and Egypt (where emigration is bound for the Gulf). These three countries offer contrasted situations: the host countries are either more (the West) or less (the Gulf) advanced in their demographic transition than the home country. He finds empirical evidence that time-series data on birth rates and migrant remittances (reflecting the intensity of the relationship between the emigrants and their home country) are strongly correlated. Correlation is negative for Morocco and Turkey, and positive for Egypt. This suggests that Moroccan and Turkish emigration has been accompanied by a fundamental change of attitudes regarding marriage and birth, while the opposite holds for Egyptian migration.
The broader conclusions from this paper are that migration may have caused a

Introduction
Migration is commonly regarded by development economists as a potentially win-win process, one susceptible to creating net wealth in both regions of origin and destination of migrants. How can the question of whether international migration is a "positive sum game" be transposed to demography? This paper advances the argument that international migration has contributed to contain the demographic explosion, more precisely that population movements from developing to more developed countries during the last decades have resulted in a smaller global world population than the one which would have been attained had no international migration taken place. In other words, it argues that international migration has contributed to reducing the risk of world overpopulation, 1 i.e., it has increased global security through demographic change.
The mechanism through which international migration is hypothesized to play on global demography is simple: most migration during the period of demographic transition 2a period during which international differentials in birth rates are peaking-has been from high to low birth-rate countries. It is assumed that, because international migrants adopt for themselves, and send back to their home countries, models and ideas that prevail in host countries, they are susceptible to be agents of the diffusion of demographic modernity.
The Middle East and North Africa (MENA) offers a particularly interesting case in point for testing the hypothesis that migrants are potential vectors of demographic change. 1 Looking at the long term, the common sense would on the contrary associate migration with the search for vital space, that is with the demographic expansion of mankind rather than the reduction of its rate of growth. As Kingsley Davis (1988) puts it "Liberal political and economic leaders tend to believe that a movement from areas of high population density to areas of low population density is […] desirable [while] their opponents point out that the Earth is already too crowded, that migration by helping to fill the last remaining open spaces, is making the crowding worse » (p. 252-3). In a long historical perspective, it is true that migration has often resulted in peopling scarcely peopled areas. However, things have changed: there are no longer 'empty' places and the world is now divided into well delineated political entities, between which strong economic differentials are the main driver of migration, disregarding population density.
Electronic copy available at: https://ssrn.com/abstract=940516 Several MENA countries witness an intense emigration, with emigrants bound either for the Gulf or for the West, according to countries of departure and time. With regard to demographic differentials encountered through migration, MENA thus offers contrasted situations: host countries of emigrants are sometimes less, sometimes more advanced in their demographic transition than their home countries. If the central hypothesis of the paper is true, then emigration from MENA countries will have modified the course of the demographic transition of origin countries of migrants in two opposite directions, according to places of destination: slowing it down where emigrants are destined to the Gulf and speeding it up where they emigrate to the West.
The paper is divided into five sections. Section 1 examines the place of migration in demographic analysis. Section 2 proposes a general framework of interaction between migration and demographic change. Section 3 is a rapid overview of the well-documented side of this framework, i.e. the impact of migration on the demography of migrants. Section 4 is a very first attempt to explore the other side of the framework, i.e. the impact of migration on the demography of non-migrants in countries experiencing significant rates of emigration, with MENA as an example. Section 5 offers a partial validation of the main hypothesis by comparing Morocco, Turkey and Egypt.

The demographic ideal of a closed population
Migration was never built up by demographers at the same level of formal elaboration as the two other components of population change, i.e. birth rates and death rates which form together the 'natural' growth of any population. Migration is absent from the core model of formal demography -known as the 'theory of stable populations' -and methods for estimating migration are much less settled than those devised for measuring fertility and mortality.
Formal demography is anchored in the tradition of biology. It models population reproduction as the result of two biological processes, natality and mortality. The modern mathematical demography founded by Alfred Lotka simply excludes external migration 3 : "By a very natural abstraction, demographic analysis envisages as a point of departure the case of a closed population, that is to say, a population whose numbers receive new accessions only through births and suffers losses only through deaths, immigration and emigration being excluded" (Lotka 1998(Lotka [1939, p. 53). Basic analytical models-to begin with the most commonly used of them, i.e. the life table which describes the extinction of a generation "in the absence of external migration"-are constructed on the assumption of a "closed" population, i.e. a population that receives or sends no external migration 4 .
True populations however are not closed. States, or nations, define populations and borders separate national populations from one another. As soon as a population is delineated by a border, border crossing becomes one of the factors of its growth and reproduction. For Lotka this is not a reason for introducing migration in the fundamental demographic equations: "Demographic statistics is concerned primarily with human populations, and particularly with certain more or less isolated populations, as for example those of a nation […] The practical problems [posed by migration across borders] are reduced more and more as the area included in the study expands, since emigration and immigration are plainly functions of the border periphery, whereas births and deaths are instead functions of the land area, and the ratio between the periphery and the internal area continuously decreases as the latter increases. Circumstances of politics and commerce further tend to accentuate that effect, so that for an entire country migration can in certain cases be almost negligible as a factor determining the growth of its population […]" (Ibid.) 5 .
States are not only frames of reference for the delimitation of any national population; they also form the actual frame of population data collection. Statistical records thus incorporate migration, implicitly in vital statistics 6 or explicitly in migration statistics. As a result, migration is a matter of interest for demographers, which they take into account as soon as they leave models for tackling real statistics. Their interest in migration can follow two very distinct purposes: either eliminating migration from vital records for its interference with biological demography, or measuring migration for its contribution to overall demography.
Eliminating the interference of external migration with the statistical observation of births and deaths has been an important concern for demographers. Following chemistry 7 , formal demography aims at studying fertility and mortality "in the pure state," and for this 4 Later on, Keyfitz (1968) introduced the notion of "interacting populations", and the tool of "immigration vector" in the mathematics of population. 5 Interestingly, theories of international migration do not put a greater emphasis on demography than the one demographic theory puts on international migration. Economic theories recognise that international differentials are key factors of migration, but they rarely consider demographic differentials (in population density or population growth) as true factors. Only few of them would endorse a statement such as "modern migration stems mainly from the difference in population growth between the developed and the less developed countries » (Davis 1988 p. 256). In fact, there are too many exceptions to make a rule of it. To take a Middle Eastern example, Lebanon is a country of emigration to Saudi Arabia, despite the fact that Lebanon has a much lower rate of natural population growth than Saudi Arabia (below 1% compared with 3.3% in [1995][1996][1997][1998][1999][2000], and Lebanon is a place of immigration for Syrians, despite the fact that Lebanon has a much higher demographic density than Syria. 6 For example, international migration affects death records and consequently the statistical observation of mortality: the death of an emigrant escapes national statistics of countries of origin so that emigration produces the same result as death in reducing the size of a generation. 7 Henry (1972) "Analyser, c'est décomposer un tout en ses parties : […] L'observation nous fournit des données à l'état brut […] Ces données brutes, qui peuvent paraître simples à un esprit superficiel, sont en réalité le fruit de combinaisons ou de mélanges fort compliqués où interviennent quantités d'éléments […] Comme en chimie, c'est à l'analyse qu'il revient d'isoler le phénomène à l'état pur […] Dans nos observations se mêlent l'effet du phénomène qui nous intéresse, la nuptialité, et des phénomènes perturbateurs, mortalité et migrations. » (p. 20-21). purpose needs to remove the blurring effect of external migration (Henry 1972). What would have been recorded numbers of births and deaths if no migration had taken place? This is the question to be solved. Because migration is a selective process and because it changes the course of life, its statistical interference with fertility and mortality is a complex one. Do those who have emigrated have the same probability of giving birth or of dying than those who have not emigrated? To which extent do birth rates and death rates obtained on incomplete statistics (they do not cover emigrants) apply to all members of the generation under consideration? How to deal with the dependence in probability of emigration on one side, and fertility or mortality on the other side? These questions have produced more interesting methodological developments in demography than additional knowledge on migration itself 8 .
Measuring migration also fully enters within the scope of demography, insofar as the overall growth of any population is the addition of its external migration to its natural growth 9 . Because migrants have a specific age profile, their contribution to the age structure of the population has also become a topic of interest in demography, recently rekindled by the worry about consequences of ageing in industrialized countries. The question of how to balance decreasing fertility rates by sustained flows of immigrants has received a certain attention from demographers (Keyfitz 1981, United Nations 2000 10 . However, for logical reasons intrinsic to migration itself the modeling of external migration never went very far in 8 An overview of the (modest) place of migration in demography is given by Keely (2000). 9 The balancing equation of population growth writes: P2 -P1 = B -D + I -E where P1 and P2 name the total population at dates 1 and 2, and B, D, I, E are respectively the numbers of births, deaths, immigrants and emigrants recorded between dates 1 and 2. 10 A much debated report of the United Nations (2000) dedicated to exploring how migration could bring an answer to ageing made use of population projections to answer the question "what level of migration from less developed countries would be required to compensate for negative demographic trends in more developed countries ? » demographic analysis 11 , and for limits belonging to the social rather than biological nature of external migration, no robust framework of determinants comparable to those applied to mortality and fertility was never devised in the demographic study of migration. As Davis (1988) puts it "international migration […] resembles mortality and fertility in being part of the fundamental balancing equation in demography, which says that any population change is a function of natural increase and net migration […] but unlike mortality and fertility, it has no biological constraints and hence no built-in limits. There is no 'normal' or 'natural' rate of migration" (p. 245).

Modeling the impact of international migration on birth rates
To sum up, demography basically deals with international migration as numbers to be added (immigration) or subtracted (emigration) to any population defined by national boundaries. No individual country has a zero balance of external migration, but the entire world has, because it obviously receives no external migration. Despite this indisputable fact we argue, however, that flows of international migration might change the total number of inhabitants on earth, as a result of the impact international migration can have on natural population growth, notably on birth rates 12 . We focus here on two particular sub-populations: migrants themselves, and the community they have left behind in home countries. 11 Henry (1972) "Dans l'état actuel de la démographie, on ne sait pas étudier les phénomènes ouverts en tant que tels […] » "L'émigration d'une région A […] concerne le membres de la population étudiée et fait sortir de cette population : […] L'immigration dans une région B résulte, elle, de l'arrivée de personnes étrangères à cette région […] les événements qui figurent au numérateur ne concernent pas les membres de la population figurant au dénominateur. […] Il n'y a pas symétrie entre l'émigration de A vers B et l'immigration en B provenant de A : […] Dans ces conditions, l'étude des mouvements migratoires est, au moins sous son aspect théorique, une étude de sortie, d'émigration » (p. 198-9). 12 Mortality is also linked with migration. For example, an interesting "Mediterranean" pattern of health has been found among migrant populations (Khlat & Darmon 2003). Including mortality in the paper would have complicated our purpose for only little added value, since mortality does not play as important a role as fertility in contemporary international demographic differentials.
For those who move, migration is susceptible to produce two distinct impacts on patterns of family building. The first is a short-term one which results from imbalances in the sex ratio of migration flows-labor migration (whether of men or of women) delays marriage and procreation, while migration of family reunification yields the opposite effect, and allows to recover birth rates deficits of previous steps in individuals' life cycle-, and the second is a long-term one resulting from the gradual adjustment of migrants to their host population, which translates into a convergence of migrants' demographic patterns with those prevailing in receiving countries. This last effect is shown on Figure 1, left arrow. It is a limited one, since it affects only migrants themselves.
Those who don't move but live in communities from where numerous migrants have departed might also see their demography affected by migration. This will happen as soon as their living conditions are transformed by the emigration of relatives or neighbors on one side, and their vision of life is changed by alternative models to which they are exposed through the emigration of members of their community on the other side. Because expatriates are increasingly forming transnational communities in close contact with the environment left behind through fast travel and cheap telecommunications, modern migrants is still part of the game in their home countries, in particular in the diffusion of models. Their possible impact on patterns of family building and procreation is shown in Figure 1, right arrow. It is an enlarged effect, not limited to migrants themselves and their close families, but extended to their entire local community at home, and possibly to the larger society through the media.
This process interacts with the first one: the more adjusted the emigrants to their host society-and the better connected to the world left behind-the more efficient their diffusion of new models and ideas in their home society.
This argument refers to the ideational frame of the demographic transition, by far less researched and less modeled than its structural frame. Much more has been written on the Electronic copy available at: https://ssrn.com/abstract=940516 decline of fertility in relation with structural transformations such as the spread of mass education, urbanization, or the shift from agriculture to services, than on the role played by culture and values in demographic change. As a consequence, empirical evidence is scanty: most, if not all, large fertility surveys designed on a highly standardized and comparative scheme 13 have disregarded ideas and values, with the exception of a handful of questions directly related with family building, such as the ideal number children, sex preference, or views about marriage. For lack of individual data on more fundamental ideas and values expressing the way people see their lives, this paper will content itself with a macro approach, and will not reach the micro level which would allow a true validation of the hypothesis that migration is a vehicle for alternative ideas 14 .

Downstream demographic adjustment of migrants
When people move, they change their environment but not their selves. In a short lapse of time, they will be subjected to the living conditions prevailing in their new environment, which are important to determine the cost of children. For material reasons, migration will most probably affect the timing of childbearing and the desired number of children. It will take longer however for immigrants to change some of their individual characteristics, such as the level of education which is recognized an important determinant of fertility, and even longer to adapt to a new culture. This will happen either later in the course of their own lives or only to their children. A quick but limited effect of migration on fertility has thus to be expected, before any deeper shift takes place.
In fact, the convergence of immigrants' fertility with natives' seems to be a slow process: for example in France-the country which hosts the largest Arab expatriate community outside the Arab world-total fertility rates among immigrants women of MENA origin have decreased during the 1980s and the 1990s, thus reducing the distance with their host population, but surprisingly at a much slower pace than in their countries of origin (Table 1). Algerian women living in France have experienced an earlier fertility decline than those left behind in Algeria (6.77 children per women against 4.22 in 1980), but since this decline has been slower among the former than the latter, Algerian emigrants have now higher fertility rates than their non-migrant fellow citizens in Algeria (3.19 against 2.97 in 2000). The same holds for Moroccan, Tunisian and Turkish women. This unexpected result is largely due to a statistical artifact resulting from two characteristics of migration: for women migration is often caused by marriage (another result of which being fertility), and it is a selective phenomenon.
To fully understand this artifact it has to be borne in mind that, after the quasi-closure of Europe to labor immigration starting from the mid 1970s, family reunification has become  2003). That births delayed in countries of origin are recovered in countries of emigration clearly emerges from the fact that, for a same generation of women, those who reside in France for more than 10 years have a much lower level of fertility than those arrived from less than ten years ( Egyptians in Australia were more advanced than Lebanese in their demographic transition, while the contrary was true for their countries of origin, a probable sign of upward social selection of migrants to Australia in Egypt and downward in Lebanon. As a result of convergence with native Australians, fertility declined from one generation of immigrants to the next among Lebanese, but not among Egyptians who had since their arrival a lower fertility than average Australians 17 .
The second remark on adjustment of migrants to their host societies is that convergence might happen faster in demography than in other family related issues. A survey among immigrants from Turkey and Morocco carried out in Belgium in the early 1990s found that in matters directly related with fertility, such as the desired family size, the preference for boys or girls, the utility of children and contraception, migration produces a decisive change. For example, the percentage of married women aged 25-29 using contraceptive was respectively 79% and 71% among Turkish and Moroccan women living in Belgium, compared with 44% and 35% in their countries of origin the same year. When it comes to marriage the choice of partner and female autonomy, however, only a "prudent shift in the code of conducts" is observed: marriage decided on free individual choice without parents interfering remains very rare and most often a source of conflict. Social and cultural changes would thus proceed at different paces according to domains (Lesthaeghe and Surkyn 1995).

Patriarchal vs. individualistic values sent back home by migrants
From the mid-1970s until the late 1980s, Arab countries were displaying a puzzling pattern of fertility differentials: contrary to what is observed at the level of the world-a negative correlation between GDP per capita and birth rates-the richest Arab countries were also those with the highest birth rates. As it will be briefly recalled in the next paragraph, this 17  has been interpreted as the result of oil wealth and the particular type of state-to-society relationship it generated (Fargues, 1993). Even more puzzling, when only non-oil Arab countries were compared with each other, some of those best endowed with what is considered universal factors of the fertility transition-such as good health or high level of female education-were keeping higher levels of fertility than countries less endowed with these same factors. In other words, a single country could be characterized at the same time by advance in well-being and delay in demographic matters 18 . This apparent anomaly was due in part to migration, which served as a vehicle for values and models.
All began in the wake of the 1973 Arab-Israeli war. Beforehand, all Arab countries except one, Lebanon, were still in a pre-transitional stage regarding fertility, with total fertility rates ranging from 6 to more than 8 . 19 A mechanism sustained in the Gulf by labour imports, since women-dominated jobs-notably in education, health and administration-could be occupied by immigrant women. 20 The oil crisis started in the mid-1980s gradually put an end to this mechanism.

Gulf Migration System Western Migration System
As soon as a surprising pattern of Mashreq-Maghreb demographic differentials became clearly revealed by large fertility surveys in the late 1980s-early 1990s, the idea was advanced that "cultural models encountered through international migration to a certain extent reinforce the geography of demographic transition : the Maghreb, which has a foothold in Europe through its émigrés, has experienced a marked decline in its birth rate whereas the Egypt of the Infitâh, strengthening its Arab exchanges by a million and a half expatriates in Why did Egyptians or Jordanians, but not Moroccans or Tunisians (or Turks, outside the Arab world) translate material aspirations raised by the contact with wealthier societies into conservative views and practices in family building related matters? What are the mechanisms determining women to gain access to the wealth in circulation, either through marriage and rising dowry (Egypt and Jordan), or rather through economic activity and paid labor (Tunisia and Morocco, or Turkey)? This paper's contention is that international migration has served as a vehicle to the patriarchal model where emigrants were bound for Saudi Arabia and the Gulf, and to the individualistic one when they were destined to Europe.
In other words, migration is hypothesized to have brought about normative changes, whether towards reinforced control of the family over its members or towards increased individual autonomy.  1965 1970 1975 1980 1985 1990 1995 2000 2005 Source: see Table 6 Standardised Index Remittances (1970 prices) Birth Rates but remittances are. Money transferred by emigrants is directly linked with increases in household income that are hypothesized to play on patterns and timing of family building. In addition, one can reasonably assume that their amount is a good proxy for the intensity of the overall relationship kept by emigrants with their home country, i.e. a remote proxy for ideas, values and models they convey to the community left behind. The contrast between Egypt ( Figure 3) and Morocco (Figure 4) is striking. The time correlation between remittances and birth rates is as highly positive in the former (+0.623) as it is negative in the latter (-0.741).

Standardized index defined as: [x -average(x)] / standard deviation (x)
This paper does not develop the particular role that can be attributed to state policies in the discontinuous history of birth control and fertility decline in Egypt (Fargues 1997) but  1975 1980 1985 1990 1995 2000 2005 Source: see Table 7 Standardised Index Time correlation is not causation, however. For asserting that migration truly contributes to determining the pace and direction of changes in birth rates, rather than simply covariates with these changes -which would happen if migration on one side, and the decline of fertility on the other side, were two independent outcomes of a same third evolution, such as an increasing openness of societies to the outside world -one has to verify that changes in birth rates vary with the degree of exposure to migration, i.e. that regions from where intense emigration has departed display more dramatic demographic changes than those with little or no emigration: the closer the agents of diffusion, the stronger their impact. Space correlations are expected to corroborate time correlations.
Electronic copy available at: https://ssrn.com/abstract=940516 In both countries, emigration and the pace of fertility transition are found to covariate in space. 21 Table 9 Proportion of the transition of fertility completed between 1982-94 Effectively, in Egypt emigration is negatively associated with the progress of education, while in Morocco a positive association is found (Table 3). Whether the antecedent is education or emigration--that is whether regional differences in education are a cause or a consequence of regional differences in emigration -cannot be established with data that were available for preparing this paper. 22 Evidence from Turkey goes in the same direction (Figure 7).   capital in migrants' countries of origin does not refer to the "brain gain" hypothesis, according to which the success encountered abroad by highly skilled emigrants would encourage their non-migrant fellow citizens to invest in education, thus resulting in more human capital than if no brain drain had taken place (Stark, 2000). It rather emphasizes the impact of migration on the average level of mass education. Evidence accumulated in various contexts shows that remittances being channeled by families, they reach the very places from where migrants come and are used for basic needs including education and health care, i.e.
for human capital investment (Newland, Kathleen and Erin Patrick 2004). In addition to this general argument, we argue that, if emigrants are living abroad in a context where education is more widely spread than in the region where they come from, then they convey proeducation values to their community of origin.

Conclusion: the global demographic benefit of migration
When people move from one country to another, they change their cultural, social and Two broader conclusions can be drawn. The first relates to global demography. In most of modern international migration, the host society has a lower level of fertility than the home society. From this point of view, migrants from Turkey and the Maghreb to Europe are the rule and those from the Mashreq to Arabia and the Gulf, the exception. The acceleration of demographic transition found in Turkey and the Maghreb to be correlated with migration to Europe suggests that, if a similar relation were to apply to any migration from high to low birth rates countries, then international migration would have produced a global demographic benefit, under the form of relaxed demographic pressures at the level of the world.
The second conclusion relates to the circulation of ideas. If it turns out that emigrants are agents of the diffusion of new ideas in matters related with family building and education, then the same may apply to a wider range of civil behaviors. Because modern migrants keep strong links with their community of origin, they are susceptible to become key conveyors of ideational change among non-migrants in these communities.     (6) Table 4 shows how migration flows can be estimated (very imperfectly) in the absence of any reliable source on migration, using the United Nations demographic database. Cumulated migration from 1950 to 1990 (amounting to -1,503 in the example of Morocco) divided by the total population in 1990 (24,654 million in Morocco) gives the rate of cumulated migration provided in Table 5 (Morocco: -6.1%). Palestine -60.0% 6.6 Syria -3.1% 6.6 Yemen -9.1% 7.6 _____________________________________________________________ Source: Cumulated migration is computed by the author applying the method described in Table 3.