Long-term determinants of Latin American international migration,

Long-term determinants of Latin American international migration, 1975-2010. Victoria Prieto Rosas 1 Joaquín Recaño Valverde 2 Abstract Latin American and The Caribbean outmigration coincided with the onset of the Demographic Transition for some countries, but little is known on how both processes are related. To what extent natural growth did contribute in explaining net migration s trend? What other processes could have inhibited its impact on international mobility? Country level data from UN Population Division, World Bank and the Quality of Governance Institute was organized into a macro-panel, where 16 countries were followed up from 1975 to 2010. A multivariate linear model with fixed effects and interactions -to account for demographic inhibitors- was estimated. Similar to what has been found for the Trans-Saharan migration system, natural increase has a negative and significant effect on net migration s evolution, even while other processes are going on. Urbanization significantly inhibited the effect of natural increase on net migration, but this was not the case for productivity increase or for industrialization. 1 Programa de Población Universidad de la República, Uruguay. E -mail: vprieto@fcs.edu.uy 2 Centre d Estudis Demogràfics Departament de Geografia Universitat Autònoma de Barcelona. E-mail: jrecano@ced.uab.es 1

(Preliminary version) 3 I. Introduction Net migration in Latin America has suffered numerous transformations in the last forty years. Undoubtedly, the most outstanding change is the turn in the direction of the flows, which transforms it into a net emigration region, leaving behind a long history of immigration. Most recent changes in the regional migration landscape are related to the increase of emigration and diversification of destinations (CEPAL 2007 and 2006); the feminization of the profile of migrants, specially towards Europe (López Lera & Oso 2007); and the emergency of a migration dynamics inherent to those more qualified migrants (Pellegrino & Martínez Pizarro 2001; Pellegrino 2008). Recently, interest in projecting migration based on variables which would be less volatile than economic variables, has retaken the debate about the effect of demographic growth on the evolution of international migration (Adepoju 2004; Zlotnik 2004; Malmberg 2006; De Haas 2012). The central hypothesis sustains that demographic growth, or in better words the cohorts size of those born during periods of high growth, could increase the migration intensity once the said cohorts reach the typical age for migration (Easterlin 1961). But the impact of demographic variables is never direct (De Haas 2010) and it is subject to the occurrence of other processes we will refer to as demographic inhibitors (Salinari & De Santis 2011). The periods of high demographic growth in Latin America & the Caribbean respond to the early stages of Demographic Transition, around the sixties, which coincided with the increase in international migration from this region. What was the role of the first Demographic Transition in Latin American international migration during the last four decades? To what extent natural growth did contribute in explaining net migration s trend? What socio-economic processes could have inhibited its impact on international mobility? These questions are answered through a multivariate longitudinal study. By using a macro-panel model with fixed effects, the net migration rates (M) from 16 Latin American countries between 1975 and 2010 were modeled. The impact of natural growth on M was analyzed in two ways: first in isolation, focusing on GDP and structural economic transformations, political instability, educational expansion and urbanization; then, through interactions with the urbanization, and increase in productivity and industrialization, respectively. These interactions account for the concept of demographic inhibitors, i.e. forces able to inhibit the demographic pressure on international migration (Davis 1963; Salinari & De Santis 2011). The results confirm a significant effect of the natural growth of birth cohorts from intermediate stages of the Demographic Transition (DT). However, the intense urbanization experimented by the region, in line with the ongoing DT (CESA 2011), absorbed part of the impact of natural growth on international migration. 3 Please do not cite without permission of the authors. 2

According to the hypothesis of development as promoter of migration, by effect of the generalization of aspirations, educational expansion in this period also shows a positive and significant effect, which supports the idea that the development, far from reducing the international migration, fosters it at intermediate stages of development (De Haas 2010). This research work empirically contrast the effect of those factors - economic, demographic and political - considered by the literature as long term determinants of international migration, paying special attention to demographic factors. The relation between net migration and natural growth is approached in two ways: i) first, through measuring the independent effect of natural growth on net migration, controlling by other factors such as economic growth, educational expansion, urbanization and political instability in a macro-panel model where temporal change (within change) of net migration rates between 1975 and 2010 was modeled, for the countries below: Argentina, Brazil, Bolivia, Chile, Colombia, Costa Rica, Dominican Republic, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Uruguay and Venezuela. ii) second, by incorporating the concept of demographic inhibitors, operationalized by way of interactions of natural growth with urbanization, with industrialization and with economic growth, respectively. II. Background According to Zelinsky (1971), migration transitions are framed in demographic transitions, but the effect of demographic variables on migration can be absorbed by other processes it interacts with. In this sense, Salinari & De Santis (2011) have tried to model the indirect effect of demographic determinants. They sustain that natural growth should have a certain impact on the evolution of net migration when contemplated together with other possible responses to demographic growth. In the study of net migration of Sub-Saharan Africa, these authors conclude that in the stages prior to urbanization or industrialization, the impact of demographic growth on migration may be inhibited by other responses such as enhancement of agricultural production, or industrialization and urbanization themselves (demographic inhibitors). Consequently, they incorporate demographic variables by interacting them with other non-demographic determinants, reaching the conclusion that without a reduction of the arable land surface and the food supply, natural growth per se does not generate migration. From this perspective, the effect of demographic determinants is only perceptible when contemplating long periods of time (more than 30 years), which allow appreciating how this effect is inhibited at a first stage by other valves (urbanization, internal migration, etc.). Of course, migration is a multivariate phenomenon and its determinants are fundamentally at meso and micro level, since the said factors involve decisions that are taken by individuals and households instead of by Governments or economies (Stark & Bloom 1985; Massey 1990; Massey et al 1998). But to analyze the effect of demographic variables on migration temporal evolution it is necessary to work from an aggregate level, at a scale of populations or countries, 3

as well as considering other fundamental set of material determinants such as those related to development. Attempts to incorporate the concept of agency and structure in migration studies at macro level, as those underneath the concepts of capabilities and aspirations we find in Sen (2000), are very recent in migration studies. De Haas (2010) suggests looking at opportunities more than at income differentials, by using the Human Development Index instead of the GDP. Most certainly, it seems quite a challenge to reduce such a complex phenomenon as development to mere economic growth. But most authors who have approached this topic in a bilateral and multivariate manner still consider economic growth (GDP s growth) instead of development (Jennissen 2003, 2004; Cebrián 2009; Mayda 2010; Hatton & Williamson 1998). Migration is here considered as a phenomenon that responds to development more than to growth or economic differentials, and therefore we incorporate the average educational level of a country to approximate the effect of generalization of migration aspirations generated by educational expansion. Education is expected to improve the access to information through modern mass media and internet, raising the awareness of opportunities elsewhere (De Haas 2010). Moreover, there is a strong component of qualified migration in the international mobility of the regions and an increase of the general educational level is expected to have a positive impact on migration for these reasons. In the seventies, political determinants played a fundamental role in Latin American emigration 4, and in this case it is plausible that they have a significant impact. Despite of the significance this variable has showed at intuitive level, its importance has varied along the different empirical evidence according to the extension of the temporal series considered (Solimano 2003; Cebrián 2009). III. Data and Methods Net migration rate, estimated as marginal component of the balancing equation, was used as the magnitude to reflect international mobility. In spite of the critics that the use of net migration has provoked (Rogers 1990; Termote 2002), practical reasons explain its widespread use in the study of developing countries international migration, given that long term time series on outflows are rarely available. For the variables considered to be potential determinants of net migration, different data sources were used. Country-level data on five-year period basis was collected from the sources depicted in Table III.1. 4 Countries that suffered severe restrictions to civil freedoms: Argentina (1976-1983); Brazil (1964-1985); Bolivia (1971-1982); Chile (1973-1990); El Salvador (1931-1979); Honduras (1963-1971 y 1972-1982); Guatemala (1954-1986); Nicaragua (1934-1979); Panama (1968-1989); Paraguay (1949-1989); Peru (1968-1980) and Uruguay (1973-1985). 4

Table III.1 Data sources in use by variable. Variable Description Period Source M t, t+5 (net migration rate) Five-year period net migration rate 1975-2010 Estimations & Projections (UN World Population Prospects 2010, September 2011). http://esa.un.org/wpp/unpp/panel_population.htm N t-25, t-20 (natural growth rate) Five-year period natural growth rate from 20-25 years prior to the M(t). 1950-1995 Estimations & Projections (UN World Population Prospects 2010, September 2011). http://esa.un.org/wpp/unpp/panel_population.htm GDP pc t, t+5. Gross Domestic Product per capita average value for five-year period 1975-2004, Angus Maddison Time Series (April 2011) http://www.ggdc.net/maddison/oriindex.htm World Bank (May 2011) http://data.worldbank.org/data-catalog/world-developmentindicators GDP pc growth rate 2005-2010 Quality of Government Data Set/ pwt_grgdpch (Quality of Government Institute, April 2011). This time series summarizes the estimations from Heston, Summers & Aten-Penn World Table. Industrialization t, t+5 (%GDP_Industry) Percentage of Porcentaje del GDP correspondiente al sector industrial. 1975-2008 Quality of Government Data Set/ wdi_ise Construida originalemente por el Banco Mundial. (Quality of Government Institute, Abril 2011). Quality of Democracy t, t+5. Mean value for five-year period of the Freedom House Index. In a scale of 1-10 measures the quality of democracy, being 0 its absence and 10 and optimal level of civil freedoms respect. 1972-2009 Quality of Government Data Set/ if_polity2 (Quality of Government Institute, April 2011). http://www.qog.pol.gu.se/data/data.htm Urbanization t, t+5. (%Urban_Pop.) Mean value for five-year period of percentage of population residing in urban areas. 1975-2008 Quality of Government Data Set / wdi_urban (Quality of Government Institute, April 2011). http://www.qog.pol.gu.se/data/data.htm This time series summarizes the estimations from World Bank and the UN. Average educational level t, t+5. Average educational level of both sexes population above 15 years old. 1975-2009 Quality of Government Data Set/ ihme_aye (Quality of Government Institute, April 2011). This time series summarizes the estimations from the Institute for Health Metrics and Evaluation (University of Washington). http://www.qog.pol.gu.se/data/data.htm Source: Own elaboration. The greatest source of heterogeneity in this case is related to the between countries differences. They differ in respect of their ethnical composition, geography, language, politics, economy, history and culture, and given that we aim to assess the determinants of the within change it is necessary to isolate the unobserved heterogeneity of the country effect by using a model with fixed effects. The Hausman test confirmed the election of a fixed effects model instead of using a random effects model. 5

Four models were estimated. The first one (Equation III.1) excluded the interaction terms that account for possible inhibitors of the demographic effect on migration. Equation III.1 ( ) = + + + + % _ + _ + % _ + + The natural growth rate was incorporated with a 20-25 years lag, because in case of existing an effect of demographic growth on migration this could only be noticeable once the birth cohorts turn the typical ages for migration. Instead, the second model does include these interactions in the following form: i) the first interaction is between natural growth rate and GDP pc growth rate, ii) the second interacts natural growth rate with by industry GDP, and iii) the last one interacts natural growth rate with the percentage of population residing in urban areas. (Equation III.2). Equation III.2 ( ) = + + + + % _ + _ + % _ + % _ + _ + % _ + + Other two models were estimated, similar to models in Equation III.1and Equation III.2, but in this case without the _ and % _ or their interactions with. Results from models 1-4 are presented in Table IV.1. To control for heterocedasticity (Wald test) robust panel corrected standard errors were estimated (Prais Winsten PCSE) and to deal with autocorrelation (Wooldridge test) a first order autocorrelation term was introduced (t-5, t). IV. Results IV.1 Demographic Transition and Net Migration The main transformations observed in this period are those related to change of migration status of several countries in the region, which replace their clearly immigration-oriented profile to a profile which is essentially emigration-oriented. As it was already stated the beginning of DT in Latin America took place in the sixties. According to Chackiel (2004), during the five-year period 1960-1965, the region reached the N highest value, around 28 per thousand, as a result of two decades of a sharp decline in mortality accompanied by a high level of fertility. The subsequent drop in fertility contributes since then to the drop in natural growth, which nowadays reaches approximately 13.3 per thousand (WPP 2010). Moreover, Argentina, Brazil, Chile, Uruguay and several Caribbean countries showed onedigit values of the N in 2005-2010. It is essential to consider the growth values detected at the moment of birth of 1950-1955 to 1980-1985 cohorts, which are the ones born in a regime of high natural growth rates. Our hypothesis supports the idea that, if the highest values of the natural growth rate of the region are the ones registered in, for instance, 1960-1965, it is foreseeable that the net migration rates 6

of 1985-1990 - when said birth cohorts were between 20 and 25 years old - reflects some negative impact (high net emigration). As seen in Graph IV.1, several countries reach their natural growth rates maximum values years later. For instance, Colombia reached a value of 25.7 per thousand in 1970-1975, and the countries which later joined the DT, such as Guatemala, Honduras and Mexico, reached their maximum values in 1980-1985. It is possible then that the natural growth s effect on net migration in countries such as these is only noticeable later on, towards the decade 2000-2010. Graph IV.1 Natural Growth Rate (per 1000). Selected countries, 1950-2010. 40.0 35.0 Dominican Rep. Venezuela Nicaragua Honduras 30.0 25.0 Guatemala Mexico Colombia 20.0 15.0 10.0 5.0 0 1950-1955 1955-1960 1960-1965 1965-1970 1970-1975 1975-1980 1980-1985 1985-1990 1990-1995 1995-2000 2000-2005 2005-2010 Am. Latina & Caribe Dominican Rep. Costa Rica El Salvador Guatemala Honduras Mexico Nicaragua Panama Argentina Bolivia Brazil Chile Colombia Paraguay Uruguay Venezuela Source: Own elaboration based on World Population Prospects data (UN 2010). This type of effects can be identified in different ways, in Graph IV.2, which shows the net migration rates (M) and natural growth rates (N) time series between 1950 and 2010. This graph intends to locate countries within the process of DT, identifying at which point of it each country was at the beginning of the decades of high levels of net migration. In this case, the Ns have a lag of t-25, t-20 years with respect to M values ( in t). The first outstanding feature of from Graph IV.2 is the heterogeneity between countries. For countries like Bolivia, both series follow opposite trends, i.e. the relation between them is of negative nature. Another example of this is Panama, where N was high until 1990 and since then the M has reversed its value (from positive to negative). The same happens in Costa Rica, Chile or Colombia, where relation is also negative, the lower the natural growth, the higher the net immigration (Graph IV.2). In Mexico and Panama the years of maximum natural growth coincided with a substantial decrease of net migration, but of course this may be a mere random coincidence or it could 7

respond to other factors such as period effects due to economic or political changes, etc. when we are not controlling for other factors as we do in the next section. Graph IV.2 Natural Growth Rate (t-25, t-20) and Net Migration Rate (t), 1975-2010. NOTE: The dates on the x axis correspond to the values of M. The values for N (y axis) are actually the values observed 20-25 years before (t-25, t-20). Source: Own elaboration based on World Population Prospects data (UN 2010). IV.2 Multivariate analysis of the long-run determinants of net migration The concept of development was operationalized, incorporating separately the variable of average years of study and the Gross Domestic Product per Capita (GDPpc), so as to differentiate the specific effect of each component of development, instead of introducing the Human Development Index as a variable. Migration features a clear economic nature, which is noticeable from the significance and sign of GDP per capita s coefficient (models 1 to 4 in Table IV.1). The effect of average years of study on population tends supports the hypothesis of migration as a response to development more than to economic growth. Educational level works as a force increasing the opportunities of migration. Of course this relation is reversed at very advanced levels of development, but at intermediate stages, as it could be the case for Latin America and The Caribbean, the relation between development and emigration is negative, regardless of the fact that it may evolve in the future into a positive relation, following an inverted U-shape (De Haas 2010:34). 8

Table IV.1 Results from Prais Winsten regression models (PCSEs). log(m) model 1 model 2 model 3 model 4 N (t-25, t-20) -.00037* -.00103 -.00034* -.00113* GDPpc.00032*.00037*.00035*.0004* Democracy (FHI).00014*.00015*.00012.00013Ɨ Educational level (years) -.00089*** -.00067* -.0008** -.00059* %Pop. Urban.00087* -.00048.00077* -.00055 r_gdppc (growth rate).000046.0001 %GDP_Industry.000074.00015 %Urban Pop.*N (t-25, t-20).00087*.00084* %GDP_Industry*N(t-25, t-20) -.00017 r_gdppc * N (t-25, t-20) -.000056 Bolivia.00063 -.00028.00058 -.00031 Brazil.00014 -.00024.00021 -.00016 Chile.00071***.00031.00073***.00036 Colombia.00055* -9.2e-05.00056* -4.5e-05 Costa Rica.00365***.00253**.00349***.00241** Dominican Rep..00039 -.00045.00034 -.00044 El Salvador -.00259*** -.00352*** -.0027*** -.00358*** Guatemala -.00096 -.00189* -.00112* -.00195** Honduras.00089-6.2e-05.00074 -.00013 Mexico -.0005 -.00127* -.00056 -.00126** Nicaragua -.00065 -.0014* -.00078 -.00144* Panama.00273***.00157.00246***.00144 Paraguay.002***.00101.00177***.0009 Uruguay -.00131*** -.00091** -.00129*** -.0009** Venezuela.00081.00027.00095.00034 Cons. -.0016377 -.000685 -.00157*** -.00067 R2.889.892.884.884 Ar(1).1372352.1344651.1510679.1605554 N_i 16 16 16 16 N_t 7 7 7 7 N total 112 112 112 112 Ɨ p <0.10; * p<0.05; ** p<0.01; *** p<0.001 Source: Own elaboration. Results also confirm the hypothesis that suggest a positive relation between the quality of democracy and net migration, according to which in absence of political inestability net migration should show positive values. The isolated impact of potential inhibitors of the demographic growth effect. Neither GDP per capita growth rate nor industrial GDP seem to contribute to predict the temporal evolution of net migration in model 1. On the other hand, the percentage of population residing in urban areas does actually represent a potential inhibitor of the natural growth effect on migration. Our hypotheses, and part of the theoretical studies (Malmberg 2006), support the expectation of a negative relation between urbanization and migration, that reflects the urban feature of 9

international emigration from very urbanized countries. However, most direct background of this work especially in terms of the methodology here in use- do find a positive and significant relation between urbanization and international migration in the case of Sub-Saharan migration system (Salinari & De Santis 2010), which is the same we find in models 1 and 3. In these cases, the positive value of urbanization may be interpreted as follows: great deal of the movements reflected by the net migration rates are due to the intraregional movements, which reflect the mobility vector of a flow from less urbanized countries to more urbanized countries with dynamic economies that act as pull factors. Bear in mind that the main poles of attraction of the region are also the countries with more percentage of population residing in urban areas (Argentina, Venezuela, Brazil, Costa Rica, Chile) and that countires with net immigration are included in the study. This hypothesis may apply for the case reflected by models 1 and 3. In any case, in models 2 and 4 - where interaction terms are incorporated - the coefficient s sign reverses and the initial hypothesis is confirmed. Given interactions are significant, it is clear that those models which do not incorporate same (1 and 3) are not sufficiently specified. The direct effect of natural growth is negative and significant in all the models, including or not interactions with other social processes. In models 1 and 2, change in the composition of the percentage of GDP due to industry production or its growth rate do not represent a determining force in the long-term evolution of international migration. This goes against the evidence found for the African migratory system, where processes such as modernization of agricultural production or extension of the arable land surface do have a significant effect in the explanation of international migration (Salinari & De Santis 2011). But is coherent with the heterogeneuous and truncated nature of the industrialization process of Latin American countries. However, the interaction term of the natural growth rate (t-20, t-25) with the percentage of population residing in urban areas does show a significant effect, being this the only relevant interaction of the three. This points to the idea that even when the direct effect of natural growht on migration is negative and significant, while urbanization is going on the slope of this relationship changes, which means that urbanization inhibits the negative direct effect of natural growth on net migration absorbing part of the demographic bulk. V. Conclusions Latin American international migration system has developed in line with the onset of the DT in many countries of the region. Also, strong changes have taken place in this period in other economic, political and social spheres, which may have directly or indirectly impacted on the evolution of international migration of the last four decades. In the multivariate model, migration is more connected to development than to economic growth, as it can be seen in the effect of educational expansion on net migration. Recent evolution of Latin American migration shows that development, far from reducing net emigration, fosters the same at intermediate levels of development (De Haas 2010b). Additionally, the significance of political determinants in Latin American migration was verified, and it could be attributable to political instability of the 70s and the 80s. 10

What was the role of Demographic Transition in the evolution of Latin American international migration? The results of the multivariate model suggest that the high demographic growth observed at the initial-intermediate stages of an accelerated DT -experienced by Latin American countries between 1960 and 1980-, had a negative effect on international migration in the last decades. However, the region is currently quite ahead in DT and its growth rates have dropped rapidly, which suggests that the effect of natural growth on migration will lose significance in the decades to come. What are the processes which inhibited their effect on international mobility? The direct effect of natural growth on migration was significant, but the intensity and speed of the Latin American urbanization has inhibited this effect. This could have happened in two ways, either through the rural-urban migration that could have postponed the emergence of international emigration in the countries where urbanization started later, or attracting population from neighbor countries in the cases in which both urbanization and economic growth took place simultaneously. This empirical exercise reflects that international migration and internal migration, approximated in this case by the urbanization variable, are closely related while DT is going on. As remarked by Zelinsky (1971), DT and mobility transition are phenomena which should not be studied in isolation. VI. References ADEPOJU, A. (2004) Trends in International Migration in and from Africa. In MASSEY, D. S. & TAYLOR, E., J. (Eds.) International migration, prospects and policies in a global market. Oxford, Oxford Scholarship online. CEBRIÁN, M. (2009) Los determinantes de los flujos migratorios internacionales: el caso español, 1995ı2007. Principios. Estudios de Economía Política, 14, 49-70. CEPAL (2006) Migración Internacional, derechos humanos y desarrollo, Santiago, CEPAL LC/W.98. CEPAL (2007) Migración internacional, derechos humanos y desarrollo: síntesis y conclusiones, Santiago, CEPAL. CESA (2011) Population Distribution, Urbanization, Internal Migration and Development, New York, Population Division-UNDESA. CHACKIEL, J. (2004) La dinámica demográfica de América Latina. Población y Desarrollo 53. Santiago de Chile, CEPAL. DAVIS, K. (1963) The Theory of Change and Response in Modern Demographic History. Population Index, 29, 345-366. 11

DE HAAS, H. (2010) Migration transitions, a theoretical and empirical inquiry into the developmental of international migration. Working paper of the International Migration Institute-Oxford University. 24. DE HAAS, H. (2010b) Migration and Development: A Theoretical Perspective. International Migration Review, 44, 227 264. EASTERLIN, R. (1961) Influences in European overseas emigration before World War I. Economic and Development and Cultural Change, 9, 331-351. HATTON, T. J. & WILLIAMSON, J. G. (1998) The age of mass migration. Causes and economic impact, New York, Oxford University Press. JENNISSEN, R. (2003) Economic Determinants of Net International Migration in Western Europe. European Journal of Population 19, 171 198. JENNISSEN, R. (2004) Macro-economic determinants of international migration in Europe. Rijksuniversiteit Groningen. LÓPEZ LERA, D. & OSO, L. (2007) La inmigración latinoamericana en España. Tendencias y estado de la cuestión. In YÉPEZ, I. & HERRERA, G. (Eds.) Nuevas migraciones latinoamericanas a Europa. Balances y desafíos. Quito, FLACSO Ecuador. MALMBERG, B. (2006) The boom and bust of net migration. A 40-year forecast. In TAMAS, K. & PALME, J. (Eds.) Globalizing migration regimes. New challenges to transnational cooperation. Burlington, Ashgate. MASSEY, D. S., ARANGO, J., HUGO, G., KOUAOUCI, A., PELLEGRINO, A. & TAYLOR, E., J. (1998) Worlds in motion. Understanding international migration at the end of the millennium, New York, Oxford University Pres. MASSEY, D. (1990) Social Structure, Household Strategies, and Cumulative Causation of Migration. Population Index, 56, 3-26. MAYDA, A. M. (2010) International migration: a panel data analysis of the determinants of bilateral flows. Journal of Population Economics, 23, 1249-1274. PELLEGRINO, A. (2008) La migración calificada en América Latina. Foreign Affairs. PELLEGRINO, A. & MARTÍNEZ, J. (2001) Una aproximación al diseño de políticas sobre la migración internacional calificada en América Latina. Población y Desarrollo 23. Santiago de Chile, CEPAL. ROGERS, A. (1990) Requiem for the net migrant. Geographical Analysis, 22, 280-300. SALINARI, G. & DE SANTIS, G. (2010) The Role of the Demographic Transition in the Formation of the North African and Trans-Saharan Migration Systems. 2010 Quetelet Seminar on Stalls, resistances and reversals in demographic transitions. Louvain. 12

SALINARI, G. & DE SANTIS, G. (2011) The role of the demographic transition in the formation of the trans-mediterranean and trans-saharan migration systems. PAA. Washington. SEN, A. (2000) Desarrollo y Libertad, Barcelona, Editorial Planeta. SOLIMANO, A. (2003) Globalización y migración internacional: la experiencia latinoamericana. Revista de la CEPAL, 80, 55-72. STARK, O. & BLOOM, D. (1985) The new Economics of Labor Migration. The American Economic Review, 75, 173-178. TERMOTE, M. (2002) The explanatory power of migration models. In FRANCK, R. (Ed.) The explanatory power of models. Netherlands, Kluwer Academic Publishers. ZELINSKY, W. (1971) The hypothesis of the Mobility Transition. Geographical Review, 61, 219-249. ZLOTNIK (2004) Population Growth and International Migration. In MASSEY, D. S. & TAYLOR, J. D. (Eds.) International Migration: Prospects and Policies in a Global Market. Oxford, Oxford University Press. 13